Topotecan, Pegylated Liposomal Doxorubicin Hydrochloride

Home , Pirarubicin, Topotecan

Topotecan, pegylated liposomal doxorubicin hydrochloride and paclitaxel for second-line or subsequent treatment of advanced ovarian cancer (report contains no commercial in confidence data)

Produced by Centre for Reviews and Dissemination, University of York

Authors Ms Caroline Main, Research Fellow, Systematic Reviews, Centre for Reviews and Dissemination, University of York, YO10 5DD Ms Laura Ginnelly, Research Fellow, Health Economics, Centre for Health Economics, University of York, YO10 5DD Ms Susan Griffin, Research Fellow, Health Economics, Centre for Health Economics, University of York, YO10 5DD Dr Gill Norman, Research Fellow, Systematic Reviews, Centre for Reviews and Dissemination, University of York, YO10 5DD Mr Marco Barbieri, Research Fellow, Health Economics, The Economic and Health Research Centre, Universitat Pompeu Fabra, Barcelona, Spain Ms Lisa Mather, Information Officer, Centre for Reviews and Dissemination, University of York, YO10 5DD Dr Dan Stark, Senior Lecturer in Oncology and Honorary Consultant in Medical Oncology, Department of Oncology, Bradford Royal Infirmary Mr Stephen Palmer, Senior Research Fellow, Health Economics, Centre for Health Economics, University of York, YO10 5DD Dr Rob Riemsma, Reviews Manager, Systematic Reviews, Centre for Reviews and Dissemination, University of York, YO10 5DD

Correspondence to Caroline Main, Centre for Reviews and Dissemination, University of York, YO10 5DD, Tel: (01904) 321055, Fax: (01904) 321041, E-mail: [email protected]

Date completed September 2004

Expiry date September 2006

Contributions of authors Caroline Main Lead reviewer responsible for writing the protocol, study selection, data extraction, validity assessment and writing the final report. Laura Ginnelly Involved in the cost-effectiveness section, writing the protocol, study selection, data extraction, development of the economic model and report writing. Susan Griffin Involved in the cost-effectiveness section, writing the protocol, study selection, data extraction, development of the economic model and report writing. Gill Norman Second reviewer involved in the clinical-effectiveness section, involved in study selection, data extraction, validity assessment and writing the final report. Marco Barbieri Involved in the cost-effectiveness section, responsible for the review and re-analysis of the company submissions. Lisa Mather Devised the search strategy and carried out the literature searches; wrote the search methodology sections of the report. Dan Stark Provided input at all stages, commented on various drafts of the report and contributed to the discussion section of the report. Stephen Provided input at all stages, commented on various drafts of the Palmer report; overall responsibility for the cost-effectiveness section of the report. Rob Riemsma Provided input at all stages, commented on various drafts of the report; overall responsibility for the clinical-effectiveness section of the report.

Conflicts of interest: Colleagues at the Centre for Health Economics (not involved in this review) have received research funding from GlaxoSmithKline and Schering-Plough.

Source of funding: This report was commissioned by the NHS R&D HTA Programme.

Relationship of reviewer(s) with sponsor: None

Acknowledgements We thank the expert advisory panel for their useful advice and constructive comments on the report. The views expressed in this report are those of the authors and not necessarily those of the NHS R&D Programme. We also wish to thank Dr Tony Ades of the MRC Health Services Research Collobaration at the University of Bristol for his help and advice on the use of the mixed comparison model applied in the economic analysis.

Figure 7: Reproduced from W. ten Bokkel Huiink, S.R. Lane & G.A. Ross on behalf of the International Topotecan Study Group. Long term survival in a phase III, randomised study of topotecan versus paclitaxel in advance epithelial ovarian carcinoma. Annals of Oncology 15: 100-102, 2004, page 101 with permission of Oxford University Press.

2 TABLE OF CONTENTS

EXECUTIVE SUMMARY. 9

1. AIM OF THE REVIEW 27

2. BACKGROUND 27 2.1. Description of underlying health problem 27 2.2. Current service provision 29 2.3. Description of new intervention 30

3. METHODS 36 3.1. Search strategy 36 3.2. Inclusion and exclusion criteria 38 3.3. Data extraction strategy 40 3.4. Quality assessment strategy 41 3.5. Methods of analysis/synthesis 41 3.5.1. Clinical effectiveness 41 3.5.2. Cost Effectiveness 42 3.5.3. Handling the company submissions 42

4. RESULTS 43 4.1.1. Quantity and quality of research available 43 4.1.2 Description of included studies 49 4.2 Assessment of clinical effectiveness 67 4.2.1. Effectiveness of PLDH versus Topotecan 67 4.2.2. Effectiveness of topotecan versus paclitaxel 87 4.2.3. Effectiveness of PLDH versus paclitaxel 100 4.2.4. Effectiveness of single agent paclitaxel compared to a combination of cyclophosphamide, doxorubicin and cisplatin 107 4.2.5. Effectiveness of paclitaxel in combination with platinum-based chemotherapy versus platinum-based therapy alone 113 4.2.6. Effectiveness of single-agent paclitaxel versus oxaliplatin 120 4.2.7. Effectiveness of paclitaxel given weekly versus every three weeks 124 4.2.8. Effectiveness of paclitaxel 175 mg/m2 compared to 250 mg/ m2 128 4.2.9. Effectiveness of oral versus intravenous topotecan 133 4.3. Other Related Systematic Reviews 138 4.4. Assessment of Economic Evaluations 139 4.4.1. Introduction 139 4.4.2. Review of Smith 139 4.4.3. Review of Ojeda 146 4.4.4. Review of Capri 149 4.4.5. Review of Prasad 152 4.4.6. Review of GlaxoSmithKline submission 156 4.4.7. Review of Schering-Plough Ltd submission 163 4.4.8. Review of Bristol-Myers Squibb Ltd submission 169 4.5. Summary of cost-effectiveness 172 4.6. Summary of findings from the cost-effectiveness review 176

5. ECONOMIC MODEL 178 5.1. Introduction 178 5.2. Methods 178 5.2.1. Analysis 1: Base case model 185

3 5.2.2. Analysis 2: Platinum-sensitive model 190 5.3. Model inputs 191 5.4. Analytic methods 204 5.5. Results 205 5.5.1. Analysis 1: Base case 206 5.5.2. Analysis 2: 213 5.6. Budget impact analysis 216 5.7. Economic model conclusions 218

6. DISCUSSION 221 6.1 Clinical effectiveness 221 6.2 Cost-effectiveness 234

7. CONCLUSIONS 237 7.1 Clinical effectiveness 237 7.2 Cost-effectiveness 239

8. REFERENCES 241

9. APPPENDICES 252 Appendix 1. Details about FIGO cancer staging 252 Appendix 2 Summary of current manufacturers information provided for health professionals 253 Appendix 3. Search strategy 258 Appendix 4. Excluded studies 265 Appendix 5. Details of quality assessment 269 Appendix 6. Data extraction tables 272 Appendix 7. Trials only available in abstract form 350 Appendix 8. Additional tables for the review of cost-effectiveness 354 Appendix 9. Quality of life searches 380 Appendix 10 WinBUGS code 389 Appendix 11. Additional tables for the economic model 408

TABLES Table 1. Summary of included RCTs 46 Table 2. Summary of the comparators included in the review of clinical effectiveness 48 Table 3. Quality checklist for included trials 66 Table 4. Summary of overall survival data based on long-term follow-up for PLDH versus topotecan. 67 Table 5. Summary of overall survival data based on long-term follow-up for PLDH versus topotecan – platinum-sensitive disease subgroup 69 Table 6. Summary of overall survival data based on long-term follow-up for PLDH versus topotecan – platinum-refractory disease subgroup 70

4 Table 7. Overall survival for sub-groups according to baseline disease characteristics (multivariate Cox regression analysis) 71 Table 8 Summary of progression-free survival data based on long-term follow-up for PLDH versus topotecan 72 Table 9. Summary of objective response rates based on long-term follow-up for PLDH versus topotecan 73 Table 10. Percentage of patients with a maintained or improved QoL score at 12 weeks follow-up for PLDH versus topotecan. 76 Table 11. Time (in months) spend in the three health states and their differences 78 Table 12. Treatment emergent adverse events in a least 10% of participants by preferred term: All grades, Grade 3, and Grade 4. 81 Table 13. Response rate by treatment group for topotecan versus paclitaxel 89 Table 14. Response in relation to platinum sensitivity for topotecan versus paclitaxel 90 Table 15. Response in relation to baseline disease status for topotecan versus paclitaxel 90 Table 16 Response duration for participants treated with topotecan or paclitaxel 91 Table 17. Time to response data for participants treated with topotecan or paclitaxel 91 Table 18. Number and percentage of patients and courses with haematological toxicity by worst CTC Grade. 93 Table 19. Related or possibly related non-haematological toxicities occurring in more than 10% of patients treated with topotecan or paclitaxel 94 Table 20. Response to Topotecan (third-line) treatment according to response to paclitaxel (second-line) 96 Table 21. Response to Paclitaxel (third-line) treatment according to response to topotecan (second-line) treatment 96 Table 22. Haematological toxicity during cross-over (third-line) treatment with topotecan versus paclitaxel 97 Table 23. Non-haematologic Toxicity to Cross-over (third-line) treatment for topotecan versus paclitaxel 98 Table 24. Summary of overall survival data for all participants for PLDH versus paclitaxel 100 Table 25. Summary of overall survival data based on PLDH versus paclitaxel – platinum-sensitive disease subgroup 101 Table 26. Summary of overall survival data based on PLDH versus paclitaxel – platinum-resistant disease subgroup 101 Table 27. Treatment emergent adverse events in a least 10% of participants by preferred term: All grades, Grade 3, and Grade 4. 103 Table 28. Response to CAP compared to paclitaxel treatment. 108 Table 29. The maximum grade toxicities during treatment with paclitaxel compared with CAP 111

5 Table 30. The proportion of patients in each treatment group who experienced moderate or severe toxic effects during treatment along with the RR and 95% CI 118 Table 31. Response to treatment with paclitaxel versus oxaliplatin in the total treatment population 121 Table 32. Severe (Grade 3 & 4) toxicity by treatment group for paclitaxel compared to oxaliplatin. 122 Table 33. Response rate by treatment group for weekly versus three weekly paclitaxel 124 Table 34. Toxicity events for weekly versus three weekly paclitaxel 126 Table 35. Hypersensitivity reactions for weekly versus three weekly paclitaxel 127 Table 36. Response rates by sensitivity to prior platinum therapy for paclitaxel 175 mg/ m2 versus 250 mg/m2 130 Table 37. Incidence of Grade 3 or 4 Toxicity other than neutropenia. 131 Table 38. Response to oral topotecan versus i.v. topotecan 133 Table 39. Response to oral and i.v. topotecan according to baseline stratifying factors 134 Table 40. Grade 3 and 4 haematological toxicities, per patient and per course by oral versus i.v. topotecan. 135 Table 41. Non-haematological toxicities, all grades and worst grade per patients for oral versus i.v. topotecan. 137 Table 42. Adverse event data from 30-49 140 Table 43: Study drugs and cycles received in the 30-49 trial. 141 Table 44: Quantities of resource use, per person 142 Table 45: US cost-analysis, per person (US$) 143 Table 46: UK cost-analysis, per person (US$) 144 Table 47: Total costs of treatment per person (Euros) 148 Table 48: Base case cost-analysis (Euros) 151 Table 49: Clinical outcomes in Prasad 153 Table 50: Quantities of resources used included in the GSK submission 158 Table 51: Results of the cost-minimisation re-analysis 160 Table 52: Results from the long-term follow-up of patients in the study 30-49 165 Table 53: Results of the Schering-Plough cost-minimisation analyses 168 Table 54. Results of LYG estimations for paclitaxel/platinum, paclitaxel monotherapy, topotecan and PLDH 171 Table 55: Results of cost-effectiveness analysis for the 4 chemotherapeutic agents under study (in £) 173 Table 56. Direct comparisons of relevant treatment comparators according to the included patient population 183 Table 57: Survival data used in analysis 1 and 2 199 Table 58: Other trial data 200 Table 59: Details of cost data 202 Table 60: Cost results broken down by category 205 Table 61: Cost-effectiveness results from analysis 1: base case model 206 Table 62: Cost-effectiveness results applying subgroup-specific treatment effects. 210 Table 63: Cost-effectiveness results applying alternative cost assumptions. 211 Table 64: Cost-effectiveness results incorporating results from 30-57 via MTC 213 Table 65: Cost results broken down by category 214

6 Table 66: Cost-effectiveness results from analysis 2 215 Table 67: Results of the budget impact analysis 217 Table 68: Summary of the median survival, progression-free survival and response rates for participants with platinum resisitant disease 230 Table 69: Summary of the median survival, progression-free survival and response rates for participants with platinum sensitive disease 233

FIGURES Figure 1. Process of study selection for clinical effectives and cost-effectiveness 43 Figure 2. Kaplan-Meier overall survival curves for PLDH versus topotecan 68 Figure 3. Kaplan-Meier survival curves for PLDH versus topotecan: platinum-sensitive subgroup 69 Figure 4. Kaplan-Meier survival curves for PLDH versus topotecan: platinum-refractory subgroup 70 Figure 5. Kaplan-Meier progression-free survival curves for PLDH versus topotecan 72 Figure 6 Relative risks for response rates for PLDH versus topotecan sub-group analysis stratified by platinum sensitivity 75 Figure 7. Relative risk of number of patients with a maintained or improved QoL score at 12 weeks follow-up for PLDH versus topotecan 77 Figure 8. Q-TwiST survival analysis: partitioned survival curves 78 Figure 9. Q-TwiST threshold utility analysis for PLDH versus topotecan 79 Figure 10. Relative risks of Grade 3 treatment-related adverse events for PLDH versus topotecan 84 Figure 11. Relative risks of Grade 4 treatment-related adverse events for PLDH versus topotecan 85 Figure 12. Percentage of participants receiving topotecan or paclitaxel surviving at least 4 years post-randomisation. 88 Figure 13. Relative risks of Grade 3 and 4 haematological toxicity for topotecan versus paclitaxel. 93 Figure 14. Relative risks of Grade 3 treatment related adverse events for PLDH versus paclitaxel 104 Figure 15. Relative risk for response to CAP versus Paclitaxel for initial treatment. 109 Figure 16. Relative risk for response to CAP versus Paclitaxel after treatment cross-over 110 Figure 17. Relative risks of treatment related toxicities for paclitaxel compared to CAP 111 Figure 18. Kaplan-Meier curves for overall survival for paclitaxel / platinum combination therapy versus platinum-based therapy alone 113 Figure 19. Kaplan-Meier curves for progression-free survival for paclitaxel / platinum combination therapy versus platinum-based therapy alone 115

7 Figure 20: Relative risks of moderate to severe toxicity for paclitaxel plus platinum chemotherapy compared to conventional platinum-based therapy alone. 118 Figure 21: Relative risks of Grade 3 or Grade 4 haematological toxicity for oral versus i.v. topotecan 136 Figure 22: Structure of the economic model 179 Figure 23: Analysis 1, overall population 207 Figure 24: Analysis 1, platinum sensitive patients 208 Figure 25: Analysis 1, platinum resistant/refractory patients 209 Figure 26: Analysis 2, platinum-sensitive patients 216

8 EXECUTIVE SUMMARY Background Ovarian cancer is the most common gynaecological cancer, with an annual incidence of 21.9 per 100,000 women in England and 26.7 per 100,000 in Wales (2000 figures). The prognosis is generally poor, due to the advanced stage of disease at detection in most cases, and the UK 5 year survival rate is only around 30%. The current guidance issued by NICE is that first-line chemotherapy should include either paclitaxel in combination with a platinum based chemotherapy regimen, or a platinum-based regimen alone (carboplatin or cisplatin). As the majority of patients ultimately relapse and require re-challenging with second-line therapy, the guidance is that patients who have received recommended first-line therapy should not be re-challenged with the same agents. Pegylated liposomal doxorubicin hydorocholoride (PLDH), topotecan and paclitaxel may therefore be considered alongside other drugs licensed for second- line therapy in advanced ovarian cancer. Participants who had not received paclitaxel as a component of first-line therapy may receive it as second-line.

Objectives of the review To examine the clinical effectiveness and cost-effectiveness of intravenous formulations of topotecan monotherapy, PLDH monotherapy, and paclitaxel used alone or in combination with a platinum-based compound for the second-line or subsequent treatment of advanced ovarian cancer.

Methods Search strategy: Seventeen databases were searched for randomised controlled trials and systematic reviews for the clinical effectiveness of PLDH, topotecan and paclitaxel and economic evaluations of the cost effectiveness of PLDH, topotecan and paclitaxel.

Inclusion/exclusion criteria: Two reviewers independently screened all titles and/or abstracts including economic evaluations. The full manuscript of any study judged to be relevant by either reviewer was obtained and assessed for inclusion or exclusion. Disagreements were resolved through discussion. For the assessment of clinical effectiveness, randomised controlled trials (RCTs) that compared topotecan monotherapy, PLDH monotherapy or paclitaxel administered alone or in combination with a platinum based compound with any other comparator including usual supportive care were included. For the assessment of cost-effectiveness, a broader range of studies was considered.

Data extraction and quality assessment: Data from included studies were extracted by one reviewer and independently checked for accuracy by a second reviewer. Individual studies were assessed for quality by one reviewer and independently checked by a second for accuracy.

Methods of analysis/synthesis: The results of the data extraction and quality assessment of the RCTs were presented in structured tables and as a narrative summary. For the cost-effectiveness section of the report, details of each identified published economic evaluation, together with a critical appraisal of its quality, were presented in structured tables.

Handling the company submissions: All the clinical effectiveness data included in the three company submissions were assessed. Where this met the inclusion criteria it was included in the clinical effectiveness review. All economic evaluations (including accompanying models) included in the company submissions were assessed and a detailed assessment of the assumptions underlying the submitted analyses were undertaken. A new model was developed to assess the costs of the alternative treatments, the differential mean survival duration and the impact of health-realted quality of life. Monte-Carlo simulation was used to reflect uncertainty in the cost-effectiveness results.

Results

A total of 2,542 titles and abstracts were screened for inclusion in the review of clinical and cost-effectiveness; 194 studies were ordered as full papers and assessed in detail. Nine RCTs were identified. In five of these trials, both the comparators were used within their licensed indications. Of these five trials, three of the trials included participants with both platinum-resistant and platinum-sensitive advanced ovarian cancer, whilst a further two trials only included participants with platinum-sensitive disease. The comparators that were assessed in the three trials that included both sub- types of participants were: PLDH versus topotecan; topotecan versus paclitaxel; and

10 PLDH versus paclitaxel. In the further two trials that included participants with the sub-type of platinum-sensitive disease, the comparators that were assessed were: single-agent paclitaxel versus a combination of cyclophosphamide, doxorubicin and cisplatin (CAP); and paclitaxel plus platinum-based chemotherapy versus conventional platinum-based therapy alone.

A further four trials were identified and included in the review in which one of the comparators in the trial was used outside its licensed indication. The comparators assessed in these trials were oxaliplatin versus paclitaxel; paclitaxel given weekly versus every three weeks; paclitaxel at two different dose levels; and oral versus intravenous topotecan. a) Clinical effectiveness Trials including participants with refractory, resistant and platinum-sensitive disease

PLDH versus topotecan: PLDH was marginally more effective than topotecan in terms of overall survival in the total trial population that included both participants with platinum-sensitive and platinum-resistant disease. However, this result appears to be driven by the more highly significant benefit of PLDH treatment in the platinum-sensitive sub-group of participants. For participants with platinum-resistant disease there was no statistically significant difference in overall survival between the PLDH and topotecan treatment groups. There were also no statistically significant differences between the PLDH and topotecan groups in terms of progression-free survival, response or quality of life.

In terms of toxicities reported during the trial, the rates of grade 3 stomatitis, PPE, mucous membrane disorder and rash were significantly higher in the PLDH treatment arm. In the topotecan arm the rates of grade 3 and 4 haematological toxicities, and grade 3 alopecia and fever were significantly higher.

Topotecan versus paclitaxel: There were no statistically significant differences between the two treatment groups in terms of overall survival, time to progression, response rate or response duration. The point estimates for all of these outcomes

11 favoured treatment with topotecan over paclitaxel. However, there was a significant difference between the two treatment groups in terms of time to response, favouring paclitaxel.

In this trial, treatment with topotecan was associated with significantly more grade 3 and 4 haematological toxicities compared to paclitaxel. In addition, grades 3 and 4 nausea, vomiting, constipation, abdominal pain, asthenia, fatigue, and fever / infection were significantly higher in this group. Treatment with paclitaxel was associated with significantly more grades 3 and 4 alopecia, arthralgia, myalgia and skeletal pain compared to the topotecan treatment arm.

PLDH versus paclitaxel: In relation to overall survival there was no significant difference between the PLDH and paclitaxel treatment groups. Treatment with PLDH was associated with significantly more grade 3 PPE, ascites, stomatitis and dyspnea compared to treatment with paclitaxel. Treatment with paclitaxel was associated only with a higher incidence of grade 3 alopecia relative to PLDH. This trial was terminated prematurely, therefore the results should be interpreted with caution.

Trials including participants with platinum-sensitive disease only

Paclitaxel versus CAP: CAP was more effective than paclitaxel in terms of both overall and progression-free survival. There were no significant differences between the two treatment regimens in terms of response. However, the incidence of grade 3 and 4 haematological toxicities and grade 2 nausea and vomiting was significantly higher in the CAP treatment arm. Treatment with paclitaxel was associated with significantly higher rates of alopecia and allergic reactions relative to treatment with CAP.

Paclitaxel in combination with platinum-based chemotherapy versus platinum based therapy alone: Paclitaxel in combination with platinum based chemotherapy was more effective than platinum monotherapy in relation to both overall survival and progression-free survival. However, there was no significant treatment benefit observed for combination therapy for response rates or overall quality of life.

12 Treatment with paclitaxel in combination with platinum was associated with significantly higher rates of grades 2 to 4 neurological toxicity and alopecia. Treatment with platinum monotherapy was associated with significantly higher rates of haematological toxicity.

Trials in which one of the comparators was used outside the licensed indication:

Paclitaxel versus oxaliplatin: There were no statistically significant differences between the paclitaxel and oxaliplatin treatment groups in terms of overall survival, time to progression, response rate, response duration or quality of life. Treatment with paclitaxel was associated with a higher incidence of severe neutropenia, whilst oxaliplatin was associated with higher rates of thrombocytopenia.

Paclitaxel given weekly versus every three weeks: There was no significant treatment benefit for either of these regimens as assessed by overall survival, time to progression, response or response duration. Treatment with paclitaxel every three weeks was associated with a significantly higher incidence of grade 3 and 4 neutropenia and alopecia, whilst treatment every week was associated with problems with nail changes.

Paclitaxel 175 mg/m2 versus 250 mg/m2: There were no statistically significant differences between participants treated with the lower dose regimen and those treated with the higher dose regimen for overall survival or progression-free survival. There was a significant benefit in favour of the higher dose regimen for response rates. However, the reporting of grade 3 and 4 haematological toxicities was more common in the higher dose treatment group.

Oral versus intravenous topotecan: There was a significant benefit in favour of intravenous topotecan for overall survival. However, no further significant differences between the two treatment regimens was found for time to progression, response rate, response duration and time to response. Neutropenia and leucopenia occurred frequently in both treatment groups, but were higher in the intravenous treatment group. The rates of grade 3 and 4 nausea, vomiting, diarrhoea and fever

13 were all significantly higher in the oral treatment regimen group compared to the i.v. treatment arm.

Cost-effectiveness A total of 4 studies met the inclusion criteria for the cost-effectiveness review. In addition, separate submissions were received from Bristol Myers Squibb, GlaxoSmithKline and Schering-Plough Ltd. The review of the economic evidence from the literature and industry submissions identified a number of significant limitations in existing studies assessing the cost-effectiveness of PLDH, topotecan and paclitaxel. A new model was developed to address the limitations identified in these sources and to provide a direct comparison of the full range of possible strategies that are relevant to the NHS. The model explored a range of uncertainties and sources of variability that were not fully addressed in existing data sources. Two separate analyses (Analysis 1 and Analysis 2) were required in order to reflect the heterogeneity identified in the different trials, and the difficulties encountered in obtaining robust estimates, using a consistent approach for the methods of evidence synthesis, of the relative treatment effects.

Analysis 1 assessed the cost-effectiveness of PLDH, topotecan and paclitaxel administered as a monotherapy. Sensitivity analysis was undertaken to explore the impact of patient heterogeneity (e.g. platinum sensitive and platinum resistant/refractory patients), the inclusion of additional trial data (30-57) and alternative assumptions regarding treatment and monitoring costs. In the base-case results for Analysis 1, paclitaxel monotherapy emerged as the cheapest treatment. When the incremental cost-effectiveness ratios (ICERs) were estimated, topotecan was ruled out on the grounds of extended dominance. Hence the options considered in the estimation of the ICERs were paclitaxel and PLDH. The ICER for PLDH compared to paclitaxel was £24,606 per QALY in the overall patient population (comprising platinum sensitive, refractory and resistant patients). The ICER was more favourable in the platinum sensitive group (£20,212 per QALY) and less favourable in the platinum refractory/resistant group (£33,430 per QALY). The cost-effectiveness results for the base-case analysis were sensitive to the inclusion of trial 30-57. Incorporating the results of trial 30-57 resulted in less favourable estimates for the ICER for PLDH vs paclitaxel compared to the base-case results. The ICER of PLDH

14 compared to paclitaxel was £58,475 per QALY in the overall patient population, £45,894 per QALY in the platinum sensitive population and £76,192 per QALY in the platinum resistant and refractory population.

The results from Analysis 2 explored the cost-effectiveness of the full range of treatment comparators for platinum sensitive patients. The treatment options considered in this model comprised: PLDH, topotecan, paclitaxel monotherapy, CAP, paclitaxel plus platinum and platinum monotherapy. Due to the less robust approaches that were employed to synthesise the available evidence and the heterogeneity between the different trials, the reliability of these results should be interpreted with some caution. Topotecan, paclitaxel monotherapy and PLDH were all dominated by platinum monotherapy (i.e. higher costs and lower QALYs). CAP was ruled out by extended dominance. After excluding these alternatives, the treatments that remained under consideration were platinum monotherapy and paclitaxel/platinum combination therapy. The ICER for paclitaxel /platinum combination compared to platinum monotherapy treatment was £3,561 per QALY.

Conclusions Clinical effectiveness Participants with platinum-resistant disease: For participants with platinum-resistant disease there was a low probability of response to treatment with PLDH, topotecan or paclitaxel. Furthermore there was little difference between the three comparators in relation to overall survival. The comparators did however differ considerably in their toxicity profiles. Given the low survival times and response rates it appears that the maintenance of quality of life, and the control of symptoms and toxicity are paramount in this patient group. As the three comparators differed significantly in terms of their toxicity profiles, patient and physician choice is also an important element that should be addressed when decisions are made regarding second-line therapy. It can also be suggested that this group of patients may benefit from being included in further clinical trials of new drugs.

15 paclitaxel and platinum combination therapy. This suggests re-challenge with combination therapy may be more beneficial than re-challenge with a single-agent chemotherapeutic regimen. In terms of single-agent compounds, the evidence suggests that PLDH is more effective than topotecan. Evidence from a further trial that compared PLDH and paclitaxel suggests that there is no significant difference between these two comparators in this trial. The three comparators did however differ significantly in terms of their toxicity profiles across the trials. Although re-challenge with PLDH may therefore be more beneficial than with topotecan, patient and physician choice as to the potential toxicities associated with each of the comparators, and the patients ability and willingness to tolerate these are of importance.

Cost effectiveness The following conclusions are possible assuming the NHS is willing to pay up to £20,000 to £40,000 per additional QALY.

• PLDH appears to be cost-effective compared to topotecan and paclitaxel monotherapy, in terms of the overall patient population and the main subgroups considered. • The cost-effectiveness results for the base-case analysis were sensitive to the inclusion of trial 30-57. Incorporating the results of trial 30-57 gave less favourable estimates for the ICER for PLDH vs paclitaxel monotherapy, compared to the base-case results. Under this scenario the results would suggest that paclitaxel monotherapy was the optimal treatment. • For platinum sensitive patients, the combination of paclitaxel and platinum appears to be cost-effective.

Research Recommendations Participants with platinum-resistant disease: On the strength of the evidence reviewed in this assessment, it can be suggested that participants with platinum- resistant disease may benefit from being included in further clinical trials of new drugs.

Participants with platinum-sensitive disease: To assess the effectiveness of combination therapy against a single agent non-platinum based compound, it can be

16 suggested that a trial that compared paclitaxel in combination with a platinum-based therapy versus single-agent PLDH would be a reasonable option.

List of abbreviations

ASCO American Society of Clinical Oncology AUC Area under the curve CBA Cost benefit analysis CEA Cost effectiveness analysis CI Confidence interval CEAC Cost-effectiveness acceptability curve CMA Cost minimisation analysis COSTART Coding Symbols for Thesaurus of Adverse Reactions Terms CR Complete response CT Computed tomography CUA Cost utility analysis ECOG European Co-operative Oncology Group EORTC European Organization of Research and Treatment of Cancer EVPI Expected value of perfect information G-CSF Granulocyte colony stimulating factor HR Hazard ratio ICER Incremental cost-effectiveness ratio ITT Intention to treat i.p. Intraperitoneal i.v. Intravenous KPS Karnofsky performance status MRI Magnetic resonance imaging NA Not applicable NCI – CTC National Cancer Institute’s Common Toxicity Criteria NHS R&D HTA NHS Research & Development Health Technology Assessment Programme NICE National Institute for Clinical Excellence Nd Not determined Ns Not statistically significant NS Not stated OR Odds ratio PLDH Pegylated liposomal doxorubicin hydrocholoride PP Per protocol PPE Palmar-plantar erythrodysesthesia PR Partial response Pt Platinum Pt-r Platinum-refractory (disease) Pt-s Platinum-sensitive (disease) QLQ-C30 Quality of Life Questionnaire-C30 QoL Quality of life QALY Quality adjusted life year RR Relative risk s.c. Subcutaneous SD Standard deviation TwiST Time without symptoms or toxicity

18 Definitions of terms

Advanced ovarian cancer Disease classified as International Federation of Gynaecologists and Obstetricians (FIGO) stages III to IV. Absolute risk reduction (ARR) The difference between the event rates in the two groups, where the adverse event rate is less in the intervention group this suggests the intervention is beneficial. Adverse effect/ adverse event An abnormal or harmful effect caused by, and attributable to, exposure to a chemical (e.g. a drug), which is indicated by some result such as death, a physical symptom or a visible illness. An event may be classified as adverse if it causes functional or anatomical damage, causes irreversible changes in the homeostasis of the organism or increases the susceptibility of the organism to other chemical or biological stress. Alopecia Baldness/ the loss of body hair. Anaemia An abnormally low level of red blood cells in the blood. Red blood cells are responsible for carrying oxygen around the body. Anthracyclines A group of antibiotics that have a tetrahydronaphthacenedione ring structure attached by a glycosidic linkage to a sugar molecule. These antibiotics have potent antineoplastic activity. They intercalate with DNA, and so adversely affect many DNA functions. Furthermore, they interact with cell membranes thereby altering their functions and generating hydrogen peroxide and hydroxy radicals, which are highly destructive to cells. Antineoplastic Inhibiting or preventing the development of neoplasms, and checking the maturation and proliferation of malignant cells. Arthralgia Joint pain. Bias Deviation of results or inferences from the truth, or processes leading to such deviation. Any trend in the collection, analysis, interpretation, publication or review of data that can lead to conclusions that are systematically different from the truth. Blinding A procedure used in clinical trials to avoid the possible bias that might be introduced if the patient and / or doctor knew which treatment the patient would be receiving. If neither the patient nor the doctor is aware of which treatment has been given, the trial is termed ‘double- blind’. If only one of the patient or doctor is aware, the trial is called ‘single-blind’. CA-125 A cell surface marker found in serum. Carcinoma A cancerous growth Censored data Censorship means that th event does not occur during the period of observation and the time of event is unknown, but these cases are incorporated inot the the analysis. Thos whose event is unknown, or who are lost to the study (right censored) or new patients introduced inot the study (left censore), add to the information on patients whose event time is known (uncensored) at each time interval.

19 Chemotherapy The use of drugs that are capable of killing cancer cells, or preventing/slowing their growth. Co-intervention In a randomised controlled trial, the application of additional diagnostic of therapeutic procedures to members of either the experimental or reference group, or to both groups. Complete response The total disappearance of all detectable malignant disease for at least 4 weeks. Confidence Interval (CI) A measure of precision of statistical estimate. Confounding (1) The masking of an actual association or (2) false demonstration of an apparent association between the study variables when no real association between them exists. Cost-benefit analysis An attempt to give the consequences of the alternative interventions a monetary value. In this way, the consequences can be more easily compared with the costs of the intervention. This involves measuring individuals’ “willingness to pay” for given outcomes, and can be difficult. Cost-effectiveness analysis The consequences of the alternatives are measured in natural units, such as years of life gained. The consequences are not given a monetary value. Cost effectiveness acceptability A graphical representation of the probability of an curve (CEAC) intervention being cost effective over a range of monetary values for society’s willingness to pay for an additional unit of health gain. Cost-minimisation analysis When two alternatives are found to have equal efficacy or outcomes (consequences). Therefore, the only difference between the two is cost. This is sometimes considered to be a sub-type of cost-effectiveness analysis. Cost-utility analysis The consequences of alternatives are measured in ‘health state preferences’, which are given a weighting score. In this type of analysis, different consequences are valued in comparison with each other, and the outcomes (e.g. life- years gained) are adjusted by the weighting assigned. In this way, an attempt is made to value the quality of life associated with the outcome so that life-years gained become quality-adjusted life-years gained. Cycle Chemotherapy is usually administered at regular intervals. A cycle is a course of chemotherapy followed by a period in which the body recovers from the adverse events of the drug(s). Cytotoxic Toxic to cells. This term is used to describe drugs that kill cancer cells or slow their growth. Debulking Removal by surgery of a substantial proportion of cancer tissue. Optimal debulking refers to the removal of the largest possible amount of tumour while limiting the damage to the surrounding normal tissue; interval debulking refers to the surgical removal of a tumour after chemotherapy, aimed at further reducing its bulk. ECOG performance status 0: fully active, able to carry on all pre-disease performance without restriction. 1: restricted in physically strenuous activity but ambulatory and able to carry out work of a light or sedentary nature, e.g., light house work, office work

20 2: ambulatory and capable of all selfcare but unable to carry out any work activities. Up and about more than 50% of waking hours. 3: capable of only limited self-care, confined to bed or chair more than 50% of waking hours. 4: completely disabled. Cannot carry on any selfcare. Totally confined to bed or chair. 5: Dead. End-point A clearly defined outcome or event associated with an individual in a medical investigation. EORTC The European Organization for Research and Treatment of Cancer (EORTC) is an organisation set up to conduct, develop, coordinate, and stimulate laboratory and clinical research in Europe to improve the management of cancer and related problems by increasing survival but also quality of life of patients. Evaluable disease Unidimensionally measurable lesions, masses with margins not clearly defined, lesions with both diameters ≤0.5 cm, lesions on scan with either diameter smaller than the distance between cuts, palpable lesions with either diameter ≤ 2cm; malignant ascites or pleural effusion in conjunction with serum levels of CA-125>100 U/mL in the absence of cirrhosis (CA-125 is a glycoprotein antigen expressed by some ovarian cancers). External validity The ability to generalise the results from a particular experiment to a larger population. First-line therapy The first chemotherapy regimen (usually administered with curative intent) given to patients who have been newly diagnosed with ovarian cancer, or who had an early stage of the disease which has been previously treated with surgery alone but has since relapsed and requires chemotherapy. Forest plot The way in which results from a meta-analysis are often presented. Results are displayed graphically as horizontal lines representing the 95% or 99% confidence intervals of the effect of each trial (strictly the 95% or 99% CIs of a relative risk of the intervention group compared with the control group). Hazard ratio Measure of relative risk used in survival studies. Heterogeneous Of differing origins or different types. Histological grade The degree of malignancy of a tumour as judged by histology. Histological type The type of tissue found in a tumour as determined by histology. Histology The examination of the cellular characteristics of a tissue. Incidence The number of new events (new cases of a disease) in a defined population, within a specified period of time. Incremental cost-effectiveness An expression of the additional cost of health gain ratio associated with an intervention relative to an appropriate comparator. Expressed as the difference in mean costs (relative to the comparator) divided by the difference in mean effects. Sometimes expressed with confidence intervals. Intention-to-treat analysis An analysis of a clinical trial where participants are method analysed according to the group to which they were

21 initially randomly allocated, regardless of whether or not they dropped out, fully complied with the treatment or crossed over and received the other treatment. Interim analysis A formal statistical term indicating an analysis of data part-way through a study. Internal validity The degree to which a study is logically sound and free of confounding variables. International Federation of This organisation defines staging in gynaecological Gynaecologists and Obstetricians cancer and collates information about treatment and (FIGO) survival from a group of collaborating European centres (including some in the UK). Kaplan-Meier curves (also called A non-parametric method of compiling life or survival product limit method) tables, developed by Kaplan and Meier in 1958. This combines calculated probabilities of survival and estimates to allow for censored observations, which are assumed to occur randomly. The intervals are defined as ending each time an event (e.g. death, withdrawal) occurs and are therefore unequal. Karnofsky performance status A performance measure for rating the ability of a person scale to perform usual activities, evaluating a patient’s progress after a therapeutic procedure, and determining a patient’s suitability for therapy. It is used most commonly in the prognosis of cancer therapy, usually after chemotherapy and customarily administered before and after therapy.

A measure is given by a physician to a patients ability to perform certain ordinary tasks: 100 – normal, no complaints; 70 – unable to carry on normal activity; 50 – requires considerable assistance; 40 – disabled; 30 – hospitalisation recommended Localised disease Disease that is confined to a small part of an organ or tissue. Leucopenia An abnormally low level of leucocytes in the blood. Leucocytes are white blood cells which help to fight infections within the body. Lymph nodes Small organs that act as filters in the lymphatic system. Lymph nodes close to a primary tumour are often the first sites to which a tumour spreads. Measurable disease The presence of lesion(s) that can be unidimensionally or bidimensionally measured by physical examination, echography, radiography or computed tomographic scan. Meta-analysis A quantitative method for combining the results of many studies into one set of conclusions. Metastasis/ metastatic cancer Cancer that has spread to a site distant from the original site. Mortality rate The proportion of deaths in a population or in a specific number of the population per unit of time. Myalgia Muscle pain Neuropathy A term to describe any disorder of the neurones or nerves of the body. Neutropenia An abnormally low level of neutrophils in the blood. Neutrophils belong to a group of white blood cells known as granulocytes, which are important in fighting infections within the body. Number needed to treat (NNT) In clinical treatment regimens, the number of patients

22 with a specified condition who must follow the specified regimen for a prescribed period in order to prevent occurrence of specified complications or adverse outcomes of the condition. Mathematically equal to 1/ (risk difference). Palliative Anything that serves to alleviate symptoms due to the underlying cancer but is not expected to act as a cure. Palmar-plantar A condition characterised by an intense, often painful erythrodysesthesia (PPE) macular reddening that primarily involves the palms of the hands and soles of the feet. The skin changes may range from a painful desquamating dermatitis, with mild erythema and hyperaemia, to severe crusting, ulceration and epidermal necrosis. The mechanism of this condition is not known but is believed to be a result of microtrauma within the tissue leading to leaky blood vessels. Paresthesia Numbness/tingling or ‘pins and needles’ sensation of the skin Partial response At least a 50% decrease in tumour size for more than 4 weeks without an increase in the size of any area of known malignant disease or the appearance of new lesions. Phase II trial A study with a small number of patients diagnosed with the disease for which the drug is being studied. In this study, the safety of the new drug is tested. Early effectiveness data are also collected for varying doses of the drug. Phase III trial A study with a large number of patients diagnosed with the disease for which the drug is being studied and is unlicensed for the indication. In this study, the drug is tested against a placebo or alternative treatment. Placebo A ‘dummy’ treatment administered to the reference group in a controlled clinical trial in order to distinguish the specific and non-specific effects of the experimental treatment (i.e. the experimental treatment must produce better results than the placebo in order to be considered effective). Platinum-based chemotherapy Treatment with platinum-based drugs such as cisplatin or carboplatin. Platinum-resistant disease Disease which responded to first line platinum-based chemotherapy but relapsed within 6 months. Platinum-sensitive disease Disease which responded to first-line platinum-based therapy but which relapsed after longer than 6 months. Prevalence The measure of the proportion of people in a population who have some attribute or disease at a given point in time or during some time period. Progressive disease Used to describe a tumour that continues to grow or where a patient develops more metastatic sites. Progression-free survival The time from the start of study drug administration to documented disease progression or death due to any cause while the participant was on study drug or during the long-term follow-up period. Prophylaxis/ prophylactic An intervention (i.e. any act, procedure, drug or treatment equipment) used to guard against or prevent an unwanted outcome. Proportional hazards model Regression method for modelling survival times. The

23 outcome variable is whether or not the event of interest has occurred, and if so, after what period of time; if not, the duration of follow-up. The model predicts that hazard or risk of the event in question at any given time. p-value In the context of significant tests, the p-value represents the probability that a given difference is observed in a study sample, when such a difference does not exist in the relevant population. Small p-values indicate stronger evidence to reject the null hypothesis of no difference. Quality-adjusted life-years A term originally developed in cancer studies to balance (QALYs) poor quality of life (possibly with long life expectancy) with good quality of life (possibly with short life expectancy). Quality of Life (QoL) A concept incorporating all the factors that might impact on an individual’s life, including factors such as the absence of disease or infirmity as well as other factors which might affect their physical, mental and social wellbeing. Quality of Life Questionnaire A self-administered quality of life questionnaire (QLQ-C30) developed by the EORTC for the measurement of health- related quality of life. The questionnaire consists of nine scales – one global QOL scale, five function scales (physical, role, emotional, cognitive, and social), and three symptom scales (fatigue, pain, nausea/vomiting) and questions on six single items (dyspnea, sleep disturbance, appetite loss, constipation, diarrhoea, and financial impact). Higher scores on the function scales indicate better functioning and QOL, whereas higher scores on the symptom scales indicate the presence of more symptoms. Random allocation A method of allocation to ensure that the treatment assignment is unpredictable. Randomised controlled trial These are designed to measure the efficacy and safety of (RCT) (also randomised clinical particular types of healthcare interventions, by randomly trial) assigning people to one of two or more treatment groups and, where possible, blinding them and the investigators to the treatment that they are receiving. The outcome of interest is then compared between the treatment groups. Such studies are designed to minimise the possibility of an association due to confounding and remove the many sources of bias present in other study designs. Relative risk (RR) Also called the 'risk ratio'. A common way of estimating the risk of experiencing a particular effect or result. A RR > 1 means a person is estimated to be at an increased risk, while a RR < 1 means a person is apparently at decreased risk. A RR of 1.0 means there is no apparent effect on risk at all, e.g., if the RR = 4.0, the result is about 4 times as likely to happen, and 0.4 means it is 4 times less likely to happen. The RR is expressed with confidence intervals: e.g., RR=3.0 (95% CI: 2.5, 3.8). This means the result is 3 times as likely to happen - anything from 2.5 times as likely, to 3.8 times as likely. It is statistically significant. On the other hand, RR=3.0 (95% CI: 0.5, 8.9), means it is also estimated to be 3 times as likely, but it is not statistically significant. The chances go from half as likely to happen (0.5 a decreased chance), to nearly 9 times as

24 likely to happen (8.9 an increased chance). Relative risk reduction (RRR) Alternative way of expressing relative risk. It is calculated as follows: RRR= (1 – RR) x100% The RRR can be interpreted as the proportion of the initial or baseline “risk” which was eliminated by a given treatment or intervention, or by avoidance of exposure to a risk factor. Recurrent disease Disease that re-appears after a period during which it has shown no measurable/detectable signs. Refractory disease Disease which does not respond or progresses during first line platinum-based chemotherapy. Risk difference The difference (absolute) in the proportion with the outcome between the treatment and control groups. If the outcome represents an adverse events and the risk difference is negative (below zero) this suggests that the treatment reduces the risk – referred to as the absolute risk reduction. Salvage therapy Any therapy given in the hope of getting a response when the “standard” therapy has failed. This may overlap with “second-line” therapy, but could also include therapy given for patients with refractory disease i.e. disease that has never responded to first-line therapy. Second-line therapy The second chemotherapy regimen administered either as a result of relapse after first-line therapy or immediately following on from first-line therapy in patients with progressive or stable disease. Depending on the circumstances patients may be treated with the same regimen again, or a different regimen. In either case this is defined as second-line therapy. Stable disease No change or less than a 25% change in measurable lesions for at least 4-8 weeks with no new lesions appearing. Staging The allocation of categories (e.g. for ovarian cancer FIGO stages I to IV) to tumours, defined by internationally agreed criteria. Tumour stage is an important determinant of treatment and prognosis. Stomatitis Inflammation/ulceration of the mouth. Taxane naïve Patients who had not received a taxane as part of first-line therapy. Thrombocytopenia An abnormally low level of platelets in the blood. Platelets play a role in the blood clotting process. Time to progression The length of time from the start of treatment (or time from randomisation within the context of a clinical trial) until tumour progression. Topoisomerase inhibitors Drugs that target the DNA topoisomerase I enzyme which is involved in the replication of DNA. This leads to the inhibition of cell division. Utility A measure of the strength of an individual’s preference for a given health state or outcome. Utilities assign numerical values on a scale from 0 (death) to 1 (optimal or ‘perfect’ health), and provide a single number that summarises health-related quality of life. Hence, utility has been described as a global measure of health-related quality of life. Sometimes ‘utility’ is only used to refer to preferences (on the 0-1 scale) that are elicited using

25 methods which introduce risky scenarios to the respondent (standard gamble), with the term ‘values’ used to refer to other type of preferences. Values An alternative measure of the strength of an individual’s preference for a given health state or outcome. In contrast to utilities, values reflect preferences elicited in a risk-less context.

1. AIM OF THE REVIEW

This review examined the clinical effectiveness and cost-effectiveness of intravenous formulations of topotecan monotherapy (Hycamtin®, GlaxoSmithKline), pegylated liposomal doxorubicin hydrochloride monotherapy (Caelyx®, Schering-Plough, UK; Doxil®, Alza Corporation, USA), and paclitaxel, used alone or in combination with a platinum-based compound (Taxol®, Bristol-Myers Squibb; Paxene®Mayne Pharma) for the second-line or subsequent treatment of advanced ovarian cancer. The patient population that the review addressed were women who had failed first-line platinum- based chemotherapy.

2. BACKGROUND 2.1. Description of underlying health problem

Epidemiology

Ovarian cancer is the most common gynaecological cancer in England and Wales, and the fourth most common cause of cancer death in women, accounting for a total of 6% of female cancer deaths. In 2000 there were 5512 new cases in England and 399 in Wales, giving age-standardised rates per 100,000 women of 17.9 (95% confidence interval: 17.4 – 18.4) and 20.6 (95% confidence interval: 18.6 to 22.7) respectively.(1) In 2002-2003, ovarian cancer accounted for 34,086 hospital episodes, totalling 104,750 bed days, in England.(2) There were 4687 deaths from ovarian cancer in England and Wales in 2000.(1)

Aetiology, pathology and prognosis Ovarian cancer is generally staged using the International Federation of Gynaecologists and Obstetricians (FIGO) criteria, under which stage I refers to malignant growth which is restricted to the ovaries, and stage IV refers to disease where distant metastasis can be detected1. Ovarian cancer is often asymptomatic in its early stages and, as a result, over 75% of patients are diagnosed with advanced (stage III or IV) disease. As a result the 5 year survival rate is significantly poorer than that for other gynaecological cancers at approximately only 30%.(2)

1 Further details of the FIGO staging criteria are available in Appendix 1.

There are three principal types of ovarian cancers, with these being determined by the primary cell type. The majority of the cancers are epithelial, and these are primarily mucinous or serous carcinomas. Epithelial cancer is by far the most common, accounting for 80% of the cancers observed, whilst stromal or germ cell tumours each account for around a further 10% of cases.

The primary risk factor for developing ovarian cancer is age, with 80% of cases occurring in women over the age of 50, and the highest incidence rates in those over the age of 70.(3, 4) Between 5 and 10% of all cases occur amongst women who are known to be at high risk of the disease due to mutations in the genes BRCA1 and BRCA2, or who carry the hereditary non-polyposis colorectal cancer (HNPCC) gene.

In the 90% of ovarian cancers which are not linked to these genetic bases, a link with incessant ovarian function has been proposed. Factors suspected to be associated with an increased risk, other than age, include an early age at menarche, a late menopause, infertility and the use of fertility drugs.(5)

The factors that are reported to be protective include parity, with nulliparous women having a higher risk than parous women and there being an inverse relationship between parity and risk(6, 7) Breastfeeding may also have a protective effect(8-10), while the use of oral contraceptives(11, 12), tubal ligation(13) and hysterectomy(13) have all been reported as associated with a reduced risk of ovarian cancer.

Prognostic factors, which may influence both survival and response to treatment, include stage of the disease at diagnosis, level of residual disease after debulking surgery, the grade and histology of the tumour, performance status, and age at diagnosis.(14)

An elevated CA125 level is used as a marker in the detection and diagnosis of ovarian cancer, as around 80% of malignant ovarian masses are associated with an elevated CA125 level. Raised levels are also correlated with disease progression. However, CA125 is not specific to ovarian tumours and raised levels also occur in association with other tumours including tumours of the breast.(15)

2.2 Current service provision

The majority of patients will be treated with surgery as a first-line intervention. However, it is not usually possible to completely excise the tumour and further therapy is required in the majority of patients. This therapy generally takes the form of first-line platinum-based chemotherapy. Current NICE guidance (16) is that first- line chemotherapy should include either paclitaxel in combination with a platinum based chemotherapy regimen, or a platinum-based regimen alone. At the present time approximately 75% of patients receive paclitaxel and platinum combination therapy as their first line chemotherapy regimen, whilst the remainder received platinum-alone or in combination with a different chemotherapeutic agent.

While first-line chemotherapy achieves a response in approximately 70% - 80% of patients, the majority of these will eventually relapse and require second-line therapy. Between 55% and 75% of those who respond to first-line therapy will relapse within two years of completing treatment. On the whole, re-challenging a patient with a platinum based regimen if they are platinum sensitive, a second or subsequent time can produce a response, but fewer patients will respond and their responses will be of shorter duration. The response to first-line therapy is predictive of the response to second and subsequent courses of treatment with platinum-based therapy. In particular, the length of treatment free interval and the extent of relapse (the number and volume of tumour sites involved) have been found to be predictive of response to second and subsequent courses of therapy.

Eventually, whether it is whilst still receiving a first cycle of platinum-based therapy or after a number of re-challenges with the same platinum-based regimen, the majority of patients will fail to respond adequately. This is when second-line therapy can be offered. Such treatment is palliative, with the aims of alleviating symptoms and prolonging survival. At this stage of treatment a cure is generally not achieved, and maintenance of quality of life is a key objective of treatment. As the agents used in second-line chemotherapy are likely to cause a range of adverse effects this consideration is critical. NICE guidance currently recommends the use of either combination paclitaxel / platinum therapy, or platinum-therapy alone for the first-line treatment of advanced ovarian cancer.(16) The guidance further recommends that

29 paclitaxel should not be administered as second-line therapy to women who have received this in combination with first-line platinum based therapy. Where patients have been treated with first-line platinum monotherapy or another chemotherapeutic regimen, paclitaxel may be considered as an option for second-line therapy. Further NICE guidance recommends that topotecan(17) and PLDH(18) be considered alongside other licensed chemotherapeutic agents for second-line therapy following the failure of platinum-based mono or combination first-line chemotherapy.

2.3. Description of new intervention The interval between the end of first-line chemotherapy and the start of second line chemotherapy (treatment-free interval) has been widely used for over 10 years to stratify patients entering clinical trials and to influence the clinical management of patients. Patients who relapse after less than 6 months are considered to have platinum-resistant disease, while those who relapse after a longer period are considered to have platinum-sensitive disease. Patients who do not respond, or who progress whilst receiving first-line therapy are classified as having platinum refractory disease.

Interventions The following section of the report summarises the product characteristics for Topotecan, PLDH and paclitaxel available from the electronic Medicine Compendium (www.medicines.org.uk/). Further details of the current manufacturers information provided for health professionals are available in Appendix 2.(19)

Topotecan Topotecan is a water-soluble analogue of camptothecin, which is derived from the oriental tree Camptotheca acuminta and is a topoisomerase I inhibitor. This class of drugs inhibit the topoisomerase I enzyme, which is involved in DNA replication and is therefore critically involved in cell replication. Topotecan is currently licensed for use in patients with metastatic ovarian cancer following the failure of first line or subsequent therapy.

It is usually administered intravenously over a 30 minute period on 5 consecutive days. The initial recommended dose is 1.5mg/m2 of body surface area. Treatment is

30 normally given every 3 weeks, and while the recommended minimum treatment is 4 cycles, treatment may continue until disease progression occurs if it is well tolerated.

Contra indictions • A history of severe hypersensitivity reaction to topotecan and/or its excipients. • Pregnancy or breastfeeding. • Severe bone marrow depression with baseline neutrophils < 1.5 x 109/l and or platelet count ≤ 100 x 109/l.

Special warnings and special precautions for use • Topotecan should only be used in units specialised in the administration of cytotoxic chemotherapy and should only be administered under the supervision of a physician experienced in the use of chemotherapy. • Since haematological toxicity is dose related, full blood counts including platelets should be subject to regular monitoring. • The use of topotecan in patients with severe renal impairment (creatinine clearance < 20 ml/minute) or severe hepatic impairment (serum bilirubin ≥ 10 mg/dl) is not recommended.

Adverse events • Haematological toxicity is a dose-limiting adverse occurrence. Severe neutropenia, thrombocytopenia and moderate to severe anaemia occur in substantial proportions of patients, but are reversible and do not appear to result in cumulative toxicity. • Non-haematological adverse events occurring in more than 5% of patients include nausea, vomiting, diarrhoea, constipation, stomatitis, abdominal pain, fatigue, asthenia and alopecia.

PLDH Pegylated liposomal doxorubicin hydrochloride (PLDH) (Caelyx®) is a stealth liposome-encapsulated form of doxorubicin. Doxorubicin is isolated from cultures of Streptomyces peucetius var: caesius and is an anthracycline. This is a class of antibiotic that has antineoplastic activity as a result of intercalation with DNA and inhibition of the enzyme topoisomerase II. This class of drugs also affect the fluidity

31 and ion transport function of cell membranes and generate hydroxyl radicals which are cytotoxic.

PLDH is licensed for the treatment of advanced ovarian cancer in women who have failed a first-line platinum-based chemotherapy regimen. PLDH is usually administered at a dose of 50 mg/m2 of body surface area once every 4 weeks for as long as disease does not progress and the treatment continues to be tolerated.

Contraindictions • A history of severe hypersensitivity reactions to PLDH and/or its excipients. • Pregnancy and breastfeeding.

Special warnings and special precautions for use

• It is recommended that all patients receiving PLDH on a regular basis receive regular and frequent electrocardiogram monitoring in conjunction with monitoring of left ventricular ejection fraction by either echocardiography or, preferably, by multiple gated arteriography. Where possible injury associated with PLDH therapy is detected the risk of myocardial injury should be assessed against the benefits of continued therapy. • Caution should be exercised in treating patients who have impaired cardiac function. • Caution should be exercised in treating patients who have previously been treated with anthracyclines.

Adverse events • The principal treatment-related adverse events which are frequently reported are palmar-plantar erythrodysesthesia (PPE) and stomatitis. • Other adverse events that occur in more than 5% of patients include nausea, asthenia, rash, vomiting, alopecia, constipation, anorexia, mucous membrane disorder, diarrhoea, abdominal pain, fever, paresthesia, pain, skin discolouration, pharyngitis, dry skin, dyspepsia and somnolence. • Clinically significant laboratory abnormalities include increases in total bilirubin and serum creatinine levels.

32 Paclitaxel (Taxol and Paxene) Paclitaxel is a taxane, a class of drugs that are derived from the Pacific yew tree taxus brevifolia. The taxanes prevent the formation of mitotic spindles interfering with the process of cell division and resulting in cell death. Paclitaxel is licensed in the UK for first-line chemotherapy of ovarian cancer in combination with cisplatin. However, in practice it is usually administered in conjunction with carboplatin for patients who have advanced disease or residual disease following surgery. It is also licensed for second-line chemotherapy after failure of standard platinum first-line therapy. It is usually administered at a dose of 175 mg/m2 of body surface area as a 3 hour intravenous infusion, followed by a platinum compound. Treatment is usually given every three weeks, with patients normally receiving six cycles. Paxene is licensed for the treatment of patients with metastatic carcinoma of the ovary after failure of platinum-containing combination therapy without taxanes.

Contraindictions • A history of severe hypersensitivity reactions to paclitaxel or any other component of the formulation, especially polyethoxylated castor oil. • Pregnancy or breastfeeding. • Baseline neutrophils < 1.5 x 109/l.

Special warnings and special precautions for use

• Paclitaxel should only be administered under the supervision of a physician experienced in the use of cancer chemotherapeutic agents. As significant hypersensitivity reactions may occur, appropriate supportive equipment should be available. • Patients must be pre-treated with corticosteroids, antihistamines and H2 antagonists. • Paclitaxel should be administered before cisplatin where they are employed in combination. • Frequent monitoring of blood counts should be carried out, as bone marrow suppression, and neutropenia in particular, is a dose-limiting toxicity. Patients should not be retreated until neutrophils recover to a level ≥ 1.5 x 109/l and platelets to a level ≥ 109/l

33 • In patients with mildly abnormal liver function there is no evidence of increased toxicity of paclitaxel when given as a 3 hour infusion. When paclitaxel is given as a longer infusion, increased myelosuppression may be seen in patients with moderate to severe hepatic impairment. Paclitaxel is not recommended for patients with severely impaired hepatic function. No data are available for patients with severe baseline cholestasis. • Particular care should be taken to avoid the intra-arterial administration of paclitaxel, as animal studies indicated that severe tissue reactions occurred subsequent to intra-arterial administration. • Severe cardiac abnormalities have been reported rarely. If significant abnormalities develop during paclitaxel administration appropriate therapy should be instated and continuous cardiac monitoring instated during subsequent therapy. Hypotension, hypertension and bradycardia have been observed during paclitaxel administration but such patients are usually asymptomatic and do not require treatment. Frequent monitoring of vital signs, especially during the first hour of infusion is recommended. • While peripheral neuropathy occurs frequently, the development of severe symptoms is unusual. In such severe cases a dose reduction of 20% is recommended for all subsequent courses of paclitaxel.

Adverse events • Haematological toxicity in the form of neutropenia and leucopenia is a dose- limiting adverse occurrence. • Other adverse events include peripheral neuropathy, alopecia, gastrointestinal disturbance, myalgia, and arthralgia. • Hypersensitivity reactions characterised by dyspnoea and hypotension requiring treatment, angiodema and generalised urticaria occur, although these can be reduced in frequency by the use of antihistamine and corticosteroid- based premedication. Where appropriate premedication has been administered <1% of patients will experience such events. Where severe hypersensitivity reactions occur paclitaxel therapy should be discontinued immediately and symptomatic therapy initiated. The patient should not be re-challenged with paclitaxel.

• Current service cost

34 Topotecan The average cost of the minimum number of four cycles is £6250 per patient, rising to £9250 for the mean number of cycles actually received (5.92 in a recent trial). These figures assume that only 2.7mg of each 4mg vial is used, as is the case for a patient with average body surface area.

PLDH The cost per cycle of therapy for a typical patient is £1627, with the overall cost depending on the number of cycles undertaken.

Paclitaxel Treatment is normally undertaken on an outpatient basis and drug costs are approximately £1100 per cycle, giving a typical total cost of £6600 per patient. This figure excludes the cost of the platinum drugs, pre-medication, cost of treating adverse events associated with the drug, wider hospital treatment and VAT.

Comparator/alternative technologies

For patients who require second-line therapy, NICE guidance advises the use of platinum-based therapy except in patients who are initially refractory to this regimen or those who become resistant to platinum-based chemotherapy. The guidance also advises that therapy including paclitaxel should be considered except where this has previously failed. Given that guidance for first line therapy currently recommends the use of platinum and paclitaxel combination therapy, a majority of patients needing re-treatment following relapse will now require alternative second line therapy. Guidance currently recommends that both PLDH and topotecan be considered as therapy in such cases. Chemotherapeutic agents currently licensed for the treatment of advanced ovarian cancer are treosulfan, hexamethylmelamin, cisplatin, carboplatin, epirubicin, cyclophoshamide and methotrexate.

35 3. Methods for literature review of effectiveness and cost-effectiveness 3.1. Search strategy

The literature searches for this systematic review aimed to update the searches of previous NICE technology appraisal reviews on the use of topotecan, PLDH and paclitaxel for the treatment of ovarian cancer.

Date limits were applied to the searches according to the date the literature searches for the previous review had been carried out:

• Topotecan - Previous searches conducted August-September 2000 - Current searches limited to publication years 2000-2004

• PLDH - Previous searches conducted May-June 2001 - Current searches limited to publication years 2001-2004

• Paclitaxel - Previous searches conducted October 2001 - Current searches limited to publication years 2001-2004

The searches aimed to retrieve both published references and ongoing studies. Ongoing research studies were limited to research that had been started after 2000.

The literature searches were carried out in April 2004. There were no limits applied by study design. The following databases were searched to identify studies of topotecan, PLDH and paclitaxel:

36 Published Studies

MEDLINE PREMEDLINE EMBASE CINAHL Database of Abstracts of Reviews of Effects (DARE) NHS Economic Evaluation database (NHS EED) Health Technology Assessment (HTA) database Cochrane Controlled Trials Register Cochrane Database of Systematic Reviews Science Citation Index Index to Scientific and Technical Proceedings BIOSIS Office of Health Economics Health Economic Evaluations Database (OHE HEED)

Ongoing Studies

National Research Register Controlled Trials.com CancerPortfolio.org Clinical Trials.gov

The literature searches retrieved 2,542 unique references after de-duplication. All references were managed using Endnote software version 6. The full details of the search strategies are given in Appendix 3. In addition a further systematic search was undertaken to identify published economic evaluations relating to the use of peglylated liposomal doxorubicin hydrocholoride (PLDH), toptecan and paclitaxel, for second-line therapy for relapsed advanced ovarian cancer. The search strategies used to identify studies can be seen in Appendix 3.

37 3.2 Inclusion and exclusion criteria

Two reviewers independently screened all titles and abstracts. Full paper manuscripts of any titles/abstracts that were considered relevant by either reviewer were obtained where possible. The relevance of each study was assessed according to the criteria set out below. Studies that did not meet all the criteria were excluded and their bibliographic details listed with reasons for exclusion in Appendix 4. Any discrepancies were resolved by consensus and if necessary a third reviewer was consulted. a) Interventions

This review covers the effectiveness of the following three alternative chemotherapeutic agents, used within their respective licensed indications:

Intravenous topotecan monotherapy (Hycamtin®, GlaxoSmithKline)

Pegylated liposomal doxorubicin hydrochloride monotherapy (Caelyx®, Schering-Plough, UK; Doxil®, Alza Corporation, USA)

Paclitaxel, used alone or in combination with a platinum-based compound (Taxol®, Bristol-Myers Squibb; Paxene®, Mayne Pharma).

The interventions were assessed only as second-line or salvage therapy. Second-line therapy was defined as the second chemotherapy regimen administered either as a result of relapse after first-line platinum-based therapy in platinum-sensitive patients or immediately following on from first-line therapy in patients with platinum insensitive disease. Salvage therapy was defined as any therapy given in hope of getting a response when the ‘standard’ therapies had failed. This may overlap with ‘second-line’ therapy, but could also include therapy given for patients with refractory disease that did not respond to first line therapy. Studies that assessed two different regimens of the same agent, either in terms of dose, cycle length, period of treatment, route of administration or combination were also included in the review. Studies in which the chemotherapeutic agent was administered as ‘maintenance’ therapy following directly on from first-line therapy without evidence of disease progression were excluded.

38 b) Comparators

The comparators that were considered included both usual supportive care or any of the following chemotherapeutic agents: cisplatin, carboplatin, treosulfan, epirubicin, hexamethylmelamin, cyclophosphamide or methotrexate. Agents that are licensed for use in the treatment of other cancers but that are currently used off-licence for the treatment of advanced ovarian cancer were only included when they were directly compared to either topotecan, paclitaxel or pegylated liposomal doxorubicin hydrochloride. c) Participants

Women with resistant or refractory advanced ovarian cancer who had experienced a progression of disease. This could be after the completion of first-line platinum based therapy, or whilst still receiving first-line therapy. The different disease sub-types were defined, according to the treatment-free interval between the end of first line chemotherapy and the start of second line therapy due to disease progression. The three different stratifications were defined as:

• Refractory disease: patients who do not respond or whose disease progresses on first line platinum-based chemotherapy.

• Resistant disease: patients who respond to first line platinum-based chemotherapy but relapse within 6 months.

• Platinum-sensitive disease: patients who respond to first line platinum-based chemotherapy but relapse after 6 months. d) Study design

Randomised controlled trials that compared topotecan monotherapy, pegylated liposomal doxorubicin hydrochloride monotherapy, or paclitaxel alone or in combination with a platinum based compound with any other second line treatment including best supportive care.

For the assessment of cost effectiveness a broader range of studies were considered including economic evaluations conducted alongside trials, modelling studies and analysis of administrative databases. Only full economic evaluations that compared two or more options and considered both costs and consequences (including cost- effectiveness, cost-utility and cost-benefit analysis) were included.

39 e) Outcomes

Data on the following outcomes were included:

- Overall survival.

- Progression-free survival.

- Response (including complete and partial response).

- Quality of life.

- Adverse effects of treatment, (haematological toxicity) including neutropenia, thrombocytopenia, leucopenia and anaemia, and non-haematological toxicity including palmar-planter erythrodysesthesia (PPE), nausea, vomiting, diarrhoea, constipation, stomatitis, abdominal pain, mucositis, rash, fatigue, asthenia, alopecia, anorexia, and any other adverse events judged to be appropriate, such as infusion associated reactions.

- Costs from all reported perspectives.

Where the evidence allowed, the use of the interventions in the sub-group of women with relapsed ovarian cancer that is potentially platinum-sensitive was considered separately from that of women who are platinum-resistant or refractory. f) Publication

A full English language paper copy or trial report of the study had to be available for it to be included in the review. Studies which were reported in abstract form only, and where no further information was available, were excluded. Foreign language papers were also excluded.

3.3 Data extraction strategy

Data relating to both study design and quality were extracted by one reviewer and independently checked for accuracy by a second. Disagreements were resolved through consensus and if necessary a third reviewer was consulted. Where multiple publications of the same study were identified, data were extracted and reported as a single study.

40 3.4 Quality assessment strategy

The quality of the individual studies was assessed by one reviewer and independently checked for agreement by a second. Disagreements were resolved through consensus, and if necessary a third reviewer was consulted. The quality of the clinical effectiveness studies was assessed according to criteria based on CRD Report No. 4.(20) The quality of the cost-effectiveness studies was assessed according to a checklist updated from that developed by Drummond et al.(21) This checklist reflects the criteria for economic evaluation detailed in the methodological guidance developed by the National Institute for Clinical Excellence. Full details of the quality assessment strategy are reported in Appendix 5.

3.5 Methods of analysis/synthesis 3.5.1 Clinical effectiveness Details of the extracted data and quality assessment for each individual study of clinical effectiveness were presented in structured tables and as a narrative description. The possible effects of study quality on the effectiveness data and review findings were discussed. Data were reported separately for each outcome measure. Where sufficient data were available, treatment effects were presented in the form of relative risks (RR) or hazard ratios as appropriate. Ideally survival data were presented as hazard ratios or median times based on Kaplan-Meier survival curves. However, this was not always possible due to a lack of appropriate data. Where data were not available risk ratios and p values were presented.

Where relative risk estimates were not presented in the original trial report they have been calculated if sufficient data were available. In some cases the data were also presented in the form of Forest plots, but without pooled estimates.

Due to the heterogeneity between the studies in terms of comparators, statistical pooling was not performed. Consequently, statistical χ2 tests of heterogeneity were not performed. The small number of studies also prevented the assessment of publication bias using funnel plots or the Egger test. However, the risk is likely to be low, considering the attempts to locate unpublished data and the fact that unpublished studies in the form of industry submissions were included in the review.

3.5.2 Cost Effectiveness For the cost-effectiveness section of the report, details of each identified published economic evaluation, together with a critical appraisal of its quality were presented in structured tables. This included studies based on patient-level data and decision models and included any studies provided by the manufacturers.

For analysis based on patient-level data, the validity of the studies was assessed for the sources of resource use and effectiveness data, the valuation methods used to cost the resource use and value patient benefits, the methods of analysis and generalisability of results. For analysis based on decision models, the critical appraisal was based on a range of questions including: i Structure of model. ii Time horizon. iii Details of key input parameters and their sources. iv Methods of analysis (e.g. handling uncertainty).

3.5.3 Handling the company submissions All the clinical effectiveness data included in the three company submissions were assessed. Where these met the inclusion criteria, they were included in the clinical effectiveness review. All economic evaluations (including accompanying models) included in the company submissions were critically appraised using the same checklist applied to the published studies. In addition, a detailed analysis was undertaken to explore the cost-effectiveness more fully. This included a rigorous assessment of the assumptions underlying the submitted economic analyses and additional sensitivity analysis to determine the robustness of these findings. Following this analysis, a new model was developed to estimate costs from the perspective of the UK NHS, and health outcomes in terms of life-years and quality-adjusted life-years (QALYs) for the full range of relevant treatment strategies. The impact of patient heterogeneity (e.g. platinum sensitive and platinum resistant/refractory patients) on the cost-effectiveness results was explored in a series of separate analyses.

42 4. RESULTS 4.1.1. Quantity and quality of research available A total of 2,542 titles and abstracts were screened for inclusion in the review of clinical and cost effectiveness. Of the titles and abstracts screened, 194 studies were ordered as full papers and assessed in detail. Five studies were not received or were unavailable at the time of the assessment. The process of study selection in shown in Figure 1.

Figure 1. Process of study selection for clinical effectives and cost-effectiveness

SEARCH: Clinical effectives and cost-effectiveness

Excluded N=170 Titles and abstracts Background or identified and screened commentary N=2542 N=92 Unavailable/not Non-systematic received review N=4 N=9 First-line therapy Full copies retrieved and N=12 inspected Wrong comparator N=189 N=12 Research, not an RCT N=32 Abstract only Studies meeting inclusion criteria N=13 N=16(reported in 19 full publications)

Related systematic reviews Cost-effectiveness studies N=4 N=4

Industry submission N=1

RCTs of clinical effectiveness meeting inclusion criteria N=9

For the assessment of the clinical effectiveness of topotecan monotherapy, pegylated liposomal doxorubicin hydrochloride monotherapy and paclitaxel administered alone

43 or in combination with a platinum based agent for the second-line or subsequent treatment of advanced ovarian cancer, nine RCTs were identified.

Five of these nine trials assessed comparators that were both used within their licensed indications. These trials were:

PLDH versus topotecan for the treatment of recurrent epithelial ovarian carcinomia.(22, 23) Topotecan versus paclitaxel for the treatment of recurrent epithelial ovarian cancer.(24-26) Pegylated liposomal doxorubicin hydrocholoride versus paclitaxel for the treatment of epithelial ovarian carcinoma.(27) Single–agent paclitaxel versus cyclophosphamide, doxorubicin, and cisplatin in patients with recurrent ovarian cancer who responded to first-line platinum- based regimen.(28) Paclitaxel plus platinum-based chemotherapy versus conventional platinum- based chemotherapy in women with relapsed ovarian cancer.(29)

In the four further trials identified, one of the comparators under investigation was used outside the licensed parameters either in terms of indication, dosage, route of administration or length of the chemotherapy cycle. These trials were:

Oxaliplatin versus paclitaxel in patients with platinum pretreated advanced ovarian cancer.(30) Single-agent paclitaxel given weekly versus every three weeks (and with per oral versus intravenous steroid premedication) for patients with ovarian cancer previously treated with platinum.(31) Paclitaxel at two dose levels, the higher dose accompanied by filgrastim at two dose levels, in platinum-pretreated epithelial ovarian cancer.(32) Oral versus intravenous topotecan in patients with relapsed epithelial ovarian cancer.(33)

44 A full summary of the nine included RCTs is presented in Table 1 and full data extraction tables are presented in Appendix 6. In addition a summary of the comparators included in the review is presented in Table 2.

45 Table 1. Summary of included RCTs Study Study design Participants Intervention Trial 30-49 Phase III, multi-centre, 474 participants with disease that recurred after or failed PLDH (50mg/m2/day every 28 days) versus. (Schering-Plough) stratified open-label RCT first-line platinum-based chemotherapy. First-line therapy topotecan (1.5mg/m2/day for 5 days every could include a platinum-taxane combination. 21 days) Gordon et al. (2001)(22) Schering-Plough (2004)(23)

Trial 039 Phase III, multi-centre, 235 participants with disease that recurred after or who Topotecan (1.5mg/m2/day for 5 days every (GlaxoSmithKline) stratified open-label RCT failed one platinum based first-line regimen. All 21 days) versus paclitaxel (175mg/m2/day participants were taxane naïve. every 21 days) ten Bokkel Huinink et al. (1997)(24) ten Bokkel Huinink et al. (2004)(25) Gore et al. (2001)(26) Trial 30-57 Phase III, multi-centre, 216 participants with disease that recurred after or who PLDH (50mg/m2/day every 28 days) versus (Schering-Plough) stratified open-label RCT. failed one platinum based first-line regimen. All paclitaxel (175mg/m2/day every 21 days) Study terminated after participants were taxane naïve. Johnson & Johnson approx 50% of the Pharmaceutical Research and planned participants had Development (2004)(27) been entered due to poor participant accrual rates.

Cantu et al. (2002)(28) Phase II, randomised pilot 97 participants with disease that recurred or progressed Paclitaxel (175 mg/m2 study after 12 months had elapsed since the end of first-line every 21 days) versus cyclophosphamide platinum based therapy. All participants were taxane naïve, (500 mg/m2), doxorubicin (50 mg/m2) and and had received 2 or 3 prior chemotherapy regimens. cisplatin (50 mg/m2) (CAP) The ICON and AGO Phase III, parallel multi- 802 participants with disease that recurred after 6 months Paclitaxel in combination with carboplatin Collaborators (2003)(29) centre RCT had elapsed since the end of first-line platinum based or cisplatin versus carboplatin or cisplatin therapy. Participants had received 1or more prior monotherapy chemotherapy lines that could have included a platinum- taxane combination.

Piccart al al. (2000)(30) Phase II, multi-centre 86 participants with disease that had recurred or progressed Paclitaxel (175 mg/m2 stratified, open-label trial within 12 months of first-line platinum based therapy. every 21 days) versus oxaliplatin (130 Participants had received either 1or 2 prior chemotherapy mg/m2 every 21 days). lines. All participants were taxane naïve. (oxaliplatin is not licensed for use in advanced ovarian cancer) Rosenberg et al. (2002)(31) Multi-centre, bi-factorial, 208 participants with disease that recurred or progressed Paclitaxel weekly (67/mg2/week) versus stratified RCT. after first-line platinum based therapy. Participants were all paclitaxel every 3 weeks (200 mg/m2 every taxane naïve and had received only 1 prior chemotherapy 21 days) line. Participants in both groups were also randomised to either oral steroids 12 and 6 hours before paclitaxel or parenteral steroids 30 minutes before paclitaxel.

(paclitaxel is licensed for use on a 3 week cycle) Omura et al. (2003)(32) Phase III, multi-centre, 449 participants with disease that recurred after or who Low dose paclitaxel (135 mg/m2 every 21 stratified RCT. failed one platinum-based first-line therapy. All days) versus medium dose paclitaxel (175 participants were taxane naïve. mg/m2 every 21 days) versus high dose paclitaxel (250 mg/m2 every 21 days).

(paclitaxel is licensed for use at a dosage of 175mg/m2) Gore et al. (2002)(33) Multi-centre, stratified, 266 participants with disease that recurred after or who i.v. topotecan open-label RCT. failed one platinum based first-line regimen. First-line (1.5mg/m2/day for 5 days every 21 days) therapy could include a platinum-taxane combination. versus oral topotecan (2.3 mg/ m2/day for 5 days every 21 days)

(topotecan is licensed for use as an i.v. infustion)

47 Table 2. Summary of the comparators included in the review of clinical effectiveness

Comparators included in the review Trial Paclitaxel Topotecan PLDH Paclitaxel Platinum CAP Oxaliplatin Oral combination topotecan (i.v)

Overall patient population Trial 30-49(22) 9 9 Trial 039(24) 9 9 Trial 30-57(27) 9 9 Platinum sensitive patients Cantu et al.(28) 9 9 ICON 4/AGO-OVAR 2.2 (29) 9 9 Unlicensed comparator Piccart et al.(30) 9 9 Rosenberg et al.(31) 99 Omura et al.(32) 99 Gore et al.(33) 9 9

48 Relevant studies reported in abstract form only. In addition to the nine included trials for which there was either a full publication or a trial report available, a further four RCTs were identified that were reported in abstract form only. No further details of the studies were obtainable from the trialists and therefore the trials were excluded from inclusion in the review. The interventions that were assessed in these trials were:- Topotecan versus oxaliplatin; Vermorken et al. (2001)(34) Paclitaxel in combination with doxorubicin versus paclitaxel alone; Torri et al. (2000)(35) Topotecan versus treosulfan; Loibi et al.(2003)(36) Paclitaxel in combination with carboplatin versus carboplatin alone; Gonzalez Martin et al. (2003)(37). Full copies of the abstracts are reproduced in Appendix 7.

Systematic reviews/meta-analysis In addition to the primary studies, we identified 4 systematic reviews and/or meta- analyses that included evaluations of PLDH, topotecan and / or paclitaxel for second- line or subsequent therapy in advanced ovarian cancer. These included the two previous Health Technology Assessment reports that this present review up-dates. (38, 39). One of the previous reports was on the use of PLDH(38) and the other on the use of topotecan.(39) Of the other two identified reviews, one investigated topotecan(40 ) and the other assessed all three comparators under consideration in the present report.(41)

Excluded studies A total of 164 studies were excluded. Of these, 23 were used as background articles for the review. The majority of the other excluded studies were non-randomised studies, contained the wrong comparators or were in first-line therapy. A full list of the excluded studies with the reasons for exclusion is presented in Appendix 4.

4.1.2 Description of included studies

The following section of the report provides a summary of the nine included RCTs. For each included study a summary of the trial has been provided followed by a description of the trial quality. The trials have been grouped, firstly according to whether both the comparators that were evaluated within the trial were used within their licensed indications and, secondly, by whether the trial assessed both participants with platinum-sensitive and platinum-resistant disease, or platinum-sensitive participants only.

4.1.2.1. PLDH versus topotecan (trial 30-49) One randomised controlled trial was identified which compared the efficacy and safety of pegylated liposomal doxorubicin hydrochloride (PLDH) to topotecan. The trial had been reported in two abstracts(42, 43) and a full publication.(22) In addition a further three reports were obtained from Schering-Plough as part of the industry submission.(23, 44, 45)

Description of the trial comparing PLDH and topotecan Trial 30-49 by Gordon et al(22) evaluated the efficacy and safety of PLDH compared to topotecan in women with epithelial ovarian carcinoma whose disease did not respond to, or recurred after, treatment with first line platinum-based chemotherapy. Entry into the trial was based on (1) histologically proven epithelial ovarian carcinoma, (2) measurable or measurable and evaluable disease, (3) recurrence of disease or disease progression indicative of failure of first-line, platinum-based chemotherapy (4) KPS≥60%. Eligible participants underwent radiological imaging (X-ray, computed tomography [CT] scan, magnetic resonance imaging [MRI]) to document baseline disease within 30 days prior to the first dose of study drug. Participants were randomised to treatment in a 1:1 ratio stratified by platinum sensitivity and the presence or absence of bulky disease (tumour mass >5 cm). Platinum-sensitive disease was defined as a response to initial platinum-based therapy followed by a progression-free interval >6 months. Platinum-refractory disease was defined as progression or stable disease during the initial platinum-based therapy or relapse within 6 months after completion of therapy. The study included 481 participants, 239 received a 1-hour i.v. infusion of pegylated liposomal doxorubicin, 50mg/m2 every 28 days and 235 who received topotecan, 1.5mg/ m2 per day as a 30- minute i.v. infusion for 5 consecutive days, repeated every 21 days. The majority of participants in both treatment arms received 4-5 chemotherapy cycles. At baseline the two treatment groups were well matched in terms of treatment-free interval, disease bulk, the number of previous chemotherapy regimens, the type of previous chemotherapy agents received, histology and performance status.

The trial was designed as a non-inferiority study with respect to time to progression. The efficacy evaluation of disease status was assessed by radiological imaging every 8 weeks, with participants who achieved a complete or partial response having their radiological imaging repeated at least 4 weeks later to confirm the initial observation of response. Haematological toxicity status was assessed on a weekly basis. In the long-term follow-up of participants reported by Schering-Plough(44) in which the analysis was conducted when 87% of patients had died and 13% of the observations were censored (patients either alive or lost to follow-up), the primary efficacy outcome was overall survival.

Quality of the trial comparing PLDH and topotecan Trial 30-49 was a reasonably good quality randomised open-label comparative trial. The only slight issue for concern in the trial was the failure to conduct a true intention to treat analysis. The analysis of the trial was based on the data from participants who received at least a partial dose of study drug. However, as this was a non-inferiority trial the analysis presented may be more conservative than a true ITT analysis. Also, a further submission by Schering-Plough(23) as part of the industry submission, did include a true ITT analysis for the primary efficacy outcome of overall survival. The results of this analysis indicated that there was no significant difference between the results for this outcome when the data was analysed on an ITT basis. The evaluation of trial 30-49 in relation to study quality is shown in Table 3. Full details of the quality checklist are available in Appendix 5.

4.1.2.2 Topotecan versus paclitaxel (trial 039) One randomised controlled trial was identified which compared the efficacy and safety of topotecan versus paclitaxel. The trial had been reported in one abstract(46) and three full publications.(24, 26) The first publication by ten Bokkel Huinink et al. (1997)(24) was a full trial publication, that reported on all the efficacy and safety parameters evaluated in the trial. The later report by ten Bokkel Huinink et al (2004)(25) reported the overall survival rates at longer term follow-up of the participants. A further publication by Gore et al (2001)(26) reported the results from third-line therapy, when participants received cross over therapy with the alternative chemotherapy regimen.

Description of the trial comparing topotecan and paclitaxel The aim of the trial was to compare the efficacy and toxicity of topotecan and paclitaxel in patients with advanced epithelial ovarian cancer who had progressed either whilst on first-line platinum-based therapy or after one cycle of platinum-based therapy.

The eligibility criteria for entry into the trial were based upon a histologically proven diagnosis of measurable disease and a European Co-operative Oncology Group (ECOG) performance status ≤2. All participants had previously received one platinum-based chemotherapy regimen without a taxane.

Two hundred and thirty five eligible participants were randomised into the trial, with stratification based upon age, ascites, and response to prior platinum chemotherapy (resistant, early, interim or late relapse). Resistant disease was defined as not having a response to initial chemotherapy or having an initial response but then progressing while still on therapy. Relapsed participants were those who had an initial response (complete or partial) and then relapsed within 3 months (early), 3 to 6 months (interim), or more than 6 months after chemotherapy was stopped (late). One hundred and seventeen participants were randomised to receive 1.5 mg/m2 topotecan as a 30- minute infusion on 5 consecutive days every 21 days and 118 to receive 175 mg/m2 paclitaxel over 3 hours every 21 days. Nine participants did not receive any treatment (five randomised to topotecan and four to paclitaxel) and were not included in the analysis. A further 24 participants (16 in the topotecan arm and 8 in the paclitaxel arm) were not evaluated for response, although these participants were included in the denominator when response rates were calculated. The most common reason that participants were not evaluated was due to adverse events. On the topotecan arm, two participants had no measurable disease, and two did not have all their lesions undergo follow-up evaluation; therefore, no evaluation of response was possible. In addition, two of the topotecan group participants had evidence of response, but the response was not allowed because a confirming scan was not performed. There were no paclitaxel group participants not evaluated for these reasons. Duration of treatment was dependent on response. The median number of treatment cycles in both the treatment arms was 5, range 1-17 in the topotecan group and 1-12 in the paclitaxel

52 group. Participants whose best response was stable disease after six courses of one regimen could be removed from the study or switched to the alternate treatment regimen.

At baseline the two treatment arms were well matched in terms of age, performance status, tumour diameter, tumour histology, histologic grade, weight and platinum sensitivity. Fifty-four percent of topotecan treated participants and 52% of paclitaxel treated participants had not responded or had relapsed within 6 months of completing first-line therapy.

Tumours were evaluated by computed tomography (CT), magnetic resonance imaging (MRI) scan, ultrasound or physical examination at baseline. All claimed responses were subject to independent radiological review. However, non-responses were not subject to independent review.

The primary efficacy parameters were response rate, duration of response and time to progression. The secondary criteria for efficacy were time to response and survival. Time to response, time to progression and survival were all measured from the time of the first dose of study drug. Response duration was measured from the time of the first documented complete or partial response to the first sign of disease progression. Health related Quality of Life (QoL) was also assessed in the trial using the European Organization for Research and Treatment of Cancer Quality of Life questionnaire (EORTC ) QLQ-C30 questionnaire.

A total of 110 participants received cross-over therapy as third-line treatment. Sixty- one participants crossed-over from paclitaxel to topotecan, and 49 participants crossed-over from topotecan to paclitaxel. The results from third-line therapy are reported separately.

Quality of the trial comparing topotecan and paclitaxel This trial was a good quality randomised open-label comparative trial. The only issue for concern in relation to quality was the failure to conduct an intention to treat analysis, as only participants who received at least one dose of study drug were actually included in the final analysis. The overall survival data reported in the trial

53 are also somewhat difficult to interpret due to the switch to alternate therapy allowed in the design of the study. An evaluation of the trial in relation to study quality is shown in Table 3.

4.1.2.3 Pegylated liposomal doxorubicin hydrochloride versus paclitaxel One randomised controlled trial was included that compared the efficacy and safety of PLDH with paclitaxel. The trial had been reported in one abstract(47) and in addition a full trial report was obtained from Schering-Plough as part of the industry submission.(27)

Description of the trial comparing PLDH and paclitaxel The trial by Schering-Plough(27) was a phase III, randomised, open-label, comparative trial that evaluated the efficacy and safety of PLDH compared to paclitaxel in women with epithelial ovarian carcinoma following failure of first-line platinum based chemotherapy. Protocol-eligible participants were women with measurable disease, who had received one prior platinum-based regimen without a taxane. The planned enrolment was for 438 participants, but only 216 were ultimately randomised (108 in each treatment arm).

Participants entering the trial were stratified by platinum-sensitivity and bulky disease and randomised in a 1:1 ratio within each stratum to receive either a 1-hour i.v. infusion of PLDH 50mg/m2 every 28 days or paclitaxel 175 mg/m2 as a 3-hour infusion every 21 days. All randomised participants underwent radiological imaging (X-ray, computed tomography [CT] scan, magnetic resonance imaging [MRI]) to document baseline disease within 30 days prior to the first dose of study drug. Participants were followed weekly to assess haematological toxicities, and radiologic imaging was repeated every 7-8 weeks to assess disease status. Participants who achieved a complete or partial response were re-evaluated 4 weeks later to confirm the initial observation of response.

At baseline the two treatment groups were well matched in terms of age, treatment- free interval, disease bulk, the number of previous chemotherapy regimens, the type of previous chemotherapy agents received, histology and performance status. All participants were to have been followed for a minimum of 1 year for survival and

54 disease progression. However, the study was terminated after approximately 50% of the planned participants had been entered into the trial due poor participant accrual. This was related to Taxol® being approved for use in combination with platinum- based therapy for the first-line treatment of ovarian cancer by the European Agency for the Evaluation of Medicinal Products.

The efficacy and safety analysis was therefore limited to overall survival only, and an examination of adverse events.

Quality of the trial comparing PLDH and paclitaxel The study was a reasonably good quality randomised open-label comparative trial. However, due to its early termination only 216 participants were randomised, 108 into each treatment arm. The planned enrolment had been for 438 participants. It is therefore likely that the trial is significantly under powered to detect any difference in treatment effect between the two treatment groups. Furthermore, due to its early termination the results of the trial are likely to be preliminary and the longer term implications of any differences observed in the treatment effect at the time of data analysis are unclear. Due to these caveats, the results of the trial should be interpreted with caution. The evaluation of trial 30-57 in relation to study quality is shown in Table 3. Full details of the quality checklist are available in Appendix 5.

4.1.2.4 Paclitaxel versus combination cyclophosphamide, doxorubicin and cisplatin (Cantu et al.) One publication of a single RCT that compared single-agent paclitaxel and a platinum containing regimen in previously treated patients with recurrent ovarian cancer was identified.(28)

Description of the trial comparing paclitaxel and CAP The aim of the randomised pilot study by Cantu et al(28) was to assess the activity, efficacy and tolerability of single-agent paclitaxel and a platinum-containing regimen in previously treated patients with recurrent ovarian cancer. The eligibility criteria for entry into the trial were based upon a histologically proven diagnosis of measurable disease, that recurred or progressed after 12 months had elapsed since the end of first- line therapy that had included cisplatin or carboplatin, and a WHO performance status

55 ≤ 2. Ninety-seven eligible participants were randomised to either single-agent paclitaxel 175 mg/m2 infused over 3 hours or a combination of cyclophosphamide 500 mg/m2 , doxorubicin 50 mg/m2, and cisplatin 50 mg/m2 (CAP). Both of these regimens were administered intravenously (i.v) at 3-week intervals. The median number of cycles on each arm was 6.

Forty-seven participants were randomised to the CAP regimen and 50 were allocated paclitaxel. However, 3 participants in the paclitaxel arm were lost to follow-up just after randomisation. The analysis therefore is based upon 94 participants (47 in each group). At baseline the two treatment groups were reasonably well matched in terms of age, histology, disease site, prior chemotherapy regimens and treatment-free interval. There were however more participants with bulky disease at the time of first diagnosis in the paclitaxel treatment group (57%) than on the CAP arm (34%). All the participants had previously been treated with two or three chemotherapy regimens that did not include a taxane. The median treatment-free interval was 30.2 months in the paclitaxel arm and 38.8 months among participants receiving CAP. Although it was not prespecified in the trial protocol, participants who failed to respond to either CAP or paclitaxel as second-line therapy were crossed-over to received the alternative regimen as third-line therapy. In total twenty-three participants who did not repond to CAP were crossed-over to the paclitaxel regimen, and thirty were crossed-over from paclitaxel to CAP.

The primary outcome of the trial was overall survival. The secondary objectives were to determine progression-free intervals and the overall response to therapy.

Quality of the trial comparing paclitaxel and CAP An evaluation of the Cantu et al(28) trial in relation to study quality is displayed in Table 3. Overall, due to the level of reporting in the trial publication it was not possible to evaluate a number of the methodological aspects of the trial. The authors did state that the trial was a randomised pilot study, but the method of randomisation was not reported. It can be suggested that due to the imbalance between the two treatment arms at base line on one important prognostic factor, that a simple randomisation without stratification had been used. However, as this was not reported it was also impossible to assess the degree to which allocation concealment had been

56 attained. It was also not possible to assess whether the outcome assessors were blinded to treatment allocation. The trial was also not analysed on an ITT basis as reported in the paper, as the three participants lost to follow-up were not included in the final analyses. Furthermore, the addition of cross-over between the two regimens at third-line therapy makes the analysis of overall survival data somewhat difficult. Full details of the quality checklist are available in Appendix 5.

4.1.2.5 Paclitaxel / platinum combination therapy versus platinum therapy alone (The ICON and AGO Collaborators)

One randomised controlled trial was included which investigated the use of paclitaxel in combination with platinum-based chemotherapy in patients with epithelial ovarian carcinoma who had relapsed after 6 months, following first-line platinum-based chemotherapy. The trial had been reported in one abstract(48) and a full publication.(29)

Description of the trial comparing paclitaxel / platinum combination to platinum alone. The ICON4/ and AGO-OVAR 2.2 trial(29) was a multi-centre trial which was run in parallel, with ICON4 being co-ordinated by the Istituto Mario Negri, Italy (IRFMN) and the Medical Research Council’s Clinical Trials Unit, UK (MRC-CTU) and AGO- OVAR-2.2 being co-ordinated by AGO, Germany. The trial compared the efficacy and safety of paclitaxel in combination with platinum-based chemotherapy to platinum-based chemotherapy alone in women with epithelial ovarian carcinoma who had relapsed after 6 months, following first line platinum-based chemotherapy. The eligibility criteria for the trial differed slightly across the three protocols. Patients in the MRC-CTU protocol were permitted to have received more than one line of previous chemotherapy; participants in the IRFMN were required to have been treatment-free for a period >12 months and to have measurable disease; participants randomised into the AGO protocol must have previously received paclitaxel in combination with platinum-based therapy (cisplatin or carboplatin), whilst those randomised into the ICON4 (IRFMN and MRC-CTU) protocol were required to have received platinum-based chemotherapy, with or without paclitaxel. Eligible participants were randomised to treatment in a 1:1 ratio. In the ICON4 protocol stratification was done by centre, age, last chemotherapy received, time since

57 completion of last chemotherapy and intended platinum treatment. In the AGO protocol, stratification was based upon time since completion of last chemotherapy, and whether the participant had undergone secondary debulking surgery. In total 802 participants were randomised, 392 were assigned paclitaxel plus platinum-based chemotherapy and 410 were assigned conventional platinum-based therapy alone. When the planned treatment doses for carboplatin were determined by the area- under-curve method of Calvert and colleagues(49) the minimum dose was 5(GFR+25)mg, and when it was determined by the Cockcroft formula, the dose was a minimum of 6(GFR+25) mg. The planned minimum dose of cisplatin (ICON4 protocol only) was 75mg/m2 if given as one agent and 50mg/m2 if given in combination with other drugs. Participants in the ICON4 protocol assigned paclitaxel plus platinum in combination were to receive 175mg/m2 paclitaxel given as a 3-hour i.v. infusion, followed by either carboplatin or cisplatin at the same dose as set out above. Participants in the AGO protocol assigned paclitaxel plus carboplatin were to receive 185 mg/m2 paclitaxel given in a 3-hour infusion, followed by carboplatin at the same dose as set out above. All drugs were to be given on day 1 of each cycle.

Treatment data were collected at each cycle of treatment, with further follow-up data collected every 3 months for the first two years, and 6 monthly from then on. Quality of life data were collected for all patients in the MRC-CTU and AGO protocols, using the European Organization for Research and Treatment of Cancer quality of life questionnaire EORTC QLQ-C30. At baseline the two treatment groups were well matched in terms of age, WHO performance status, the type of last chemotherapy regimen received, the number of previous chemotherapy lines and the treatment-free interval.

The primary outcome measure of the trial was overall survival, and the secondary outcomes were progression-free survival and quality of life.

Quality of the trial comparing paclitaxel / platinum combination to platinum alone. ICON4 / AGO-OVAR-2.2 was a good quality randomised, multi-centre, parallel trial. The evaluation of the ICON4 / AGO-OVAR-2.2 trial in relation to study quality is shown in Table 3. Full details of the quality checklist are available in Appendix 5.

Trials in which one of the comparators was used outside its licensed indication.

4.1.2.6 Single agent paclitaxel versus oxaliplatin (Piccart et al) One randomised phase II study by Piccart et al.(30) was identified that investigated the efficacy of single-agent paclitaxel compared to oxaliplatin.

Description of the trial comparing paclitaxel to oxaliplatin The aim of this multi-centre open label, randomised phase II study was to evaluate the efficacy of oxaliplatin compared to single-agent paclitaxel in a relapsing progressive ovarian cancer patient population. Entry into the trial was based upon a histologically proven diagnosis of epithelial ovarian carcinoma in patients who had progressed or stabilised after prior treatment. Participants were required to have received at least 1 but no more than 2 prior platinum-based chemotherapy regimens and to have relapsed within a year of the end of treatment. All participants were also required to have a WHO performance status of 0-2.

Eligible participants were stratified by centre, performance status (0 or 1 versus 2), platinum-free interval (0-6 months versus 6-12 months), and the number of prior platinum-based regimens (1 versus 2). Out of the 86 participants randomised, 41 received paclitaxel 175 mg/m2 administered as a 3-hour i.v infusion every 21 days and 45 received oxaliplatin 130mg/m2 as a 2-hour i.v. infusion every 21 days. The median number of cycles administered per patient was 6 in the paclitaxel arm and 4 in the oxaliplatin arm (range:1-8 in both treatment arms).

Participants underwent complete examination (clinical, gynaecological, histological and haematological) at baseline and then before each treatment cycle. Anti-tumoural activity was assessed by MRI or CT scan every two cycles.

At baseline participants were reasonably balanced between the treatment arms in terms of age, disease stage and number of sites involved, and the number and type of prior chemotherapy regimens. However, an imbalance was found in the histological distribution, with 17 patients (41%) of participants in the paclitaxel arm and 33 (73%) in the oxaliplatin arm having a serous tumour type. 31 participants (76%) in the

59 paclitaxel arm and 32 (71%) in the oxaliplatin arm were classified at baseline as being platinum-resistant.

The primary efficacy endpoint in the trial was the objective confirmed response rate. The secondary efficacy outcomes were time to progression (TTP), time to treatment failure, and overall survival. Quality of life as measured by the EORTC-QLQ-C30 was also assessed as a further outcome.

Quality of trial comparing paclitaxel to oxaliplatin The trial conducted by Piccart et al(30) was a good quality randomised phase II study. The evaluation of the trial in relation to study quality is shown in 3. Further full details of the quality checklist are available in Appendix 5.

4.1.2.7 Paclitaxel weekly versus every three weeks (Rosenberg et al) One report of an RCT by Rosenberg et al(2002)(31) that evaluated single agent paclitaxel given weekly versus every three weeks and with per oral versus intravenous steroid pre-medication met the inclusion criteria.

Description of the trial comparing paclitaxel weekly to every three weeks. The aim of this randomised bifactorial trial was to evaluate the efficacy and toxicity of paclitaxel given at the same dose intensity and administered weekly or every 3 weeks, and to assess the safety of intravenous steroids versus standard per oral premedication. The participants who were eligible to enter the trial needed to have a histologically proven diagnosis, measurable disease, a KPS≥60 and to have had no more than one prior platinum-containing chemotherapy regimen. All the participants were taxane naïve.

Prior to study entry participants were stratified according to platinum sensitivity, before being randomised according to a bifactorial design to receive either paclitaxel 67mg/m2/week over a period of 3 hours (one course was defined as 3 weeks of therapy) or 200 mg/m2 every 3 weeks over a 3 hour period. In addition according to the bi-factorial design participants were also randomised to receive either oral or intravenous steroid premedication. The premedication consisted of oral

60 dexamethasone 20mg administered 12 and 6 hours before paclitaxel, or dexamethasone 20mg given intravenously 30 minutes prior to paclitaxel.

Appropriate radiological examinations were performed at baseline and then every 6 weeks during the treatment period. Haematological evaluation was undertaken on a weekly basis.

At baseline the two treatment groups were well matched and there were no significant differences between the groups in terms of age, WHO performance status, the treatment-free interval, tumour size or paresthesia at inclusion.

The primary efficacy endpoint of interest was the response rate. The secondary efficacy endpoints were response duration, progression-free and overall survival and, in addition, adverse reactions were also studied. The efficacy analysis was undertaken on an ITT basis, and the safety evaluation included all the participants who actually received paclitaxel.

Quality of the trial assessing two different paclitaxel regimens The randomised bifactorial trial by Rosenberg et al(31) was a reasonably good quality trial. However, according to the sample size estimation published in the paper, the trial as it was conducted may not have been adequately powered to detect a difference between the two treatment groups on the primary outcome measure of response rate. It is stated in the report that 318 participants were needed to detect a relative difference ≥ 54% with 80% power. However, only 208 participants were actually enrolled into the trial. Table 3 shows an evaluation of the trial in relation to study quality. Full details of the quality checklist are also available in Appendix 5.

4.1.2.8 Paclitaxel at doses of 175 mg/ m2 and 250 mg/ m2 (Omura et al) One RCT by Omura et al.(32)that evaluated single agent paclitaxel at two different dose levels was identified and included in the review.

Description of the trial comparing paclitaxel at different dose intensities The purpose of the phase III, multi-centre trial by Omura et al(32) was to determine whether increasing the dose of paclitaxel increases the probability of clinical response,

61 progression-free survival or overall survival in women who have persistent or recurrent ovarian cancer, and whether doubling the dose of prophylactic filgrastim accompanying the higher paclitaxel dose decreases the frequency of neutropenic fever. The eligibility criteria for the trial were based upon participants having histologically confirmed epithelial ovarian cancer, a Gynaecologic Oncology Group performance status of 0,1 or 2, and having received no more than one prior platinum-based chemotherapy regimen and no prior taxane. In the original study protocol, participants had to have platinum-resistant clinically measurable disease, defined as progression during first-line therapy or within 6 months of completing therapy, a best response of stable disease after six courses of platinum-therapy, or stable disease with rising CA125 levels whilst on platinum. However, after the study had begun, the eligibility criteria were expanded, due to declining enrolment, to include participants with platinum-sensitive disease and participants without clinically measurable disease. Platinum-sensitive participants were defined as participants who had an initial response to platinum-therapy lasting at least 6 months, followed by progression or recurrence.

At the initiation of the trial the study regimens included paclitaxel administered at 135 mg/m2, 175 mg/ m2 or 250 mg/ m2 by 24-hour i.v. infusion every 21 days. Participants were randomly assigned to treatment, stratified upon clinically measurable disease, platinum-sensitivity and co-operative trial group. Four hundred and forty-nine participants were enrolled onto the study. Out of these, 77 were assigned to the lowest dose regimen before the arm was closed due to poor participant accrual. There were 184 participants assigned to the 175 mg/ m2 dose of paclitaxel without filgrastim, of whom 164 (89%) were deemed eligible, and 188 participants assigned to the 250 mg/ m2 dose of paclitaxel with filgrastim, of whom 166 (88%) were eligible. Within the high dose paclitaxel group, participants were also randomised to receive either 5 or 10 ug/kg filgrastim per day subcutaneously. The reasons for ineligibility in the two treatment groups included inappropriate disease site (n=34), improper prior treatment (n=7), inadequately documented histology (n=3), second primary cancer (n=3), inadequate documentation of recurrence (n=2), borderline tumour histology (n=1), and wrong disease stage (n=1).

62 At baseline the two treatment arms were reasonably well matched in terms of the prognostic factors of age, performance status, measurable disease status and platinum- sensitivity. In the 175 mg/ m2 arm 76% (125/184) of the participants were categorised as having platinum-resistant disease, and 24% (39/184) as being platinum-sensitive. In the 250 mg/ m2 treatment arm 79% (132/188) of participants were platinum- resistant and 21% (34/188) had platinum-sensitive disease. However, there were slightly more participants with mucinous and clear-cell type histology randomised to the higher dose regimen.

Overall in the trial, there were no significant differences between the two groups in the number of treatment cycles administered. Fifty-eight percent of the participants on the paclitaxel 175 mg/ m2 arm and 55% of participants on the 250 mg/ m2 arm received six or more cycles of therapy. Over these first six courses of treatment, approximately 76% and 70% of the planned ideal dose was delivered to participants on the 175 mg/ m2 regimen and 250 mg/ m2 regimen respectively. Although dose reductions occurred in both treatment arms, a difference in total dose and dose- intensity was maintained during the first six courses of treatment.

Response was assessed throughout the trial before every other cycle of therapy. The primary treatment outcomes assessed were progression-free and overall survival. These analyses were based on all participants deemed eligible for inclusion in the trial. All analysis of response rate was based upon participants with measurable disease at baseline, whilst analyses of adverse events were based upon all participants who received any study treatment. No results were reported for the 77 participants who were assigned to the lowest dose paclitaxel regimen of 135 mg/ m2.

Quality of the trial assessing paclitaxel at different dose intensities An evaluation of the Omura et al(32) trial in relation to study quality is displayed in Table 3. Overall, due to the level of reporting in the trial publication it was not possible to evaluate a number of the methodological aspects of the trial. Of particular note in relation to the trial quality there are two important aspects, firstly the sample size upon which the reported results are based and secondly, the related aspect of the protocol amendment to expand the eligibility criteria. The publication states that the planned sample size for the trial was 540 participants, however the reported results are

63 based upon only 265 participants. It is therefore likely that the trial may be significantly underpowered to detect any difference in treatment effect between the two treatment arms for which results are reported. This may be related to the change in the protocol, that allowed for an expansion in the eligibility criteria for participant entry. A number of participants were enrolled into the trial, that were later deemed to be ineligible for participation. These participants were not included in the analysis. The trial results are therefore not based upon an intention to treat analysis, and no description of the number or reasons for withdrawal from the trial are reported.

4.1.2.9 Oral versus intravenous topotecan (Gore et al) One publication of a single RCT that compared oral versus intravenous topotecan in patients with relapsed epithelial ovarian cancer met the inclusion criteria.(33)

Description of the trial comparing oral and intravenous topotecan The aim of this open-label, multi-centre RCT was to compare the efficacy, safety and tolerability of oral topotecan versus standard intravenous topotecan in participants with relapsed epithelial ovarian cancer. The eligibility criteria for entry into the trial were based upon a histologically proven diagnosis of measurable disease, that was originally FIGO stage III or IV, and a European Co-operative Oncology Group (ECOG) performance status ≤2. All participants had progressed on first-line therapy or had relapsed within 12 months of completing initial treatment. Initial treatment must have included a platinum based therapy, which could have been administered in combination with a taxane. Only one previous chemotherapy regimen was allowed.

Two hundred and sixty-six eligible participants were randomised, with stratification based upon response to prior platinum chemotherapy, tumour size, and whether or not the previous regimen had included a taxane. One hundred and thirty-five participants were randomised to receive oral topotecan at a dose of 2.3 mg/m2/day for 5 days and 131 to receive intravenous (i.v) topotecan at a dose of 1.5 mg/m2/day for 5 days. The treatment cycle was 21 days in both arms. The median number of treatment cycles was 4 (range: 1-23) in the oral topotecan treatment group and 6 (range: 1-26) in the i.v. topotecan group.

64 At baseline the two treatment arms were well matched in terms of age, initial FIGO disease stage, performance status, tumour size and platinum sensitivity. In each group 30% of the participants were platinum-refractory and 70% were either platinum- resistant or platinum-sensitive.

Tumours were evaluated by computed tomography (CT), magnetic resonance imaging (MRI) scan, chest X-ray or photography at baseline. Lesions evaluated at baseline by CT or MRI scan were then re-assessed at the end of every second treatment cycle, and those evaluated by chest X-ray or photography were re-assessed at the end of every treatment course throughout the trial. All claimed complete or partial responses were re-assessed at a further four weeks and were subject to independent radiological review. All adverse events were collected throughout the trial.

Time to response, time to progression and survival were all measured from the time of the first dose of topotecan. Response duration was measured from the time of the first documented complete or partial response to the first sign of disease progression. The response rate was also evaluated by serial measurement of CA125 values. Response was defined as a 50% decrease in the two samples, confirmed by a further sample, or a serial decrease over three samples of greater than 75%. The final sample analysed had to be at an interval of at least 28 days after the previous sample.

Quality of the trial assessing oral and intravenous topotecan The trial by Gore et al(33) was a good quality randomised, multi-centre, open-label trial. The trial was also well reported within the publication. An evaluation of the trial of oral versus intravenous topotecan in relation to study quality is shown in Table 3 and full details of the quality assessment checklist are available in Appendix 5.

65 Table 3. Quality checklist for included trials Quality criteria Trial 30-49 Trial 039 Trial 30-57 Cantu et ICON 4 Piccart et Rosenberg Omura et Gore et al. al. al. et al. al. Was the method used to assign participants to the treatment Yes Yes Yes ? Yes Yes Yes Yes Yes groups really random? Was the allocation of treatment concealed? Yes Yes Yes ? Yes Yes Yes ? Yes Were details of baseline comparability presented in terms of Yes Yes Yes Yes Partial Yes Yes Partial Partial treatment-free interval, disease bulk, number of previous regiments, age, histology and performance status? Was baseline comparability achieved for treatment-free Yes Yes Yes Partial Yes Partial Yes Yes Yes interval, disease bulk, number of previous regiments, age, histology and performance status? Were the eligibility criteria for study entry specified? Yes Yes Yes Yes Yes Yes Yes Partial Yes Were any co-interventions identified that may influence the No No No No No No No No No outcomes for each group? Were the outcome assessors blinded to treatment allocation? No Yes No ? ? ? ? ? Yes Were the individuals who administered the intervention No No No No ? No No ? No blinded to the treatment allocation? Were the participants who received the intervention blinded No No No No ? No No No No to the treatment allocation? Was the success of the blinding procedure assessed? NA ? N/A ? ? ? N/A ? ? Were at least 80% of the participants originally included in Yes Yes No2 Yes Yes Yes Yes Partial Yes the randomisation process, followed up in the final analysis? Were the reasons for any withdrawals stated? Yes Yes Yes No Yes Yes Partial No4 Yes Was an intention to treat analysis (ITT) included? Partial1 No Yes No Yes3 Yes Yes No Yes 1 The published study report is not based on a true ITT analysis in which all of the participants randomised to take part in the trial are included in the analysis. The manufacturers submission which includes an analysis of the long term follow-up data does include an ITT analysis. 2 Trial was terminated due to lack of patient accrual. 3Except for toxic effects. 4 Only ineligible participants described. Yes (item adequately addressed); No (item not adequately addressed); Partial (item partially addressed); ? unclear or not enough information; NA not applicable. 4.2 ASSESSMENT OF CLINICAL EFFECTIVENESS 4.2.1 Effectiveness of PLDH versus Topotecan The following section of the report summarises the results reported for the long-term follow-up of trial 30-49 requested from Schering-Plough.(44) The objective of the study was to compare the efficacy of pegylated liposomal doxorubicin hydrochloride (PLDH) versus topotecan in terms of survival and progression-free survival when 90% of participants had died or were lost to follow-up. a) Overall survival

Overall survival was defined in the protocol as the time from the start of study drug administration to death. There was a statistically significant benefit in terms of overall survival observed for the PLDH treated arm compared to the topotecan treated group, HR=1.216 (95% CI: 1.00, 1.48). The reduction in the risk of death for participants treated with PLDH compared to those treated with topotecan was 18%. The median overall survival was 62.7 weeks (range: 1.7 – 258.3) for the PLDH group compared to 59.7 weeks (range: 1.6 – 247.1) for the topotecan arm. The 1-, 2- and 3- year survival rates for participants treated with PLDH were 56.3% (95% CI: 50.0%, 62.6%), 34.7% (95% CI: 28.6%, 40.8%), and 20.2% (95% CI: 14.9%, 15.5%) respectively, and 54.0% (95% CI: 47.6%,60.3%), 23.6% (95% CI: 18.1%, 29.2%), and 13.2% (95% CI: 8.8%, 17.7%) respectively for the topotecan treated arm. A summary of the overall survival data is displayed in Table 4 and the corresponding Kaplan-Meier survival curves displayed in Figure 2.

Table 4. Summary of overall survival data based on long-term follow-up for PLDH versus topotecan. Intervention N % Mediana Rangeb p value HRc 95% CI censored (weeks) for HR PLDH 239 16.7 62.7 1.7 – 258.3 0.050 1.216 1.00, 1.48 Topotecan 235 8.9 59.7 1.6 – 247.1+ a Kaplan-Meier estimates + indicates a censored observation b Stratified log-rank test c A hazard ratio (HR) greater than 1 indicates an advantage for PLDH

Figure 2. Kaplan-Meier overall survival curves for PLDH versus topotecan2

A further analysis of the results for overall survival by the sub-groups of patients with platinum-sensitive or platinum-resistant disease indicated that for participants with platinum-sensitive disease (46% of the study population) there was a pronounced overall survival benefit for participants treated with PLDH compared to those treated with topotecan; HR = 1.432 (95% CI: 1.066, 1.923); p=0.017. This corresponded to a 30% reduction in the risk of death in participants treated with PLDH. The median survival was 107.9 weeks (range: 6.9 – 258.3) for PLDH treated participants compared with 70.1 weeks (range: 1.6 – 258.3) for the topotecan treated participants. The 1-, 2- and 3-year survival rates for participants with platinum-sensitive disease were 74.1% (95% CI: 65.8%, 82.4%), 51.2% (41.6%, 60.7%) and 28.4% (95% CI: 19.6%, 37.1%) respectively for participants treated with PLDH compared to 66.2% (95% CI: 57.4%, 75.1%), 31.0% (95% CI: 22.2%, 39.7%) and 17.5% (95% CI: 10.2%, 24.7%) respectively for participants treated with topotecan. A summary of the overall survival in the platinum-sensitive disease sub-group is displayed in Table5.and the corresponding Kaplan-Meier survival curves are displayed in Figure 3.

2 Reproduced with permission from Schering-Plough.(44.).

Table 5. Summary of overall survival data based on long-term follow-up for PLDH versus topotecan – platinum-sensitive disease subgroup Intervention N % Mediana Rangeb p value HRc 95% CI for censored (weeks) HR PLDH 109 22.0 107.9 6.9 – 258.3 0.017 1.432 1.066, 1.923 Topotecan 110 10.9 70.1 1.6 – 247.1+ a Kaplan-Meier estimates + indicates a censored observation b Stratified log-rank test c A hazard ratio (HR) greater than 1 indicates an advantage for PLDH

Figure 3. Kaplan-Meier survival curves for PLDH versus topotecan: platinum-sensitive subgroup3

The results of the analysis for the sub-group of patients with platinum-resistant disease (54% of the study population) showed that there was no statistically significances differences in survival between the two treatment groups; HR= 1.069 (95% CI: 0.823, 1.387); p=0.618. The median overall survival was 38.3 weeks (range: 1.7 – 253.9) for the PLDH treated participants and 42.1 weeks (range: 1.6 – 239.3) for the topotecan treated participants. The 1-, 2-, and 3-year survival rates for the participants with platinum-resistant disease treated with PLDH were 41.5% (95% CI: 32.8%, 50.1%), 21.1% (95% CI: 14.1%, 28.2%) and 13.8% (95% CI: 7.6%, 20.0%) respectively compared with 43.2% (95% CI: 34.5%, 51.9%), 17.2% (95% CI: 10.5%, 23.8%), and 9.5% (95% CI: 4.2%, 14.7%) respectively for the topotecan treated group.

3 Reproduced with permission from Schering-Plough.(44)

69 Table 6 summarises the long-term survival data for the participants with platinum- refractory disease and the corresponding Kaplan-Meier survival curves are displayed in figure 4.

Table 6. Summary of overall survival data based on long-term follow-up for PLDH versus topotecan – platinum-refractory disease subgroup Intervention N % Mediana Rangeb p value HRc 95% CI for censored (weeks) HR PLDH 130 12.3 38.3 1.7 – 253.9+ 0.618 1.069 0.82, 1.387 Topotecan 125 7.2 42.1 1.6 – 239.3 a Kaplan-Meier estimates + indicates a censored observation b Stratified log-rank test c A hazard ratio (HR) greater than 1 indicates an advantage for PLDH

Figure 4. Kaplan-Meier survival curves for PLDH versus topotecan: platinum- refractory subgroup4

A multivariate Cox regression analysis was performed to assess the influence of possible prognostic factors on the treatment effect for overall survival. The results of this analysis for sub-groups according to baseline disease characteristics are displayed in Table7. The variables included in the regression model were treatment, platinum sensitivity (sensitive or resistant), bulky disease (yes / no), baseline KPS (<80,≥80),

4 Reproduced with permission from Schering-Plough.(44.)

70 treatment-free interval after last dose of first-line therapy, and the presence or absence of ascites at baseline.

Table 7. Overall survival for sub-groups according to baseline disease characteristics (multivariate Cox regression analysis) Variable Group N HRa 95% CI for HR ITT 474 1.216 1.00, 1.478 Age <65 years 294 1.322 1.022, 1.710 ≥65 years 180 1.077 0.786, 1.477 Baseline KPS <80 76 0.871 0.531, 1.427 ≥80 394 1.242 0.999, 1.543 Drug-free interval ≤ 6 months 211 1.103 0.826, 1.474 >6 to ≤ 18 months 201 1.284 0.945, 1.744 >18 months 62 1.191 0.633, 2.137 Bulky disease Present 213 1.131 0.849, 1.506 Absent 261 1.294 0.991, 1.691 Platinum sensitivity Sensitive 219 1.432 1.066, 1.923 Refractory 255 1.069 0.823, 1.387 Baseline ascites Present 142 0.978. 0.689, 1.389 Absent 330 1.387 1.088, 1.768 Hazard ratios are for the comparison of treatment groups within the subgroup. a A HR>1 indicates an advantage for PLDH

The results of the sub-group analysis for overall survival by potential prognostic baseline variables indicated that age less than 65 years, platinum-sensitive disease and the absence of ascites are disease characteristics associated with improved survival. None of the other potential baseline variables were significantly associated with an improvement in overall survival for PLDH in comparison to topotecan. b) Progression-free survival (PFS) Progression-free survival was defined as the time from the first day of study drug dosing to documented disease progression or death due to any cause while the participant was on the study drug or during the long-term follow-up period.

There were no statistically significant differences between the PLDH treated participants and the topotecan treated participants in terms of progression-free survival. The median PFS was 16.1 weeks (range: 1.3 – 162.4) for the PLDH group and 16.9 weeks (range: 0.4 – 178.6) for the topotecan treated participants. The corresponding HR=1.118 (95% CI: 0.93, 1.35) Table 8 provides a summary of the PFS data for both groups of participants and Figure 5 shows the corresponding Kaplan-Meier survival curves .

Table 8. Summary of progression-free survival data based on long-term follow-up for PLDH versus topotecan Intervention N % Mediana Rangeb p value HRc 95% CI for censored (weeks) HR PLDH 239 16.1 16.1 1.3 – 162.4+ 0.241 1.118 0.93, 1.35 Topotecan 235 16.9 16.9 0.4 – 178.6 a Kaplan-Meier estimates + indicates a censored observation b Stratified log-rank test c A hazard ratio (HR) greater than 1 indicates an advantage for PLDH

Figure 5. Kaplan-Meier progression-free survival curves for PLDH versus topotecan5

A further analysis of the results for PFS by the sub-groups of patients with platinum- sensitive or platinum-resistant disease was undertaken. The results of the analysis for patients with platinum-sensitive disease showed a median PFS of 27.3 weeks (range: 2.4 – 151.9) for the PLDH treated participants and 22.7 weeks (range: 0.4 – 155.9) for the topotecan treated group. The HR of 1.287 (95% CI: 0.98, 1.69) indicated that there were no statistically significant differences between the two treatment groups in this sub-group of platinum-sensitive patients on this outcome measure.

No further significant differences were observed between the two treatment groups in the platinum-resistant sub-group of participants. The median PFS was 9.1 weeks (range: 1.3 – 162.4) in the PLDH treated group compared to 13.6 (range: 1.4 – 178.6) in the topotecan treated participants; HR = 0.99 (95% CI: 0.77, 1.28).

5 Reproduced with permission from Schering-Plough.(44).

72 A further multivariate Cox regression analysis was performed to assess the influence of possible prognostic factors on the treatment effect for PFS. Again the variables included in the regression model were treatment, platinum sensitivity (sensitive or refractory), bulky disease (yes / no), baseline KPS (<80, =>80), treatment-free interval after last dose of first-line therapy, and the presence of absence of ascites at baseline.

The results of this regression analysis indicated that there were no statistically significant differences between the treatment groups in relation to the effect of any of the potentially prognostic variables on PFS. c) Response

A responder was defined as a participant with at least a durable (complete or partial) response. A durable response was the participant’s maximum confirmed response. A complete response (CR) was defined as the complete disappearance of all known measurable and assessable disease on two separate measurements at least 4 weeks apart. A partial response (PR) was defined as a 50% reduction in the sum of products of the perpendicular diameters of all measurable lesions for at least 4 weeks. Total response data included both complete and partial responses.

The total response rate was similar for the two treatment groups. Forty-seven (19.7%; 95% CI: 14.6%, 24.7%) of the participants in the PLDH treated group and 40 (17%; 95% CI: 12.2%, 21.8%) of the topotecan treated participants had either a PR or CR as their best confirmed response. Seventy-seven (32.2%) of the PLDH treated group and 95 (40.4%) of the topotecan-treated participants had stable disease as their best response. A summary of the objective response data is displayed in Table 9.

Table 9. Summary of objective response rates based on long-term follow-up for PLDH versus topotecan

PLDH (n=239) Topotecan (n=235) Overall response Response rate % 95% CI for Response rate % 95% CI for (n) response rate (n) response rate Totala 19.7% (47) 14.6%, 24.7% 17.0% (40) 12.2%, 21.8% Complete 3.8% (9) 4.7% (11) Partial 15.9% (38) 12.3% (29)

73 The response rates were also analysed according to the sub-groups of platinum- sensitive and platinum-resistant participants. For participants with platinum-sensitive disease 32 (29.4%; 95% CI: 20.8%, 37.9%) of the 109 participants treated with PLDH and 31 (28.2%; 95% CI: 19.8%, 36.6%) of the 110 participants in the topotecan treated group had confirmed PR or CR as their best response. Forty (36.7%) of the PLDH treated participants and 42 (38.2%) of the topotecan treated participants had stable disease as their best response. For participants with platinum-resistant disease 15 (11.5%; 95% CI: 6.0%, 17.0%) of the 130 participants in the PLDH group and 9 (7.2%; 95% CI: 2.7%, 11.7%) of the 125 topotecan treated participants had confirmed objective PR or CR as their best response. Thirty-seven (28.5%) of the PLDH treatment group and 53 (42.2%) of the topotecan treated arm had stable disease as their best response. Figure 6 shows the data relating to the incidence of complete, partial, total and stable responses. It can clearly be seen from figure 6 that there were no statistically significant differences between the PLDH and topotecan arms in terms of response rates (complete, partial, total or stable) for any participants in the trial.

74 Figure 6. Relative risks for response rates for PLDH versus topotecan sub-group analysis stratified by platinum sensitivity

All participants: Complete response 1.24 (0.54, 2.87)

All participant: Partial response 0.78 (0.50, 1.21)

Total response 0.87 (0.59, 1.26)

Platinum-sensitive: CR 1.23 (0.52, 2.92)

Platinum-sensitive: PR 0.94 (0.56, 1.57)

Platinum-sensitive: Total 1.01 (0.67, 1.54)

1.05 (0.11, 9.98) Platinum-resistant: CR

Platinum-resistant: PR 0.49 (0.21, 1.12)

Platinum-resistant: Total 0.52 (0.24, 1.15)

0.1 0.2 0.5 1 2 5 10 Favours PLD Favours topotecan

d) Quality of life Quality of life was assessed in the trial using the EORTC QLQ-C30 questionnaire. The questionnaire consists of nine sub-scales – one global QoL scale, five function scales (physical, role, emotional, cognitive and social), and three symptom scales (fatigue, pain and nausea/vomiting), in addition to six questions on single items that cover dyspnea, sleep disturbance, appetite loss, constipation, diarrhoea and financial impact. Higher scores on the function scales indicate better functioning and QoL, whereas higher scores on the symptoms scales indicate the increased presence of symptoms.

Assessments were made at baseline and at the start of each treatment cycle until 24 weeks follow-up. Due to the difference in the cycle length of the two treatment regimens, the first time point at which data could be collected from the two study groups was at week 12. At baseline the questionnaire was completed by 82% of participants, however less than 50% of participants in either treatment arm completed the questionnaire at 12 weeks.

75 At baseline the function and symptom scale scores were similar for the two treatment groups. At 12 week follow-up 28.5% (68/239) of the participants in the PLDH treated group and 23.4% (55/235) of the participants in the topotecan treated group had improved or stable global QoL scores; RR=0.82 (95% CI: 0.61, 1.12). In the PLDH treatment group 20.5% (49/239) had a worsened global QoL score compared with 20.4% (48/235) of participants in the topotecan treatment group; RR=0.97 (95% CI: 0.70, 1.42). Neither of these observations was statistically significant.

The numbers of patients with a maintained or improved score at 12 weeks for each of the sub-scales is shown in Table 10. The corresponding relative risks were calculated and are presented in Figure7.. The number of patients with a maintained or improved pain sub-scale score showed a statistically significant difference between the PLDH treated participants and the topotecan treated group, in favour of topotecan; RR=1.26 (95% CI: 1.08, 1.50). However, there were no other statistically significant differences between the two groups observed on any of the other sub-scale scores.

Table 10. Percentage of patients with a maintained or improved QoL score at 12 weeks follow-up for PLDH versus topotecan. QoL sub-scale PLDH Topotecan Physical functioning 56% (66/118) 56% (61/107) Role functioning 65% (77/118) 58% (63/109) Emotional functioning 67% (80/119) 74% (80/108) Cognitive functioning 73% (87/119) 73% (79/108) Social functioning 69% (82/119) 64% (69/108) Global QoL 58% (68/117) 52% (54/104) Fatigue 57% (67/118) 56% (61/109) Nausea/vomiting 72% (86/119) 71% (77/109) Pain 64% (76/119) 81% (88/109)

Figure 7. Relative risk of number of patients with a maintained or improved QoL score at 12 weeks follow-up for PLDH versus topotecan

Physical function 1.02 (0.81, 1.28)

Role function 0.89 (0.72, 1.10

Emotional function 1.10 (0.93, 1.31)

Cognitive function 1.00 (0.85, 1.17) Social function 0.93 (0.77, 1.12) Global QoL 0.89 (0.70 1.13)

Fatigue 0.99 (0.78, 1.24)

Nausea/Vomiting 0.98 (0.83, 1.15)

Pain 1.26 (1.08, 1.50)

0.5 1 2

Favours PLDH Favours topotecan

Quality-adjusted survival analysis (Q-TwiST)

Interventions may vary not only in terms of their effect on disease status, but also in terms of their effect on quality of life. Quality-adjusted survival analysis is an approach that compares treatments, taking into account both the quality and quantity of the life of a patient. A Q-TwiST analysis was undertaken to compare PLDH versus topotecan.

TwiST (Time without Symptoms or Toxicity) is the period of time during which the average patient experiences no symptoms or toxicity, and a higher TwiST is desirable. Figure 9 shows the overall survival curve for each treatment group partitioned into the three health states: TOX (time a participant reported a Grade 3 or higher toxicity); PROG (time from relapse until death or until15 months following randomisation, whichever occurred first); and TwiST (time a patient was not in progression or toxicity). The areas between the consecutive curves represent the average time participants spent in each particular health state.

77 Figure 8. Q-TwiST survival analysis: partitioned survival curves6

The average times spent in the three health states of TOX, PROG and TwiST for the two treatment arm and their differences are presented in Table 11.

Table 11. Time (in months) spent in the three health states and their differences Health states PLDH Topotecan Differences (95% CI) TOX 0.84 1.54 -0.70 (-1.04, -0.36) TwiST 4.65 3.51 1.14 (0.46, 1.82) PROG 5.07 5.53 -0.46 (-1.31, 0.39)

Compared to topotecan, PLDH had a smaller TOX value (-0.70) and PROG value (- 0.46) whilst the time spent in TwiST was higher (+ 1.14). This indicates that the participants in the PLDH treatment group spent more time in the good health state of TwiST and less time in the poor health states of TOX and PROG relative to the topotecan treatment group.

Ultimately Q-TwiST combines the states of TOX, TwiST and PROG into a single measure by summing the time spent in these three health states weighted by their relative utilities. The utility values vary between 0 and 1, where 1 represents perfect health and 0 denotes a state as bad as death. The utility for TwiST is always taken to be 1. Utility values less than 1 for the health states of TOX and PROG penalise treatments for toxicity and disease progression periods.

Figure 9 shows that for all combinations of the utility values, Q-TwiST would always favour PLDH relative to topotecan, with this difference being statistically significant

6 Reproduced with permission from Schering-Plough.(45)

78 for the area below the curve. This suggests PLDH provided both lower treatment toxicity and better treatment effect compared to topotecan, and thus improved the Q- TwiST.

Figure 9. Q-TwiST threshold utility analysis for PLDH versus topotecan7

The Q-TwiST analysis suggests that even though there is only a marginally significant difference observed in overall survival in favour of PLDH, when quality of life outcomes such as toxicity and progression are also taken in to account, PLDH has further advantages over topotecan. However, Q-TwiST analysis is a simplistic model that is largely based on the assumption that a day with any toxicity is valued the same (awarded the same utility value) regardless of the type of adverse event experienced. A full analysis of quality adjusted survival should weight patients’ quality of life according to the actual health states and toxicities they experience, including the period without either toxicity or progression.

7 Reproduced with permission from Schering-Plough.(45)

79 e) Adverse events Extensive data on adverse events were reported in the submission from Schering- Plough.(44) However, only the data on treatment-related effects experienced by at least 10% of participants will be reported and discussed in this section.

Overall in this trial 18% (43/239) of the PLDH treated participants and 16% (37/235) of the participants in the topotecan group discontinued treatment due to adverse events. In addition three participants treated with topotecan died as a result of treatment-related adverse events in the form of neutropenia and sepsis. No treatment related deaths were recorded in the participants treated with PLDH. Table 12 and Figures 10 and 11 provide further details of the adverse events experienced in the two treatment groups.

80 Table 12. Treatment emergent adverse events in a least 10% of participants by preferred term: All grades, Grade 3, and Grade 4.

Body system PLDH (n=239) Topotecan (n=235) Adverse eventa All grades Grade 3 Grade 4 All grades Grade 3 Grade 4 n (%) n (%) n (%) n (%) n (%) n (%) Body as a whole Asthenia 96 (40.2%) 17 (7.1%) 0 121 19 (8.1%) 0 (51.5%) Abdominal pain 80 (33.5%) 24 (10.0%) 1 (0.4%) 89 (37.9%) 19 (8.1%) 4 (1.7%) Fever 51 (21.3%) 2 (0.8%) 0 72 (30.6%) 8 (3.4%) 5 (2..1%) 1 Pain 50 (20.9%) 4 (1.7%) 1 (0.4%) 40 (17.0%) 4 (1.7%) 0 Mucous membrane 34 (14.2%) 9 (3.8%) 0 8 (3.4%) 0 0 disorder Back pain 28 (11.7%) 4 (1.7%) 0 24 (10.2%) 2 (0.9%) 0 Infection 28 (11.7%) 5 (2.1%) 0 15 (6.4%) 2 (0.9%) 0 Headache 25 (10.5%) 2 (0.8%) 0 35 (14.9%) 0 0 Digestive system Nausea 110 12 (5.0%) 1 (0.4%) 148 16 (6.8%) 3 (1.3%) (46.0%) (63.0%) Stomatitis 99 (41.4%) 19 (7.9%) 1 (0.4%) 36 (15.3%) 1 (0.4%) 0 Vomiting 78 (32.6%) 17 (7.1%) 2 (0.8%) 103 18 (7.7%) 5 (2.1%) (43.8%) Constipation 72 (30.1%) 6 (2.5%) 0 107 11 (4.7%) 2 (0.9%) (45.5%) Diarrhoea 50 (20.9%) 5 (2.1%) 1 (0.4%) 82 (34.9%) 9 (3.8%) 1 (0.4%) Anorexia 48 (20.1%) 6 (2.5%) 0 51 (21.7%) 3 (1.3%) 0 Dyspepsia 29 (12.1%) 2 (0.8%) 0 33 (14.0%) 0 0 Intestinal obstruction 27 (11.3%) 19 (7.9%) 4 (1.7%) 26 (11.1%) 14 (6.0%) 7 (3.0%) Haemic and lymphatic system Anaemia 96 (40.2%) 13 (5.4%) 1 (0.4%) 177 59 (25.1%) 10 (4.3%) (75.3%) Leucopenia 88 (36.8%) 21 (8.8%) 3 (1.3%) 151 83 (35.3%) 36 (15.3%) (64.3%) Neutropenia 84 (35.1%) 19 (7.9%) 10 (4.2%) 193 33 (14.0%) 146 (82.1%) (62.1%) Thrombocytopenia 31 (13.0%) 3 (1.3%) 0 153 40 (17.0%) 40 (17.0%) (65.1%) Metabolic / nutritional disorder Peripheral oedema 27 (11.3%) 5 (2.1%) 0 41 (17.4%) 6 (2.6%) 0 Nervous system Paresthesia 24 (10.0%) 0 0 21 (8.9%) 0 0 Dizziness 10 (4.2%) 0 0 24 (10.2%) 0 0 Respiratory system Pharyngitis 38 (15.9%) 0 0 42 (17.9%) 1 (0.4%) 0 Dyspnea 36 (15.1%) 8 (3.3%) 2 (0.8%) 55 (23.4%) 7 (3.0%) 3 (1.3%) Cough increased 23 (9.6%) 0 0 27 (11.5%) 0 0 Skin and appendages PPE 121 55 (23.0%) 2 (0.8%) 2 (0.9%) 0 0 (50.6%) Rash 68 (28.5%) 10 (4.2%) 0 29 (12.4%) 1 (0.4%) 0 Alopecia 46 (19.2%) 3 (1.3%) 0 123 15 (6.4%) 0 (52.3%) a At each level of summarisation, body system and COSTART preferred term, a participant is counted once for one or more adverse events at that level. b Investigators reported Grade 3 alopecia even though the NCI CTC lists criteria only for Grade

81 Treatment related adverse events occurred in 92.9% (222/239) of the PLDH treated participants and 98.7% (232/235) of the topotecan treated group. The most common treatment related adverse events for the PLDH treated participants were PPE (50.6%), stomatitis (40.6%), nausea (36.8%), leucopenia (36.4%), anaemia (36.0%), neutropenia (35.1%), and asthenia (32.6%). In addition more than 75% of the total instances of the following adverse events were considered by the investigators to be related to treatment with PLDH: mucous membrane disorder, thrombocytopenia, paresthesia, rash and alopecia.

The most common treatment related adverse events for the topotecan treated participants were neutropenia (81.3%), anaemia (71.9%), thrombocytopenia (64.7%), leucopenia (63.4%), nausea (54.9%), alopecia (52.3%), asthenia (44.3%), and vomiting (34.9%). More than 75% of the total instances of the following adverse events were considered by the investigators to be related to treatment with topotecan: mucous membrane disorder, stomatitis, PPE and alopecia.

Grade 3 adverse events: The incidence of adverse events was quite different, both in type and severity between the two treatment arms.

The incidence of the following type of grade 3 events was significantly higher in the PLDH treatment arm compared to the topotecan treatment group:- • mucous membrane disorder (3.8% versus 0%) RR=0.05 (95% CI: 0.006, 0.56) • stomatitis ( 7.9% versus 0.4%) RR=0.056 (95% CI: 0.01, 0.31), • PPE (23% versus 0%) RR=0.009 (95% CI: 0.001, 0.087) • rash (4.2% versus 0.4%) RR=0.11 (95% CI: 0.017, 0.61)

Conversely the incidence of the following events was higher in the topotecan group relative to the PLDH group:- • fever (3.4% versus 0.8%) RR = 4.07 (95% CI: 1.00, 16.82) • anaemia (25.1% versus 5.4%) RR=4.62 (95% CI: 2.64, 8.16) • leucopenia (35.3% versus 8.8%) RR =4.02 (95% CI: 2.6, 6.27) • neutropenia (14.0% versus 7.9%) RR = 1.7 (95% CI: 1.04, 3.00) • thrombocytopenia (17.0% versus 1.3%) RR =13.56 (95% CI: 4.54, 40.99) • alopecia (6.4% versus 1.3%) RR = 5.09 (95% CI: 1.60, 16.27).

There were no statistically significant differences in the number of grade 4 adverse events reported in the PLDH group compared to the topotecan treatment arm. However, the incidence of reports of pain, stomatitis and PPE did remain higher in the PLDH group, although these failed to reach statistical significance. There were, however, significant differences in the number of grade 4 adverse events reported in the topotecan group relative to the PLDH group, with the adverse event profile being very similar to that observed for grade 3 events. These differences were again observed for: • fever (2.1% versus 0) RR=10.17 (95% CI: 1.00, 105.11) • anaemia (4.3% versus 0.4%) RR=10.17 (95% CI: 1.70, 61.42) • leucopenia (15.3% versus 8.8%) RR=12.20 (95% CI: 4.07, 37.04) • neutropenia (35.3% versus 8.8%) RR=14.85 (95% CI: 8.18, 27.36) • thrombocytopenia (17.0% versus 0%) RR=13.56 (95% CI: 4.54, 41.00)

83 Figure 10. Relative risks of Grade 3 treatment related adverse events for PLDH versus topotecan

Asthenia 1.14 (0.61, 2.11) Abdomial pain 0.81 (0.46, 1.42) Fever 4.07 (0.99, 16.82) Pain 1.02 (0.28, 3.67) Mucous membrane disorder 0.057 (0.06, 0.59) Back pain 0.51 (0.11, 2.35) Infection 0.41 (0.09, 1.80) Headache 0.25 (0.02, 2.91) Nausea 1.36 (0.67, 2.77) Stomatitis 0.05 (0.01, 0.31) Vomiting 1.08 (0.57, 2.02) Constipation 1.86 (0.73, 4.80) Diarrhoea 1.83 (0.65, 5.15) Anorexia 0.51 (0.14, 1.83) Dyspepsia 0.25 (0.02, 2.91) Intestinal 0.75 (0.39, 1.44) Anemia 4.62 (2.64, 8.16) Leukopenia 4.02 (2.60, 6.27) Neutropenia 1.76 (1.04, 3.00) Thrombocytopenia 13.56 (4.54, 41.00) Peripheral oedema 1.22 (0.40, 3.72) Paresthesia 1.02 (0.06, 17.57) Dizziness 1.02 (0.06, 17.57) Pharyngitis 2.03 (0.15, 27.31) Dyspnea 0.89 (0.34, 2.33) Cough 1.02 (0.06, 17.57) PPE 0.01 (0.00, 0.09) Rash 0.10 (0.02, 0.61) Alopecia 5.09 (1.60, 16.27)

1.00E-05 0.001 0.01 0.10.2 0.5 1 2 5 10 100

PLD caused more adverse events Toptecan caused more adverse events

84 Figure 11. Relative risks of Grade 4 treatment-related adverse events for PLDH versus topotecan

Asthenia 1.02 (0.06, 17.57) Abdomial pain 4.07 (0.62, 26.96)

Fever 10.17 (0.99, 105.11) Pain 0.51 (0.04, 6.82) Mucous membrane disorder 1.02 (0.06, 17.57) Back pain 1.02 (0.06, 17.57) Infection 1.02 (0.06, 17.57) Headache 1.02 (0.06, 17.57) Nausea 3.05 (0.44, 21.22) Stomatitis 0.51 (0.04, 6.82) Vomiting 2.54 (0.58, 11.28) Constipation 4.07 (0.36, 46.76) Diarrhoea 1.02 (0.11, 9.71) Anorexia 1.02 (0.06, 17.57) Dyspepsia 1.02 (0.06, 17.57) Intestinal obstruction 1.78 (0.56, 5.63) Anemia 10.17 (1.70, 61.42)

Leukopenia 12.20 (4.07, 37.04) Neutropenia 14.85 (8.18, 27.36) Thrombocytopenia 81.36 (8.55, 785.73) Peripheral oedema 1.02 (0.06, 17.57) Paresthesia 1.02 (0.06, 17.57) Dizziness 1.02 (0.06, 17.57) Pharyngitis 1.02 (0.06, 17.57) Dyspnea 1.53 (0.31, 7.58) Cough 1.02 (0.06, 17.57) PPE 0.25 (0.02, 2.91)

Rash 1.02 (0.06, 17.57) Alopecia 1.02 (0.06, 17.57)

0.01 0.1 0.2 0.5 1 2 5 10 100 1000 PLD caused more adverse events Topotecan caused more adverse events

85 Summary of effectiveness data for PLDH versus topotecan One RCT was identified which investigated the efficacy and safety of PLDH compared to topotecan in women with epithelial ovarian carcinoma whose disease did not respond to or recurred after treatment with first line platinum-based chemotherapy.

Overall survival: The overall survival rates favoured PLDH compared to topotecan with median survival rates of 62.7 weeks versus 59.7 weeks respectively; HR=1.216 (95% CI: 1.00, 1.48); p=0.050.

The overall survival benefit from treatment with PLDH versus topotecan was most pronounced in the platinum-sensitive sub-group of patients with median survival rates of 107.9 weeks versus 70.1 weeks respectively; HR = 1.432 (95% CI: 1.07, 1.92); p=0.017.

For participants with platinum-refractory disease there were no statistically significant differences in overall survival between the two treatment groups. Median survival was 38.3 in the PLDH group and 42.1 weeks in the topotecan group; HR = 1.07 (95% CI: 0.82, 1.39); p=0.619

For PFS and response, the point estimates favoured PLDH over topotecan but as all confidence intervals crossed unity this benefit was not significant.

Progression-free survival There were no significant differences between the two groups. Median survival was 16.1 weeks in the PLDH group versus 16.9 weeks in the topotecan group (HR=1.118; 95% CI: 0.93, 1.35)

Response No significant differences were observed between the groups in terms of response rates. 19.7% of the PLDH group and 17% of the topotecan group had either a complete or partial response.

Quality of Life At 12 weeks, there were no significant differences between the groups in the number of participants who had a maintained or improved score on the EORTC QLQ-C30 overall. However, a significant difference on the pain sub-scale was observed in favour of topotecan, RR=1.26 (95% CI: 1.08, 1.50).

Adverse events The most common adverse events for the PLDH group were: PPE (50.6%), stomatitis (40.6%), nausea (36.8%), leucopenia (36.4%), anaemia (36.0%), neutropenia (35.1%) and asthenia (32.6%).

The most common adverse events in the topotecan group were: neutropenia (81.3%), anaemia (71.9%), thrombocytopenia (64.7%), leucopenia (63.4%), nausea (54.9%), alopecia (52.3%), asthenia (44.3%), and vomiting (34.9%).

There were no statistically significant differences in the number of grade 4 adverse events reported in the PLDH group compared to the topotecan treatment arm. However, the incidence of reports of pain, stomatitis and PPE did remain higher in this group, although these failed to reach statistical significance. There were however, significant differences in the number of grade 4 adverse events reported in the topotecan group relative to the PLDH group, with the adverse event profile being very similar to that observed for grade 3 events. These differences were observed for:- • fever (2.1% versus 0) RR=10.17 (95% CI: 1.00, 105.11) • anaemia (4.3% versus 0.4%) RR=10.17 (95% CI: 1.70, 61.42) • leucopenia (15.3% versus 8.8%) RR=12.20 (95% CI: 4.07, 37.04) • neutropenia (35.3% versus 8.8%) RR=14.85 (95% CI: 8.18, 27.36) • thrombocytopenia (17.0% versus 0%) RR=13.56 (95% CI: 4.54, 41.00)

4.2.2 Effectiveness of topotecan versus paclitaxel This section of the report summarises the results reported in the three publications of trial 039, which compared topotecan to paclitaxel.(24-26). Where long-term follow- up data has been reported, this has been used as opposed to data reported from an analysis at an earlier time point in the trial. The results for the effectiveness outcomes of overall survival and time to progression are therefore taken from the publication of the long-term follow-up data.(25). The results reported from further treatment after cross-over to the alternative drug as third-line therapy are taken from the publication by Gore et al.(26). All other results are taken from the earlier full publication of the trial results.(24) a) Overall survival(25) (HR<1 favours topotecan)

There were no statistically significant differences found between the topotecan treatment arm and the paclitaxel treatment group in terms of overall survival at long- term follow up. At this time the median follow-up time was 58.5 weeks for the topotecan group (range: 0-86) and 52.6 weeks for the paclitaxel treatment group (range: 0-117). The median survival was 63.0 weeks (95% CI: 47.0, 71.9) (range: <1 – 238.4+ weeks; 20.5% censored) in the topotecan treatment arm and 53.0 weeks (95% CI: 42.3, 68.7) (range: <1 – 226.3+; 12.3% censored) in the paclitaxel treatment group. The corresponding adjusted HR, after adjustment for stratification factors was 0.914 (95% CI: 0.681, 1.226); p= 0.44. Figure12 shows the percentage of participants receiving topotecan or paclitaxel surviving at least 4 years post-randomisation.

87 Figure 12. Percentage of participants receiving topotecan or paclitaxel surviving at least 4 years post-randomisation.8

A further analysis of the results for overall survival by the sub-groups of participants with platinum-sensitive or platinum-resistant disease was undertaken. The results indicated that within the sub-group of participants with platinum-sensitive disease (late relapse) there were no significant differences between the two treatment arms in terms of survival, unadjusted HR=1.010 (95% CI: 0.663, 1.541). For participants with disease classified as being platinum-resistant (refractory, early and interim relapse) there were also no statistically significant differences between the two treatment arms in terms of overall survival, unadjusted HR=0.738 (95% CI: 0.498, 1.093) b) Time to Progression(25) There were no statistically significant differences between the two treatment arms in terms of time to progression. The median time to progression was 18.9 weeks (95% CI: 12.1, 23.7) (range: <1 – 92.6+ weeks; 25% censored) in the topotecan treated group compared to 14.7 (95% CI: 11.9, 18.3) (range: <1 – 137.3+ weeks; 12.3% censored) in the paclitaxel treatment group. The corresponding adjusted HR, after adjustment for stratification factors was 0.811 (95% CI: 0.603, 1.092); p= 0.08.

Further analysis of the results for time to progression by the sub-groups of participants with platinum-sensitive or platinum-resistant disease indicated that for both disease sub-groups there were no statistically significant differences between the two

8 Please refer to end of the acknowledgements for reference.

88 treatment groups in terms of time to progression. For the platinum-sensitive disease sub-group the unadjusted HR was 0.823 (95% CI: 0.538, 1.261). The unadjusted HR for the participants with platinum-resistant disease was 0.749 (95% CI: 0.501, 1.121). c) Response(24) Response rate In total 202 participants were evaluated for response, 96 in the topotecan group and 106 in the paclitaxel group. All participants who were not fully assessed for efficacy or who were not evaluated were considered to be non-responders.

There were no statistically significant differences in the response rates between the topotecan treatment group and the paclitaxel treatment group. The overall response rate was 20.5% for the participants treated with topotecan and 13.2% for participants treated with paclitaxel (p = 0.138). For the 23 participants who responded to topotecan, five (4.5%) had a complete response and 18 (16.1%) had a partial response. For the 15 participants treated with paclitaxel who had a response, 3 (2.6%) had a complete response and 12 (10.5%) had a partial response. Table 13 shows the response rates by treatment group for topotecan versus paclitaxel

Table 13. Response rate by treatment group for topotecan versus paclitaxel

Response to treatment Topotecan (n=112) Paclitaxel (n=114) p-value Responder Number % Number % CR 5 4.5 3 2.6 PR 18 16.1 12 10.5 Total 23 20.5 15 13.2 0.138 95% CI 13 – 28.3 7.0 – 19.4 Non-responders Stable disease 33 29.5 38 33.3 Progressive disease 39 34.8 56 49.1 Not assessable 17 15.2 5 4.4 Total 89 79.5 99 86.8

The response to the two different drug regimens was further explored in relation to prior platinum sensitivity (resistant, early relapse, interim relapse or late relapse). The results of the analysis for response by these different sub-groups is displayed in Table 14.

In the sub-group of participants who were classified as having platinum-resistant disease, 3 out of 34 (8.8%) in the topotecan treatment group had a partial response,

89 compared with 1 out of 33 (3%) in the paclitaxel treatment arm. The rates of response to topotecan compared to paclitaxel for early, interim, or late relapse were 16.7% (1/6) versus 10.0% (1/10), 20.0% (4/20) versus 12.5% (2/16) and 28.8% (15/52) versus 20.0% (11/55) respectively.

Table 14. Response in relation to platinum sensitivity for topotecan versus paclitaxel Resistant Early Interim Late Relapse Total Relapse Relapse Platinum No % No % No % No % No % sensitivity Topotecan n=34 n=6 n=20 n=52 n=112 CR 0 0.0 0 0.0 1 5.0 4 7.7 5 4.5 PR 3 8.8 1 16.7 3 15.0 11 21.2 18 16.1 Total (CR + PR) 3 8.8 1 16.7 4 20.0 15 28.8 23 20.5 Paclitaxel n=33 n=10 n=16 n=55 n=114 CR 0 0.0 0 0.0 0 0.0 3 5.5 3 5.7 PR 1 3.0 1 10.0 2 12.5 8 14.5 12 10.7 Total (CR + PR) 1 3.0 1 10.01 2 12.5 11 20.0 15 13.4

A further analysis of response to treatment relative to baseline disease characteristics indicated that in both the treatment groups higher response rates were observed in participants without ascites at baseline, and in participants with smaller tumour burden (<5 cm ), better performance status scores and those who responded to first- line therapy. Table 15 displays the results for response to treatment in relation to baseline disease status .

Table 15. Response in relation to baseline disease status for topotecan versus paclitaxel Baseline status Topotecan Response (%) Paclitaxel Response (%) Age, years ≤ 40 0 0 41-64 19.7 12.0 ≥65 23.7 16.7 Ascites Present 18.9 7.5 Absent 21.3 16.2 Performance status 0 22.0 14.3 1 25.5 13.2 2 5.0 11.8 Tumour burden (cm) <5 cm 33.3 18.0 5≤10cm 10.9 12.5 First-line response Responders 15.2 10.5 Non-responders 5.4 2.6

90 Duration of response Response duration was measured from the time of the first documented complete or partial response to the first sign of disease progression.

The median duration of response was 32.1 weeks (range: 5.4 – 53.1) for participants treated with topotecan who evidenced a response, compared to 19.7 weeks (range: 6.3 – 24.3) for participants treated with paclitaxel. These differences in the duration of response between the two groups were not statistically significant (p = 0.221). Table 16 shows the duration of response for the two treatment groups.

Table 16. Response duration for participants treated with topotecan or paclitaxel

Response variable (weeks) Topotecan Paclitaxel Response duration n=23 n=15

Median 32.1 weeks 19.7 Range 5.4 – 53.1 6.3 – 24.3 Risk ratio 0.416 p-value 0.2218

Time to response There was a statistically significant difference between the topotecan and paclitaxel treatment arms in terms of time to response, in favour of paclitaxel. The median time to documented radiologic response for participants in the topotecan treatment arm was 9.0 weeks (range: 3.1 – 19.0) compared to a median of 6 weeks (range: 2.4 – 12.3) for participants treated with paclitaxel. Table 17 displays the results for the outcome of time to response.

Table 17. Time to response data for participants treated with topotecan or paclitaxel

Response variable (weeks) Topotecan Paclitaxel Time to response n=23 n=15 Median 9.0 6.0 Range 3.1 – 19.0 2.4 – 12.3 Risk ratio 0.476 p-value 0.0409

d) Quality of Life Quality of life was assessed in the trial using the EORTC QOL-C30 questionnaire. However, no results for this outcome were reported other than that the results were similar in both treatment groups.

91 e) Adverse Events Toxicity was summarised as both the worst CTC grade experienced by each patient and as the worst CTC grade experienced within each course for each participant.

Two participants in the topotecan treatment group died due to treatment related toxicity. Both of these deaths were attributed to topotecan-induced sepsis. There were no deaths that were attributed to paclitaxel-induced myelosuppression. A further 7% of participants treated with topotecan and 4% of participants in the paclitaxel treatment group were withdrawn from the study due to treatment related toxicity. The primary reasons for withdrawal in the topotecan-treated participants were febrile neutropenia, infection, and sepsis. Neurotoxicity was the primary reason for withdrawal for paclitaxel-treated participants.

Haematological Toxicity The majority of categories of reported grade 3 and 4 haematological toxicities occurred significantly more often in the topotecan group compared to the paclitaxel treatment arm. The only incidence of haematological toxicity that was higher in the paclitaxel group was for grade 3 neutropenia. This was observed in 30.4% of the paclitaxel treatment arm compared to 15.3% of the topotecan treated participants, RR=1.98 (95% CI: 1.19, 3.34). All other haematological toxicities were higher in the topotecan treatment group, although this was not significantly higher for grade 4 anaemia. Table 18 shows the number and percentages of participants who experienced grade 3 and 4 haematological toxicities by participant and course for each treatment group. The relative risks of experiencing a grade 3 or 4 haematological adverse event by treatment group are displayed in Figure 13.

92 Table 18. Number and percentage of patients and courses with haematological toxicity by worst CTC Grade. Grade 3 Grade 4 Participants n (%) Course n (%) Participants n (%) Course n (%) Toxicity T P T P T P T P Leucopenia 56/110 20/112 234/590 51/585 38/110 2/112 57/590 2/586) (50.9%) (17.9%) (39.7%) (8.8%) (34.5%) (1.8%) (10.0%) (0.3%) Neutropenia 17/111 34/112 166/590 123/584 89/111 24/112 211/590 50/584 (15.3%) (30.4%) (28.1%) (21.1%) (80.2%) (21.4%) (35.8%) (8.6%) Thrombocytopenia 25/111 2/112 89/591 2/565 30/111 3/112 59/591 3/585 (27.5%) (1.8%) (15.1%) (0.3%) (17.0%) (2.7%) (10.0%) (0.5%) Anaemia 42/111 4/112 89/591 8/585 4/111 3/112 6/591 3/585 (37.8%) (3.6%) (15.1%) (1.4%) *(3.6%) (2.7%) (1.0%) (0.5%)

Figure 13. Relative risks of Grade 3 and 4 haematological toxicity for topotecan versus paclitaxel.

Grade 3 leucopenia 0.35 (0.23, 0.54)

Grade 3 neutropenia 1.98 (1.19, 3.34)

Grade 3 thrombocytopenia 0.08 (0.02, 0.29)

Grade 3 anaemia 0.09 (0.04, 0.24)

Grade 4 leucopenia 0.05 (0.01, 0.19)

Grade 4 neutropenia 0.27 (0.18, 0.38)

Grade 4 thrombocytopenia 0.10 (0.03, 0.29)

Grade 4 anaemia 0.74 (0.19, 2.91)

0.01 0.1 0.2 0.5 1 2 5 Toptoccan caused adverse events Paclitaxel caused more adverse events

Non-haematological toxicities The majority of non-haematological adverse events were mild to moderate in severity (Grade 1 / 2). The most frequently reported toxicities considered related or possibly related to treatment in both groups were alopecia and gastrointestinal disturbances, including nausea, vomiting, diarrhoea and constipation. The Grade 3 / 4 adverse events reported more commonly in the topotecan group compared to the paclitaxel group were nausea (9.8% versus 1.8%), vomiting (9.9% versus 2.7%), constipation (5.4% versus 0%), abdominal pain (5.4% versus 3.5%), asthenia (5.4% versus 3.5%), fatigue (8.0% versus 6.1%) and fever/infection (0.9% versus 0%). The Grade 3 / 4

93 adverse events reported more commonly in the paclitaxel treatment arm relative to the topotecan treatment arm were alopecia (0.9% versus 0%), arthralgia (2.6% versus 0.9%), myalgia (2.6% versus 0%) and skeletal pain (5.3% versus 0%). A summary of the non-haematological toxicities that occurred in more than 10% of participants in the trial is displayed in Table 19.

Table 19. Related or possibly related non-haematological toxicities occurring in more than 10% of patients treated with topotecan or paclitaxel

Adverse event Topotecan (% patients) Paclitaxel (% patients) Grade 1 / 2 Grade 3 / 4 Grade 1 / 2 Grade 3 / 4 Alopecia 75.9 0.0 92.1 0.9 Nausea 67.9 9.8 43.0 1.8 Vomiting 53.6 9.9 28.1 2.7 Fatigue 33.1 8.0 25.4 6.1 Constipation 37.5 5.4 30.7 0.0 Diarrhoea 33.9 6.3 37.8 0.9 Abdominal pain 21.5 5.4 36.0 3.5 Fever 27.7 0.9 17.7 0.0 Stomatitis 23.2 0.9 14.0 0.9 Dyspnea 17.8 6.3 13.2 5.3 Asthenia 17.0 5.4 9.6 3.5 Arthralgia 5.5 0.9 28.9 2.6 Myalgia 3.6 0.0 25.4 2.6 Neuropathy 0.9 0.0 15.8 0.0 Skeletal pain 4.5 0.0 11.4 5.3 Flushing 4.5 0.0 14.1 0.0 Paresthesia 0.9 0.0 29.0 0.0

Effectiveness of third-line cross over therapy for topotecan versus paclitaxel

This section of the report is based upon the publication by Gore et al (2001)(26) that reports the results for response to third-line therapy for topotecan versus paclitaxel.

A total of 110 participants crossed over to the alternative drug regimen as third-line therapy within the trial due to either failure to respond to second-line therapy, relapse after an initial response to second-line therapy or toxicity. Sixty-one participants crossed over from paclitaxel to topotecan and 49 participants crossed over from topotecan to paclitaxel.

At cross over there were no major differences in the demographic characteristics of the two groups of participants. However, there were statistically significant differences in the two groups’ potential chemo-responsiveness. The topotecan group

94 contained fewer participants who had evidenced a partial response to second-line therapy compared to the paclitaxel group (3% versus 14%), and more participants in this group had progressive disease (57% versus 45%). Participants who were in the topotecan treatment group received a median of 3 courses of treatment (range: 1 – 23) and those in the paclitaxel treatment arm a median of 4 course (range: 1 – 12). The planned dose was maintained in 95% of courses of topotecan and 94% of courses of paclitaxel. f) Overall survival: There were no statistically significant differences between the topotecan and paclitaxel treatment arms observed in terms of overall survival on third- line therapy. The median survival time from the initiation of cross-over therapy was 40 weeks (range: 1 – 123) for participants who received topotecan as third-line therapy compared to 48 weeks (range: 2 – 86) for participants who received paclitaxel. g) Time to progression: There were no differences between the two treatment groups in time to progression. The median time to progression was 9 weeks in both of the treatment arms. h) Response: There were no significant differences between the participants in the two treatment groups in relation to response rate on third-line therapy. The overall response rate to topotecan was 13% (8/61; 8 partial responses) and to paclitaxel was 10% (5/49; 2 complete responses and 3 partial responses); p=0.6838. Table 20 and Table 21 show the responses to topotecan and paclitaxel according to participants response to previous second-line chemotherapy. In total seven out of the eight participants who responded to topotecan as a third-line therapy had disease that had been refractory to paclitaxel. Only one of the participants who responded to third-line paclitaxel had responded to second-line topotecan.

Table 20. Response to Topotecan (third-line) treatment according to response to paclitaxel (second-line) treatment

Patient response to paclitaxel (third-line) treatment (no.) CR PR SD PD NE Response to topotecan (second-line) treatment PR, n= 7 1 0 2 4 0 SD, n=27 0 2 4 10 1 PD, n=22 1 1 6 10 4 NE, n=3 0 0 0 1 2 * PR=partial response; SD=stable disease; PD=progressive disease; NE=not evaluable.

Table 21. Response to Paclitaxel (third-line) treatment according to response to topotecan (second-line) treatment

Patient response to topotecan (third-line) treatment (no.) CR PR SD PD NE Response to paclitaxel (second-line) treatment PR, n= 2 0 0 1 1 0 SD, n=17 0 4 4 10 5 PD, n=35 0 3 3 24 5 NE, n=1 0 1 0 0 0 * PR=partial response; SD=stable disease; PD=progressive disease; NE=not evaluable.

Response according to participants’ sensitivity to first-line platinum therapy: Response to third-line treatment was also analysed according to participants sensitivity to first-line platinum based therapy. Only participants who had a treatment-free interval of more than 6 months (platinum-sensitive diseases) responded to third-line topotecan, but four participants who relapsed within 6 months of first-line platinum-based therapy (platinum-resistant disease) responded to paclitaxel as the third-line therapy. i) Adverse events:

Haematologic Toxicities The toxicity profiles between the two treatment groups for participants who received third-line cross over therapy were similar to those recorded in the randomised study. The incidence of haematological toxicities was higher in the topotecan treatment group compared to the paclitaxel treatment arm. Grade 4 neutropenia was recorded in 81% of participants in the topotecan group compared with 23% in the paclitaxel arm. Grade 4 thrombocytopenia was also higher in the topotecan arm, 22% versus 2% in the paclitaxel group. Both grade 3 and 4 anaemia were also reported more frequently

96 in the topotecan group relative to the paclitaxel treatment group, grade 3: 27% versus 0% and grade 4: 2% versus 0%. Table 22 shows the percentage of haematological toxicities recorded during third-line cross over therapy for both the topotecan and paclitaxel treatment groups.

Table 22. Haematological toxicity during cross-over (third-line) treatment with topotecan versus paclitaxel Third-line treatment (% of participants) Toxicity Topotecan (n=61) Paclitaxel (n=49) Neutropenia Grade 1 0 15 Grade 2 3 15 Grade 3 14 31 Grade 4 81 23 Thrombocytopenia Grade1 47 17 Grade 2 12 0 Grade 3 8 2 Grade 4 22 0 Anaemia Grade 1 8 52 Grade 3 62 42 Grade 3 27 2 Grade 4 2 0

Non-haematologic Toxicities Non-haematological toxicities associated with both topotecan and paclitaxel were generally mild or moderate (grade 1 / 2). The haematological toxicities reported for this stage of treatment are displayed in Table 23. The toxicities associated with topotecan were generally gastrointestinal disturbances, and were similar to those observed in the randomised phase of the study. The incidences of nausea, vomiting, abdominal pain and fatigue were greater with topotecan (59%, 36%, 25% and 23% respectively) than with paclitaxel (39%, 27%, 20% and 16% respectively). However, the participants who received paclitaxel were premedicated with dexamethasone, which may have reduced the incidence of nausea and vomiting within this group. The incidences of arthralgia, myalgia, and paresthesia were greater with paclitaxel (37%, 33%, and 31% respectively) than with topotecan (15%, 18%, and 8% respectively)

Table 23. Non-haematologic Toxicity to Cross-over (third-line) treatment for topotecan versus paclitaxel (% of participants)

Grade 1 / 2 Toxicity Grade 3 / 4 Toxicity Adverse event Topotecan (n=61) Paclitaxel (n=49) Topotecan (n=61) Paclitaxel (n=49) Nausea 59 39 13 10 Vomiting 36 27 12 8 Diarrhoea 33 35 3 2 Abdominal pain 25 20 8 8 Constipation 26 35 5 0 Fatigue 23 16 8 0 Alopecia 21 41 0 0 Myalgia 16 31 0 2 Arthralgia 13 37 2 0 Paresthesia 8 31 0 0

98 Summary of effectiveness data for topotecan versus paclitaxel One RCT was included that compared the efficacy and safety of topotecan to paclitaxel in patients with advanced epithelial ovarian cancer who had progressed during or after one platinum-based chemotherapy regimen.

There were no statistically significant differences between the groups on any outcome measure, apart from time to response. This favoured paclitaxel.

Overall survival: There were no significant differences between the two treatment groups in terms of overall survival. The median survival time was 63 weeks in the topotecan group compared to 53 weeks in the paclitaxel group; HR=0.914 (95% CI: 0.681, 1.226).

Time to progression. There were no significant differences between the two treatment arms. Median time to progression was 19.8 weeks in the topotecan group versus 14.7 weeks in the paclitaxel group; HR=0.811 (95% CI: 0.603, 1.092).

Response rate No significant differences were observed between the groups in terms of response rates. 20.5% of the topotecan group and 13.2% of the paclitaxel group had either a complete or partial response.

Response duration There were no significant differences between the groups. Median duration of response was 32.1 weeks in the topotecan group compared with 19.7 weeks in the paclitaxel group.

Time to response There was a statistically significant difference in time to response in favour of paclitaxel. The median time to response was 9.0 weeks in the topotecan group compared with 6.0 weeks in the paclitaxel group.

Quality of Life No results were reported for Quality of Life.

Adverse events The reported grade 3 / 4 haematological toxicities (leucopenia, neutropenia, thrombocytopenia and anaemia) were significantly higher in the topotecan group compared to the paclitaxel arm, apart from grade 3 neutropenia which was higher in the paclitaxel treatment group.

The Grade 3 / 4 adverse events reported more commonly in the topotecan group compared to the paclitaxel group were nausea, vomiting, constipation, abdominal pain, asthenia, fatigue and fever/infection. The Grade 3 / 4 adverse events reported more commonly in the paclitaxel treatment arm relative to the topotecan treatment arm were alopecia, arthralgia, myalgia and skeletal pain.

Results reported from cross-over to third-line therapy

Overall survival: There were no significant differences between the two treatment groups. Median survival time was 40 weeks in the topotecan group compared to 48 week in the paclitaxel arm.

Time to progression: The median time to progression was 9 weeks in both treatment arms.

Response: There were no significant differences in the response rates in the two treatment groups. 13% of the topotecan group and 10% of the paclitaxel group had a complete or partial response.

Adverse events: The toxicity profiles between the two treatment groups for participants who received third-line cross over therapy were similar to those recorded in the randomised study. The incidence of haematological toxicities was higher in the topotecan treatment group compared to the paclitaxel treatment arm. Non-haematological toxicities associated with both topotecan and paclitaxel were generally mild or moderate (grade 1 / 2). The incidences of nausea, vomiting, abdominal pain and

fatigue were greater with topotecan than with paclitaxel. The incidence of arthralgia, myalgia, and paresthesia were greater with paclitaxel.

4.2.3. Effectiveness of PLDH versus paclitaxel

The results presented in this section of the report summarise those presented in a trial report submitted by Schering-Plough as part of the industry submission.(27) However, this trial was terminated prematurely, therefore the results should be interpreted with caution. a) Overall survival The primary outcome in this trial was overall survival. The median overall survival was 46.6 weeks (range: 2.3 – 263.7+) in the PLDH treated participants compared to 56.3 weeks (range: 1.4 – 211.4) in the paclitaxel treated participants. The HR of 0.931 (95% CI: 0.702, 1.234) indicated that there were no statistically significant differences between the two treatment groups in terms of overall survival. A summary of the overall survival data for all participants is displayed in Table 24.

Table 24. Summary of overall survival data for all participants for PLDH versus paclitaxel Intervention N % Median Range p value HRa 95% CI censored (weeks) for HR PLDH 108 7.4 46.6 2.3 – 263.7+ 0.0618 0.932 0.702, 1.234 Paclitaxel 108 6.5 56.3 1.4 – 211.4 + indicates a censored observation a A hazard ratio (HR) greater than 1 indicates an advantage for PLDH

A further analysis of the results for overall survival by the sub-groups of patients with platinum-sensitive or platinum-resistant disease indicated that for participants with platinum-sensitive disease the median overall survival was 65.4 weeks (range: 3.9 – 263.7+) for PLDH treated participants and 57.0 weeks (range: 14 – 172.3) for paclitaxel treated participants. The corresponding HR of 1.051 (95% CI: 0.663, 1.667) showed that once again there was no statistically significant difference between the two treatment arms in relation to overall survival in this sub-group of platinum-

100 sensitive patients. A summary of the overall survival in the platinum-sensitive disease sub-group is displayed in Table 25. Table 25. Summary of overall survival data based on PLDH versus paclitaxel – platinum-sensitive disease subgroup Intervention N % Median Range p value HRa 95% CI for censored (weeks) HR PLDH 44 13.6 65.4 3.9 – 263.7+ 0.833 1.051 0.66, 1.67 Paclitaxel 41 7.3 57.0 1.4 – 172.3 + indicates a censored observation a A hazard ratio (HR) greater than 1 indicates an advantage for PLDH

The results of the analysis for the sub-group of patients with platinum-resistant disease showed that the median overall survival was 36.7 weeks (range: 2.3 – 241.1+) for the PLDH treated participants and 54.3 weeks (range: 1.7 – 211.4+) for the paclitaxel treated participants. There were no statistically significant differences between the two groups in terms of survival, as indicated by a HR of 0.865 (95% CI: 0.61, 1.24). Table 26 summarises the long-term survival data for the participants with platinum-resistant disease.

Table 26. Summary of overall survival data based on PLDH versus paclitaxel – platinum-resistant disease subgroup Intervention N % Median Range p value HRa 95% CI for censored (weeks) HR PLDH 64 3.1 36.7 2.3 – 242.1+ 0.427 0.865 0.61, 1.24 Paclitaxel 67 6.0 54.3 1.7 – 211.4+ + indicates a censored observation a A hazard ratio (HR) greater than 1 indicates an advantage for PLDH

b) Progression-free survival No data were reported for this outcome. c) Response The data for this outcome measure was not analysed and therefore no data is reported. d) Quality of Life There were no results reported for the Quality of Life assessment. e) Adverse events Overall in this trial, 16.7% (18/108) of the PLDH treated participants and 6.5% (7/108) of the participants in the paclitaxel group discontinued treatment due to adverse events.

101

The five most commonly reported treatment emergent adverse events for the PLDH treated participants were nausea (51.9%), PPE (50.9%), stomatitis (48.1%), alopecia (43.5%) and asthenia (38.9%). For the paclitaxel treated participants the five most commonly reported adverse events were alopecia (87.0%), nausea (43.5%), paresthesia (43.5%), constipation (38.0%), and asthenia (33.3%).

The treatment emergent adverse events that occurred in at least 10% of participants in either treatment group for all grades, Grade 3 and Grade 4 are reported in Table 27, whilst the relative risks of experiencing a Grade 3 toxicity are displayed in Figure 14.

102 Table 27. Treatment emergent adverse events in a least 10% of participants by preferred term: All grades, Grade 3, and Grade 4.

Body system PLDH (n=108) Paclitaxel (n=108) Adverse events All grades Grade 3 Grade 4 All grades Grade 3 Grade 4 Body as a whole Asthenia 42 (38.9%) 4 (3.7%) 0 36 (33.3%) 6 (5.3%) 1 (0.9%) Abdominal pain 34 (31.5%) 12 (11.1%) 0 35 (32.4%) 7 (6.5%) 0 Fever 28 (25.9%) 7 (6.5%) 0 8 (7.4%) 3 (2.8%) 0 Pain 24 (22.2%) 1 (0.9%) 0 24 (22.2%) 3 (2.8%) 0 Infection 23 (21.3%) 2 (1.9%) 1 (0.9%) 10 (9.3%) 1 (0.9%) 0 Headache 12 (11.1%) 1 (0.9%) 0 13 (12.0%) 2 (1.9%) 0 Ascites 11 (10.2%) 6 (5.6%) 0 8 (7.4%) 1 (0.9%) 0 Back pain 11 (10.2%) 1 (0.9%) 0 14 (13.0%) 1 (0.9%) 0 Cardiovascular system Vasodilation 5 (4.6%) 1 (0.9%) 0 13 (12.0%) 1 (0.9%) 0 Digestive system Nausea 56 (51.9%) 6 (5.6%) 1 (0.9%) 47 (43.5%) 2 (1.9%) 0 Stomatitis 52 (48.1%) 11 (10.2%) 0 12 (11.1%) 1 (0.9%) 0 Vomiting 37 (34.3%) 10 (9.3%) 2 (1.9%) 34 (32.5%) 4 (3.7%) 0 Constipation 30 (27.8%) 4 (3.7%) 0 41 (38.0%) 5 (4.6%) 0 Diarrhoea 23 (21.3%) 3 (2.8%) 0 24 (22.2%) 3 (2.8%) 0 Anorexia 18 (16.7%) 1 (0.9%) 0 11 (10.2%) 0 0 Dyspepsia 14 (13.0%) 1 (0.9%) 0 11 (10.2%) 0 0 Haemic & lymphatic system Neutropenia 18 (16.7%) 6 (5.6%) 1 (0.9%) 23 (21.3%) 10 (9.3%) 3 (2.8%) Anaemia 17 (15.7%) 3 (2.8%) 0 23 (21.3%) 5 (4.6%) 0 Leucopenia 15 (13.9%) 5 (4.6%) 1 (0.9%) 21 (19.4%) 9 (8.3%) 0 Metabolic/nutritional disorder Peripheral oedema 14 (13.0%) 0 0 15 (13.9%) 1 (0.9%) 0 Musculoskeletal system Myalgia 4 (3.7%) 1 (0.9%) 0 31 (28.7%) 7 (6.5%) 0 Arthralgia 2 (1.9%) 0 0 23 (21.3%) 2 (1.9%) 0 Nervous system Paresthesia 15 (13.9%) 0 0 47 (43.5%) 4 (3.7%) 0 Somnolence 11 (10.3%) 3 (2.8%) 0 17 (15.7%) 2 (1.9%) 0 Respiratory system Dyspnea 18 (16.7%) 6 (5.6%) 1 (0.9%) 15 (13.9%) 1 (0.9%) 0 Pharyngitis 8 (7.4%) 0 0 18 (16.7%) 0 0 Skin & appendages PPE 55 (50.9%) 16 (14.8%) 1 (0.9%) 13 (12.0%) 0 0 Alopecia 47 (43.5%) 3 (2.8%) 0 94 (87.0%) 20 (18.5%) 1 (0.9%) Rash 15 (13.9%) 2 (1.9%) 0 19 (17.6%) 1 (0.9%) 0

103 Figure 14. Relative risks of Grade 3 treatment related adverse events for PLDH versus paclitaxel

Asthenia 1.50 (0.47, 4.84)

Abdomial pain 0.58 (0.24, 1.38) Fever 0.43 (0.12, 1.48) Pain 3.00 (0.44, 20.78) Infection 0.50 (0.07, 3.77) Headache 2. 00 (0.27, 15.15) Ascites 0.17 (0.03, 1.03) Back pain 1.00 (0.11, 9.51)

Vasodilation 1.00 (0.11, 9.51) Nausea 0.33 (0.08, 1.41) Stomatitis 0.09 (0.02, 0.53) Vomiting 0.40 (0.14, 1.17) Constipation 1.25 (0.37, 4.21) Diarrhoea 1.00 (0.24, 4.26) Anorexia 0.50 (0.04, 6.68) Dyspepsia 0.50 (0.04, 6.68) Neutropenia 1.67 (0.65, 4.29) Anaemia 1.67 (0.45, 6.20) Leukopenia 1.80 (0.66, 4.99) Peripheral oedema 2.00 (0.15, 26.79) Myalgi 7.00 (1.15, 43.28) Arthralgia 4.00 (0.35, 45.88) Paresthesia 8.00 (0.77, 84.04) Somnolence 0.67 (0.14, 3.28) Dyspnea 0.17 (0.03, 1.03) Pharyngitis 1.00 (0.06, 17.24) PPE 0.03 (0.00, 0.30) Alopecia 6.67 (2.20, 20.66) Rash 0.50 (0.07, 3.77)

0.001 0.01 0.1 0.2 0.5 1 2 5 10 100 PLD caused more adverse events Paclitaxel caused more adverse events

104 Grade 3 and 4 adverse events The incidence of Grade 4 adverse events was relatively low in both of the treatment arms. Neutropenia was the only grade 4 event reported in both of the treatment groups, 0.9% in the PLDH group compared to 2.8% in the paclitaxel treatment arm. The other grade 4 toxicities reported in the PLDH group were infection (0.9%), nausea (0.9%), vomiting (1.9%), leucopenia (0.9%), dyspnea (0.9%) and PPE (0.9%). None of these toxicities were reported at the grade 4 level of severity in the paclitaxel group. Both grade 4 asthenia (0.9%) and alopecia (0.9%) were reported in the paclitaxel arm, with there being no incidence of these within the PLDH group.

At the grade 3 level of toxicity classification, the following toxicities were reported significantly more often in the PLDH group compared to the paclitaxel treatment arm:- • PPE (14.8% versus 0%), RR=0.031 (95% CI: 0.003, 0.297) • ascites (10.2% versus 0.9%), RR=0.17 (95% CI: 0.027, 1.029) • stomatitis (10.2% versus 0.9%), RR=0.091 (95% CI:0.02, 0.53) • dyspnea (5.6% versus 0.9%), RR=0.17 (95% CI: 0.03, 1.03)

The only toxicity that occurred more frequently in the paclitaxel treatment group compared to the PLDH group was alopecia (18.5% versus 2.8%), RR=6.67 (95% CI: 2.20, 20.66).

105

Summary of effectiveness data for PLDH versus paclitaxel One RCT was identified that evaluated the efficacy and safety of pegylated liposomal doxorubicin hydrochloride compared to paclitaxel in women with epithelial ovarian carcinoma following failure of first-line platinum based chemotherapy. This trial had been terminated early due to poor participant accrual.

Overall survival: There were no significant differences between the two treatment groups in terms of overall survival. The median survival time was 46.6 weeks in the PLDH group compared to 56.3 weeks in the paclitaxel group; HR=0.0932 (95% CI: 0.702, 1.234)

There were no further significant differences in overall survival between the two treatment arms when this was assessed in the sub-groups of participants with platinum- resistant and platinum-sensitive disease.

Adverse events: The most commonly reported adverse events in the PLDH treatment group were nausea, PPE, stomatitis, alopecia and asthenia. The most frequently reported events were alopecia nausea, paresthesia, constipation and asthenia in the paclitaxel treated participants.

The incidence of Grade 4 adverse events was relatively low in both of the treatment arms. At the grade 3 level of toxicity classification, PPE, stomatitis, dyspnea and ascites occurred more frequently in the PLDH treatment arm compared to the paclitaxel group. Alopecia occurred more frequently in the paclitaxel treatment group compared to the PLDH group.

106 4.2.4 Effectiveness of single agent paclitaxel compared to a combination of cyclophosphamide, doxorubicin and cisplatin (CAP) a) Overall survival Overall survival was defined as the time from randomisation to death by any cause. Participants still alive at the time of analysis were censored at the time of last follow- up. At a median follow-up of 49 months (range 40 to 54) 61 participants had died. Twenty-seven of these were in the CAP treatment group (57% of group) and 34 in the paclitaxel treatment arm (72% of group). Median survival times were 34.7 months for the CAP treatment participants and 25.8 months for the paclitaxel treatment group.

Comparison of the Kaplan-Meier curves showed an insignificant trend in favour of CAP; HR = 0.70 (95% CI: 0.42, 1.15); p=.160. However, after an adjustment for prognostic factors using a Cox multivariate regression analysis, with residual tumour, treatment-free interval and age adjusted for, the HR was 0.58 (95% CI: 0.34, 0.98); p=0.043 in favour of CAP. The regression analysis showed that the presence of a larger residual tumour (>2 cm) at the time of first surgery was associated with lower mortality; HR=0.49 (95% CI: 0.28, 0.86); p=0.013. b) Progression-free survival A progression-free interval was defined as the time from randomisation to first appearance of progressive disease. Eastern Cooperative Oncology Group criteria were used to assess response to chemotherapy.

Eighty-five patients had disease progression or had died (41 in the CAP arm and 44 in the paclitaxel treatment group) at a median follow-up time of 49 months (range 40 – 54 months). Of these, 24 had disease progression and were still alive at the time of analysis, 54 had disease progression and died, whilst 7 died without evidence of progression.

The median progression-free interval was 15.7 months in the CAP group compared to 9 months in the paclitaxel treated participants. Comparison of the Kaplan-Meier curves for progression-free intervals by treatment, showed an insignificant trend in favour of CAP; HR=0.65 (95% CI: 0.41, 1.01); p=0.08.

107 Again, after adjustment for prognostic factors (residual tumour, treatment-free interval and age) by Cox regression analysis, the HR was 0.60 (95% CI; 0.37, 0.97); p=0.038 in favour of CAP compared to paclitaxel. There were no other significant associations between either age, residual tumour size or treatment-free interval and the risk of progression. c) Response

In total 94 participants were assessable for response. The overall response for both groups combined was 50% (22 complete responses [CR] and 25 partial responses [PR]). Twenty-four participants (26%) had stable disease, whilst 23 participants (24%) evidenced disease progression during treatment. Table 28 shows the responses to CAP compared to paclitaxel.

Table 28. Response to CAP compared to paclitaxel treatment.

Response CAP (n=47) Paclitaxel (n=47) Total No. % No. % No. % CR 14 30 8 17 22 23 PR 12 25 13 28 25 27 SD 14 30 10 21 24 26 PD 7 15 16 34 23 24 Total 47 47 94 Not evaluated - - - - 3 3 Response after cross-over CR 7 23 3 13 10 9 PR 7 23 2 9 9 17 SD 7 23 7 30 14 26 PD 9 30 11 48 20 38 Total 30 23 53

Fourteen participants in the CAP treatment group (30%) versus 8 participants (17%) in the paclitaxel group had CR, RR=0.57 (95% CI: 0.27, 1.20). Twelve participants in the CAP arm (25%) compared to 13 participants in the paclitaxel treated group (28%) had a PR, RR=1.08 (0.56, 2.11). In relation to stable disease, 14 participants in the CAP arm (30%) had stable disease compared to 10 participants (21%) in the paclitaxel arm, RR=0.71 (95% CI: 0.35, 1.42). None of these differences in response rates were significantly different between the two treatment groups. However, a significant difference was observed in relation to disease progression. Seven (15%) of the participants in the CAP treatment arm had progressive disease compared to 16 (34%) of the participants in the paclitaxel arm, RR=2.29 (95% CI: 1.07, 5.03).

108 Overall, a total of 53 patients were crossed over because of treatment failure. Twenty-three participants who did not respond to CAP were crossed over to paclitaxel treatment, and three of these participants (13%) achieved a CR, whilst a further 2 (9%) showed a PR. Thirty participants who initially did not respond to paclitaxel crossed over to the CAP treatment arm. In these participants 7 (23%) achieved a CR and 7 (23%) achieved a PR. There were no statistically significant differences observed in the response rates between the two groups after cross-over of regimen.

Figure 15 shows the relative risk of responding to CAP versus paclitaxel by response to initial treatment, whilst figure 16 displays the relative risk of responding to CAP versus paclitaxel by response after treatment cross-over.

Figure 15. Relative risk for response to CAP versus Paclitaxel for initial treatment.

Complete response 0.57 (0.27, 1.20)

Partial response 1.08 (0.56, 2.11)

Stable disease 0.71 (0.35, 1.42)

Progressive disease 0.44 (0.20, 0.93)

0.1 0.2 0.5 1 2 5 Favours CAP Favours paclitaxel

109

Figure 16. Relative risk for response to CAP versus Paclitaxel after treatment cross- over

Complete response 0.43 (0.13, 1.42)

Partial response 0.29 (0.07, 1.14)

Stable disease 1.00 (0.39, 2.55)

Progressive disease 0.82 (0.38, 1.75)

0.01 0.1 0.2 0.5 1 2 5 Favours CAP Favours paclitaxel

d) Quality of Life This outcome was not assessed in the trial, and therefore no data were reported. e) Adverse events The median number of treatment cycles administered was six on both trial arms. The major reason for stopping treatment was disease progression. Table 29 and Figure 17 display the maximum grade toxicities experienced during treatment.

110 Table 29. The maximum grade toxicities during treatment with paclitaxel compared with CAP

Toxicity CAP (n=47) Paclitaxel (n=47) No. % No. % p-value Leucopenia (grade 3-4) 16 34 2 4 0.001 Neutropenia (grade 3-4) 17 36 6 13 0.009 Thrombocytopenia (grade 3-4) 6 13 - - 0.012 Nausea and vomiting (grade 2-3) 24 51 8 17 0.004 Alopecia 28 60 41 87 0.010 Allergic reactions 1 2 7 15 0.085 Sensory neuropathy (≥grade 2) 3 6 5 11 0.002 Myalgia (grade 2) 2 4 9 19 0.025 Other (cardiac toxicity, renal 1 2 4 8 >0.05 toxicity, stomatitis)

Figure 17. Relative risks of treatment related toxicities for paclitaxel compared to CAP

Leukepenia 0.125 (0.033, 0.448)

Neutropenia 0.353 (0.153, 0.781)

Thrombocytopenia 0.083 (0.008, 0.808)

Nausea and Vomiting 0.333 (0.166, 0.640)

Alopecia 1.464 (1.148, 1.954)

Allergic reactions 7.000 (1.193, 42.857)

Sensory neuropathy 1.667 (0.465, 6.058)

Myalgia (grade 2) 4.500 (1.177, 17.915)

Other 4.000 (0.629, 26.149)

0.001 0.01 0.1 0.2 0.5 1 2 5 10 100 CAP caused more adverse events Paclitaxel caused more adverse events

As is shown in Table 30 and Figure 18, the toxicity profiles were quite different between the paclitaxel and CAP treatment arms. CAP was associated with significantly higher rates of grade 3-4 haematological toxicity compared with paclitaxel: • leucopenia (34% versus 4%), RR=0.13 (95% CI: 0.033, 0.45) • neutropenia (36% versus 13%), RR=0.35 (95% CI: 0.15, 0.78) • thrombocytopenia (13% versus 0%), RR=0.083 (95% CI: 0.008, 0.808)

111 Grade 2-3 nausea and vomiting was also significantly higher in the CAP treatment group, 51% compared to 17% in the paclitaxel treated arm, RR=0.33 (95%CI: 0.17, 0.64). The rates of alopecia were high in both groups, but significantly higher rates were associated with paclitaxel treatment (87%) compared with CAP treatment (60%), RR=1.46 (95% CI: 1.48, 1.95). Allergic reactions were also significantly higher in the paclitaxel group, 15% compared to 2% in the CAP treatment arm, RR=7.00 (95% CI: 1.19, 42.86). The incidence of sensory neuropathy, myalgia and other toxicities (cardiac, renal and stomatitis) were all higher in the paclitaxel treated group than in the CAP arm, with rates of 11% versus 6%, 19% compared to 4% and 8% versus 2% respectively. All other toxicities were rare in both treatment arms.

Summary of effectiveness data for paclitaxel versus CAP

One RCT was included that compared single-agent paclitaxel and a platinum-containing regimen (CAP) in participants who had been previously treated, with recurrent ovarian cancer.

Overall survival: There was a significant difference between the two treatment groups in terms of overall survival in favour of CAP; HR=0.58 (95% CI: 0.34, 0.98). The median survival times were 34.7 months for the CAP treatment group compared with 25.8 months for the paclitaxel treatment group.

Progression-free survival: There was a statistically significant difference in progression- free survival between the CAP treatment arm and the paclitaxel treatment group in favour of CAP; HR=0.60 (95% CI: 0.37, 0.97). The median progression-free interval was 15.7 months in the CAP treatment group and 9 months in the paclitaxel treatment group.

Respons e: There were no significant differences between the treatment groups in response rates. 55% of participants in the CAP arm and 45% in the paclitaxel arm had a complete or partial response.

Quality of life: This outcome was not assessed, and therefore no data were reported.

Adverse events: Grade 3-4 haematological toxicities were reported significantly more often in the CAP treatment arm compared to the paclitaxel treatment group. Grade 2-3 nausea and vomiting was also significantly higher in the CAP treatment group. The rates of alopecia were high in both groups, but significantly higher rates were associated with paclitaxel treatment. Allergic reactions were also significantly higher in the paclitaxel group. The incidence of sensory neuropathy, myalgia and other toxicities (cardiac, renal and stomatitis) were all higher in the paclitaxel treated group than in the CAP arm. All other toxicities were rare in both treatment arms.

112 4.2.5 Effectiveness of paclitaxel in combination with platinum-based chemotherapy versus platinum-based therapy alone

The following section of the report summarises the trial by the ICON and AGO Collaborators(29)

a) Overall survival The primary outcome was overall survival which was defined as the time from randomisation to death from any cause. Patients known to still be alive at the time of the analysis were censored at the time of their last follow-up. At a median follow-up time of 42 months, 530 (66%) of participants in the trial had died. Figure 18 shows the Kaplan-Meier curves for overall survival.

Figure 18. Kaplan-Meier curves for overall survival for paclitaxel / platinum combination therapy versus platinum-based therapy alone9

1.0 p r 0.9 o p 0.8 o 0.7 r t 0.6 i o 0.5 n 0.4 a l 0.3 i v 0.2 e Pac- 0.1 Plat 0.0 01234 Patients at risk Years from randomisation 392 306 167 96 43 Pac-Plat Plat 410 295 150 68 33

The survival curves indicated a HR of 0.82 (95% CI: 0.69, 0.97) in favour of the participants in the paclitaxel and platinum-based combination group relative to the platinum-based monotherapy group for overall survival. This translates into an absolute difference in 2-year survival of 7% in favour of the paclitaxel combination therapy compared to the conventional monotherapy treatment group (57% versus 50%

9 Reporduced from powerpoint presentation by Lederman, J.A on behalf of all ICON & AGO Collaborators; ASCO Gynecologic care: Advances in management. 1st June 2003.

113 [95% CI for difference: 1%, 12%]). For median survival the hazard ratio translates into a difference of 5 months (29 versus 24 months [95% CI for difference: 1, 9]).

Effect of paclitaxel plus platinum chemotherapy on overall survival in sub- groups An analysis was undertaken to explore the effect of the two different treatment regimens on survival in the different sub-groups defined by randomisation group, time since completion of last chemotherapy, number of previous lines of chemotherapy, previous exposure to taxanes, intended platinum agent (carboplatin or cisplatin), age and WHO performance status. This analysis indicated that there were no significant differences between the paclitaxel plus platinum chemotherapy treatment group and the conventional treatment group in any of the sub-groups in relation to overall survival. However, many of the sub-groups were small and therefore may have lacked the power to detect any real differences between the groups. There were however, a number of statistically insignificant trends and interactions observed within the sub-group data. Trends were observed within the sub-groups of age (<55, 55-65, >65) and the number of previous lines of chemotherapy (1, 2, >2). The interactions were observed within the sub-groups of randomisation group (ICON4 MRC CTU, ICON4 Italy, AGO), WHO performance (WHO = 0, WHO >0), intended platinum agent (carboplatin, cisplatin), time since completion of last chemotherapy cycle in months (≤12, >12) and previous exposure to taxanes (yes, no). b) Progression-free survival The secondary outcomes included progression-free survival (PFS) which was defined as the time from randomisation to first appearance of progressive disease or death from any cause. Again patients known to be alive and without progressive disease at the time of analysis were censored at their last follow-up. At a median follow-up time of 42 months 717 (89%) of patients had developed progressive disease or died. Figure 19 shows the Kaplan-Meier curves for progression-free survival.

114 Figure 19. Kaplan-Meier curves for progression-free survival for paclitaxel / platinum combination therapy versus platinum-based therapy alone10

1.0

0.9

0.8 Proportion 0.7 alive and progression- 0.6 free 0.5

0.4

0.3 0.2 Pac-Plat 0.1 Plat 0.0 01234 Years from randomisation Patients at risk Pac-Plat 392 179 52 25 17 Plat 410 157 45 17 7

The survival curves showed a HR of 0.76 (95% CI: 0.66, 0.89) in favour of paclitaxel

plus platinum chemotherapy relative to platinum-based chemotherapy alone for

progression-free survival. This translated into an estimated absolute difference in 1-

year progression-free survival of 10% (40% versus 50% [95% CI for difference: 4%,

15%]). Further analysis by the trial collaborators indicated that the results translated

into an absolute difference in median progression-free survival of 3 months in favour

of the combination chemotherapy group relative to the conventional monotherapy

group (9 months versus 12 months ([(95% CI for difference: 1, 4]).

Effect of paclitaxel plus platinum chemotherapy on progression-free survival in

sub-groups

An analysis was undertaken to explore the effect of the two different treatment

regimens on progression-free survival in the different sub-groups. These were again

defined by randomisation group (ICON4 MRC CTU, ICON4 Italy, AGO), WHO

10 10 Reporduced from powerpoint presentation by Lederman, J.A on behalf of all ICON & AGO Collaborators; ASCO Gynecologic care: Advances in management. 1st June 2003.

115 performance (WHO = 0, WHO >0), intended platinum agent (carboplatin, cisplatin), time since completion of last chemotherapy cycle in months (≤12, >12) and previous exposure to taxanes (yes, no). Once again the analysis showed that there were no significant differences between the paclitaxel plus platinum chemotherapy treatment group and the conventional treatment group in terms of progression-free survival in any of the sub-groups. There were again, however, a number of statistically insignificant trends and interactions shown within the sub-group data. Trends were observed within the sub-groups of age and the number of previous lines of chemotherapy regimens. The interactions were observed within the sub-groups of randomisation group, WHO performance , intended platinum agent, time since completion of last chemotherapy cycle in months and previous exposure to taxanes. c) Response rate There were no statistically significant differences observed between the two treatment groups in relation to response rate, defined as a complete or partial response. However, 66% (78/119) of participants in the paclitaxel plus platinum chemotherapy group had a response to treatment compared with 54% (69/128) of participants in the conventional monotherapy group. This showed a difference of 12% (95% CI: 0.1%, 24%) in response rate in favour of the paclitaxel plus platinum group. d) Quality of Life (QoL) Quality of life was assessed using the EORTC QLQ-C30 questionnaire. In total 482 (90%) of 536 patients completed the questionnaire at baseline, before receiving any study drug. All the scales were balanced across the two treatment groups at this time. The majority of the participants had little or no functional difficulties and few moderate or severe symptoms at baseline.

In the first 6 months post randomisation there were no significant differences between the treatment groups for the worst scores or areas under the curve for all of the five functional scales, for eight out of the nine symptoms scales, and for the global health

116 status. There was a significant difference observed between the two groups on the nausea and vomiting symptom scale. This symptom was worse for the participants receiving the conventional platinum-based monotherapy (p = 0.0014 for worst score; p=0.005 for area under the curve). However, this difference appeared to be transient, lasting only for the first 15 weeks of treatment. e) Adverse events

Data were only reported for the moderate and severe toxic effects experienced during treatment. The incidence of grade 2-4 neurological toxicities was significantly higher in the paclitaxel plus platinum chemotherapy regimen group compared to the conventional platinum-based therapy group (20% versus 1%), RR=19.10 (95% CI: 7.37, 49.93). The rates of alopecia were also significantly higher in this group (86%) compared to the conventional monotherapy regimen group (25%), RR=3.47 (95% CI: 2.92, 4.17).

The rates of haematological toxicity were significantly higher in the conventional platinum chemotherapy group. In this group, 46% of participants experienced haematological toxicity compared to 29% in the combination therapy group, RR=0.63 (95% CI: 0.52, 0.75). The incidence of grades 2-4 nausea and vomiting was also higher in the conventional platinum-based therapy group, 40% compared to 35% in the combination therapy group. However, this difference in the rates was not statistically significant. The rates of other reported toxic effects (infection, mucositis and renal) were low in both treatment groups.

Table 30 shows the incidence of moderate or severe toxic effects experienced in both the treatment groups, whilst Figure 20 displays the relative risks of experiencing a moderate or severe toxicity during treatment.

117 Table 30. The proportion of patients in each treatment group who experienced moderate or severe toxic effects during treatment along with the RR and 95% CI

Toxic effect Conventional Paclitaxel plus Relative 95% CI platinum based platinum risk chemotherapy chemotherapy (RR) (n=410) (n=392) Neurological (grade 2-4) 4 (1%) 76 (20%) 19.10 7.37, 49.93 Not yet known 31 15 Haematological* 182 (46%) 111 (29%) 0.63 0.52, 0.75 Not yet known 16 8 Infection* 53 (14%) 64 (17%) 1.24 0.89, 1.73 Not yet known 24 15 Renal* 37 (9%) 31 (8%) 0.86 0.55, 1.35 Not yet known 16 8 Mucositis (grade 2/3) 21 (6%) 26 (7%) 1.24 0.72, 2.16 Not yet known 31 15 Nausea & vomiting (grade 153 (40%) 131 (35%) 0.87 0.72, 1.04 2-4) Not yet known 29 15 Alopecia (grade 2-4) 95 (25%) 322 (86%) 3.47 2.92, 4.17 Not yet known 28 19

Figure 20. Relative risks of moderate to severe toxicity for paclitaxel plus platinum chemotherapy compared to conventional platinum based therapy alone.

Neurological 19.10 (7.37, 49.93)

Haematological 0.63 (0.52, 0.75)

Infection 1.24 (0.89, 1.73)

Renal 0.86 (0.55, 1.35)

Mucositis 1.24 (0.72, 2.16)

Nausea & Vomiting 0.87 (0.72, 1.04)

Alopecia 3.47 (2.92, 4.17)

0.5 1 2 5 10 100

Platinum alone adverse events Paclitaxel / platinum adverse events

118

Summary of the effectiveness of paclitaxel in combination with platinum-based chemotherapy versus platinum-based therapy alone

Overall survival: There was a significant difference in overall survival observed between the two groups in favour of the paclitaxel and platinum-based combination group, HR=0.82 (95% CI: 0.69, 0.97). This showed an absolute difference in the 2-year survival rates of 7% in favour of this group, 57% versus 50%. For median survival the HR translated into a difference of 5 months, 29 versus 24 months.

There were no significant differences between the two treatment regimens in any of the sub-groups in relation to overall survival.

Progression-free survival: There was a significant difference in progression-free survival rates between the two treatment regimens that favoured the paclitaxel plus platinum-based chemotherapy group relative to the platinum-based chemotherapy group; HR=0.76 (95% CI: 0.66, 0.89). This translated into an estimated absolute difference in 1-year progression-free survival of 10% ( 40% versus 50%), and an absolute difference in median progression-free survival of 3 months in favour of the combination therapy group (9 months versus 12 months)

There were no significant differences between the two therapy regimen groups in terms of progression-free survival in any of the sub-groups.

Response rate: There were no statistically significant differences between the two treatment groups in response rates, complete or partial response. 66% of the participants in the paclitaxel plus platinum chemotherapy group had a response compared with 54% in the conventional monotherapy group. This showed an insignificant difference of 12% between the two treatment groups.

Quality of Life: In the first 6 months post randomisation there were no significant differences between the two treatment groups for the worst scores or areas under the curve for the functional scales, the global health status or 8 out of 9 symptoms scales. There was a significant difference between the groups on the nausea/vomiting sub-scale which was transient, lasting for the first 15 weeks of treatment only. This symptom was worse for the participants in the conventional platinum-based monotherapy group.

Adverse events: The rates of grade 2-4 neurological toxicities and alopecia were significantly higher in the paclitaxel plus platinum chemotherapy regimen group compared to the conventional monotherapy group. The incidence of haematological toxicity was significantly higher in the conventional platinum-bases monotherapy group compared to the combination therapy group. The incidence of grades 2-4 nausea and vomiting was also higher in this group, but this was not statistically significant.

119 4.2.6 Effectiveness of single-agent paclitaxel versus oxaliplatin

The following section of the report summarises the trial by Piccart et al.(30) a) Overall survival At the time of cut-off 45 participants had died out of the 86 randomised in total, 25 / 41 in the paclitaxel arm and 20 I / 45 in the oxaliplatin treatment arm. The estimated median overall survival was 37 weeks in the paclitaxel treatment arm compared with 42 weeks in the oxaliplatin treatment group. b) Time to progression The median time to progression in the paclitaxel treated participants was 14 weeks compared to 12 weeks in the oxaliplatin arm. At the time of analysis 9/41 participants in the paclitaxel arm had not progressed relative to 8/45 in the oxaliplatin treatment group. In both treatment arms the median times to treatment failure were comparable, 13 weeks for the paclitaxel arm and 12 weeks for the oxaliplatin treatment arm. c) Response rate The primary efficacy endpoint in this trial was the confirmed response rate. Confirmed response, verified by two independent radiologists, was defined as partial response (PR) or complete response (CR) that was observed in at least two consecutive evaluations that were at least 4 weeks apart. Overall response rate was defined by the total number of participants in each treatment arm.

All participants who received at least two treatment cycles were considered by the investigators to be assessable for response. A total of 5 participants, 2 in the paclitaxel arm and 3 in the oxaliplatin arm were not assessable. Four of these participants were ineligible to be included in the trial and one died 6 days after the first dose of oxaliplatin due to causes unrelated to treatment. Table 31 shows the response to treatment with paclitaxel versus oxaliplatin in the total treatment population.

120 Table 31. Response to treatment with paclitaxel versus oxaliplatin in the total treatment population Paclitaxel (n=41) Oxaliplatin (n=45) Response category No. of participants % No. of participants % CR 0 0 0 0 PR 7 17 7 16 NC 18 44 20 44 PD 14 34 15 33 Not assessable 2 5 3 7 Total 41 100 45 100 Abbreviation: NC, no change

Overall the results showed that there were 7 partial responses to treatment in each of the trial arms. The response rate in the paclitaxel arm was 17% (95% CI: 7%, 32%) compared with 16% (95% CI: 7%, 29%) in the oxaliplatin treated group. In the participants that were categorised as potentially platinum-sensitive 2 out of 10 (20%) in the paclitaxel treatment group, and 5 out of 13 (38%) in the oxaliplatin arm had a PR. In the participants that were categorised at baseline as being platinum-resistant 5 out of 31 (16%) in the paclitaxel arm compared with 2 out of 32 (6%) in the oxaliplatin group had a PR.

Response duration The median duration of response was 33 weeks in the paclitaxel treated participants compared with 31 weeks in the oxaliplatin treatment arm. At the time of cut-off 8 out of the 14 participants who had achieved a PR had progressed, 3/7 in the paclitaxel arm and 5/7 in the oxaliplatin arm. d) Quality of Life

Sixty-six participants completed the QoL assessment at baseline, but only 47 completed questionnaires at the end of the second treatment cycle (6 weeks) and 31 at the end of the fourth treatment cycle (12 weeks). The mean QoL score increased by more than 10 points between baseline and cycle 4 for the participants in the paclitaxel treatment group, regardless of study withdrawal. In the oxaliplatin group, the mean QoL score decreased, but by less than 10 points through cycle 2, and then returned to baseline levels thereafter for the majority of participants.

121 e) Adverse events All of the 86 participants in the trial received at least one treatment cycle and therefore were assessable for the toxicity analysis. A summary of the haematological and non-haematological toxicities associated with treatment are given in Table 32.

Table 32. Severe (Grade 3 & 4) toxicity by treatment group for paclitaxel compared to oxaliplatin. Paclitaxel (n=41) Oxaliplatin (n=45) Toxicity Grade 3 Grade 4 Grade 3 Grade 4 No. % No. % No. % No. % Haematological Neutropenia 6 15 3 7 - - - - Anaemia - - 1 2 1 2 - - Thrombocytopenia - - - - 2 4 - - Liver function AST ------ALT 2 5 ------Gastrointestinal Nausea 1 2 NA NA 2 4 NA NA Vomiting 1 2 - - 3 7 - - Diarrhoea - - - - 2 4 - - Neurosensory 3 7 NA NA 4 9 NA NA Other Lethargy 3 7 NA NA 3 7 NA NA Pain 5 12 - - 2 4 - - * NA = not applicable

Haematological toxicities: severe neutropenia (grade 3 and 4) occurred in the paclitaxel treatment group in 9 patients (22%), whereas grade 3 thrombocytopenia occurred only in 2 patients (4%) in the oxaliplatin treatment arm. Severe anaemia occurred once in each treatment group. In the paclitaxel group it was rated as being of grade 4 severity and in the oxaliplatin group of grade 3.

Gastrointestinal: All instances of nausea and vomiting were rated as being mild to moderate (grade 3). One occurrence of nausea and one of vomiting occurred in the paclitaxel treatment arm in comparison with two occurrences of nausea and 3 of vomiting in the oxaliplatin treatment group. Diarrhoea was only reported in the oxaliplatin treatment group, of which there were 2 reports.

Neurosensory: Grade 3 neurosensory toxicity was reported in 3 participants in the paclitaxel arm and in 4 participants in the oxaliplatin treatment group.

122 Lethargy and pain: Lethargy was reported in 3 participants in each of the treatment groups. Pain was reported in 5 participants in the paclitaxel treatment arm compared to 2 participants in the oxaliplatin treatment arm.

Summary of the trial comparing single-agent paclitaxel and oxaliplatin One phase II RCT was identified that evaluated the efficacy of oxaliplatin compared to single-agent paclitaxel in a relapsing progressive ovarian cancer patient population.

Overall survival: There were significant differences between participants treated with paclitaxel and those treated with oxaliplatin in overall survival. The estimated median overall survival was 37 weeks in the paclitaxel treatment arm compared with 42 weeks in the oxaliplatin treatment group.

Time to progression: The median time to progression was 14 weeks in the paclitaxel treatment group compared to 12 weeks in the oxaliplatin arm. This difference was not statistically significant. There were also no differences between the two treatment groups in terms of time to treatment failure. This was 13 weeks in the paclitaxel arm and 12 weeks in the oxaliplatin treatment arm.

Response rate: There were no complete responses in either of the treatment groups. In the paclitaxel group 17% of participants had a partial response compared to 16% in the oxaliplatin group.

Response duration: There were no significant differences between the treatment groups for response duration. The median response duration was 33 weeks in the paclitaxel arm compared with 31 weeks in the oxaliplatin arm.

Quality of Life: The mean QoL score increased by more than 10 points between baseline and cycle 4 for participants in the paclitaxel treatment group. In the oxaliplatin group, the mean QoL score decreases slightly through cycle 2, and then returned to baseline levels. These data from the QoL outcome assessment, should be interpreted with caution given the very small number of participants upon which in it based.

Adverse events: severe neutropenia was more common in the paclitaxel treatment group compared to the oxaliplatin group (22% versus 4%). Thrombocytopenia was more common in the oxaliplatin group compared to the paclitaxel treatment arm (4% versus 0%). All other adverse events were similar between the two treatment arms.

123 4.2.7 Effectiveness of paclitaxel given weekly versus every three weeks This section of the report summarises the resultws reported in Rosenberg et al.(31) a) Overall survival The median survival for the weekly group was 13.6 months (95% CI: 10.5, 18.7) compared to 14.7 months (12.3, 19.1) for the every 3 weeks group. This difference in the median survival times between the two treatment regimens was not statistically significant. b) Time to progression There were no statistically significant differences between the weekly and every 3 weeks group in terms of time to progression. The median time to progression was 6.1 months (95% CI: 5.0, 8.0) in the weekly group versus 8.1 months (95% CI: 6.4, 9.7) for the every 3 weeks group. c) Response The primary outcome in the study was response rate, which was categorised according to WHO tumour response criteria.

There were no significant differences between the two groups in terms of response rate. Thirty-seven participants in the weekly group and 38 in the every 3 week group responded to treatment. Table 33 displays the tumour response rate for all participants in the ITT analysis by group.

Table 33. Response rate by treatment group for weekly versus three weekly paclitaxel Response Paclitaxel 67mg/m2/week Paclitaxel 200 mg/m2/3 wks (n=105) (n=103) CR1 13 (12.4% 17 (16.5%) PR2 24 (22.8%) 21 (20.4%) Overall response rate 37 (35.2%) 38 (36.9%) SD 43 (41.0%) 33 (32.0%) PD 15 (14.3%) 19 (18.5%) NE 9 (8.6%) 11 (10.7%) Not treated 1 (0.9%) 2 (1.9%) 1 Three participants in each group had unconfirmed complete response. 2 Seven and six participants in the weekly and every three week group respectively had unconfirmed partial response.

124 Response duration The overall median response duration did not differ significantly between the two treatment groups. The median response duration was 9.4 months (95% CI: 6.2, 13.9) for the weekly group compared to 12.4 months (95% CI: 9.1, 14.3) for the every 3 weeks group (p=0.57). The complete response duration was also shorter in the weekly group at 4.5 months (95% CI: 3.6, 10.7) compared with 7.8 months (95% CI: 4.2, 10.2) in the every 3 weeks group. However, again this did not reach a level of statistical significance (p=0.84). d) Quality of Life This outcome was not assessed in the trial, and therefore no data were reported. e) Adverse events The safety analysis was based on 205 participants who received at least one dose of paclitaxel. There were no treatment-related deaths reported in the trial. Table 35 shows the adverse events reported in the trial.

Haematologic toxicity: Significantly more participants in the paclitaxel every 3 weeks group experienced grade 3 or 4 neutropenia, 45/101 (45%) compared to 19/104 (18%) in the weekly group. Thrombocytopenia was uncommon in both of the treatment groups.

Non-haematological toxicity: Grade 1-3 Grade 1-3 non-haematological toxicity was frequently observed in both of the treatment arms. Neuropathy occurred in 84/104 (81%) of participants in the weekly treatment group and in 86/101 (85%) of the participants in the every 3 weeks group. Alopecia occurred in 85/104 (82%) and 91/101 (90%) of participants in the weekly and 3 weekly groups respectively. Whilst arthralgia/myalgia occurred frequently in both of the groups, it occurred significantly more often in the every 3 weeks group 85/101 (84%) compared to the weekly group 61/104 (59%), RR=1.30 (95% CI: 1.17, 1.69) Problems with nail changes (discoloration and/or loosening from the nail bed) occurred significantly more often in the weekly treatment arm 37/104 (36%) compared to the every 3 weeks treatment arm 2/101 (2%). Participants in both of the

125 treatment arms reported nausea and vomiting relatively frequently. In the weekly group 48/104 (46%) and in the every 3 weeks group 42/101 (42%) reported this as a side effect of treatment. Grade 3 Grade 3 neuropathy was observed significantly more frequently in the every 3 weeks treatment arm 29/101 (29%) compared to the weekly treatment group, 11/104 (11%), RR=2.71 (95% CI: 1.46, 5.12). Likewise, alopecia was also observed significantly more often in this group, 80/101 (79%) compared to 48/104 (46%) in the weekly treatment arm. Problems with nail changes were observed significantly more often in the weekly treatment group 9/104 (9%) compared with no occurrences of this problem in the 3 weekly treatment group. There were no significant differences between the two arms in relation to the number of episodes of nausea/vomiting or arthralgia/myalgia.

Table 34. Toxicity events for weekly versus three weekly paclitaxel

Paclitaxel 67 Paclitaxel 200 mg/m2/3 wks p-value mg/m2/week (n=104) (n=101)

Haematological toxicity (lowest value per patient) Grade 1-4 Haemoglobin 81 (78%) 65 (64%) 0.04 WBC 74(71%) 79 (78%) 0.27 Neutrophils 63 (61%) 80 (79%) <0.01 Platelets 1 (1%) 5 (5%) 0.12 Grade 3-4 Haemoglobin 4 (4%) 4 (4%) 1.0 WBC 17 (16%) 17 (17%) 1.0 Neutrophils 19 (18%) 45 (45%) <0.001 Platelets 0 1 (1%) 0.49 Non-haematological toxicity (worst value per patient) Grade 1-3 Neuropathy 84 (81%) 86 (85%) 0.72 Alopecia 85 (82%) 91 (90%) 0.11 Arthralgia/Myalgia 61 (59%) 85 (84%) <0.001 Nausea/Vomiting 48 (46%) 42 (42%) 0.57 Nails 37 (36%) 2 (2%) <0.001 Grade 3 Neuropathy 11 (11%) 29 (29%) <0.001 Alopecia 48 (46%) 80 (79%) <0.001 Arthralgia/Myalgia 5 (5%) 8 (8%) 0.40 Nausea/Vomiting 4 (4%) 3 (3%) 1.0 Nails 9 (9%) 0 <0.01

126 Hypersensitivity reactions The number and severity of hypersensitivity reactions were assessed for the 106 participants who received the standard pre-medication schedule of oral steroids 12 and 6 hours prior to paclitaxel, and the 99 participants who received parenteral steroids 30 minutes prior to paclitaxel administration. There were no statistically significant differences between the two groups, either overall or for grade 3-4 reactions. The paclitaxel infusion was discontinued in approximately 3% of participants in both arms due to hypersensitivity reactions. Table 35 shows the number of hypersensitivity reactions experienced in both treatment arms.

Table 35. Hypersensitivity reactions for weekly versus three weekly paclitaxel

Oral steroids 12 and 6 Parenteral steroids 30 hours prior to paclitaxel minutes prior to paclitaxel (n=106) (n=99) Grade 1-4 Skin 17 (16.0%) 18 (18.2%) Generalised urticaria 1 (0.9%) 0 Dyspnea 4 (3.8%) 4 (4.0%) Respiratory distress requiring 1 (0.9%) 1 (1.0%) treatment Hypotension 1 (0.9%) 1 (1.0%) Grade 3-4 Skin 2 (2.8%) 4 (4.0%) Generalised urticaria 1 (0.9%) 0 Dyspnea 0 1 (1.0%) Respiratory distress requiring 1 (0.9%) 0 treatment

127 Summary of the trial of paclitaxel given weekly versus every 3 weeks

One RCT was included that evaluated the efficacy and toxicity of paclitaxel given at the same dose intensity and administered weekly or every 3 weeks in women who had received one prior platinum-containing chemotherapy regimen without a taxane.

Overall survival: There were no significant differences between the two treatment regimens in terms of overall survival. The median survival was 13.6 months in the weekly group compared with 14.7 months in the every 3 weeks group.

Time to progression: The median time to progression was 6.1 months in the weekly group compared to 8.1 months in the every 3 weeks group. This difference between the two treatment groups was not statistically significant.

Response: There were no significant differences between the two treatment regimens on this outcome measure. 35% of participants in the weekly regimen group responded to treatment compared with 35% in the every 3 weeks group.

Response duration: The overall median response duration did not differ significantly between the two treatment groups. Response duration was 9.4 months in the weekly group compared with 12.4 months in the every 3 weeks group. Complete response duration also did not differ significantly. This was 4.5 months in the weekly group compared with 7.8 months in the every

3 weeks group.

Quality of Life: Quality of life was not assessed as an outcome measure in this trial.

Adverse events: Grade 3 and 4 neutropenia and alopecia were significantly more common in the every 3 weeks group compared to the weekly group. Problems with nail changes were observed significantly more often in the weekly group compared with the every 3 weeks group.

4.2.8 Effectiveness of paclitaxel 175 mg/m2 compared to 250 mg/ m2 The following section of the report summarises the results of the trial by Omura et al(32) a) Overall survival: Overall survival was defined as the time from randomisation until the date of death, or last contact if the date of death was unknown.

There were no statistically significant differences between the two treatment groups in terms of overall survival. The estimated median survival time for participants in the 175 mg/ m2 group was 13.1 months compared with 12.3 months for participants in the 250 mg/ m2 treatment arm. The estimated HRs, from a proportional hazards model that included co-variate adjustments for initial performance status, platinum sensitivity, co-operative group and measurable disease, showed that the adjusted survival rate was 3% lower on the 250 mg/ m2 regimen; HR=0.972 (95% CI: 0.774, 1.22).

128 b) Progression-free survival: There were no significant differences observed between the two treatment regimens of 175 mg/ m2 paclitaxel and 250 mg/ m2 in terms of time to progression. The median time to progression was 4.8 months for participants treated with the lower dose regimen compared to 5.5 months for participants in the higher dose treatment group. c) Response rate: For participants with measurable disease 36 out of 131 (27%) (95% CI: 20%, 36%) in the 175 mg/ m2 regimen arm demonstrated an objective response (partial and complete response) compared to 49 out of 134 (36%) (95% CI: 29%, 46%) in the 250 mg/ m2 regimen treatment group. The overall odds ratio of responding to the higher-dose regimen of 250 mg/ m2 was 1.89 (95%CI: 1.07, 3.31); p=0.027, times greater than the odds of responding to the low-dose regimen, after adjusting for histologic cell type (papillary serous versus clear-cell or mucinous versus other cell types), co-operative group, performance status, and prior platinum sensitivity. However, further analysis indicated that the odds ratio of responding to the 250 mg/ m2 regimen was not consistent across the sub-groups of participants as defined by their classification of platinum sensitivity at baseline. There was a significant treatment - sub-group interaction (p=0.041). The adjusted relative odds ratio of responding to the 250 mg/ m2 regimen was 2.59 (95% CI: 1.36, 4.95) for the 213 participants who had platinum-resistant disease, whilst the odds ratio for the 52 participants with platinum-sensitive disease was 0.630 (95% CI: 0.191, 2.07). The response rates by treatment group and prior platinum sensitivity are displayed below in Table 36.

129 Table 36. Response rates by sensitivity to prior platinum therapy for paclitaxel 175 mg/ m2 versus 250 mg/m2 Paclitaxel dose 175 mg/2 (n=104) 250 mg/m2 (n=109) Total no. of patients No. of % No. of % Response patients patients Platinum-resistant Complete response 5 5 13 12 18 Partial response 18 17 27 25 45 No response 81 78 69 63 150 Sub-total 104 109 213 Platinum-sensitive Complete response 4 15 4 16 8 Partial response 9 33 5 20 14 No response 14 52 16 64 30 Sub-total 27 25 52 Total 131 134 265

Within the participants assigned to the high-dose 250 mg/m2 regimen there were no significant differences between the number of participants assigned to 5 µg/kg filgrastim who responded and those assigned to the 10 µg/kg regimen. Twenty-four (35%) of 68 participants (95% CI: 24%, 48%) assigned to the 5 µg/kg filgrastim regimen responded compared to 25 (37.9%) of 66 participants (95% CI: 26%, 51%) assigned to the 10 µg/kg dose. d) Quality of Life This outcome was not assessed in the trial, and therefore no data were reported. e) Adverse events Adverse events were assessed in all participants who received any study treatment.

There were no significant differences in terms of the rates of neutropenic fever after the first cycle of treatment between the participants in the 175 mg/m2 dose regimen group without filgrastim and the 175 mg/m2 dose group with filgrastim. Twenty-two percent of participants in the lower dose arm experienced neutropenic fever compared with 19% of participants in the 250 mg/m2 dose regimen group.

Neutropenic fever after the first course of therapy occurred in 19% (16 of 83 participants) and 18% of participants (15 of 82 participants) on the 5µg/kg and 10µg /kg regimens, respectively. The 95% CI for the difference between the two filgrastim

130 doses for the number of cases of neutropenia after the first course of therapy were – 11% and 13%.

The incidence of other reported toxicities was relatively rare, but was always higher in the in the 250 mg/m2 group compared to the 155 mg/m2 regimen arm. Grade 3 and 4 thrombocytopenia (15% versus 7%), neuropathy (16% versus 7%) and myalgia/arthralgia (10% versus 3%) were all reported significantly more often on the 250 mg/m2 regimen with filgrastim relative to the lower dose treatment group. However, comparing 5 and 10 µg/kg of filgrastim, there were no significant differences in these toxicities. Table 37 displays the incidence of Grade 3 or 4 toxicities other than neutropenia.

Table 37. Incidence of Grade 3 or 4 Toxicity other than Neutropenia.

Paclitaxel regimen Toxicity 175 mg/m2 (%) 250 mg/m2 + G-CSF p-value (%) Anaemia 7 15 0.102 Thrombocytopenia 5 15 0.009 Nausea and vomiting 5 10 0.211 Neuropathy 7 16 0.024 Myalgia/arthralgia 3 10 0.022

131

Summary of the effectiveness of paclitaxel 175 mg/m2 compared to 250 mg/m2

One RCT was included in the assessment of clinical effectiveness that assessed whether increasing the dose of paclitaxel increases the probability of response, progression-free survival or overall survival in women who have persistent or recurrent ovarian cancer, and whether doubling the dose of prophylactic filgrastim accompanying the higher paclitaxel dose decreases the frequency of neutropenic fever.

Overall survival: There were no significant differences in overall survival between the two different dose regimen groups; HR=0.972 (95% CI: 0.774, 1.22) The median survival time 2 2 was 13.1 months in the 250 mg/m group compared with 12.3 months in the 175 mg/m regimen group.

Progression-free survival: The median time to progression was 4.8 months for the lower dose regimen group compared with 5.5 months for participants in the higher dose treatment arm. This difference was not statistically significant.

Response rate: 27% of participants in the 175 mg/m2 group demonstrated an objective response compared to 36% in the 250 mg/m2 arm. This difference was statistically significant. The OR of responding to the higher-dose regimen was 1.89 times greater than the odds of responding to the low-dose regimen. This was not consistent across sub-groups of participant platinum-sensitivity. The adjusted OR of responding to the 250 mg/m2 regimen was 2.59 for participants with platinum-resistant disease, and 0.630 for participants with platinum-sensitive disease.

Quality of Life: Quality of life was not assessed in this trial.

Adverse events: Grade 3 and 4 thrombocytopenia, neuropathy and myalgia/arthralgia were all significantly more common in the 250 mg/m2 treatment arm compared to the lower dose regimen arm. There were no significant differences between the groups at the higher dose level that received either 5 or 10 µg/kg of filgrastim in terms of these toxicities.

132

4.2.9 Effectiveness of oral versus intravenous topotecan

The following section summarises the results reported in the publication by Gore et al.(2002)(33) a) Overall survival: There was a statistically significant benefit in favour of i.v. topotecan for overall survival compared to participants treated with oral topotecan. The median survival was 58 weeks (range: 0.3-120.0) in the i.v. topotecan arm compared to 51 weeks (range: 1.6- 109.0) in the oral topotecan arm. This corresponds to a risk ratio of death (oral versus i.v.) of 1.361 (95% CI: 1.001, 1.850) in favour of the i.v. topotecan treatment group. b) Time to progression: There were no statistically significant differences observed between the two treatment groups for time to progression. The median time to progression was 13 weeks (range: 1.6-76.6) for the oral topotecan treated group compared to 17 weeks (range: 0.1-91.6) for the i.v. treated group. c) Response: There were no statistically significant differences observed between the two treatment groups in terms of partial or complete response. Seventeen participants (13%) (95% CI: 7.6%, 19.1%) in the oral topotecan group compared to 26 (20%) (95% CI: 13.0%, 26.7%) in the i.v. topotecan arm responded to treatment. Out of these responses, 2 participants in the oral treatment arm compared to 4 participants in the i.v. arm obtained a complete response. Table 38 shows the response to treatment by oral and i.v formulations.

Table 38. Response to oral topotecan versus i.v. topotecan

Response Treatment group Oral topotecan (n=135) i.v. topotecan (n=131) Complete response (CR) 2 (1%) 4 (3%) Partial response (PR) 15 (11%) 22 (17%) Overall Response (OR) 17 (13%) 26 (20%) 95% CI 7.6%, 19.1% 13%, 26.7% Stable disease (SD) 39 (29%) 35 (27%) Progressive disease (PD) 65 (48%) 59 (45%) Not evaluated 14 (10%) 11 (8%)

133 Further analysis of the response rates according to the baseline stratifying factors of platinum-sensitivity, tumour size and first-line treatment was undertaken. The results indicated that there were no significant differences between the two treatment arms in the response rates for participants who were classified as platinum-refractory, - resistant, or –sensitive at baseline. Furthermore, no differences in terms of treatment activity was observed between the two groups when this was assessed according to initial tumour size (<5 cm versus >=5 cm) or first-line therapy. The results of the response to topotecan according to the baseline stratifying factors are displayed in Table 39.

Table 39. Response to oral and i.v. topotecan according to baseline stratifying factors

Treatment group Oral topotecan (n=135) i.v. topotecan (n=131) n Response (No) n Response (No) Platinum-sensitivity Refractory 40 3 (8%) 39 2 (5%) Resistant 37 3 (8%) 36 4 (11%) Sensitive 58 11 (19%) 56 20 (36%) Tumour size <5cm 66 10 (15%) 65 15 (23%) ≥5cm 68 7 (10%) 61 10 (16%) First-line treatment Platinum/paclitaxel 53 8 (15%) 54 12 (22%) All patients 135 17 (13%) 131 25 (19%)

Response duration: There were no statistically significant differences between the two treatment groups observed for response duration. The median response duration was 34 weeks (range: 13.1-62.3) for participants in the oral topotecan arm compared to 26 weeks (range: 6.6-52.7) for participants in the i.v. treatment group.

Time to response: The median time to response was 12 weeks (range: 5.6-18.1) in the oral topotecan treated participants compared to 8 weeks (range: 5.1-25.4) in the i.v. treated participants. This difference was not statistically significant. d) Quality of Life This outcome was not assessed in the trial, and therefore no data were reported.

134 e) Adverse events

Haematological toxicity There were seven deaths in the trial that were reported as being due to haematological toxicity. Two of these occurred in the oral treatment group and five in the i.v. treatment group.

The most common haematological toxicities that occurred in both of the treatment groups were neutropenia and leucopenia, although these occurred more frequently in the i.v. topotecan group compared to the oral topotecan arm. Grade 4 neutropenia occurred in 50% of the participants in the oral treatment group compared to 81% of the participants in the i.v. treatment arm, RR=1.69 (5% CI: 1.42, 2.06). Grade 4 leucopenia also occurred more frequently in the i.v. treatment group, although this was not statistically significant. Leucopenia was reported in 21% of the participants in the oral treatment arm compared to 31% in the i.v. treatment group, RR=1.47 (95% CI: 0.97, 2.24) The occurrence of Grade 4 thrombocytopenia was similar between the two treatment arms. Thrombocytopenia occurred in 20% of participants in the oral treatment arm compared to 18% in the i.v. treatment group. Anaemia occurred in 4% of participants in the oral topotecan treatment group compared to 8% in the i.v. arm. Table 40 shows the Grade 3 / 4 haematological toxicities by treatment group and Figure 21 .displays the relative risks for these toxicities.

Table 40. Grade 3 and 4 haematological toxicities, per patient and per course by oral versus i.v. topotecan.

Treatment group Oral topotecan i.v. topotecan Grade 3 Grade 4 Grade 3 Grade 4 Patients (n=135) (n=131) Neutropenia 40 (30%) 67 (50%) 15 (11%) 110 (84%) Leucopenia 59 (44%) 28 (21%) 78 (60%) 40 (31%) Thrombocytopenia 30 (22%) 27 (20%) 27 (21%) 23 (18%) Anaemia 51 (38%) 5 (4%) 43 (33%) 10 (8%) Courses (n=729) (n=778) Neutropenia 190 (36%) 106 (15%) 249 (32%) 393 (51%) Leucopenia 163 (22%) 31 (4%) 371 (48%) 68 (9%) Thrombocytopenia 70 (10%) 42 (6%) 90 (12%) 29 (4%) Anaemia 85 (12%) 7 (1%) 78 (10%) 10 (1%)

135 Figure 21. Relative risks of Grade 3 or Grade 4 haematological toxicity for oral versus i.v. topotecan

Grade 3: Neutropenia 0.39 (0.22, 0.66)

Grade 3: Leucopenia 1.36 (1.08, 1.74)

Grade 3: Thrombocytopenia 0.93 (0.59, 1.46)

Grade 3: Anaemia 0.87 (0.63, 1.20)

Grade 4: Neutropenia 1.69 (1.42, 2.06)

Grade 4: Leucopenia 1.47 (0.97, 2.24)

Grade 4: Thrombocytopenia 0.88 (0.53, 1.44)

Grade 4: Anaemia 2.06 (0.76, 5.64)

0.2 0.5 1 2 5 10 Oral topotecan Intravenous topotecan

Non-haematological toxicities The predominant non-haematological toxicities that occurred were gastrointestinal and fever. These were all higher in the oral topotecan group compared to the i.v. treatment group. The majority of these adverse events were reported as being mild to moderate in severity, with Grade 3 / 4 toxicities occurring in 10% or less of participants. In the case of Grade 3 / 4 toxicities, nausea, diarrhoea, vomiting and fever occurred more frequently in the oral topotecan group compared to the i.v. treatment group: nausea 9% versus 5%, RR=0.52 (95% CI: 0.21, 1.28); diarrhoea 10% versus 5%, RR = 0.48(95% CI: 0.19, 1.17); vomiting 7% versus 3%, RR=0.41 95% CI: 0.14, 1.21) and fever 10% versus 5%, RR = 0.52 (95% CI: 0.22, 1.20)] respectively. Table 41 shows the non-haematological toxicities by all grades and worst grades per patient for the two treatment arms.

136 Table 41. Non-haematological toxicities, all grades and worst grade per patients for oral versus i.v. topotecan. Treatment group Oral topotecan (n=135) i.v. topotecan (n=131) All grades Grades 3/4 All grades Grades 3/4 Nausea 92 (68%) 12 (9%) 80 (61%) 6 (5%) Diarrhoea 76 (56%) 13 (10%) 40 (31%) 6 (5%) Vomiting 74 (55%) 10 (7%) 52 (40%) 4 (3%) Alopecia 72 (53%) 10 (7%) 68 (52%) 8 (6%) Fatigue 50 (37%) 5 (4%) 50 (38%) 5 (4%) Abdominal pain 49 (36%) 9 (7%0 39 (30%) 9 (7%) Constipation 47 (35%) 4 (3%) 42 (32%) 2 (2%) Fever 38 (28%) 14 (10%) 31 (24%) 7 (5%))

Summary of the effectiveness of oral versus intravenous topotecan

One RCT was identified that compared the efficacy, safety and tolerability of oral topotecan versus standard intravenous topotecan in participants with relapsed epithelial ovarian cancer who had received one previous chemotherapy regiment that could have included a taxane.

Overall survival: There was a significant difference in overall survival in favour of the i.v. topotecan group. The median survival was 58 weeks in the i.v. topotecan group compared to 51 weeks in the oral topotecan arm. This corresponds to a risk ratio of death of 1.361 in favour of the i.v group.

Time to progression: The median time to progression was 13 weeks in the oral topotecan arm compared to 17 weeks in the i.v. treatment arm. This difference was not statistically significant.

Response rate: There were no statistically significant differences between the two treatment groups in terms of partial or complete response. 13% of participants in the oral topotecan group and 20% in the i.v. responded to treatment.

There was also no significant difference between the two treatment groups in terms of response in the sub-groups of participants who were classified as platinum-sensitive, refractory or resistant at baseline.

Response duration: The median response duration was 34 weeks in the oral regimen group and 26 weeks in the i.v. regimen arm. This difference in response duration was not statistically significant.

Time to response: There was no significant difference between the two treatment groups for time to response. The median time to response was 12 weeks in the oral treatment regimen group compared to 8 weeks for the i.v. treatment group.

Quality of Life: Quality of life was not assessed in this trial.

Adverse events: Two deaths occurred in the oral treatment group and 5 in the i.v. group due to haematological toxicity. Neutropenia and leucopenia occurred frequently in both treatment arms, although the rates were higher in the i.v. treatment group. The rates of grade 3 /4 nausea, vomiting, diarrhoea and fever were all significantly higher in the oral treatment regimen compared to the i.v. treatment group.

137 4.3 Other related systematic reviews

Four systematic reviews were identified that had included evaluations of PLDH, topotecan and/or paclitaxel for second line treatment of advanced ovarian cancer. The first of these reviews was a previous HTA report on the effectiveness of PLDH.(38) The report had included the trial of PLDH versus topotecan included in the present review(22) The previous report concluded that, overall there were no statistically significant differences between pegylated liposomal doxorubicin hydrochloride and topotecan with regards to the main clinical outcomes. However, statistically significant differences were observed in terms of the incidence of adverse events. The second HTA review included was a previous report on the effectiveness of topotecan. The report included both the trial of PLDH versus topotecan and the trial of topotecan versus paclitaxel included in this review. The report concluded that, there were no significant differences between topotecan and paclitaxel, and between topotecan and PLDH with regards to the main outcome measures. Statistically significant differences were however observed in terms of the incidence of adverse effects. A third review(41) that included both the trial of paclitaxel versus CAP(28) and the PLDH versus topotecan trial(22) included in this review, stated that one trial detected a statistically significant difference between treatments in progression-free survival, which was longer with CAP than with paclitaxel. However, another trial did not show a difference between PLDH and topotecan overall in women with recurrent ovarian cancer, but a sub-group analysis detected a significant survival advantage for PLDH over topotecan in women with platinum-sensitive disease. The last identified related systematic review, included the trial of topotecan versus paclitaxel(24). The report describes the results of the trial, and concludes that there are no statistically significant differences between the two comparators in terms of effectiveness.

138 4.4 ASSESSMENT OF ECONOMIC EVALUATIONS 4.4.1 Introduction

The systematic literature search detailed in Section 3.1.1 identified 17 studies that were assessed for inclusion in the cost-effectiveness review. Of these a total of 4 studies met the inclusion criteria for the cost-effectiveness review. In addition, separate submissions were received from Bristol Myers Squibb, GlaxoSmithKline and Schering-Plough Ltd.

The following sections provide a detailed overview of the cost-effectiveness evidence from each of these sources and an assessment of the quality and relevance of the data from the perspective of the UK NHS. Full data extraction tables and a quality assessment checklist are provided in Appendix 8. An overall summary of the cost- effectiveness evidence is also provided at the end of this chapter.

4.4.2 Review of Smith. (2002)

Overview The analysis by Smith(50) compared the costs of topotecan with PLDH from a US and UK perspective. A cost-minimisation analysis was undertaken, assuming equivalence in clinical outcomes based on the results of trial 30-49(22). The costs comprised: acquisition costs of the study drugs, the costs of drug administration and the costs of managing adverse events. Resource use data obtained from the trial was supplemented with expert opinion on the management of adverse events in the separate localities. Two separate analysis were conducted, a European (UK) and a North American analysis (US), based on expert opinion from two different sets of clinical experts. No details were provided on the composition of the expert panel or how the estimates were derived.

Summary of effectiveness data The assumption of equivalence in clinical outcomes between topotecan and PLDH was based on the original (short-term) results of trial 30-49(22). This study reported that there were no statistically significant differences in the main clinical outcomes between the PLDH and topotecan. Based on these initial trial results PLDH was

139 assumed to be at least as effective as topotecan and a cost-minimisation analysis was therefore undertaken. However, it should be noted that a subgroup analysis performed on platinum-sensitive patients in trial 30-49, showed a statistically significant increase in PFS and overall survival in the PLDH group. This issue was not addressed by using a cost-minimisation framework. In addition, the toxicity profiles of the two agents were very different with a higer percentage of patients in the topotecan group experiencing grade 4 adverse events (table 42). Consequently, trial 30-49 does not provide a sufficient basis for assuming equivalence in overall health-related quality of life. Therefore, the appropriateness of a cost-minimisation analysis could be debated (see comments).

Table 42. Adverse event data from 30-49(22) Adverse event No. (%) patients PLDH (N=239) Topotecan (N=235) All Severities Grade 3/4 All Severities Grade 3/4 Severity Severity Neutropenia 84 (35%) 29 (12%) 191 (81.3%) 180 a (77%) Anaemia 86 (36%) 12 (5%) 169 (72%) 66 (28%) Thrombocytopenia 31 (13%) 3 (1%) 152 (65%) 80 (34%) Leucopoenia 87 (36%) 24 (10%) 149 (63%) 117 (50%) Alopecia 45 (19%) 3 (1%) 123 (52%) 15 (6%) PPE 121 (51%) 57 (24%) 2 (1%) 0 Stomatitis 97 (41%) 20 (8%) 35 (15%) 1 (<1%) a Includes three fatal occurrences of neutropenic sepsis

Summary of resource utilisation and cost data Resource utilisation and cost data were derived from a combination of patient-level data from trial 30-49 and expert opinion relating to the management of adverse events. The Smith study only considered costs incurred during the treatment period and hence assumed that resource utilisation and costs beyond this period would be identical between the drugs. The acquisition costs of the drugs were based on the dosage (mean dose per infusion/patient, 1.50 mg/m2 for topotecan versus 50 mg/m2 for PLDH) and number of treatment cycles (5.75 and 4.87, respectively) obtained from clinical trial data (see table 43).

140 Table 43. Study drugs and cycles received in the 30-49 trial (22). Total mg Mean mg per Total cycles Mean cycles Total doses drug used patient per patient PLDH 94447 395.18 1164 4.87 1164 Topotecan 15653 66.60 1349 5.74 6673

Similarly, the resource use associated with the drug administration was taken from the trial and included an outpatient visit for each dose and a specialist visit at the beginning of each cycle. In addition, eight adverse events associated to drug toxicity were considered: stomatitis/pharyngitis, PPE, nausea/vomiting, diarrhoea, anaemia, thrombocytopenia, neutropenia and sepsis/fever. These adverse events were chosen on the basis of patient perception, frequency and clinical importance. The adverse events occurred at four severity levels, from grade 1 (mild) to grade 4 (life- threatening). Adverse events were recorded in the trial each time they occurred, even when they were simultaneous (i.e. when more than one event occurred within a given time period). To avoid potential double counting in the cost analysis, a number of assumptions were made. In those instances where a patient experienced two or more of the same event in a cycle, the most severe event was selected. When overlapping time periods were indicated for neutropenia, sepsis or fever, the most resource- consuming event was included in the cost analysis. Finally, for patients experiencing several events in a sequential unbroken time-period, events were discarded if they did not allow the patient to complete the maximum estimated number of visits or length of stay.

For some resource use items relating to adverse events, namely hospitalisations and some drug treatments, data was not available directly from the trial. In these instances, the resources required to manage adverse events were estimated by a panel of oncologists from the UK (for the European analysis) and the US (for the North American analysis). To account for protocol induced resource use, hospitalisations for adverse events were restricted to grades 3 and 4. Estimates of the resource use for the UK and the US analysis can be seen below in table 44.

141

Table 44. Quantities of resource use, per person (50) Topotecan PLDH Difference (P-T) US UK US UK US UK Cycles of therapy 5.75 5.75 4.87 4.87 -0.88 -0.88 Units of platelets 1.10 0.30 0.05 0.02 -1.05 -0.28 Units of packed 1.56 4.15 0.26 0.65 -1.30 -3.49 red cells Units of 291091 20786 25670 690 -265421 -20096 erythropoietin Colony 3520 1494 762 31 -2758 -1463 stimulating factor Hospital stays 0.24 0.53 0.08 0.12 -0.17 -0.41 Office visits 12.7 4.2 4.6 3.2 -8.1 -1.0

For the UK analysis it was assumed that all patients in trial 30-49 received care in the UK, and similarly in the US analysis it was assumed that all patients received care in the US, that is the total number of adverse events were included in the cost-analysis, not just those experienced by UK patients in the UK analysis and by US patients in the US analysis. For resource use items where there were differences between resource use in the UK and the US, extrapolation methods were used to predict the amount of resources that would have been used if patients had been treated in the relevant setting.

Costs were reported in US$ for both the UK and the US analyses. To aid comparison we have also converted total costs to £GBP. The sources of cost data for the UK analysis were the British National Formulary (BNF) for drug costs, the National Blood Authority for blood products, a national database of hospital trusts for inpatient stays, a specific UK trust for the cost of ICU and a UK cancer care centre for the cost of outpatient visits and chemotherapy administration. For the US analysis, costs of clinician visits were taken from Medicare fees, medication costs from the red book and, hospital stays, laboratory tests and blood products from the hospital fee lists from an academic centre.

142

Summary of cost-effectiveness The results of the cost-minimisation analysis can be seen below in tables 45 and 46.

Table 45. US cost-analysis, per person in US Dollars (£GBP*) (50) Topotecan PLDH Difference (P-T) Study drug 10058 12962 2904 Administration 8377 1438 -6939 Total drug + administration 18435 14400 -4035 Stomatitis/pharyngitis 30 101 71 PPE 0 104 104 Nausea/vomiting 83 49 -34 Diarrhoea 58 34 -24 Colony stimulating factors 1936 419 -1517 Neutropenia visits and medication 1820 78 -1742 Total neutropenia 3756 497 -3259 Sepsis and fever 111 56 -55 Erythropoietin 3493 308 -3185 Transfusions 1346 140 -1206 Anaemia and thrombocytopenia 342 18 -342 visits and medication Total anaemia and 5181 466 -4715 thrombocytopenia Hospital stays 566 188 -378 Total cost 28,220 15,895 -12325 (£15,735) (£8,863) (-£6,872) 95% CI (25,750 to30,974) (14,515 (-9445 to -15,415) to17,306) *Exchange rate: 1.00 £GBP = 1.79 US Dollars

143

Table 46. UK cost-analysis, per person in US Dollars(£GBP*) (50) Topotecan PLDH Difference (P-T) Study drug 7286 11381 4095 Administration 5701 1513 -4187 Total drug + administration 12986 12894 -92 Stomatitis/pharyngitis 70 175 105 PPE 1 189 188 Nausea/vomiting 298 151 -147 Diarrhoea 60 36 -24 Colony stimulating factors 538 11 -526 Neutropenia visits and medication 218 25 -193 Total neutropenia 756 36 -720 Sepsis and fever 105 46 -59 Erythropoietin 242 8 -234 Transfusions 1190 181 -1009 Anaemia and thrombocytopenia 0 0 0 visits and medication Total anaemia and 1432 189 -1243 thrombocytopenia Hospital stays 1197 280 -917 Total cost 16,906 13,997 -2909 (£9,429) (£7,805) (-£1,622) 95% CI (15,617 to18,847) (12,863 (-779 to -3415) to15,392) *Exchange rate: 1.00 £GBP = 1.79 US Dollars

The results of the Smith study indicate that, in both the US and UK analyses, PLDH was significantly less costly than topotecan. Confidence intervals around total costs and cost differences were calculated using bootstrap methods. These showed the cost differences to be statistically significant at the 5% level. In the US analysis the mean cost per patient was approximately $12,325 (£6,872) higher for topotecan compared to PLDH ($28,220 and $15,895 respectively). Although the mean costs reported in the UK analysis were also higher for topotecan, the difference in costs was lower than that reported in the US analysis ($2,909 [£1,622] vs. $12,325 [£6,872]). Despite higher drug acquisition costs for PLDH, the overall treatment costs incurred by patients receiving PLDH were lower in both the UK and US analyses due to lower

144 administration costs per treatment cycle (1 outpatient visit per cycle for PLDH versus 5 for topotecan) and lower costs of managing adverse events. The differences in the costs between the UK and US analyses were due to the different assumptions made by the 2 expert panels for the management of side effects and, in particular, the use of colony stimulating factors (G-CSF), erythropoietin and blood transfusions.

The robustness of the base case analysis to alternative assumptions was confirmed using sensitivity analysis. Sensitivity analysis of an extreme scenario favouring topotecan was undertaken by applying the minimum number of days for hospitalisations and the minimum number of outpatient visits estimated by the experts. Also, the cost of hospitalisations, study drug infusion and outpatient visit associated with adverse events was decreased. In addition, an analysis restricted to local patients (no extrapolation between settings) was also undertaken. Both these sensitivity analysis concluded that PLDH was cost saving in the majority of bootstrap replicates (at least 83%). On the basis of these results the authors concluded that PLDH is a dominant treatment since it is associated with lower costs and similar (or better) outcomes compared to topotecan.

Comments There are important differences in the total costs estimated in the UK and US analysis, namely the US analysis shows a much greater cost saving associated with PLDH than the UK analysis. This is due to the increased costs of managing adverse events (which were more common in the topotecan group) in the US and, in particular, the use of erythropoietin for the treatment of anaemia, which is not frequently used in the UK. This study helps to illustrate the importance of local practice when determining the most cost-effective treatment for a given disease.

There are, however, a number of potential limitations associated with this study. In particular, the use cost-minimisation analyses as an analytic framework. Although the initial (short-term) results of trial 30-49(22) did not show any statistically significant differences between the two drugs in terms of overall survival and PFS, significant differences were found for subgroup analyses and drug toxicity. Consequently trial 30-49 does not provide a sufficient basis for assuming equivalence in overall health- related quality of life, particularly given the differences in toxicity profiles between

145 the two drugs. In addition certain important cost items, such as patient costs, were excluded from the cost-analysis. However given that topotecan is administered over 5-days, as opposed to 1-day for PLDH, it is likely that inclusion of such costs would have further strengthened the cost difference between the 2 drugs.

Secondly the use of expert opinion to supplement resource use data is not ideal. As highlighted by the authors the use of expert opinion can lead to underestimates of the variability in items. In addition the assumptions of only one severe (neutropenia, sepsis or fever) adverse event and only one event of each type per cycle, to avoid double counting, probably resulted in a rather conservative estimate of the cost of adverse events. The extrapolation of data from one setting to another assumes that resource use in one country can be used to accurately predict resource use in another country, which may not necessarily hold true in practice. The authors did, however, perform a sensitivity analysis using local patients for the UK and US analysis. Cost- savings were still associated with PLDH in this scenario. Finally, the UK analysis actually included patients from a number of European countries, some of which may differ to the UK in terms of treatment practice. The authors did not address this issue in a sensitivity analysis.

4.4.3 Review of Ojeda et al 2003 Overview The analysis by Ojeda(51) compared the costs of topotecan with PLDH using a similar approach to that reported in Smith(50) A cost minimisation analysis was performed, assuming equivalence in efficacy based on the initial (short-term) results of trial 30-49(22). Costs considered in the analysis were: costs of the study drugs, costs of drug administration and the costs of managing adverse events (medications and hospitalisations). A Spanish health service perspective was used. Expert opinion was used to supplement resource use data from trial 30-49, in particular relating to the resources required to manage adverse events that were not available from the trial.

Summary of effectiveness data The effectiveness evidence for PLDH and topotecan was derived from trial 30-49(22). Study characteristics and results have been described earlier (see “summary of effectiveness data” in the study by Smith(50)).

146

Summary of resource utilisation and cost data Resource use for the 2 treatments was determined using data from trial 30-49 supplemented by expert opinion for the resources associated with the management of adverse events. In contrast to the study by Smith, this analysis includes nine adverse events associated with drug toxicity: stomatitis/pharyngitis, PPE, nausea/vomiting, diarrhoea, anaemia, thrombocytopenia, neutropenia, sepsis and fever. Sepsis and fever were considered separately as opposed to combining these into a single category as in the Smith analysis. Adverse events occurred at four severity levels, from grade 1 (mild) to grade 4 (life-threatening), however unlike the Smith analysis hospitalisations associated with grades 1 and 2 adverse events are also included in the cost-analysis. Likewise the assumptions used to avoid double counting of adverse events (as assumed by Smith(50)) were not used in this analysis.

The costs associated with drugs were taken from the Spanish catalogue of medicinal products and unit costs of procedures and tests were obtained from the Spanish database of sanitary costs and published literature. The cost year was 2001.

147

Summary of cost-effectiveness The results of the cost-minimisation analysis can be seen below in table 47. Table 47. Total costs of treatment per person in Euros (£GBP*)(51) PLDH Topotecan Difference (P-T) Drug cost 7785.43 4256.66 Administration cost 862.27 4263.29 Total drug and 8647.70 8519.94 127.76 administration cost Anaemia 334.25 1687.19 Thrombocytopenia 4.51 337.51 Neutropenia 45.32 859.82 Sepsis 48.49 209.61 Fever 15.53 36.41 Stomatitis/pharyngitis 163.94 23.00 Nausea/vomiting 69.48 115.84 Diarrhoea 210.01 35.37 PPE 75.9 0.00 Total cost adverse events 967.02 3304.75 -2337.73 Total cost 9,614.71 11,824.69 -2,209.97 (£6,571.05) (£8,079.85) (-£1,509.9) *Exchange rate: 1.00 £GBP = 1.46 Euro

The analysis showed that the total cost per patient for PLDH was Euro 2,210 (£1,510) less than for topotecan (Euro 9,615 and Euro 11,824.69 respectively). This was largely due to differences in the number of adverse events between the 2 treatments, and hence the management costs of these events. In particular, topotecan was associated with higher costs of anaemia (Euro 1,353) and neutropenia (Euro 815). The statistical significance of the differences in total costs were not reported.

A sensitivity analysis was undertaken to assess the impact of the most significant (not defined) variables on the results of the analysis. Namely by increasing/reducing hospitalisations and outpatients visits associated with adverse events by 50%, assuming the same number of cycles administered for both drugs, assuming anti- emetic drugs required during administration of topotecan have zero cost. A worse- case scenario was also conducted assuming all of the previous sensitivity analysis in a

148 direction that favoured topotecan and using lower doses (unlicensed) for both drugs. All sensitivity analysis showed PLDH was associated with a lower total costs compared to topotecan (representing savings of between Euro 320 to 3,058), although the extent of this difference varied greatly. Based on the results of this analysis the authors conclude that PLDH dominantes topotecan.

Comments In addition those limitations mentioned in the "comments" section of the Smith, et al review, namely the use of a cost-minimisation framework and the use of expert opinion, the study by Ojeda(51) also has one other important limitation that is worth noting. The costing of adverse events in this study does not appear to have followed the principles of the analysis by Smith(50). In particular, patients can still have more than one hospitalisation and more than one of a particular event per cycle. Thus there is a potential for double-counting in these cost estimates and as such the costs of adverse events may have been overestimated. This analysis has also included the hospitalisation costs of grades 1 and 2 adverse events, which as argued by Smith, are likely to be protocol induced and may not be representative of the treatment of adverse events in practice. Despite the differences in costing methods, the cost savings associated with PLDH reported in the UK analysis by Smith produced very similar results to the analysis by Ojeda (Euro 2,410 and Euro 2,210). This could be due to a more conservative treatment of adverse events in Spain, which in turn could have implications concerning the generalisability of the UK estimates to all European patients in the study by Smith.

4.4.4 Review of Capri Overview

The analysis by Capri(52) compared the costs of topotecan with PLDH. A cost minimisation analysis was performed, assuming equivalence in efficacy as demonstrated by the short-term results of trial 30-49(22). The costs considered in the analysis were the costs of the study drugs, costs of drug administration and the costs of managing adverse events (medications and hospitalisations). An Italian health service perspective was used. Expert opinion was used to supplement resource use data from trial 30-49 for the resources required to manage adverse events, that were not available from the trial.

149

Summary of effectiveness data The effectiveness evidence for PLDH and topotecan was derived from the short-term results of trial 30-49. Study characteristics and results have been described earlier (see “summary of effectiveness data” in the study by Smith(50).

Summary of resource utilisation and cost data Resource use for the 2 treatments was determined using data from the trial 30-49 (50) supplemented by expert opinion for the resources associated with the management of adverse events. Expert opinion was provided by 5 Italian oncologists, using the Delphi method. The analysis includes nine adverse events associated to drug toxicity were considered: stomatitis, PPE, nausea/vomiting, diarrhoea, anaemia, thrombocytopenia, neutropenia, sepsis and fever. As in the Smith analysis, costs of adverse events were evaluated at four severity levels, from grade 1 (mild) to grade 4 (life-threatening). In contrast to the Smith analysis, hospitalisations associated with grades 1 and 2 adverse events are also included in the cost-analysis. Also the authors do not make any assumptions that would ensure that adverse events were not double counted (as assumed by Smith). Costs of each adverse event are presented in Appendix 8, table K.

Only direct medical costs were included in the analysis and were based on prices as of July 2002, reported in Euros. The costs associated with drugs were taken from the Italian National Formulary. National changes were used to cost medical visits and laboratory tests and diagnostic-related group (DRG) reimbursements rates were used to cost hospitalisations.

Summary of cost-effectiveness An assumption of equivalence in efficacy between topotecan and PLDH was made (based on the initial results of trial 30-49), and a cost-minimisation analysis was performed. The results of the cost-minimisation analysis can be seen below in table 48.

150

Table 48. Base case cost-analysis in Euros (£GBP*)(52) PLDH Topotecan Total cost Mean cost per Total cost Mean cost per patient patient Drug cost 1,656,419 6931 908,077 3864 Administration cost 164,118 687 951,011 4047 Drug and administration 1,820,537 7617 1,859,088 7911 costs Nausea/vomiting 16,143 68 27,862 119 PPE 2386 10 0 0 Stomatitis 36,428 152 4139 18 Anemia 187,667 785 1,009,692 4297 Neutropenia 29,023 121 645,037 2745 Diarrhoea 6818 29 12,146 52 Thrombocytopenia 0 0 117,834 501 Fever 2730 11 16,810 72 Sepsis 4342 18 17,475 74 Total adverse events 285,536 1195 1,850,995 7,877 Total cost 2,105,073 8,812 3,710,083 15,788 (£1,438,096) (£6,019) (£2,535,109) (£10,787) *Exchange rate: 1.00 £GBP = 1.46 Euro

Mean total costs were Euro 6,976 (£4,768) higher in the topotecan arm than the PLDH arm (Euro 15,788 and Euro 8,812 respectively). Although PLDH was associated with higher study drug costs, this was more than offset by the higher administration costs and the costs of treating adverse events (e.g. neutropenia) in the topotecan group.

Sensitivity analysis using the minimum and maximum (as determined by the expert panel) levels of resource use for each adverse event was undertaken. For the analysis using the minimum level of resource use, total costs were reduced to Euro 12,014 for topotecan and Euro 8,281 for PLDH. For the analysis using the maximum level of resource use total costs for topotecan were Euro 18,847 and Euro 9,314 for PLDH. Based on the results of the base case and sensitivity analysis the authors conclude that PLDH is associated with lower costs than topotecan, and therefore represents the most efficient treatment.

151

Comments The study by Capri(52) suffers from some of the limitations discussed in the "comments" section of review of Smith(50), namely the use of a cost-minimisation framework and the use of expert opinion. In addition the study appears to have used the same methods of costing adverse events as described in the review of the study by Ojeda(51). As a result the costing of severe adverse events (neutropenia, fever and sepsis) has not taken account of the fact that patients may already be hospitalised for another adverse event in the same time period, therefore potentially over-estimating the total costs of adverse events. It is however understandable that this type of analysis could not be undertaken by Capri, et al and Ojeda, et al, as they did not have access to patient-level data, as in the Smith analysis. As in the Ojeda analysis, however, the authors could have considered the exclusion of hospital costs relating to grades 1 and 2 adverse events, as these are not likely to receive hospital-based treatment outside of a trial setting.

4.4.5 Review of Prasad et al Overview The analysis by Prasad(53) aimed to compare the costs of topotecan with gemcitabine

(GEMZAR™) in platinum and paclitaxel resistant patients. A cost consequence analysis (cost and effects were not formally combined) was performed from a US perspective, using survival and resource use data collected in a non-randomised cohort study. Costs included in the cost analysis were study medications, hospitalisations and other clinical visits, laboratory tests and costs of managing adverse events. Outcomes were progression-free survival, overall survival, response and toxicities experienced.

Summary of effectiveness data This study was a non-randomised follow-up of 51 (46 for the cost-analysis) patients receiving topotecan and 56 patients receiving gemcitabine. The non-randomised nature of this study meant that it was not included in our main review of effectiveness evidence. All patients were receiving chemotherapy as 2nd line or greater for persistent or recurrent platinum-resistant ovarian cancer. Patients were identified from retrospective hospital records and from prospective recruitment via the clinic.

152

Topotecan was administered over 5-days every 21-days with a median dose intensity of 1.64 mg/m2 per week and a median number of 4 cycles. Gemecitabine was administered every 7 days for 21 days, followed by 7-days of no therapy. Patients received a median dose intensity of 497 mg/m2 and a median number of 5 cycles.

The assessment of response was based on CT scans and/or Rustin criteria. The occurrence of either an objective response or stable disease was defined as a clinical benefit. Progression free survival was calculated as the difference between the initiation of therapy and documented disease progression. Likewise, overall survival was calculated as the difference between initiation of therapy and death/last follow- up. Kaplan Meier methods were used to estimate progression-free and overall survival. Potential predictors of progression-free survival and overall survival were investigated using the Cox proportional hazards model. Multivariate logistic regression was used to analyse the covariates associated with clinical benefit.

The outcomes of the study can be seen in table 49 below. Response and survival times (time to progression and overall survival) were greater in the topotecan group than the gemcitabine group e.g. median time to progression was 1.8 months in the gemcitabine group compared to 3.6 months in the topotecan group.

Table 49. Clinical outcomes in Prasad(53) Chemotherapy Outcome Value (95% CI) Gemcitabine Clinical benefit 13/56 (23.2%) (11.9%-34.5%) Median survival (months) 8.2 (range 5.9-12.1) Median time to progression (months) 1.8 (range 1.5-2.6) Topotecan Clinical benefit 28/50 (56%) (42%-70%) Median survival (months) 16.8 (range 13.9-24.9) Median time to progression 3.6 (3-5.5) (months)

A number of variables were thought to be associated with the clinical outcomes, namely age, prior salvage therapy, time from last treatment and duration of 1st clinical

153 remission. Regression analysis showed no significant association between any of the these variables and clinical benefit, overall survival and disease progression. However for the gemcitabine group there was some positive association reported between increased time from last treatment and increased time to progression and longer overall survival.

With regards to toxicities, topotecan and gemcitabine patients were reported to be equally likely to experience events, with around 50% of patients in both treatment arms experiencing grade 3 or 4 toxicities. The data presented on toxicities was rather limited and data was only reported in those instances where there were differences between the 2 groups.

Summary of resource utilisation and cost data Charge data on study medications, hospitalisations, clinical visits, laboratory tests, radiotherapy, administration, blood products and adverse events were collected directly from hospital database records for patients in the study. Charges were then converted to costs using appropriate ratios of charges to costs. The cost year was not stated and costs were in US$. Data on resource use items were not presented.

Summary of cost-effectiveness Although both total costs and clinical outcomes were recorded in the study, these were not combined to provide an estimate of cost-effectiveness; instead the focus of the study was on the differences in costs between the 2 groups.

The mean costs per cycle per patient were significantly higher for topotecan compared to gemcitabine ($7832 compared to $2732). This was largely due to the higher study drug acquisition and administration costs for topotecan. Individual cost items, however, were not presented so it is not possible to verify this. Mean costs per course delivered were $13,937 (Range $1,824 to $82,340) for gemcitabine and $28,098 (Range $5,384 to $157,170) for topotecan. The cost of individual items, such as management of adverse events was not reported. Instead the authors only reported the total costs for both groups. The total cost of drugs in the topotecan arm was $14,161 (£7,897) higher than the gemecitabine arm ($28,098 and $13,937 respectively). Sensitivity analysis was not undertaken.

154

Survival in the topotecan group was greater than the gemecitabine group (median survival = 16.8 months in the topotecan group compared to 8.2 months in the gemcitabine group). It is not clear if this difference was statistically significant. No attempt was made to combine the cost and outcome data in the form of a cost- effectiveness ratio. Given the results derived from this study, the authors conclude that although patients receiving topotecan had a greater clinical benefit the costs of treatment were considerably higher.

Comments The economic evaluation by Prasad et al (53) used different methods to assess the cost-effectiveness of 2nd line ovarian cancer treatments, to those used in the studies by Smith.(50), Ortega(51) and Capri(52). In particular estimates of resource use were obtained directly from patients in the study, and hence expert opinion was not required. Although this is preferable to the methods used by Smith, Ojeda and Capri, the non-randomised nature of the study does not make it a reliable source of data for clinical and resource outcomes. That is, it is likely that there was a degree of patient selection for topotecan and gemcitabine. Indeed the authors note that this is likely to be the reason for the large cost differences seen between the 2 drugs.

The authors did not undertake any type of sensitivity analysis. Given the large ranges that are reported for total costs, it is likely that individual cost items are associated with some uncertainty. The use of probabilistic analysis, explicitly accounting for this uncertainty would have produced results that were more useful for decision makers, that is an estimate of the degree of uncertainty associated with a treatment being the most cost-effective (or cost-saving in this instance) option.

Following on from this, it is apparent from the outcome data reported that there were some important differences between the 2 treatments. The authors, however, do not make any attempt to combine the estimates of survival (progression free and overall) with the cost differences observed. It is therefore not possible to ascertain if topotecan is cost-effective.

155 Finally the reporting of important data in this study is particularly poor. No data on resource use or individual cost items were reported. It is therefore difficult to interpret the sources of the costs differences and thus the validity of the total costs estimated.

4.4.6 Review of the GlaxoSmithKline submission

Overview The GSK submission compared the cost-effectiveness of topotecan and PLDH, using new evidence made available since the previous NICE guidance was issued.(17) The previous submission compared topotecan and PLDH based on the short-term results of trial 30-49(22). Assuming equivalence in efficacy between topotecan and PLDH, a cost-minimisation analysis was performed(50). This analysis showed that PLDH was significantly less costly than topotecan, because of lower administration costs and lower costs associated with the management of adverse events.

Recently, an audit of current practice for relapsed ovarian cancer in UK has been undertaken by GSK.(54) The purpose of the audit was to provide an overview of the current management of ovarian cancer and to investigate the outcomes associated with second or subsequent line chemotherapy. In this new submission a cost-minimisation analysis was conducted based on the clinical results of the previous pivotal trial but using data for some resource use items (e.g. mean dosage with associated administration costs or strategies to manage adverse events) obtained from the audit. In addition, the potential savings to the NHS via a home administration of topotecan were estimated and the results of a survey that assessed the patient convenience with topotecan home delivery were presented.

Summary of effectiveness data The effectiveness evidence for PLDH and topotecan was derived from the trial 30-49. Study characteristics and results have been described earlier (see “summary of effectiveness data” in the review of the study by Smith(50).

Summary of resource utilisation and costs data Data for resource utilisation from the trial 30-49 was combined with the audit of current practice for relapsed ovarian cancer in UK. The submission justified this

156 approach on the basis that the audit was considered more representative of UK management of patients with epithelial ovarian cancer than a clinical trial.

Audit data was obtained from nine UK centres. Each centre completed a centre registration form agreeing to participate in the data collection exercise. Patients were included in the audit if they were: 18 years of age or older, had a diagnosis of ovarian cancer and required treatment with a 2nd line or subsequent chemotherapy. The number of patients available for the analysis was 198. More than 75% of patients had received surgery as initial treatment and the most frequently prescribed drugs for 1st line chemotherapy were carboplatin or carboplatin with paclitaxel (around 80% of patients). All patients entering the study were prospectively followed up every 6- months and their status at each follow-up documented. The results of the audit were used to estimate the number of treatment cycles with topotecan and PLDH, the associated administration costs and the resource use and costs relating to the management of adverse events.

The results of the audit showed that the mean dosage and the number of cycles for topotecan and PLDH were lower than that reported in the trial (see table 46). The mean dose per cycle of topotecan was 1.49mg/m2 with a mean of 3.88 cycles, while the mean dose per cycle of PLDH was 49.07 with a mean of 3.54 cycles (more details are presented in L and M in Appendix 8). A lower mean number of cycles together with a lower mean dose are likely to have implications for both drug and administration costs and the risk of adverse events. Therefore, new estimates of costs were made using data for drug dosage and treatment cycles obtained from the audit.

In costing adverse events, particular attention was given to the incidence of neutropenia and the associated use of colony stimulating factor (G-CSF) and hospitalisations. Neutropenia was the most common adverse event (table 44) for topotecan patients in the trial 30-49 with an incidence of 77% (for grades 3/4). 29.1% of patients in trial 30-49 received G-CSF with a mean dosage of 1494 µg. In the UK audit, only 17.39% of patients suffered from neutropenia and only 4.35% of patients received G-CSF. The mean units of G-CSF used in the trial (1494 µg) were thus multiplied by the factor 0.149 (4.35/29.1) giving 223 µg mean units of G-CSF. The same process was applied for office visits and other medications for treating

157 neutropenia, in order to reflect that only 4.35% of patients received some treatment (rather than 29.1% reported in trial 30-49 ).

In addition, the audit findings showed that the majority of inpatient stays were due to neutropenia. The rate of hospitalisations not related to neutropenia was taken from the 30-49 trial (39%), however the rate of hospitalisation for neutropenia obtained from the trial 30-49 was substituted by the rate found in the audit. The mean number of inpatient hospitalisations in the trial was 0.53 days, and 61% of hospitalisations were due to neutropenia. Inpatient hospitalisations (per patient) due to neutropenia, as recorded in the trial, were therefore 0.323 days (0.53x0.61). In the audit, however, inpatient hospitalisation stay (per patient) due to neutropenia was 0.163 days. The total mean inpatient hospitalisation stay used in this re-analysis was therefore estimated as 0.37 days (0.53 – 0.323 + 0.163), that is inpatient stay in the trial minus inpatient stay due to neutropenia in the trial plus inpatient stay due to neutropenia in the audit. Table 50 presents the alternative resource use estimates used in the GSK submission, as compared to the original analysis of the 30-49 trial(50) reported in the Smith study(50).

Table 50. Quantities of resources used included in the GSK submission (only those different from the 30-49 trial Topotecan PLDH Mean dose per infusion/patient, 1.49 49.07 (mg/m2) Cycles of therapy 3.88 3.54 Colony stimulating factor (µg) 223 N/A Hospital stay (days) 0.37 N/A

For adverse events other than neutropenia, given the lack of data from the audit, it was felt appropriate to adjust the costs associated with adverse events for each treatment used in the Smith study(50) to reflect the lower cycle lengths (3.88 versus 5.75 for topotecan; 3.54 versus 4.87 for PLDH). Therefore, for topotecan patients, costs for all adverse events were multiplied by a rate of 0.675 (3.88/5.75); and for PLDH patients, all adverse event costs were multiplied by 0.727 (3.54/4.87). The rationale for this adjustment is that fewer cycles may reduce the incidence of adverse events associated with both therapies.

158

The resources required to administer topotecan in hospital were obtained directly from the trial 30-49. This was multiplied by the cost per cycle estimated by Smith. However, since less cycles were reported in the audit (3.88 cycles instead of 5.75), the total costs of administration were reduced. The same approach was applied for PLDH (3.54 cycles instead of 4.87). In the Smith study the administration costs for PLDH did not appear to include cardiac monitoring. Therefore, it was assumed that patients receiving PLDH would need to receive one ECG and one echocardiogram or multigated angiography per cycle (assumed at a rate 2:1) and these costs were added to the administration costs.

Finally unit costs were obtained from the NHS Reference Costs (2003),the National Tariffs (2004), and from the Smith study. A unique price year was not reported. Unit costs and their sources were presented transparently and in detail in the submission.

Summary of cost-effectiveness

An assumption of equivalence in efficacy between topotecan and PLDH was made (based on the results of the trial 30-49 (22), thus a cost-minimisation analysis was performed. This analysis was based on the Smith(50) study but updated to reflect the results of the UK audit. As described previously, drug costs, administration costs, costs of neutropenia management and related hospitalisations were directly calculated using resource use data found in the audit, applying UK unit costs. The costs associated with adverse events were calculated by multiplying the number of events reported in the Smith study by 0.675 (i.e. 3.88/5.75) for topotecan and by 0.727 (i.e. 3.54/4.87) for PLDH.. The results of the cost-minimisation analysis by GSK can be seen in Table 51

159 Table 51. Results of the GSK cost-minimisation re-analysis

Topotecan PLDH Study Drug Costs £3,737.52 £5,799.26 Administration Costs £2,680.50 £1,701.89 Total drug administration cost (per £6,418.02 £7,501.15 patient) Colony stimulating factors £57.40 £5.79 Neutropenia office visits & other medications £23.27 £12.98 Total Neutropenia £80.66 £18.77 Erythropoietin £116.22 £4.15 Transfusions £304.28 £93.98 Total Anemia & Thrombocytopenia £420.49 £98.13 PPE £0.48 £98.13 Stomatitis £33.74 £90.86 Diarrhoea £28.92 £25.71 Sepsis & Fever £50.61 £23.88 Nausea & Vomiting £143.63 £78.40 Total Other AE Management costs £257.38 £316.99 Adverse Event Costs £758.54 £433.90 Hospital Stays £596.47 £145.38 Total adverse event cost (per patient) £1,355.01 £579.28 Total treatment cost (per patient) £7,773.03 £8,080.43

The use of the audit reduced the total cost of topotecan and PLDH compared with the Smith study (see tables 45 and 46). The reduction in treatment cycles from 5.75 to 3.88 for topotecan and from 4.87 to 3.54 for PLDH led to a reduction in the study drug costs from £5,204 to £3,738 for topotecan and from £8,129 to £5,799 for PLDH. Similarly, the administration costs for topotecan decreased from £4,072 to £2,681. The administration costs of PLDH did however rise from £1,081 to £1,702 due to the inclusion of cardiac monitoring costs.

With regard to adverse events, there was an important reduction in costs associated with neutropenia for topotecan, due to the lower percentage of patients receiving G- CSF in the audit compared to trial 30-49 and the lower rate of hospitalisations and office visits. Costs associated with other adverse events were lower, because of the assumption that fewer cycles may reduce the incidence of adverse events associated

160 with both therapies. The total cost per patient receiving topotecan was slightly lower than the cost per patient of treatment with PLDH (£7,773 versus £8,080). This is contrary to the results of the Smith study where topotecan was associated with higher treatment costs (£12,076 versus £9,997).

An additional scenario was presented to estimate whether there were potential savings to the NHS in administering topotecan via home delivery was undertaken. Home delivery was assumed save costs associated with outpatient appointments (or ward attendance) to receive infusion and nurse consultations. Given a frequency of 5 infusions and 5 nurse consultations per cycle, it was estimated that home delivery of topotecan would save the NHS £575 per cycle (per patient) and £2300 over 4 cycles. However, the cost of a nurse consultation does not appear to have been added in the calculations presented.

Comments

The use of an audit was justified by the need to obtain results that were more representative of UK management of patients with epithelial ovarian cancer than those available in a clinical trial setting. However, the analysis undertaken in the submission presents some important limitations.

Firstly, although the audit may be more representative of UK current practice, it does not guarantee the same internal validity of a clinical trial. Secondly, many assumptions were made due to the lack of data from the audit associated to some important items. For example, it was assumed that the reduction in adverse events (except neutropenia) would occur at the same rate of the reduction in the cycle treatments for topotecan and PLDH. The rationale for this assumption is clear, although the appropriateness of applying exactly the same rate is not fully justified in this re-analysis. Also, the most important adverse event for topotecan was neutropenia with high associated costs of management. The results of the audit showed that a lower percentage of patients experienced neutropenia compared to trial patients and it was assumed that the associated use of G-CSF, hospitalisations and office visits was reduced at the same rate. However, there was no information on G- CSF dose and visits due to neutropenia for patients included in the audit. Given the

161 importance of the costs associated with neutropenia for topotecan, more information regarding the treatment of this particular adverse event for patients included in the audit would have been useful. The audit results were, in fact, only used for resource use and costs estimates, but not for effectiveness evidence. The authors partly justified this approach, mentioning that the efficacy of topotecan with reduced dosage had been shown in a recent study by Rodriguez and colleagues.(55) However, it is not clear why effectiveness data for topotecan and PLDH from the audit were not used in this analysis or at least presented. The assumption that reduced dosage in UK clinical practice would alter only the costs results without any impact on the effectiveness needs further investigation

Secondly, the analysis assuming the home delivery of topotecan, appears to suggest that patients visit a specialist at the beginning of each 5-day cycle, but then have no further contact with the medical profession for the remaining 4-days of therapy. The costs of community nurse visits should therefore be added to the cost of topotecan at home, calculated in the GSK submission (estimated in our own analysis to require five 1-hour visits at £56/hour = £280). It is also worth noting that the estimates of efficacy (in this case assumption of equivalence between topotecan and PLDH) were based on administration of topotecan on an outpatient basis, and one should therefore be cautious when extrapolating the efficacy estimates for topotecan to administration in an alternative setting.

Thirdly, as discussed in the review of Smith(50) the use of a cost-minimisation framework is not ideal. Although significant differences between PLDH and topotecan were not apparent there were differences for platinum sensitive patients and drug toxicity. The submission highlighted the impact that different adverse events could have on patients on quality of life. By using a cost-minimisation framework the impact on QoL has not been considered. Furthermore, in a 3-year follow-up of patients included in trial 30-49, it was reported that survival was higher for PLDH compared with topotecan (hazard ratio: 1.216; 95% CI: 1.000-1.478; p=0.05) and significantly prolonged in platinum-sensitive patients (hazard ratio: 1.432; 95% CI: 1.066-1.923; p=0.0017).

162 In addition, an assessment of uncertainty was not undertaken. Mean values were reported for each cost category and the issue of the variability around these values was not addressed. No statistical analyses were performed in order to determine the significance of the cost differences, and confidence intervals were not presented. Given that the reduction in number of cycles for topotecan and PLDH found in the audit was the key factor in driving drug costs, administration costs and adverse events costs, a sensitivity analysis on this parameter appears necessary and relevant. In the submission the number of treatment cycles for topotecan and PLDH was based on a weighted average of the number of cycles over all lines of chemotherapy (3.88 and 3.54). We have performed a one-way sensitivity analysis just using the number of cycles for topotecan and PLDH as 2nd line treatment (4.19 and 3, respectively; see Appendix 8, tables 50a and 50b). In this sensitivity analysis, drug costs, administration costs and costs associated with adverse events (except neutropenia) were recalculated applying the same methodological approach used in the submission, but assuming 4.19 treatment cycles for topotecan and 3 treatment cycles for PLDH. The methods to estimate costs associated with neutropenia applied in this submission were used in this sensitivity analysis. The results of this sensitivity analysis show that the total cost per patient for topotecan increases to £8,328 while the total cost per patient for PLDH decreases to £6,845. This demonstrates the high variability around the mean values used in the analysis and raises some doubts as to the validity of the final results.

Finally, sub-group analyses in trial 30-49 (22) indicated PLDH as the preferred option for platinum-sensitive patients. This issue was not addressed or acknowledged in this submission and sub-group analyses were not undertaken.

4.4.7 Review of Schering-Plough Ltd submission

Overview In the Schering-Plough submission, an economic evaluation was performed to assess the cost-effectiveness of PLDH compared with topotecan. Effectiveness evidence was taken from the short-term results of trial 30-49(22) that showed equivalence in outcomes between the two drugs. In addition, results of long-term survival analysis were presented, showing a trend favouring PLDH both in terms of overall survival

163 and for a sub-group of platinum-sensitive patients. On the basis of these results it was assumed that PLDH was at least as effective as topotecan, and a cost-minimisation analysis was performed. The rationale for this analysis was that if PLDH was less costly than topotecan, than it should be considered the preferred option. The cost- minimisation analysis presented in this submission was based on the results of a previously published study by Smith(50).

Summary of effectiveness data

The effectiveness evidence for PLDH and topotecan was derived from trial 30-49. Study characteristics and results have been described earlier (see “summary of effectiveness data” in the study by Smith.(50).

This submission also included long-term survival data for PLDH and topotecan, not available at the time of the previous manufacturer submission; this new evidence was presented and discussed. In particular, overall survival results for PLDH and topotecan were reported based on a 3-year follow-up of an intent-to-treat population from the trial 30-49. Survival rates and median survival (weeks) were reported. Overall survival was estimated using the Kaplan-Meier method and hazard ratios (HR) were presented. Sub-group analyses for platinum-sensitive patients and platinum-refractory patients were also performed and discussed. The main results of this analysis are presented in table 52(HR>1.0 favours PLDH).

164 Table 52. Results from the long-term follow-up of patients in the study 30-49 (in brackets 95% CI)

All patients Platinum-sensitive Platinum-refractory patients only patients only PLDH Topotecan PLDH Topotecan PLDH Topotecan

Median survival 62.7 59.7 107.9 70.1 38.3 42.1 (weeks)

Hazard ratio 1.216 (1.000,1.478) 1.432 (1.066, 1.923) 1.069 (0.823,1.387) (median) Survival rate (1 56.3% 54.0% 74.1% 66.2% 41.5% 43.2% year) (50.0,62.6) (47.6,60.3) (65.8,82.4) (57.4,75.1) (32.8,50.1) (34.5,51.9) Survival rate (2 34.7% 23.6% 51.2% 31.0% 21.1% 17.2% years) (28.6,40.8) (18.1, 29.1) (41.6,60.7) (22.2,39.7) (14.1,28.2) (10.5,23.8) Survival rate (3 20.2% 13.2% 28.4% 17.5% 13.8% 9.5% years) (14.9,25.5) (8.8,17.1) (19.6,37.1) (10.2,24.7) (7.6,20.0) (4.2,14.7)

The long-term analysis shows that PLDH is associated with a survival benefit over topotecan. The median survival weeks in a 3-year follow-up were 62.7 for PLDH versus 59.7 for topotecan (p=0.05). The trend in favour of PLDH was more evident when platinum-sensitive patients were considered. A statistically significant difference in median survival was found and a higher percentage of patients were alive at 2 and 3-years follow-up. These differences were less relevant in the patient- refractory group.

Adverse events associated with PLDH and topotecan were taken directly from trial 30-49. More patients in the topotecan group suffered from grade 3 and grade 4 adverse events, particularly in the case of neutropenia. A higher proportion of patients in the PLDH group suffered from PPE and stomatitis, but these were managed successfully with dose modifications.

Finally, the impact of the two therapies on the patients’ health-related quality of life was discussed. Results from trial 30-49 (where presented) assessed quality of life using the self-administered QLQ-30 questionnaire. Overall, the impact of PLDH and topotecan on patients’ quality of life was similar but some differences were found for

165 single items. In particular, the percentage of patients with reduced pain was higher in the topotecan group than in the PLDH group. This difference was possibly due to the higher number of patients experiencing PPE in the PLDH group (see table N in Appendix 8).

It was concluded that PLDH was associated with survival benefit in the long-term, similar quality of life and lower severe adverse events compared to topotecan. Thus, PLDH should be considered dominant if less costly than topotecan.

In addition, comparative data on PLDH versus paclitaxel were presented, based on the trial 30-57. This was a Phase III study initially enrolling 460 patients with relapsed ovarian cancer that did not respond to a first-line platinum-based chemotherapy. Paclitaxel and PLDH were comparable in terms of overall survival (56.1 weeks versus 45.7, p=0.44), overall PFS (22.4 versus 21.7, p=0.15), response rate (22.4% versus 17.8%, p=0.34) and drug toxicity. However, it was noted that the 30-57 trial results should be interpreted with caution. The trial was terminated prematurely because paclitaxel became a part of standard first-line therapy. Thus, in the submission an economic analysis comparing PLDH and paclitaxel was not conducted.

Summary of resource utilisation and costs data Resource use data were taken mainly from trial 30-49 (See Smith (50) for details). In addition, nine adverse events associated to drug toxicity were considered: stomatitis/pharyngitis, PPE, nausea/vomiting, diarrhoea, anaemia, thrombocytopenia, neutropenia, sepsis and fever. These adverse events occurred at four severity levels, from grade 1 (mild) to grade 4 (life-threatening). Adverse events were recorded in the trial each time they occurred, even when they were coincident. Thus, for two or more events occurring at the same time, the most resource-consuming event was selected and only one adverse event per patient at each cycle was included.

Only limited data were available from the trial for some resource items associated with adverse events (e.g. G-CSF dosage, transfusions, clinical visits, hospitalisations). It was also felt that the management of adverse events may vary between countries and locations. Expert opinion was therefore used to estimate resource utilisation related to the management of adverse events for topotecan and PLDH. No data on

166 resource consumption for the long-term follow-up were presented. Findings for some of the resources used from trial 30-49 have been shown in Table 45.(Smith, et al analysis)(50).

Unit costs were taken from the British National Formulary (for drugs; year 2000), from the CIPFA database (in-patient stays; year 1999), from tariffs at a UK cancer centre (administration costs) and from the literature (ICU visits). A unique price year was not reported.

Summary of cost-effectiveness Based on the assumption that PLDH was at least as effective as topotecan, a cost- minimisation analysis was conducted. Mean per patient drug costs were higher for PLDH than topotecan (£8,129 versus £5,204, respectively). However, this was compensated for by lower administration costs for PLDH (£1,058 versus £4,043), due to higher dosing frequency and number of treatment cycles for topotecan. The total drug costs were therefore similar between the two therapies (£9,187 for PLDH versus £9,247 for topotecan). In addition, topotecan was associated with more severe and resource consuming adverse events (see table 45, Smith(50)). In particular, important differences were found for the costs of managing anaemia and neutropenia. Therefore, the mean total cost per patient was lower for PLDH than for topotecan (£9,957 versus £12,610, respectively). Confidence intervals (95%) were calculated based on a normal distribution assumption and checked with the bootstrap method. In the base case analysis PLDH was significantly less costly than topotecan.

This submission reports on a sensitivity analysis previously described by Smith, where an analysis of extremes favouring topotecan was performed. Even in this best- case scenario for topotecan, PLDH was significantly less costly, with a total per patient cost of £10,078 versus £12,286. The results of these analyses (basecase and analysis of extremes) are presented in Table 53.

167 Table 53. Results of the Schering-Plough cost-minimisation analyses

Base case Analysis of extremes PLDH Topotecan PLDH Topotecan Drug cost £8,129 £5,204 £8,129 £5,204 Administration £1,058 £4,043 £1,058 £4,043 Sub-total £9,187 £9,247 £9,187 £9,247 Stomatitis/Pharyngitis £93 £26 £161 £50 PPE £99 £0 £172 £1 Nausea/Vomiting £222 £355 £222 £355 Diarrhoea £34 £56 £34 £56 Anaemia £250 £1,361 £250 £1,361 Thrombocytopenia £0 £272 £0 £272 Neutropenia £35 £892 £21 £637 Sepsis £22 £213 £17 £159 Fever £15 £186 £14 £149 Sub-total £770 £3,362 £892 £3,038 TOTAL £9,957 £12,610 £10,078 £12.286 (95% CI) (9,067-10,847) (11,512-13,708) (£9,176-10,980) (£11,216-13,356)

The Schering-Plough submission concluded, “even if the clinical results of PLDH were equal to those of topotecan, PLDH would be a preferred therapy as its overall cost is lower”. Given that long-term follow-up findings showed survival advantages over topotecan, PLDH was considered as the dominant option.

Comments The economic analysis conducted in the Schering-Plough submission was based on resource use data from a clinical trial. The patient population appears representative of individuals with relapsed ovarian cancer, given that approximately 60% of patients had received a taxane as first-line therapy. Also, the study design assures high internal validity. In addition, details on resource use and unit costs were presented transparently and the replication of the study appears feasible. However, some limitations and issues should be highlighted.

In the clinical trial complete economic data were not collected for management of adverse events, hospitalisations and outpatient visits. Expert opinion was therefore

168 used when limited trial data were available. It is likely that the variability in patient- level data was minimised using this approach. This issue was partly addressed conducting an extreme analysis that favoured topotecan but given that the costs associated with adverse events are the main driver for differences in costs between the two drugs, it seems important to remark this point. Also, it is not clear whether the administration costs for PLDH included cardiac monitoring. The inclusion of these costs would increase the total costs per patient in the PLDH group.

The general limitations of cost-minimisation analyses and the usefulness of more recent and advanced techniques have already been underlined (see comments of the GSK section). In this submission, the use of bootstrap methods has been mentioned but little information was given. For example, in the Smith(50) analysis a curve representing the cumulative probability distribution of the cost difference between the two drugs (obtained performing 1000 bootstrap replicates) was presented and the probability of PLDH being cost saving was estimated. The use of this or similar methods in the submission would have given more information to decision-makers for the cost comparison between PLDH and topotecan.

4.4.8 Review of Bristol-Myers Squibb Ltd Submission

Overview In this submission four alternative 2nd line chemotherapies for patients with advanced ovarian cancer were compared: paclitaxel monotherapy, topotecan, PLDH and paclitaxel in combination with a platinum agent. The rationale for this analysis was that while paclitaxel as monotherapy has shown similar results in terms of response rate, PFS and overall survival, compared to topotecan and PLDH, paclitaxel in combination with platinum had demonstrated survival and tolerability benefits over platinum monotherapy. Given the lack of head to head comparisons for the 4 therapies under study, a model was constructed to estimate costs and effects associated with these alternatives. Life years gained over 3 years were calculated for the four chemotherapies based on clinical trial survival curves obtained from the literature. An incremental cost-effectiveness analysis was performed, including drug costs and administration costs for the 4 agents. One-way sensitivity analyses on key parameters were conducted in order to assess the robustness of base case findings.

169

Summary of effectiveness data Effectiveness evidence for the comparison between PLDH and paclitaxel as monotherapy were taken from trial 30-49 and data for the comparison between paclitaxel monotherapy and topotecan were obtained from trial 039 (22). In addition efficacy and tolerability of paclitaxel in combination with platinum has been recently shown in 2 parallel, international trials (ICON4)(29). In these trials, 802 platinum- sensitive ovarian cancer patients were randomised to receive paclitaxel (175 mg/m2, as a 3 hour infusion every 3 weeks) in combination with a platinum agent (carboplatin or cisplatin) or conventional platinum therapy (carboplatin or cisplatin). Kaplan-Meier curves were used to estimate PFS and overall survival and hazard ratios (HR) were presented. With a median follow-up of 42 months, significant differences in favour of combined therapy were found both for PFS (HR=0.76, 95% CI=0.66-0.99; p=0.0004) and overall survival (HR=0.82, 95% CI=0.69-0.97; p=0.02). Sub-group analyses confirmed a trend favouring paclitaxel plus platinum. Similar results were found for patients’ quality of life and drug toxicity. It was concluded that a paclitaxel/platinum therapy is likely to improve survival and PFS in relapsed platinum-sensitive ovarian cancer patients compared to conventional platinum therapy. Full details of this trial are available in section 4.2.

No direct comparisons of paclitaxel monotherapy, paclitaxel in combination with platinum therapy, topotecan and PLDH, are available from the trial data. In order to compare effectiveness evidence from the 3 trials (30-49, 039 and ICON4), a model was developed. Life-years gained with the 4 chemotherapeutic agents were estimated on the basis of the survival curves found in the trials. The ICON4 trial was used to provide survival data for paclitaxel combination, 039 was used for topotecan and paclitaxel monotherapy and trial 30-49 was used for PLDH. Survival curves were available from the ICON 4 and 039 trial reports. For PLDH, survival curves observed in trial 30-49 were obtained from a previous Schering-Plough submission to NICE (38). Life-years gained for the 4 agents were calculated as follows:

170 (Proportion of patients alive yearn + Proportion of patients alive yearn+1)/(2* 100) That is, the proportion alive between two periods was averaged and than divided by 100 to obtain per patient values.

Given the different follow-up period for the four agents (4.5 year for paclitaxel/platinum and for topotecan, 4 years for paclitaxel monotherapy, 3 years for PLDH), survival curves were truncated at 3 years to allow direct comparability. This analysis showed that paclitaxel/platinum was associated with the highest number of LYG (1.83) followed by PLDH (1.45), topotecan (1.37) and paclitaxel monotherapy (1.295) in a 3-year time-horizon. Results of LYG estimations and proportion of patients alive for each year are presented in Table 54.

Table 54. Results of LYG estimations for paclitaxel/platinum, paclitaxel monotherapy, topotecan and PLDH

Paclitaxel/ Paclitaxel Topotecan PLDH Platinum Years Proportion LYG Proportion LYG Proportion LYG Proportion LYG alive alive alive alive 0 100 100 100 100 1 78 0.89 50 0.75 56 0.78 56 0.78 2 43 0.60 22 0.36 22 0.39 30 0.43 3 24 0.33 15 0.18 18 0.20 18 0.24 Total 1.83 1.29 1.37 1.40

Summary of resource utilisation and costs data Only drug costs were considered in the analysis, including pre-medication costs, study drugs costs and administration costs. Data on resource use for drug dosage were taken from the trial data. For topotecan a mean dose of 1.5mg/m2 per cycle (for 5 consecutive days) was assumed, while for paclitaxel monotherapy a mean dose per cycle of 175mg/m2 was used. Both dosages were based on the 039 trial. The mean doses for PLDH were taken from the 30-49 trial, where patients received 50mg/m2 per cycle. Finally, for the combination therapy, carboplatin was assumed as the platinum agent of choice, given that 71% of women in the ICON 4 study had received this drug. In addition, for patients receiving paclitaxel (as monotherapy or in combination with carboplatin) pre-medication is required. Pre-medication drugs included

171 cimetidine (300 mg injection), dexamethasone (2x20mg tablets) and chlorpheniramine (10 mg injection). A total of 6 treatment cycles for each chemotherapeutic agent were considered. The rationale for the choice of this number of cycles was not given.

Administration costs were included assuming delivery on an outpatient basis. PLDH, paclitaxel monotherapy and paclitaxel/platinum each required one outpatient administration per cycle, for topotecan, however, 5 outpatient visits were necessary given the different administration process (over 5-days).

Unit costs for the study drugs (including pre-medications) were obtained from the British National Formulary(56) and the cheapest available formulation was used. Costs per outpatient visit (£68.00) were taken from the PSSRU (2000).(57) A unique price year was not given.

From the total per patient costs (per year) for the 4 chemotherapeutic agents that were calculated, paclitaxel monotherapy was the least expensive therapy (£7,157.76) followed by paclitaxel plus carboplatin (£9,527.52) and PLDH (£10,169.88). Topotecan was the most expensive drug (£11,475.00) mainly due to higher administration costs. Details on single costs items and methods of calculations are given in Table 50 in Appendix 8.

4.5 Summary of cost-effectiveness An incremental analysis was performed to compare the 4 chemotherapeutic agents in terms of costs and effectiveness. Total costs for the 4 drugs over 3 years were calculated by tripling the annual costs associated with each agent on the assumption of 6 treatment cycles per year. LYG estimated from trials survival curves were used as benefit measure (see table 54). A 6% discount rate was applied for costs and 1.5% for outcomes. The perspective of the analysis was that of the NHS.

Table 56 shows the results of the cost-effectiveness analysis. In the base-case analysis paclitaxel in combination with platinum was dominant compared to topotecan and PLDH (less costly and associated with a higher number of LYG). The options under consideration in the base-case analysis of the ICER were therefore paclitaxel in

172 combination with platinum and paclitaxel monotherapy. The incremental cost per LYG of paclitaxel/platinum compared to paclitaxel monotherapy was £12,120.

Table 55. Results of cost-effectiveness analysis for the 4 chemotherapeutic agents under study (in £) Treatment Total costs (3 Life years Average Incremental years) gained Cost/LYG cost/LYG Paclitaxel/ Platinum 25,287.79 1.784 14,176.47 X

Liposomal Doxorubicin 26,992.73 1.415 19,078.05 Dominated

Topotecan 30,297.51 1.338 22,646.89 Dominated

Paclitaxel 18,998.01 1.265 15,020.21 £12,120.07

One-way sensitivity analyses were conducted on the discount rate used for benefits, varying from 0% to 6%. The results of the base case did not vary substantially. Also, uncertainty in the benefit measure was considered varying the values of LYG by 15%. However, the impact of this variation was assessed only on average cost-effectiveness ratios and not in the incremental ratios, thus providing little additional information. Based on these results it was recommended that paclitaxel in combination with a platinum agent should be used as 2nd line treatment for platinum-sensitive relapsed ovarian cancer patients. This option was considered dominant compared to PLDH and topotecan and cost-effective compared with paclitaxel alone.

Comments The main objective of this submission was to assess the cost-effectiveness of paclitaxel in combination with a platinum agent compared to 3 other recommended therapies (PLDH, topotecan and paclitaxel alone). Effectiveness data for paclitaxel/carboplatin were obtained from a randomised clinical trial (ICON4). A model was constructed to compare survival estimates from the 3 trials (039, 30-49 and ICON4). The use of a model was necessary due to the lack of data for direct comparisons among the four therapies. However, the economic evaluation conducted presents some problems and limitations.

173

The main limitation of the study appears to be associated with the different characteristics of the patient population in the trials considered for the effectiveness analysis. The ICON4 study, from which survival rates were obtained for paclitaxel/platinum, included only platinum-sensitive patients who relapsed more than 6 months after the completion of a previous platinum chemotherapy. Instead, the patient population of the clinical trials from which survival rates were obtained for topotecan (trial 039), paclitaxel monotherapy (trial 039) and PLDH (mainly from trial 30-49) included both platinum-sensitive and platinum-refractory patients. It is well known that survival rates associated with platinum-sensitive patients are (on average) much higher than those for platinum-resistant and platinum-refractory patients (as illustrated, for example, in Table 54). It therefore seems inappropriate to compare the survival rates found in the ICON4 study with those of the 2 other clinical trials. It would have been more appropriate to compare ICON4 with the sub-group of platinum-sensitive patients in the 30-49 and 039 trials. Using the survival rates for PLDH found in trial 30-49 (table 54) for platinum-sensitive patients only, we estimated a total of 1.89 LYG over 3 years for PLDH. This value is higher than the figure found for paclitaxel/platinum in this submission (1.83). Also, considering only platinum-sensitive patients in trial 039 trial, LYG associated with paclitaxel alone would rise to 1.74 and to 1.56 for topotecan. It should however be noted that, while in trial 30-49 censored patients were excluded from the analysis and survival rates were calculated considering only patients alive and dead for every year, survival rates used in this submission for paclitaxel/platinum were obtained considering all patients at risk for each year (thus, including censored patient). This approach tends to underestimate the actual survival rate for paclitaxel/platinum because censored patients are treated as dead. In addition, LYG were calculated by taking the average number of patients alive between two periods (and than divided by 100 to obtain per patient values). This approach assumes that all deaths occurred exactly at the middle of the two periods considered (at month 6 of each year). This is clearly not the case in practice and the results obtained should be considered as an approximation of the real per patient LYG. It is not clear whether the approach used in this submission was due to lack of patient-level data or it was a methodological choice.

174 With regards to the cost-analysis undertaken, only resource use associated with study drugs and their administration were considered. No justification was given for the exclusion of adverse events costs. Given the relevance of this category of costs (for example, adverse event costs were the main reason for cost differences between topotecan and PLDH in the Smith analysis), their exclusion raises an issue as to the completeness of the economic analysis. In addition by assuming that all patients receiving paclitaxel combination received carboplatin as their platinum agent. This may underestimate the total costs of administration since a proportion of patients in ICON4 received cisplatin as opposed to carboplatin which is administered on an inpatient basis.

No justification was given for the number of treatment cycles included in the analysis (6). This was higher than the number of treatment cycles reported, for example, in trial 30-49 for both topotecan (5.75) and PLDH (4.87). This seems particularly relevant for topotecan that is associated with administration costs 5-times higher than its alternatives. It would also have been appropriate to adjust costs occurred in subsequent years by patient mortality. This adjustment does not appear to have been conducted in the submission. Moreover, total per patient costs over 3 years were calculated assuming 6 treatment cycles per year for each agent. Given the low response rate and the high discontinuation rate found in the clinical trial, this assumption does not seem plausible.

Finally, the issue of the uncertainty around the model parameters was only partly addressed. Only mean values for total costs and LYG were given. No statistical analyses were performed and confidence intervals were not presented. Sensitivity analyses for incremental ratios were performed by varying the discount rate for health benefits. LYG were varied by 15%, but its impact was assessed only for average cost- effectiveness ratios. No sensitivity analyses were conducted on other important items as drug dosage and number of treatment cycles. There are therefore some doubts as to the validity of the base case results.

In summary, the dominance of paclitaxel in combination with platinum over topotecan and PLDH appears questionable and some potential confounding factors have been found. Further analyses with head-to-head effectiveness data for the 4

175 therapies under study (when available) or at least a re-analysis based on comparable patient populations would appear necessary to corroborate the results of this submission.

4.6 Summary of findings from the cost-effectiveness review The review of economic evidence from the literature and industry submissions has highlighted a number of significant limitations. Of the published cost-effectiveness evidence reviewed, only the study by Smith was assessed from the perspective of the UK NHS. This study demonstrated significant cost savings with the use of PLDH compared to topotecan. Similar findings were reported in Ojeda et al.(51) and Capri et al.(52) from the perspective of non-UK health providers. The study by Smith was also used as the basis for the submissions by GlaxoSmithKline and Schering-Plough Ltd. Both of these submissions were based on cost-minimisation analysis but used alternative assumptions related to the management of adverse events and the drug administration costs associated with topotecan. However, none of these studies directly compared the full range of possible strategies that are relevant to the NHS (paclitaxel, PLDH, topotecan). Consequently it is not possible to make any direct comparison of the relative cost-effectiveness of these alternative treatments from this evidence.

The submission by BMS was the only study to attempt to make a direct comparison of the cost-effectiveness of the main licensed treatments in the NHS. However, as outlined earlier the proposed approach is subject to a number of important limitations and no adjustments were made for the different characteristics of the patient population in the trials. Consequently the dominance of paclitaxel in combination with platinum over topotecan and PLDH appears questionable.

In addition, the quality of life associated with disease states (progression-free or progressive) and toxicities has not been incorporated into any of the analysis undertaken previously. Ideally, a generic measure of health outcomes (e.g. quality- adjusted life-years [QALYs]) should be used to enable the cost-effectiveness results to be compared to other interventions in different disease areas.

176 Finally, the handling of uncertainty in all of the studies did not include a simultaneous assessment of all the uncertainty in variables included, that is a probabilistic sensitivity analysis. In those studies that did undertaken sensitivity analysis this was restricted to the use of 1-way analyses using minimum and maximum values restricted by the information provided by the expert panels.

In summary, the existing evidence relating to the cost-effectiveness of paclitaxel, topotecan and PLDH for the 2nd line treatment of advanced ovarian cancer has a number of limitations which make it insufficient to inform decision making regarding the most appropriate treatment for women treated in England and Wales. The following chapter therefore presents a new decision analytic model that has been developed to address this issue more formally.

177 5. ECONOMIC MODEL

5.1 Introduction

The review of cost-effectiveness studies in Chapter 4 identified a number of important limitations in the existing studies for assessing the cost-effectiveness of PLDH, topotecan and paclitaxel, in 2nd line advanced ovarian cancer. These limitations meant that it was not possible to make a reliable comparison of the relative cost- effectiveness of the alternative treatments on the basis of existing evaluations. To address the limitations and to facilitate a direct comparison of the relative cost- effectiveness of all relevant comparators, a new decision analytic model was developed. The new model provides a framework for the synthesis of data from the clinical effectiveness review and the industry submissions to identify the optimal treatment for advanced 2nd line ovarian cancer. The following sections outline the structure of the model in detail and provide an overview of the key assumptions and data sources used to populate the model.

5.2 Methods

Overview The model has been developed to estimate costs from the perspective of the UK NHS, and health outcomes in terms of life-years gained (LYG) and quality-adjusted life- years (QALYs) for the full range of relevant treatment strategies. An overview of the basic structure of the model is provided in Figure 22.

178

Figure 22. Structure of the economic model

mean_surv

Progressive Stable Dead Disease Disease mean_ttp (mean_surv - mean ttp)

Key:

mean_ttp = mean time to progression mean_surv = mean (overall) survival time

The model evaluates overall survival in relation to two distinct periods: the progression-free period, and the time from progression to death (calculated as the difference between overall survival and progression-free survival). The estimates of LYG were testimated using the following formula:

LYG = mean_ttp + (mean_surv – mean_ttp) where: Mean_ttp = mean time to progression (in years) Mean_surv = mean (overall) survival time (in years)

Overall survival is then quality-adjusted using separate utility weights for the two periods of time during which the average patient is stable (i.e. progression-free) or in progression. The estimates of QALYs are calculated as follows:

QALYs = mean_ttp*utility_sd + ((mean_surv – mean_ttp)*utility pd) where: Utility_sd = utlity weight for stable disease Utility_pd = utility weight for progressive disease

179 Although the structure of the model is relatively simple, the lack of direct data on the comparative survival for each of the relevant treatments under consideration presents a significant analytic challenge. In these situations it is important that the approach used to synthesise the available data are based on explicit and defensible methods which can used to combine data from all relevant sources. Due to the relative complexity of these approaches, the methods are described in detail in the following sections.

The costs included in the analysis comprised the costs of study drugs, pre-medication, monitoring, drug administration and the cost of managing adverse events. Due to the lack of data reported on the differential impact of the alternative treatments on the long-term management of patients (including subsequent lines of chemotherapy), the analysis of costs is confined to the initial treatment period only and hence discounting is not applied to these costs.

The model is probabilistic, that is relevant input parameters are entered as probability distributions to reflect their imprecision, and Monte Carlo simulation is used to reflect this uncertainty in the model’s results.(58) The impact of patient heterogeneity (e.g. platinum sensitive and platinum insensitive/refractory patients) on the adoption decision is explored in a series of separate analyses. This approach ensures that uncertainty in the decision, due to the imprecision in parameter inputs, can be separated from the uncertainty of whether an intervention is cost-effective for particular sub-groups of patients. In addition, sensitivity analysis was undertaken to determine the robustness of the results to alternative assumptions applied in the main analysis. A 2003-2004 price base is used, and annual discount rates of 1.5% for health outcomes are applied.(59)

Choice of outcome measures for the model The model builds on the trial-based evidence summarised in the accompanying systematic review of the effectiveness data (see Chapter 4.2). The primary health outcomes reported in the clinical trials included: response rate (complete or partial); duration of response; toxicity events; progression-free survival (PFS) and overall survival. Survival data were typically presented as median number of weeks (PFS and overall survival), Kaplan-Meier survival curves and hazard ratios between treatments.

180 Only limited evidence was found linking response to survival data, and survival data were not presented separately in the trials for responders and non-responders. Consequently, in the absence of comparative survival data reported by response, the primary outcomes assessed in the model were progression-free survival and overall survival reported for all patients. The use of survival data reported for all patients in the model enables a more comprehensive range of comparisons to be made and thus ensures a more systematic approach to study inclusion. The model evaluates overall survival in relation to two distinct periods: the progression-free period, and the time from progression to death (calculated as the difference between overall survival and progression-free survival).

As previously noted, PFS and overall survival were reported in several ways, including median number of weeks, Kaplan-Meier survival curves and hazard ratios. For the purposes of the model it was necessary to identify the most appropriate summary measure that would enable comparisons to be made between the alternative treatments. The hazard ratio represents the most accurate of these measures for comparing survival between treatments, as it is specifically designed to allow for censoring and time to an event. Furthermore, the use of the (log) hazard ratio and its variance allows studies to be pooled using conventional meta-analytic approaches. However, since the some of the studies did not report the hazard ratio, alternative methods were explored which would allow the ratio (and variance) to be estimated from other data reported in these studies. For those studies that did not report the hazard ratio, it is possible to estimate the statistic from either the median number of weeks (PFS and overall survival) or the published Kaplan-Meier curves, however both methods are subject to limitations. In order to estimate the hazard ratio from the median number of weeks’ survival, it is necessary to make an assumption about the distribution of survival times, for example that they follow an exponential distribution with a constant hazard. If the actual survival curves do not follow an exponential distribution, then it is possible that the use of the approach will not reflect the actual survival difference between treatments.

In order to estimate the hazard ratio from survival curves, one must measure the area under the survival curves, which must be extrapolated beyond the published curves to eliminate right- censoring. This method depends on the quality of the published

181 curves, i.e. the accuracy with which they can be measured, and the validity of the assumed distribution used to extrapolate those curves. Due to the limitations of these approaches and potential bias that the use of alternative approaches might introduce in estimating the relative treatment effects, the base case model (analysis 1) only included those studies which reported hazard ratios for survival data.

Trial data utilised and treatment strategies compared The clinical effectiveness review in chapter 4.2 identified 8 published RCTs, and obtained the trial report of one further RCT (30-57) in which paclitaxel, topotecan or PLDH were evaluated as second-line treatment for advanced ovarian cancer. Among these 9 studies, treatment effects were available in the form of hazard ratios from 7 (22, 24, 27-29, 32, 33), this includes hazards relating to the longer-follow-up of the 039 trial, provided by GSK via personal communication (see appendix 9 for details).

A total of 4 studies(30-33) were excluded from the model because the comparator groups (unlicensed treatments) provided no evidence on the relationship between the three main treatments under consideration. The excluded studies included the two trials that did not report hazard ratios.(30, 31) In each of these 4 trials each unlicensed comparator was uniquely represented, which means that the separate pair-wise comparisons cannot be linked to provide indirect evidence about the relative treatment effects of licensed comparators (see table 56). Consequently, only 5 of the 9 studies identified in the clinical review were eligible for inclusion in the model.

Table 59 presents the direct comparisons between relevant comparators for the 5 studies included. Of these studies, three reported results for the overall patient population (including platinum sensitive, platinum resistant and refractory patients),(22, 24, 27) and two reported results in platinum sensitive patients only.(28, 29) In the two studies in platinum sensitive patients, platinum and CAP were identified as relevant comparators (licensed for this indication) in this patient population and were thus included in the economic model.

182

Table 56. Direct comparisons of relevant treatment comparators according to the included patient population Treatments compared Paclitaxel Paclitaxel Topotecan PLDH Platinum CAP Trial combination

Overall patient population 039 9 9 30-49 9 9 30-57 9 9

Platinum sensitive patients ICON4 9 9 Cantu 9 9

Table 56 shows that in the overall patient population there are three different pair- wise comparisons of three treatments of interest (i.e. paclitaxel vs topetecan, PLDH vs topetecan and paclitaxel vs PLDH), but no one trial that compares all treatments simultaneously. In the effectiveness review, no attempt was made to formally synthesise the results for each drug across the trials, due to between-study heterogeneity. While this is a legitimate approach for the effectiveness review, there are potentially significant limitations from a decision-making perspective, particularly if the individual pair-wise comparisons give conflicting results. Furthermore, the existing evidence does not provide direct trial data on all the possible comparisons between the relevant comparators under consideration. Consequently is it not possible to base a single, coherent, comparative cost-effectiveness assessment of all treatments on the basis of the separate pair wise comparisons reported in the literature. It is possible to incorporate all this evidence simultaneously in the form of a mixed treatment comparison (MTC) model.(60, 61) Such a model provides an explicit analytical framework to identify the most cost-effective treatment strategy given the combined weight of evidence from all the relevant clinical trials. There are several examples in the literature of statistical models for combining mixed comparison evidence to provide a consistent set of treatment effect estimates, relative to a common baseline and this approach has been used in previous assessment models.(62) Using a similar approach, a model was developed to estimate a set of hazard ratios relative to a common baseline, using the Bayesian inference software program

183 WinBUGS (see section ‘mixed treatment comparison for the inclusion of trial 30-57’ below for further details of this analysis).(63)

The following sections summarise the approach used to derive a common baseline and describes in detail the MTC models developed to assess the relative effectiveness of the relevant treatment alternatives in each of the patient groups. The approaches used to address the heterogeneity in patient populations, reported in the trials, are also outlined.

Patient sub-groups

A patients’ likelihood of responding to second line therapy and hence their survival prospects are largely governed by their sensitivity or resistance to platinum-based therapies.(64) This source of heterogeneity makes a direct comparison between the results from the various trials problematic. While the patient population of the clinical trials (039, 30-49 and 30-57) included both platinum-sensitive and platinum- refractory patients, Cantu (2002) and ICON 4 included only platinum-sensitive patients who relapsed more than 6 months after the completion of a previous platinum chemotherapy. Failure to adequately reflect this heterogeneity in the model could introduce potential bias in the results due to the confounding effect of the different populations. Furthermore, the data for platinum and paclitaxel combination therapies, as assessed in ICON4, cannot be direct or indirectly linked to the data for paclitaxel, topotecan or PLDH. Since paclitaxel combination therapy and platinum therapy were not compared to any of the other comparators included in the economic model, there is no way to link their relative effectiveness to topotecan, paclitaxel or PLDH using only direct hazard ratios.

To address this issue, two separate analyses have been undertaken (Analysis 1 and Analysis 2). Analysis 1 is restricted to a comparison of the three trials (039,30-49 and 30-57) that included both platinum sensitive and resistant/refractory patients. This analysis provides a comparison of the cost-effectiveness of paclitaxel monotherapy, PLDH and topotecan in the overall population. As discussed in the effectiveness review section, due to the early termination of trial 30-57 the results of this trial are likely to be preliminary and as such the longer-term implications of any differences observed in the treatment effect at the point are unclear. Due to these limitations the

184 results of trial 30-57 are excluded from the base-case analysis. Sensitivity analysis is used to explore the impact of including the results of trial 30-57. A series of subgroup analyses are also undertaken to explore the cost-effectiveness of these interventions in the different patient populations.

Although this approach does not allow a comparison of the full-range of relevant comparators for platinum sensitive patients (paclitaxel and platinum combination, platinum monotherapy and CAP are excluded), it ensures that the potential confounding caused by differences between the inclusion criteria of the populations of the trials, are minimised.

Analysis 2 then broadens the model to include the full range of relevant comparators by relaxing the requirement for direct hazard ratios and by incorporating those licensed comparators that were not formally included in the systematic review (CAP and platinum). This approach enables data from ICON4 to be incorporated into the analysis at the expense of a less robust estimate of treatment effect. Analysis 2 focuses on data specific to platinum sensitive patients and uses subgroup specific data from trials 039 and 30-49.

5.2.1 Analysis 1: Base case model The base case model estimates the costs and quality-adjusted survival in an overall patient population, including platinum sensitive, platinum resistant and refractory patients. Although the evidence shows that baseline survival differs between these three groups, with platinum sensitive patients having the most favourable prognosis, there is less evidence to support a difference in relative treatment effect among these subgroups, as the clinical trials have failed to demonstrate a statistically significant difference in hazard ratios between these groups. Hence in the subgroup analysis, we also apply the overall treatment effect to subgroup specific baseline data to estimate the cost-effectiveness in particular patient groups. The use of subgroup-specific treatment effects is explored in a sensitivity analysis to determine whether this changes the results of the model.

It is important to note that the 30-49 and 30-57 trials did not report outcomes separately for platinum resistant and refractory patients, and instead treated these as

185 one group. Consequently the subgroup analysis is restricted to consideration to only 2 groups, platinum sensitive patients and the combined group of platinum resistant and refractory patients.

Calculating baseline survival for Analysis 1 In order to build the model, it was necessary to select one of the three treatments to provide a baseline PFS and overall survival against which the hazard ratios of the other two treatments could be compared. Trial 30-57 was ruled out as a provider of baseline information, as it was terminated early, and the length of follow-up was not available and was likely to have been 1 year or less. The data concerning median weeks’ survival from trial 039 was available for a longer period of follow-up, of around 4 years. However it was also limited, that is it would not allow specification of subgroup-specific baselines. Trial 30-49 provided 4 years follow-up, by which time 87% of patients had died, and allowed specification of subgroup-specific baselines, and baselines stratified by other covariates. Consequently trial 30-49 was chosen as the source of baseline data, and topotecan acted as the common comparator between the two completed trials (30-49 and 039).

Since none of the trials provided estimates of the absolute hazard of progression or death, it was necessary to estimate the baseline hazard using median weeks and an exponential approximation. This approach has been used in a previous technology assessment report looking at treatment for advanced breast cancer.(65) The baseline hazard (λ) and its variance are calculated according to the following formulae:

λ = -LN(0.5)/t Var(λ) = λ2/r

Where t = median weeks survival; r = number of events.

Using this approach, the baseline hazard (λ) can then be converted into a mean survival time, for progression-free survival and overall survival, by simply taking the inverse of the hazard (1/λ). This represents the mean survival times for topotecan. The mean PFS and overall survival for the 2 treatment comparators are then estimated by applying the hazard ratio (relative to topotecan) to the baseline hazard, in order to

186 estimate the absolute hazard for each of the comparators. Mean survival times for the comparators are then estimated using a similar approach to that described for topetecan (i.e by taking the inverse of the absolute hazards for each of the comparators). The following sections describe in detail the approach used to synthesise the effectiveness data to obtain pooled estimates for the hazard ratio for PLDH and paclitaxel relative to topotecan.

Mixed treatment comparison for the inclusion of trial 30-57 in Analysis 1

For the base-case model (described above) topotecan acts as a common comparator in both trials (039 and 30-49), therefore the hazard ratio for PLDH and paclitaxel (relative to toptecan) can be obtained directly from the reported trial data. Hence, although there is no direct comparison between PLDH and paclitaxel in trials 039 and 30-49, an indirect comparison can be made between PLDH and paclitaxel in the economic model by applying the hazard ratios associated with each intervention relative to this common baseline. However, the analysis is more complex when incorporating evidence from 30-57 as part of the sensitivity analysis.

Ideally, statistical inference concerning a comparison of three treatments, say A and B and C, would ideally be based on a direct ‘head-to-head’ RCT. However, in practice it is rare to find a RCT that directly compares all the relevant comparators. More often, one finds a mixture of comparisons looking at different subsets of the full set of relevant comparators. In the case of 3 pair-wise comparisons, AB, AC and BC, a traditional and simplistic method might be to compare direct treatment effects against a common baseline, say A, and as a consequence discard the information provided by the BC comparison. Indirect comparisons of A and C based, for example, on comparisons of AB and BC, are said to represent a lower level of evidence. However, it is evident that, based on the principle of transitivity, if the true differences between

AB, AC, and BC are θAB, θAC, and θBC, then we expect:

θAC = θAB + θBC

Thus, reasonable inferences can be made about the AC comparison with few additional assumptions over those, which are routinely made in simple meta-analyses. These assumptions are, firstly, the simple transitivity assumption outlined above.

187 Secondly, that the differences are taken on an appropriate scale; for example, the log- odds scale. Thus the information provided by the BC comparison need not be discarded, and can be used to update the direct comparisons of AB and AC. Higgins & Whitehead,(61) in particular, have shown how the use of ‘external’ AB and BC evidence can substantially reduce uncertainty about an AC comparison of primary interest.

Based on these general principles, a Bayesian meta-analysis of (log) hazard ratios assuming fixed treatment effects was conducted using Markov Chain Monte Carole (MCMC) implemented in WinBUGS. The analysis assumes that the (log) hazard ratios, observed in the clinical trials, are normally distributed about a true underlying effect size, θ, according to the precision (=1/variance), τ2, also observed in the trials. The underlying treatment effects are given independent vague priors, N(0, 0.001). The term "vague " is used to denote that prior information in the form of expert opinion or prior data is not included in the analysis, and hence these parameters are assigned very diffuse distributions.The main assumption of the analysis is that, had paclitaxel been included as a comparator in trial 30-49, or PLDH included as a comparator in trial 039, the observed relative treatment effect of paclitaxel compared to PLDH would have been the same as that observed in trial 30-57.

2 Log(HRtop_pac) ~ N(θtop_pac, τ top_pac); θtop_pac ~ N(0, 0.001) 2 Log(HRtop_PLDH) ~ N(θtop_PLDH, τ top_PLDH); θtop_PLDH ~ N(0, 0.001) 2 Log(HRpac_PLDH) ~ N(θpac_PLDH, τ pac_PLDH); θpac_PLDH = θtop_PLDH - θtop_pac

Where top_pac = topotecan versus paclitaxel; top_PLDH = topotecan versus PLDH; pac_PLDH = paclitaxel versus PLDH.

Thus the model updates the hazard ratios of topotecan versus paclitaxel and PLDH from trials 039 and 30-49 with the information on the comparison of paclitaxel and PLDH observed in trial 30-57. The WinBUGS code for the model is provided in appendix 9. This model was used to estimate overall survival for patients receiving topotecan, paclitaxel and PLDH as second-line treatment for ovarian cancer. An important assumption of the model is that trial 30-57 is given the same weight as the 2 completed RCTs. Trials 039 and 30-49 provide hazard ratios estimated after approximately 4 years’ follow-up. The hazard ratios from 30-57 were estimated after

188 a much shorter period of follow-up, of around 1 year although the precise length is unknown. They therefore incorporate a higher degree of censored data. Another key assumption of the model is that the hazard ratio is independent of the period of follow-up, hence the difference observed after 1-year follow-up is assumed to be equal to that would be observed at 4 years. Finally, we note that trial 30-57 was underpowered due to stopping recruitment early, however this will be incorporated in the model in the form of higher variance around the treatment effect.

Trial 30-57 only provided hazard ratios for overall survival, and so only trials 039 and 30-49 were used to estimate PFS in both the base-case and sensitivity analysis for Analyis 1.

2 Log(HRtop_pac) ~ N(θtop_pac, τ top_pac); θtop_pac ~ N(0, 0.001) 2 Log(HRtop_PLDH) ~ N(θtop_PLDH, τ top_PLDH); θtop_PLDH ~ N(0, 0.001)

Where top_pac = topotecan versus paclitaxel; top_PLDH = topotecan versus PLDH

The hazard ratios for PFS and overall survival of paclitaxel and PLDH compared to topotecan are then multiplied by the respective absolute hazard for the baseline to calculate the absolute hazard of PFS and overall survival for paclitaxel and PLDH. The absolute hazards (λ) for each treatment are then converted into mean PFS and overall survival by taking the inverse of the hazard (1/λ).

In addition to the inclusion of trial 30-57, a further sensitivity analysis of the baseline model, was conducted. This explored the cost-effectiveness for particular sub-groups (platinum sensitive and platinum resistant/refractory) using the treatment effects reported for all patients and also the sub-group specific treatment effects. Sensitivity analysis around particular cost assumptions, namely the inclusion of additional ECG cardiac monitoring costs for PLDH and home administration of topotecan, were also conducted.

189 5.2.2 Analysis 2: Platinum sensitive model

Calculating baseline survival for Analysis 2

Since prognosis is better in platinum sensitive patients, it would not be appropriate to compare the absolute hazard associated with platinum combination therapy in ICON4 to the absolute hazard of topotecan estimated from a population including platinum resistant and refractory patients. Thus we used the subgroup-specific survival data reported for topotecan in trial 30-49 for platinum sensitive patients.

Treatment strategies compared Analysis 1, the base case model, focused on paclitaxel, topotecan and PLDH from trials 039 and 30-49. As shown in Table 56 there are two further trials, providing 3 more relevant comparators. Both of these trials (ICON4 and Cantu) looked specifically at one of the patient subgroups, and recruited only platinum sensitive patients. In the trial comparing paclitaxel and CAP (Cantu), the patient population was restricted to those patients who had relapsed greater than 12 months after completion of therapy, rather than 6 months. The more common 6-month cut-off was used to define the platinum sensitive population in the trials recruiting an overall patient population (22, 24), and ICON4 also included patients relapsing between 6 and 12 months after completion of therapy. Of the 3 additional comparators, paclitaxel combination therapy was the only one formally included in the systematic review, by virtue of the presence of paclitaxel, and hence it can be compared on an equal basis to paclitaxel alone, topotecan and PLDH. However, as can be seen from Table 56, there is no common comparator between ICON4 and any other trial, and consequently it was not possible to include it in a MTC as described in Analysis 1. In order to estimate a treatment effect of platinum combination therapy compared to topotecan, it was necessary to break randomisation. Specifically, it was necessary to use data reported on the median weeks’ survival and an exponential approximation to estimate the absolute hazard associated with paclitaxel combination and topotecan respectively, and then take the ratio of these to provide the relative treatment effect. Thus the estimated treatment effect of paclitaxel combination compared to topotecan in Analysis 2 is not as robust as the treatment effects estimated in Analysis 1.

190 Indirect estimate of treatment effect In the same way in which the baseline is estimated in Analysis 1, the absolute hazards, λ, are calculated using median weeks’ survival and an exponential approximation. The log absolute hazard is assumed to be normally distributed around a true underlying absolute effect, θ, which is given a vague prior, N(0, 0.001). Relative treatment effects for paclitaxel combination compared to topotecan can then be estimated as the difference between the respective log absolute hazards.

2 LN(λpacpt) = LN(-LN(0.5)/tpacpt) ~ N(θpacpt, τ pacpt); θpacpt ~ N(0, 0.001) 2 LN(λtop) = LN(-LN(0.5)/ttop) ~ N(θtop, τ top); θtop ~ N(0, 0.001)

θtop_pacpt = θtop - θpacpt

Where pacpt = platinum combination; top = topotecan; top_pacpt = topotecan versus paclitaxel combination; t = median weeks survival.

Using this method, one can also get a relative treatment effect for platinum compared to topotecan. With respect to CAP, the trial provides us with a hazard ratio of CAP compared to paclitaxel, which can be used directly in the mixed treatment comparison to maintain randomisation and the use of the most reliable estimate of treatment effect, the hazard ratio. The WinBUGS code for Analysis 2 is provided in Appendix 9. To maintain correlation between the posterior survival estimates, the simulated output were exported directly into Excel.

5.3 Model inputs Data used to populate the 2 analyses came from a variety of sources; these are listed in detail below.

Survival

Five of the nine trials identified in the clinical review were used to inform the survival estimates used in the models.(22, 24, 27-29) The data available from the trials was utilised differently for the 2 main analyses (detailed above) and the various sensitivity analysis. The (log) hazards for progression –free and overall survival, used in the 2 main analyses are presented below in table 57. The estimates used for the various sensitivity analyses are shown in table P in Appendix 11.

191

As detailed above, survival data was utilised in the following ways: Analysis 1 used overall and progression free survival hazard ratios generated from the 039, 30-49, to compare topotecan, paclitaxel and PLDH, for the overall trial population, platinum sensitive patients and platinum resistant/refractory patients (assuming a common treatment effect across subgroups). Sensitivity analyses were undertaken to incorporate evidence from the 30-57 trial, and to explore the impact of using subgroup specific treatment effects. Finally Analysis 2 incorporates additional evidence available from the ICON4 and Cantu et al. trials to explore the impact of extending the cost-effectiveness analysis to include the full range of treatment comparators that are relevant in the platinum sensitive group (topotecan, paclitaxel monotherapy, PLDH, paclitaxel + platinum and CAP).

Quality of life

The survival analysis described in previous sections details the approaches and inputs used to generate mean estimates for the survival times (overall survival and progression-free survival) associated with each intervention. For the purposes of estimating QALYs it was necessary to quality-adjust these survival times using an appropriate utility or preference score(National Institute for Clinical Excellence, 2004).(66)

In addition to the differences in relative survival estimated for each treatment, the different toxicity profiles and impact on progression of the alternative chemotherapy drugs may have important implications for the quality of life (QoL) for patients. As previously reported in the effectiveness review, quality of life data was reported in both the ICON4 and 30-49 trials, using the EORTC cancer quality of life questionnaire.(67) This questionnaire encompasses 6 domains and 8 symptoms scales and is designed to assess the impact of cancer treatments on patients QoL. For the purposes of the economic model this data presents a number of potential limitations. Firstly, the reported data only provides a profile of a persons’ QoL and is therefore unsuitable for inclusion into an economic evaluation, where a single index preference weighted score is required. Secondly, QoL estimates using the reported EORTC data were not available for all treatments considered.

192 Due to the limitations of QoL data reported in the main trials a separate systematic search was undertaken to identify potential sources of utility data in ovarian cancer. Full details of the search strategy are reported in Appendix 9. In total, 5 studies were identified that reported utility estimates for advanced ovarian cancer: Tengs,(68) Ortega,(69) Bennett,(70) Grann(71) and Calhoun.(72) 4 of these studies provided utility estimates relating to specific toxicity events experienced as a result of chemotherapy.(69-72) However, these events were not consistent with the toxicity events reported in the trials, and therefore could not be linked to the trial data identified in the clinical review. Since it was not possible to accurately reflect the difference in toxicity events between the treatments using a utility-adjustment, an overall utility weight was applied to the survival estimates to reflect the different length of time spent on the progression free and progressive states differed between drugs.

Although an overall estimate for stable disease (0.63) in advanced ovarian cancer was reported in Tengs(68) no separate utility estimates were identified a part of the systematic search for the utility associated with progressive disease. In the absence of direct estimates, we looked for alternative estimates from other related areas in order to provide a proxy measurement. The Tengs study provided a comprehensive list of utlity values in a range of other female cancers which provided estimates of the utility decrement between stable and progressive disease. Although these utilities were not in ovarian cancer patients, it was possible to calculate the utility decrement, between the progression-free and progressive state, and apply this decrement as a proxy estimate of progression-free QoL for ovarian cancer patients (0.63) obtained from the literature.(68) We selected a single study in breast-cancer as the best proxy measure, since the estimate of stable disease was identical to that reported for ovarian cancer patients. In addition, the study by Brown(73) provided estimates utilities using the time-trade-off technique(74) a method with strong theoretical foundations that has been well validated for estimating utilities for health states in cancer. The utility for the progressive period was then calculated by applying the utility decrement of 0.29, estimated as the difference between the utitlity of stable and progressive disease reported in the study by Brown, to the overall estimate of 0.63 for stable disease.

193 Cost Analysis The costs included in the model are those considered to be the key components of treatment costs associated with advanced ovarian cancer; and which are likely to differ between the various treatments considered. These include the acquisition costs of the drug, the costs of monitoring and administration and the costs of managing the adverse events. For the analysis of adverse events it was necessary to synthesise the data from the various trials. The approach is described in detail in the following section.

Adverse events

All patients, in all trials experienced some form of adverse event, however the grade and type of adverse event is often related to the regime received. Topotecan is particular is associated with the occurrence of neutropenia and PLDH with PPE. Each of the adverse events has different resource use implications. In addition the resource use associated with adverse events increases dramatically for high grade (3 and 4) events. It was therefore necessary to cost the adverse events observed in the trials, for each of the treatment comparators.

The proportions of patients experiencing adverse events neutropenia, thrombocytopenia, PPE, diarrhoea, nausea/vomiting, stomatitis/pharyngitis, anaemia, sepsis and fever, during treatment, were available from the 30-49, 039, 30-57, ICON4 and Cantu trials, although some of the data were missing for Cantu and ICON4. Data for the platinum therapies CAP (from Cantu) and carboplatin/cisplatin (from ICON4) was not reported for some events, and therefore an additional trial. (Bolis, 2001)(75) reporting on adverse events for platinum therapy was used to supplement these. As reported by Smith.(50) the use of the proportions of adverse events is likely to involve double counting for certain resource use items such as hospitalisations and assumes that each patient only sustained one of each type of adverse event. However without access to patient level data it was not possible to account for simultaneous adverse events and multiple events for all comparators. Although a detailed patient level analysis was conducted by Smith(50) this analysis was also only relevant for the comparison between topotecan and PLDH. In order to apply a consistent method, which would enable the full range of comparators to be assessed, it was necessary to use data on adverse events that was reported using a common format.

194

Data on grades 1 and 2 adverse events were particularly poor. This is probably because the trials concentrated on reporting adverse events that would require a change in therapy or discontinuation. Grade 1 and 2 events, were therefore not included in the cost-analysis. These are likely to incur little or no cost, as the majority of hospitalisations for grades 1 and 2 adverse events will only occur as part of trial protocol, see Smith(50) as such the remaining cost of adverse events 1 and 2 is negligible (zero in many cases).

Given that we are only including grades 3 and 4 adverse events, the possibility of multiple events and more than 1 adverse event (requiring hospitalisation) in the same time period, is likely to be less than if we were to include all adverse events (grades 1- 4).

The proportion of grades 3 and 4 adverse events, from the trials, were therefore used to calculate the total cost of adverse events for each of the 6 treatments, using resource use data reported in the Schering submission (see below). As with the survival data a mixed comparisons model was used to estimate the probabilities of adverse events using data from the 6 trials included in the analysis. For Analysis 1 adverse events data from the 30-49,039 and 30-57 trials were used and for Analysis 2, adverse events data from 30-49, 039, 30-57, Cantu, ICON 4 and Bolis trials were used.

In a similar manner to approach used to analyse survival data, the analysis of adverse events was undertaken using a Bayesian meta-analysis to account for the mixed and indirect comparisons for the various treatments reported in the different trials. The Winbugs model used to estimate probabilities of adverse events, using the trial data assumes a regression-like structure, with the logit of the probability of an adverse event for any treatment k, depending on a ‘baseline’ term (for topetecan) µi in trial i, i = 1,2, …. 5, and a fixed treatment effect δk. The trial-specific baselines are drawn from a common random normal distribution, whose parameters must be estimated from the data, given vague priors. Formally, this can be expressed as:

k k logit ( pi ) = µi + δ

µi ~ N(µ.b,τ.b) ; µ.b ~ N(0,0.0001), τ.b ~ gamma(0.01,0.01)

195

The treatment effects δk are also given independent vague priors, N(0,0.0001). A binomial likelihood is assumed from the available data points:

k k k ri ~ Bin(pi , ni ), where k denotes all treatment indices in study i .

The Winbugs code for analysis 1 (topotecan, paclitaxel and PLDH) and analysis 2 (topotecan, paclitaxel, paclitaxel combination, platinum and CAP) are reported in Appendix 10. Unlike the survival data adverse events data from 30-57 was used for the base case model (analysis 1) and analysis 2. This is because adverse events were observed only during the treatment period; thus there is no problem of short follow up, as observed with the 30-57 survival data. The mean probabilities of adverse events estimated using the mixed comparisons model, for Analyses 1 and 2, can be seen in table 62.

Unit costs and resource use

The cost inputs used in the model are listed below in table 59. Drug costs were taken from the British National Formulary.(56) Other sources included industry submission data, earlier published estimates in the area and national unit cost databases. Resource use associated with adverse events was taken from information supplied as part of the Schering submission.

Drug costs were calculated according to the dosages reported in the trials, this was also the licensed dose. Pre-medication as reported in the trials was also included. This consisted of the intravenous administration of metoclopramide, chlorpheniramine and cimetidine, for relevant drugs, prior to administration of the study drug. Saline was used to administer the iv concentrate; this was costed according to millilitres as specified by a clinical expert (50-500mL @ 0.6p per mL). The total cost of drugs were calculated by multiplying the cost per cycle by the average number of cycles received, reported in table 58.

Chemotherapy was assumed to be administered on an outpatient basis for all regimes not containing cisplatin, that is topotecan, PLDH and paclitaxel. A single visit was

196 required for each cycle, apart from for topotecan, which is administered over 5-days thus necessitating 5 outpatient visits per cycle. For CAP, the 11% of patients in ICON4 receiving cisplatin platinum therapy and the 10% of patients in ICON4 receiving cisplatin plus paclitaxel; on the advice of a clinical expert; chemotherapy was assumed to be administered on an inpatient basis. This is because cisplatin is potentially severely nephrotoxic and thus to administer it safely it is necessary to create a sodium chloride duiresis. This is most often achieved by intravenous infusion of saline solution (usually 0.9%) and a diuretic (most commonly mannitol). This diuresis is maintained during the cisplatin infusion and for several hours afterwards. This whole process may take between 7 and 12 hours for moderate and high doses of cisplatin, and therefore require an overnight admission to hospital.

The home administration of topotecan was included as a sensitivity analysis, as it is not common practice in the UK and does not represent the method of administration used in the trials. Home administration of topotecan; as advised by GSK; requires a daily visit by a district nurse and a consultation with a specialist physician each cycle. The total cost of administration was calculated by multiplying the cost per cycle by the average number of cycles received, reported in table 58.

The cost of monitoring patients for toxicity events, in terms of a chemical pathology monitoring, haematological toxicity monitoring and biochemical monitoring, was added to all comparators in the model. The total cost of monitoring was calculated by multiplying the cost per cycle by the average number of cycles received, reported in table 58. The potential elevated risk of cardiac toxicity in the PLDH strategy was highlighted in the industry submissions,(76, 77) and the Summary of Product Characteristics (SPC) for PLDH recommends that all patients should routinely undergo frequent ECG monitoring. However after clinical consultation it was concluded that tests for cardiac function (e.g. ECG, echocardiogram and MCUG) would not be routinely used in practice for these patient groups for PLDH. This is due to the lower cardiac toxicity observed with PLDH compared to doxorubicin and its use in the palliative setting, when life expectancy is short and patients and clinicians are often prepared to give concerns about late toxicity less weight than in the curative setting. Hence in the base-analysis no additional costs are assigned to PLDH for cardiac monitoring. The robustness of this assumption is explored in a sensitivity

197 analysis that included the cost of an ECG for each cycle of PLDH. The use of echocadiography and multigated angiography (MUGA) were not included since the use of these more costly and invasive tests (mainly used for the evaluation of left ventricular function) is only considered mandatory before each additional administration than exceeds a lifetime dose of PLDH of 450mg/m2 , which exceeds the lifetime dose considered in this model.

Grades 3 and 4 adverse events were costed using resource use data reported in the Schering-Plough submission.(23) In this submission resource use associated with various adverse events by level of severity were estimated by a group of clinical experts. The adverse events that incurred a cost were consistent with the adverse events reported by trials, namely neutropenia, thrombocytopenia, anaemia, stomatitis/pharyngitis, PPE, nausea/vomiting, diarrhoea, sepsis and fever. Some of these resource use items are associated with uncertainty, the distributions used to reflect this uncertainty is shown in table Q in Appendix 11. These were again taken from data reported in the Schering-Plough submission(23).

198 Table 57. Survival data used in analysis 1 and 2

Name Overall trial Platinum sensitive Platinum Source population resistant/refractory Analysis 1 Overall Topotecan ‘v’ Paclitaxel 0.914 (0.681,1.226) 1.01 (0.663,1.541) 0.738 (0.498,1.093) 039 survival Topotecan ‘v’ PLDH 1.216 (1,1.478) 1.432 (1.066,1.923) 1.069 (0.823,1.387) 30-49 hazards Progression Topotecan ‘v’ Paclitaxel 0.811 (0.603,1.092) 0.823 (0.538,1.261) 0.749 (0.501,1.121) 039 free survival Topotecan ‘v’ PLDH 1.118 (0.928, 1.347) 1.287 (0.977,1.694) 0.992 (0.77,1.279) 30-49 hazards

Analysis 2 Overall Topotecan ‘v’ Paclitaxel - 1.01 (0.663,1.541) - 039 survival Topotecan ‘v’ PLDH - 1.432 (1.066,1.923) - 30-49 hazards Paxlitaxel combination ‘v’ - Paxlitaxel + pt = - ICON4 platinum11 0.005(0.004,0.006) Pt = 0.006(0.005, 0.007) Paclitaxel ‘v’ CAP - 1.724 (1.02,2.94) - Cantu Progression Topotecan ‘v’ Paclitaxel - 0.823 (0.538,1.261) - 039 free survival Topotecan ‘v’ PLDH - 1.287 (0.977,1.694) - 30-49 hazards Paxlitaxel combination ‘v’ - Paxlitaxel + pt= - ICON4 platinum 0.012(0.01, 0.013) Pt =0.015 (0.013, 0.017) Paclitaxel ‘v’ CAP - 1.66 (1.03,2.70) - Cantu

11 Median survival was used to calculate absolute hazards for 2 treatments. Hazards ratio’s compared to Topotecan baseline was then calculated. Absolute hazards derived from trial data are reported in the table for ICON4 comparators.

Table 58. Other trial data Name Value Distribution Source Average number of Topotecan 5.72 Fixed Mean of 039 and 30-49 cycles Paclitaxel 4.82 Fixed 039 PLDH 4.87 Fixed 30-49 Paclitaxel + platinum 5.56 Fixed ICON4 Platinum 5.26 Fixed ICON4 CAP 6 Fixed Cantu

Topotecan Paclitaxel PLDH Paclitaxel + Platinum CAP Sources(s) platinum Probability of Stomatitis/pharyngitis 0.01 0.01 0.09 - - - 039, 30-49, experiencing adverse grade 3 30-57 events (analysis 1) Stomatitis/pharyngitis 0.01 0.00 0.01 - - - grade 4 PPE grade 3 0.00 0.01 0.20 - - - PPE grade 4 0.00 0.01 0.01 - - - Neutropenia grade 3 0.13 0.21 0.09 - - - Neutropenia grade 4 0.58 0.13 0.06 - - - Thrombocytopenia 0.19 0.01 0.01 - - - grade 3 Thrombocytopenia 0.21 0.01 0.00 - - - grade 4 Anaemia grade 3 0.29 0.04 0.05 - - - Anaemia grade 4 0.04 0.02 0.051 - - - Diarrhoea grade 3 0.01 0.02 0.02 - - - Diarrhoea grade 4 0.01 0.01 0.01 - - -

200 Nausea/vomiting 0.16 0.04 0.13 - - - grade 3 Nausea/vomiting 0.04 0.01 0.02 - - - grade 4 Sepsis grade 3 0.01 0.01 0.00 - - - Sepsis grade 4 0.00 0.00 0.00 - - - Fever grade 3 0.06 0.02 0.04 - - - Fever grade 4 0.02 0.01 0.00 - - - Probability of Stomatitis/pharyngitis 0.01 0.01 0.01 0.01 0.01 0.01 039, 30-49, experiencing grade 3 30-57, Stomatitis/pharyngitis 0.00 0.01 0.01 0.01 0.01 0.00 Cantu, adverse events grade 4 ICON4, (analysis 2) PPE grade 3 0.00 0.01 0.20 0.01 0.01 0.20 Bolis PPE grade 4 0.00 0.01 0.01 0.01 0.01 0.01 Neutropenia grade 3 0.11 0.18 0.07 0.18 0.16 0.23 Neutropenia grade 4 0.52 0.09 0.04 0.09 0.17 0.38 Thrombocytopenia 0.18 0.01 0.01 0.01 0.13 0.12 grade 3 Thrombocytopenia 0.20 0.01 0.00 0.01 0.01 0.02 grade 4 Anaemia grade 3 0.3 0.04 0.05 0.04 0.05 0.07 Anaemia grade 4 0.04 0.02 0.00 0.02 0.05 005 Diarrhoea grade 3 0.05 0.02 0.03 0.02 0.03 0.03 Diarrhoea grade 4 0.01 0.01 0.01 0.01 0.01 0.01 Nausea/vomiting 0.15 0.37 0.13 0.08 0.09 0.06 grade 3 Nausea/vomiting 0.04 0.01 0.02 0.01 0.02 0.03 grade 4 Sepsis grade 3 0.01 0.01 0.01 0.01 0.01 0.01 Sepsis grade 4 0.01 0.01 0.01 0.01 0.01 0.01

201 Fever grade 3 0.07 0.03 0.04 0.03 0.02 0.04 Fever grade 4 0.02 0.01 0.01 0.01 0.02 0.00

Table 59. Details of cost data

Name Value Distribution Source (s) Drug costs Topotecan, per cycle £790 Fixed BNF (including Paclitaxel, per cycle £660 Fixed 30-49, 039, ICON4 and Cantu trials for dosages pre- PLDH per cycle £1630 Fixed Paclitaxel + platinum £1029 Fixed medication) Platinum £376 Fixed CAP £118 Fixed Drug Topotecan, per cycle £450 Fixed Schering submission and clinical administration Paclitaxel, per cycle £90 Fixed advice PLDH, per cycle £90 Fixed Paclitaxel + platinum £113 Fixed Platinum £111 Fixed CAP £293 Fixed Home Per cycle £387 Fixed GSK submission, PSSRU and administration clinical advice of Topotecan Monitoring Topotecan, per cycle £8.85 Fixed BNF Paclitaxel, per cycle £8.85 Fixed PLDH, per cycle £8.85 Fixed Paclitaxel + platinum £8.85 Fixed Platinum £8.85 Fixed CAP £8.85 Fixed ECG (PLDH sensitivity analysis) £65.24 Fixed GSK submission Adverse Stomatitis/pharyngitis grade 3 £143.81 See resource use Schering submission events Stomatitis/pharyngitis grade 4 £2013.77 Table 35 in PPE grade 3 £93.38 Appendix 10

202 PPE grade 4 £2153.50 Neutropenia grade 3 £54 Neutropenia grade 4 £1419.65 Thrombocytopenia grade 3 £0 Thrombocytopenia grade 4 £1016 Anaemia grade 3 £389.8 Anaemia grade 4 £682.8 Diarrhoea grade 3 £1475.92 Diarrhoea grade 4 £1475.92 Nausea/vomiting grade 3 £477.11 Nausea/vomiting grade 4 £1882.15 Sepsis grade 3 £1748.23 Sepsis grade 4 £4931.69 Fever grade 3 £1748.23 Fever grade 4 £4931.69

203 5.4 Analytic methods The output from the meta-analyses undertaken in WinBUGs were imported directly into Microsoft ® Excel 2000. These were then combined with data on resource use and cost in order to obtain the mean estimates for the outcomes of interest and their associated uncertainty. The results are presented in two ways. Firstly, mean costs and QALYs for the various comparators are presented and their cost-effectiveness compared, estimating incremental cost-effectiveness ratios (ICERs) as appropriate, using standard decision rules. The ICER examines the additional costs that one strategy incurs over another and compares this with the additional benefits. When more than two interventions are being compared the ICERs are calculated using the following process:

i) The strategies are ranked in terms of cost (from the least expensive to the most costly). ii) If a strategy is more expensive and less effective than the previous strategy, then this strategy is said to be dominanted and is excluded from the calculation of the ICERs. iii) The ICERs are calculated for each successive alternative, from the cheapest to the most costly. If the ICER for a given strategy is higher than that of the next more effective strategy, then this strategy is ruled out on the basis of extended dominance.

Finally, the ICERs are recalculated excluding any strategies that are ruled out by principles of dominance or extended dominance.

Given that mean costs and QALYs gained are estimated with uncertainty, the output from the simulations were then used to generate cost-effectiveness acceptability curves (CEACs) for the alternative analyses. These curves detail the probability that each intervention is cost-effective over a range of potential maximum values that the health services is prepared to pay for an additional QALY(78).

5.5 Results

Costs As noted earlier, the total cost does not vary between sub-groups, as cost parameters were not available by specific subgroups, and they apply only to the active treatment phase during which all patients are assumed to be alive in the model. Table 60 presents this cost data broken down into the cost of the study drug, the cost of administration, the cost of monitoring for adverse events and the total cost of managing any adverse events, and this applies to all analyses.

Table 60. Cost results broken down by category

Topotecan Paclitaxel PLDH Cost of drug £4,223 £3,180 £7,936 Cost of administration £2,405 £434 £438 Cost of monitoring £47 £43 £43 Cost of adverse events £1,773 £490 £484 Total cost £8,448 £4,146 £8,902

Paclitaxel emerges as the cheapest treatment strategy in nearly every category and overall. Topotecan is the next cheapest option, costing roughly twice as much as paclitaxel per patient. This is due to slightly higher drug costs, higher costs of administration (as each cycle is given over 5 days, instead of 1 day as is the case with paclitaxel and PLDH), and higher costs of adverse events due mainly to the higher incidence of neutropenia. PLDH is the most expensive treatment strategy, and this is due almost entirely to the higher acquisition cost of the drug. In terms of the cost of managing adverse events, PLDH is associated with the lowest cost of managing adverse events (approximately £1,289 less than topotecan and £4 less than paclitaxel).

In the submission by GSK, the monitoring cost for PLDH included a cardiac monitoring component not attributed to the other drugs (ECG followed by either echocardiogram or MCUG). As discussed earlier, after consultation with our clinical advisor we assumed that additional cardiac monitoring for PLDH was unlikely to occur in practice. As a sensitivity analysis, the results of the model were re-estimated, including the cost of an ECG before every cycle of PLDH. In this scenario the totoal cost of PLDH increased from £8,902 to £9,220.

205

The submission by GSK also described a novel method of home administration of topotecan. In the GSK submission the costs of community nurse visits were not considered in their costings. In this sensitivity analysis we assumed that provision of topetecan at home would require five 1-hour visits (at £56 per hour), plus the costs of a consultation with a consultant at every cycle. The total costs associated with topotecan administered at home was estimated to be £8111, representing a saving of £337 compared to the estimates for topetecan applied in the base-case analysis.

5.5.1 Analysis 1 – base case Table 61 presents the analysis of the incremental cost-effectiveness ratio (ICER) in the overall patient population, and also where those same treatment effects are applied to a platinum sensitive or platinum resistant and refractory baseline. In Analysis 1 the treatments compared are topotecan, paclitaxel and PLDH.

Table 61. Cost-effectiveness results from analysis 1: base case model Quality- Probability cost-effective for a PFS OS adjusted maximum WTP: Treatment Cost ICERa (wks) (wks) survival £10,000 £30,000 £50,000 (wks) Analysis 1 – overall patient population Topotecan 24.5 86.0 34.2 £8,448 ED 0.00 0.02 0.02

Paclitaxel 20.1 79.7 30.9 £4,146 - 0.99 0.37 0.20

PLDH 27.5 104.8 40.9 £8,902 £24,606 0.01 0.61 0.78 Analysis 1 – treatment effects applied to platinum sensitive baseline Topotecan 33.2 101.6 41.8 £8,448 ED 0.00 0.02 0.03

Paclitaxel 27.2 94.1 37.6 £4,146 - 0.95 0.29 0.16

PLDH 37.2 123.6 49.8 £8,902 £20,212 0.05 0.69 0.81 Analysis 1 – treatment effects applied to platinum resistant/refractory baseline Topotecan 19.8 61.2 25.1 £8,448 ED 0.00 0.01 0.02

Paclitaxel 16.3 56.7 22.6 £4,146 - 1.00 0.57 0.29

PLDH 22.2 74.6 30.0 £8,902 £33,430 0.00 0.42 0.69 a Cost per quality adjusted life year (QALY). PFS = progression-free survival; OS = overall survival; WTP = willingness to pay per QALY.

Using the decision rules for calculating the ICERs described earlier, topotecan is ruled out by extended dominance by PLDH. The options under consideration in the base- case analysis of the ICER are therefore PLDH and paclitaxel. The ICER for PLDH

206 compared to paclitaxel is £24,606 per QALY in the overall patient population. Topetecan continues to be ruled out on the basis of extended dominance in the subgroup analyses undertaken for platinum sensitive and platirnum resistant/refractory patients. The ICER for PLDH compared to placlitaxel is £20,212 in the platinum sensitive population, and £33,430 in the platinum resistant and refractory population.

In the overall population and in each of the subgroups, as the willingness to pay (WTP) per quality-adjusted life year (QALY) increases, so does the likelihood that PLDH is the most cost-effective treatment strategy. In the overall patient population, at a WTP of £10,000 per QALY, PLDH has only a 1% probability of being the most cost- effective treatment strategy, rising to 61% at a WTP of £30,000 per QALY, and 78% at a WTP of £50,000 per QALY. The CEACs for the overall population, showing the probability that each comparator is cost-effective over a range of threshold values, is provided in Figure 23.

Figure 23. Analysis 1, overall population

CEAC Analysis 1 - overall population

0.9

0.8

0.7

0.6

0.5 Topotecan Paclitaxel 0.4 PLDH

0.3 Probability cost-effective

0.2

0.1

0 £0 £10,000 £20,000 £30,000 £40,000 £50,000 £60,000 -0.1 Threshold for cost-effectiveness

207 The pattern observed for the overall population is similar in the platinum sensitive and platinum resistant and refractory populations, although the the estimate of the ICER varies markedly. PLDH looks more favourable in a platinum sensitive population because the relative treatment effects estimated for PLDH are applied to a higher baseline survival (and a lower baseline survival for resistant refractory patients; hence the less favourable ICER in this subgroup. The decision uncertainty, in the form of a CEAC, for platinum sensitive patients and refractory patients, can be seen in Figures 24 and 25 respectively.

Figure 24. Analysis 1, platinum sensitive patients

CEAC Analysis 1 - platinum sensitive population

0.9

0.8

0.7

0.6

0.5 Topotecan Paclitaxel 0.4 PLDH

0.3 Probability cost-effective

0.2

0.1

0 £0 £10,000 £20,000 £30,000 £40,000 £50,000 £60,000 -0.1 Threshold for cost-effectiveness

208 Figure 25. Analysis 1, platinum resistant/refractory patients

CEAC Analysis 1 - platinum resistant/refractory population

0.9

0.8

0.7

0.6

0.5 Topotecan Paclitaxel 0.4 PLDH

0.3 Probability cost-effective

0.2

0.1

0 £0 £10,000 £20,000 £30,000 £40,000 £50,000 £60,000 -0.1 Threshold for cost-effectiveness

Patients in the platinum resistant and refractory subgroup may not be elgible to receive paclitaxel as second-line therapy if they received it as part of their first-line treatment. Thus in platinum resistant and refractory patients who received paclitaxel as first-line therapy, the relevant comparators may be limited to topotecan and PLDH. In this scenario the ICER of PLDH compared to topotecan in these patients is £4,850 per QALY, and so PLDH appears to be the most cost-effective treatment strategy.

Sensitivity analysis – subgroup specific treatment effects In the base case analysis we applied the treatment effects estimated from the full trial populations, including a mixture of platinum sensitive, platinum resistant and refractory patients, to subgroup specific baseline data. In a sensitivity analysis we apply subgroup specific treatment estimates taken from the trials and apply those to subgroup specific baseline survival. Table 62 presents the analysis of the ICER in the platinum sensitive and platinum resistant and refractory subgroups.

209 Table 62. Cost-effectiveness results applying subgroup-specific treatment effects. Quality- Probability cost-effective for a PFS OS adjusted maximum WTP: Treatment Cost ICERa (wks) (wks) survival £10,000 £30,000 £50,000 (wks) Sensitivity analysis – platinum sensitive Topotecan 33.1 101.3 41.7 £8,448 ED 0.00 0.00 0.01

Paclitaxel 27.8 104.3 40.9 £4,146 - 0.73 0.22 0.14

PLDH 43.0 145.7 58.4 £8,902 £14,080 0.27 0.78 0.85 Sensitivity analysis – platinum resistant/refractory Topotecan 19.8 61.2 25.1 £8,448 ED 0.00 0.19 0.29

Paclitaxel 15.2 46.3 19.1 £4,146 - 1.00 0.47 0.22

PLDH 19.8 65.9 26.6 £8,902 £33,116 0.00 0.35 0.49 a Cost per quality adjusted life year (QALY). PFS = progression-free survival; OS = overall survival; WTP = willingness to pay per QALY.

The results are similar to those observed in Analysis 1, with topotecan ruled out by extended dominance by PLDH. The ICER for PLDH compared to paclitaxel is reduced to £14,080 per QALY in the platinum sensitive population (compared to £20,212 in the base-case analysis), and is £33,116 per QALY in the platinum resistant and refractory population (compared to £33,430 in the base-case analysis).

Both trial 039 and trial 30-49 were stratified by platinum sensitivity in order to ensure even numbers in each group. Clearly, there is greater uncertainty around the subgroup-specific treatment effects as they are estimated on a smaller population. This increased uncertainty is reflected in the model, which is fully probabilistic and so incorporates the wider confidence intervals around these subgroup effects. Despite this increased uncertainty, the results of this sensitivity analysis show that PLDH has a higher probability of being the most optimal treatment strategy at a WTP of £30,000 per QALY in platinum sensitive patients, because of the more favourable expected improvement in survival. The same is not true in the platinum resistant and refractory subgroup where we are more uncertain about the most optimal treatment strategy at a WTP of £30,000 per QALY.

For patients who have received placliataxel as part of their first-line treatment, the ICER of PLDH compared to topotecan increases to £19,584 per QALY.

210 Sensitivity analysis – changing cost assumptions for PLDH and topotecan

In the base-case analysis patients receieving PLDH were assumed to receive no additional tests to monitor cardiac function. In a sensitivity analysis we added the cost of an ECG test to each cycle of PLDH recieved. We also undertook a sensitivity analysis to look at the home administration of topotecan via a community nurse. The ICER’s for these sensitivity analyses regarding cost assumptions (described above) are presented in Table 63 below.

Table 63. Cost-effectiveness results applying alternative costs assumptions. Quality- Probability cost-effective for a PFS OS adjusted maximum WTP: Treatment Cost ICERa (wks) (wks) survival £10,000 £30,000 £50,000 (wks) Analysis 1 – overall patient population- cost of ECG added to each cycle of PLDH

Topotecan 24.5 86.0 34.2 £8,448 ED 0.00 0.02 0.03

Paclitaxel 20.1 79.7 30.9 £4,146 - 1.00 0.40 0.21

PLDH 27.5 104.8 40.9 £9220 £26,250 0.00 0.58 0.76

Analysis 1 – overall patient population- home administration of topotecan

Topotecan 24.5 86.0 34.2 £8,111 ED 0.00 0.02 0.03

Paclitaxel 20.1 79.7 30.9 £4,146 - 0.99 0.37 0.19

PLDH 27.5 104.8 40.9 £8,902 £24,606 0.01 0.61 0.77

For both sensitivity analysis the results are similar to those observed in the overall population of Analysis 1. Topotecan is still ruled out on the grounds of extended dominance by PLDH, as such the ICER for paclitaxel/PLDH in sensitivity analysis of home administration of topotecan remains unchanged. Adding in additional cardiac monitoring costs for PLDH increased the ICER compared to paclitaxel to £26,250 per QALY (compared to £24,406). The results of the model would therefore appear to be robust to these changes in costs assumptions.

Sensitivity analysis – incorporation of trial 30-57 Included in the cost-effectiveness review in chapter 4.2 alongside the 8 published trials was a trial report for an aborted study comparing paclitaxel to PLDH, trial 30-57. The evidence from this study appeared to contradict the inferences one could draw from trial 039 and trial 30-49, by showing that paclitaxel may be superior to PLDH in

211 terms of overall survival (although not statistically significant). Looking only at trial 039 and trial 30-49, the conclusion may have been that as topotecan and paclitaxel are shown to be of similar effectiveness in 039, and as PLDH is shown to be superior to topotecan in 30-49, PLDH would be superior to paclitaxel in a direct comparison. In a sensitivity analysis we incorporated the hazard ratio for overall survival from 30-57 via a MTC in order to assess the implications of including this new evidence. As mentioned earlier, the MTC gives trial 30-57 equal weight to the two completed RCTs, and assumes that the hazard ratio observed at approximately 1 year is equal to that observed after a longer period of follow-up. Table 64 presents the analysis of the ICER in the overall patient population, and also where those same treatment effects are applied to a platinum sensitive or platinum resistant and refractory baseline.

212 Table 64. Cost-effectiveness results incorporating results of 30-57 via a MTC Quality- Probability cost-effective for a PFS OS adjusted maximum WTP: Treatment Cost ICERa (wks) (wks) survival £10,000 £30,000 £50,000 (wks) Sensitivity analysis – overall patient population Topotecan 24.5 86.0 34.2 £8,448 D 0.00 0.01 0.03

Paclitaxel 20.1 92.1 34.6 £4,146 - 1.00 0.81 0.55

PLDH 27.5 98.1 38.9 £8,902 £58,475 0.00 0.18 0.42 Sensitivity analysis – treatment effects applied to platinum sensitive baseline Topotecan 33.1 101.6 41.8 £8,448 D 0.00 0.02 0.04

Paclitaxel 27.2 108.6 42.0 £4,146 - 1.00 0.70 0.45

PLDH 37.2 115.7 47.4 £8,902 £45,894 0.00 0.29 0.51 Sensitivity – treatment effects applied to platinum resistant/refractory baseline Topotecan 19.8 61.2 25.1 £8,448 D 0.00 0.00 0.02

Paclitaxel 16.2 65.5 25.3 £4,146 - 1.00 0.93 0.69

PLDH 22.2 79.8 28.5 £8,902 £76,192 0.00 0.06 0.29 PFS = progression-free survival; OS = overall survival; WTP = willingness to pay per QALY. a Cost per quality adjusted life year (QALY).

As expected, the inclusion of 30-57 alters the results of the model by making paclitaxel look more effective than topotecan in terms of overall survival, and by reducing the effectiveness of PLDH. This results in topotecan being dominated by paclitaxel, which is now associated with marginally higher quality-adjusted survival. Due to the improved effectiveness of paclitaxel and the less favourable estimate for PLDH, the ICER for PLDH compared to paclitaxel is higher than in analysis 1 at £58,475 per QALY in the overall patient population, £45,894 in the platinum sensitive population and £76,192 in the platinum resistant and refractory population. If 30-57 is to be given the same weight as trials 039 and 30-49, then paclitaxel is the most cost- effective treatment strategy at WTP values below £45,894 per QALY.

For paclitaxel experienced platinum resistant and refractory patients, the ICER for PLDH compared to topotecan is £6,938 and so PLDH still appears to be cost-effective.

5.5.2 Analysis 2 Analysis 2 relaxed the need to use reported hazard ratios, and also introduced comparators not formally included in the systematic review. Analysis 2 also considered only a platinum sensitive population, as the additional comparators were only assessed in this subgroup. The relevant treatment strategies in Analysis 2 are

213 topotecan, paclitaxel, PLDH, paclitaxel combined with platinum, platinum alone and CAP. The costs for the 6 comparators are presented, broken down by category, in table 65. The cost for the 3 comparators included in analysis 1 are slightly different, owing to the different data used to estimate adverse events (described above)

Table 65. Cost results broken down by category

+ Paclitaxel Topotecan Paclitaxel Paclitaxel Platinum platinum PLDH CAP

Cost of drug £4,223 £3,180 £7,936 £5,724 £1,979 £709 Cost of £2,405 £434 £438 £626 £474 £1,758 administration Cost of monitoring £47 £43 £43 £49 £47 £53 Cost of adverse £1,654 £403 £433 £429 £770 £992 events Total cost £8,330 £4,066 £8,851 £6,828 £3270 £3,512

In Analysis 2, platinum therapy is associated with the lowest overall costs. Altough the acquisition costs of CAP are markedly lower than the other treatments, the additional costs of administration and the management of adverse events means that the total costs are higher that platinum therapy. Paclitaxel and platinum combination is more expensive than paclitaxel alone, this is due almost entirely to the higher cost of the drug itself. PLDH appears the most expensive treatment, closely followed by topotecan.

The inclusion of the Cantu and ICON4 trial relates specifically to the platinum sensitive population, in which these trials were conducted. Data from these trials is therefore added to the platinum sensitive data from the 039 and 30-49 trials, reported in Analysis 1. Table 66 presents the analysis of the incremental cost-effectiveness ratio (ICER) in the platinum-sensitive patient population. In Analysis 2 the treatments compared are topotecan, paclitaxel monotherapy, PLDH, paclitaxel + platinum, platinum and CAP.

214

Table 66. Cost-effectiveness results from Analysis 2 Quality- Probability cost-effective for a PFS OS adjusted maximum WTP: Treatment Cost ICERa (wks) (wks) survival £10,000 £30,000 £50,000 (wks) Analysis 2 – platinum sensitive patient population Topotecan 33.1 101.4 41.7 £8,330 D 0.00 0.00 0.00

Paclitaxel 28.0 105.1 41.2 £4,066 D 0.00 0.00 0.00

PLDH 43.0 145.8 58.5 £8,851 D 0.00 0.01 0.02 Paclitaxel + 82.0 178.8 81.2 £6,828 £3,561 0.15 0.60 0.63 platinum Platinum 63.5 149.7 66.3 £3,383 - 0.35 0.01 0.00

CAP 47.9 176.7 69.5 £3,512 ED 0.50 0.38 0.35 a Cost per quality adjusted life year (QALY). PFS = progression-free survival; OS = overall survival; WTP = willingness to pay per QALY.

Topotecan, paclitaxel monotherapy, PLDH are all dominated by platinum, that is the have higher costs and lower QALYs than platinum therapy. Although CAP is not dominated by platinum, it is ruled out by extendend dominance (i.e. the ICER for CAP vs platinum is higher than placlitaxel/platinum combination vs CAP). After excluding dominated alternatives, the treatments under consideration are platinum and paclitaxel/platinum combination therapy. The ICER for paclitaxel /platinum combination compared to platinum treatment is £3,561 per QALY.

The CEACs for analysis 2, showing the probability that each comparator is cost- effective over a range of threshold values, is provided in Figure 26. At a maximum WTP of £30,000 per QALY the probability that the combination of paclitaxel and platinum is cost-effective is 0.6.

215

Figure 26. Analysis 2, platinum sensitive patients

CEAC Analysis 1 - platinum sensitive population

0.9

0.8

0.7

0.6 Topotecan Paclitaxel 0.5 PLDH Paclitaxel + platinum 0.4 Platinum CAP 0.3 Probability cost-effective

0.2

0.1

0 £0 £10,000 £20,000 £30,000 £40,000 £50,000 £60,000 -0.1 Threshold for cost-effectiveness

5.6 Budget impact analysis

In order to estimate the budget impact of the economic model recommendations we must consider the number of people who will receive 2nd line therapy for advanced ovarian cancer and the cost of switching, from therapies they would have received to the therapy determined optimum by Analysis 1 of the economic model. Analysis 1 concluded that PLDH was the most cost-effective therapy for an overall population, platinum sensitive and platinum resistant/refractory patients. Given the limitations of Analysis 2, we have not calculated the budget impact, for platinum sensitive patients, based on the conclusions of this model.

Population receiving 2nd line therapy Assuming an annual incidence of 6000 (NICE guidance on PLDH), and the percentage that receive 1st line chemotherapy is 75%(NICE guidance on PLDH), there are 4500 women per year that will potentially go onto 2nd line therapy. The percentage that actually relapses from 1st line therapy is estimated at 65% (mean of 55-75% stated in NICE guidance). The population that will be affected by a change in 2nd line therapy is:

Number of women requiring 2nd line therapy = (6000 * 0.75) * 0.65 = 2925

216

Costs of switching The GSK audit shows that patients are receiving 23 different treatments as 2nd line therapy. For many of the treatments dosage, number of cycles received and toxicity events are not stated, and it is therefore not possible to comprehensively cost 2nd line therapy using all the data reported by GSK. We therefore assume patients can only receive the 3 drugs considered in this appraisal (as in Analysis 1 of the economic model) by pro-rating up the percentages reported for PLDH, paclitaxel and topotecan. For our budget-impact analysis, 2nd line therapy currently consists of: 10.53% receiving PLDH, 73.68% receiving topotecan and 15.79% receiving paclitaxel.

Given that PLDH is the most cost-effective drug in Analysis 1, all patients receiving paclitaxel and topotecan should instead receive PLDH. The cost of 2nd line therapy for the 3 drugs is £8,902 for PLDH, £4,146 for paclitaxel and £8,448 for topotecan. Switching to PLDH will therefore cost an additional £454 for those patients who would have received topotecan and an additional £4756 for those patients who would have received paclitaxel, as 2nd line therapy.

Total budget impact Table 67 shows the results of the budget impact analysis.

Table 67. Results of the budget impact analysis Percentage receiving therapy Number on therapy Cost of 2nd line (£) Total cost (£) PLDH 10.53 308 8902 2,740,879 Paclitaxel 15.79 462 4146 1,914,797 Tamoxifen 73.68 2155 8448 18,207,663 2925 22,863,339

Given a yearly incidence and using the percentages receiving PLDH, paclitaxel and topotecan as 2nd line therapy, reported above, the current yearly total cost of 2nd line ovarian cancer in England and Wales is £22,863,339. If all patients were to receive PLDH, that is the percentage receiving PLDH therapy was 100%, the yearly cost of 2nd line ovarian cancer chemotherapy treatment would be £26,038,350. The budget impact of changing to PLDH as 2nd line therapy is therefore £3,175,010 per year.

217

This analysis does not consider a patients eligibility to receive PLDH as 2nd line therapy. Given that PLDH is not licensed as 1st line therapy in ovarian cancer, it is assumed that no patients will be ineligible to receive PLDH 2nd line based on their 1st line treatment. There may however be other factors that determine which 2nd line treatment is offered and it therefore possible that not all of the 2925 women that have recurrent disease each year will actually receive PLDH, even if it is the most cost- effective treatment option.

5.7 Economic model conclusions The purpose of the economic model was to explore a range of uncertainties and sources of variability that were not fully assessed in either the published literature or the industry submissions. Two main analyses were undertaken to explore the cost- effectiveness of PLDH, topotecan and paclitaxel in advanced ovarian cancer. Two separate analyses (Analysis 1 and Analysis 2) were required in order to reflect the heterogeneity identified in the different trials and the difficulties encountered in obtaining robust estimates using a consistent approach in the methods of evidence synthesis, of the relative treatment effects associated with particular treatments.

In the base-case results for Analysis 1, paclitaxel monotherapy emerged as the cheapest treatment. When the ICERs were estimated, topotecan was ruled out on the grounds of extended dominance. Hence the options considered in the analysis were paclitaxel and PLDH. The ICER for PLDH compared to paclitaxel was £24,606 per QALY in the overall patient population (comprising platinum sensitive, refractory and

218 resistant patients). The ICER was more favourable in the platinum sensitive group (£20,212 per QALY) and less favourable in the platinum refractory/resistant group (£33,430 per QALY). Clearly the conclusions about cost-effectiveness will depend on the NHS’ threshold willingness to pay for additional health gain (in terms of QALYs). Assuming the NHS is willing to pay between £20,000 to £40,000 per additional QALY, then PLDH appears to be cost-effective compared to topotecan and paclitaxel monotherapy in terms of the overall patient population and the main subgroups considered. These conclusions were robust to alternative cost assumptions and the use of subgroup specific treatment effects.

The cost-effectiveness results for the base-case analysis were sensitive to the inclusion of trial 30-57. This trial compared paclitaxel to PLDH and provided additional direct evidence information on the relative effectiveness of these 2 treatments. Incorporating the results of trial 30-57 resulted in less favourable estimates for the ICER for PLDH vs paclitaxel compared to the base-case results. The ICER of PLDH compared to paclitaxel was £58,475 per QALY in the overall patient population, £45,894 per QALY in the platinum sensitive population and £76,192 per QALY in the platinum resistant and refractory population. Using the same range of values to represent the NHS’ willingness to pay for an additional QALY (£20,000-£40,000), these results would suggest that paclitaxel was the optimal treatment.

Although the results of Analysis 1 appear sensitive to the inclusion of trial 30-57, these findings should be interpreted with due caution for a number of reasons. Firstly, since trial 30-57 was terminated prematurely, the results have only been reported for a limited follow-up period. This contrasts with the longer-term follow up available from trial 039 and 30-49 used in the base-case analysis. Until comparative follow-up data is available, the findings from trial 30-57 should only be considered provisional. Secondly, since paclitaxel is now being increasingly used as part of first-line therapy (in combination with platinum), paclitaxel monotherapy may not be considered a relevant comparator for platinum resistant and refractory patients. For this patient group, the relevant treatment comparators would be limited to topotecan and PLDH. The ICER for PLDH compared to topotecan suggests that that PLDH is cost-effective in all scenarios considered in Analysis 1.

219 The results from Analysis 2 explored the cost-effectiveness of the full range of treatment comparators for platinum sensitive patients. The treatment options considered in this model comprised: PLDH, topotecan, paclitaxel monotherapy, CAP, paclitaxel plus platinum and platinum monotherapy. Due to the less robust approaches that were employed to synthesise the available evidence and the heterogeneity between the different trials, the reliability of these results should therefore be interpreted with some caution. Topotecan, paclitaxel monotherapy and PLDH were all dominated by platinum monotherapy (i.e. higher costs and lower QALYs). CAP was ruled out by extended dominance. After excluding these alternatives, the treatments that remained under consideration were platinum monotherapy and paclitaxel/platinum combination therapy. The ICER for paclitaxel /platinum combination compared to platinum monotherapy treatment is £3,561 per QALY. Consequently, for platinum sensitive patients, the combination of paclitaxel and platinum appears to be cost-effective.

220 6. DISCUSSION 6.1 Summary of Clinical-effectiveness

In total nine RCTs were included in the assessment of clinical effectiveness. In five of these trials the comparators in both trial arms were used within their licensed indications. In the further four trials identified, one of the trial arms assessed a comparator that was used outside its licensed indication either in terms of cycle length, dosage, route of administration or type of disease. The discussion of the clinical effectiveness of the three comparators assessed in this review, PLDH, topotecan and paclitaxel will therefore mainly focus on the five trials in which the comparators were used within their licensed indications.

Summary of clinical effectiveness data Three out of the five trials included participants with both platinum-resistant and platinum-sensitive relapsed advanced ovarian cancer(24, 27, 44), whilst a further two trials included only participants with platinum-sensitive disease.(28, 29) The comparators that were assessed in the three trials that included both sub-types of participants were, PLDH versus topotecan(44), topotecan versus paclitaxel(24) and PLDH versus paclitaxel.(27) In the further two trials that included participants with the sub-type of platinum-sensitive disease only, the comparators that were assessed were, single-agent paclitaxel versus a combination of cyclophosphamide, doxorubicin and cisplatin (CAP)(28) and paclitaxel plus platinum-based chemotherapy versus conventional platinum-based therapy alone.(29)

221 Trials including participants with refractory, resistant and platinum-sensitive disease

PLDH versus topotecan. This trial was a multi-centre open-label comparative trial of good quality. In total 474 participants were treated with either PDLH 50 mg/m2 every 28 weeks or topotecan 1.5 mg/m2/day for 5 consecutive days every 3 weeks. Two hundred and thirty-nine participants were treated with PDLH and 235 with topotecan. At baseline 46.5% of the participants were classified as having platinum- sensitive disease and 53.5% as having platinum-resistant disease. Approximately 73% of the included participants had received a paclitaxel and platinum-based combination therapy as first-line treatment.

At long-term follow-up, when 90% of the participants had died or were lost to follow- up, PLDH was marginally more effective than topotecan for the outcome of overall survival in the total trial population, HR=1.216 (95% CI: 1.00, 1.48). The reduction in the risk of death for participants treated with PLDH was 18% compared to those treated with topotecan. However, the lower boundary of the 95% confidence interval suggests that though the point estimate favours treatment with PLDH, the size of this benefit may be small. Further analysis, which assessed the effectiveness of the two comparators in sub-groups of participants classified at baseline as having platinum- sensitive or platinum-resistant disease, indicated that this treatment effect was not consistent across the two sub-groups. In the platinum-sensitive disease sub-group there was a more pronounced survival benefit for participants treated with PLDH compared to those treated with topotecan, HR=1.432 (95% CI: 1.066, 1.923). This corresponds to a 30% reduction in the risk of death for the participants treated with PLDH. However, in the platinum-resistant disease sub-group there were no statistically significant differences between the two groups shown for overall survival, HR=1.069 (95% CI: 0.82, 1.387). This suggests that any survival benefit observed for PLDH compared to topotecan is limited to participants with platinum-sensitive disease and that the marginally significant benefit observed in the total trial population is driven by the more pronounced benefit observed in this sub-group. For the further outcomes of progression-free survival and response there were no statistically significant differences between the two treatment regimens either in the total trial population or in either of the sub-groups. The point estimates for both

222 progression-free survival and response rates favoured PLDH over topotecan but, as all confidence intervals crossed unity, this benefit was not significant.

In terms of the toxicity states reported in the trial, there were three treatment related deaths due to sepsis in the topotecan group, but no treatment related deaths observed in the PLDH treatment arm. The incidence of grade 3 toxicities of stomatitis, PPE, mucous membrane disorder and rash was significantly higher in the PLDH treatment arm compared to the topotecan treatment group. However, there were no significantly higher rates of grade 4 adverse events reported in this group compared to the topotecan arm. The topotecan treatment arm had significantly higher reported rates of grade 3 haematological toxicities (anaemia, leucopenia, neutropenia and thrombocytopenia), alopecia and fever compared to the PLDH treatment group. In addition to this, grade 4 haematological toxicities were also significantly higher in this group.

Overall summary of the trial: Whilst the long-term follow-up data from this trial indicated that PLDH is marginally more effective than topotecan in terms of overall survival rates, it appears that this benefit is limited to the sub-group of participants with platinum-sensitive disease. In addition a number of caveats must be considered when interpreting these trial data. Firstly, at an earlier time of analysis there were no significant differences observed between the two treatment groups in terms of overall survival, although a trend was noticed in favour of treatment with PLDH for platinum-sensitive participants. Secondly, this trial was not realistically powered to detect differences in the treatment effect between the comparators in sub-groups of participants.

Topotecan versus paclitaxel. This trial was a good quality randomised open-label comparative trial. In total 112 participants were treated with topotecan and 114 with paclitaxel. The treatment cycle in both of the trial arms was 21 days. At baseline 47% of the participants were classified as having platinum-sensitive disease and 53% as having platinum-refractory disease. None of the participants had received a taxane as part of first-line therapy.

223 At 4 years post randomisation there were no statistically significant differences between the two treatment groups in terms of overall survival, time to progression, responses rate or response duration. The point estimates for all of these outcomes favoured treatment with topotecan over paclitaxel, but as the confidence intervals for all of the outcomes cross unity this benefit was not statistically significant. There were also no significant differences between the two treatment groups in the sub- groups of participants who were classified as having platinum-sensitive or platinum- resistant disease at baseline. There was, however, a significant difference observed in terms of time to response, in favour of paclitaxel. The median time to response was 9.0 weeks in the topotecan group compared with 6 weeks in the paclitaxel group. The results from the post hoc analysis of third-line cross over therapy, when participants crossed over to the alternative treatment regime, also showed that there were no significant differences between the two treatment groups in terms of overall survival, time to progression or response. This suggests that the two treatment regimens are non-cross resistant in only a small number of participants, and therefore sequential treatment with these two regimens may not be beneficial in the majority of patients.

In relation to toxicity states, in this trial two participants in the topotecan treatment group died due to treatment related toxicity. Both of these deaths were attributed to topotecan-induced sepsis. There were no deaths attributed to paclitaxel-induced myelosuppression. The incidence of grade 3 and 4 haematological toxicities was significantly higher in the topotecan group compared to the paclitaxel treatment arm, apart from grade 4 anaemia. The only incidence of haematological toxicity that was higher in the paclitaxel group was grade 3 neutropenia. For grade 3 / 4 non- haematological toxicities, nausea, vomiting, constipation, abdominal pain, fatigue and fever/infection were reported more commonly in the topotecan group. For the paclitaxel group grade 3 / 4 alopecia, arthralgia, myalgia and skeletal pain were reported more frequently.

Overall summary of the trial: At a follow-up time of 4 years post randomisation, there were no significant differences between the two treatment regimens in overall survival, time to progression, responses rate or response duration. The point estimates favoured topotecan on all of these outcomes, but remained statistically insignificant. The only outcome that was statistically significant was time to response, which

224 favoured paclitaxel. The incidence of adverse events in both of the treatment arms, indicated that topotecan was associated with a higher incidence of haematological toxicities compared to paclitaxel. There were no significant differences in the number of grade 3 /4 non-haematological adverse events reported in both of the trial arms.

PLDH versus paclitaxel. This trial was a reasonably good quality randomised, open- label, comparative trial. In total 108 participants were randomised to treatment with PLDH and 108 to paclitaxel. All participants in the trial had received one prior line of platinum-based chemotherapy and were taxane naïve. However due to poor participant accrual it was terminated early when only approximately 50% of the intended participants had been accrued. Due to this fact the trial data was only analysed for the primary outcome of overall survival

There were no significant differences in overall survival between the two treatment groups. A further analysis of the results for overall survival in the sub-groups of participants classified as having either platinum-sensitive or platinum-resistant disease indicated that there were also no significant differences between the two treatment groups in either of these sub-groups. The incidence of grade 4 adverse events was relatively low in both treatment arms. At the grade 3 level of toxicity PPE, ascites, stomatitis and dyspnea were observed significantly more often in the PLDH group compared to the paclitaxel group. The only toxicity that occurred significantly more often in the paclitaxel group was alopecia.

Overall summary of the trial: Whilst there were no significant differences observed between the two treatment groups, it is likely that, due to the low participant numbers, the trial was significantly under-powered to detect any differences in treatment effect between the two groups. Furthermore due to its early termination the results of the trial are likely to be preliminary and the longer term implications of the relative efficacy of these two comparators is unclear. However, from the available data it would appear that paclitaxel has a favourable toxicity profile compared to PLDH.

225 Trials including participants with platinum-sensitive disease only

Paclitaxel versus CAP. This trial was a relatively small randomised pilot study. In total 47 participants were treated with single-agent paclitaxel 175 mg/m2 every 21 days and 47 with the combination of cyclophosphamide 500 mg/m2, doxorubicin 50 mg/m2 and cisplatin 50 mg/m2 every 21 days (CAP). All the participants had previously been treated with two or three chemotherapy regimens that did not include a taxane. The treatment-free interval for all the participants was reasonably long, being 30.2 months in the paclitaxel arm and 38.8 months among the participants receiving CAP.

At a median follow-up of 49 months, after adjusting for prognostic factors CAP was more effective than paclitaxel for the outcomes of overall survival; HR=0.58 (95% CI: 0.34, 0.98) and progression-free survival, HR=0.60 (5% CI: 0.37, 0.l97). However, there were no significant differences between the two treatment groups in terms of response. The response rate to CAP was 55% compared to 45% in the paclitaxel treatment arm. The further analysis of third-line cross over therapy, when participants crossed over to the alternative treatment regime, showed that there were no significant differences between the two treatment regimes in terms of response. The analysis of the adverse events that were experienced in both the treatment groups showed that CAP was associated with significantly higher rates of grade 3 and 4 haematological toxicity compared to paclitaxel. It was also associated with significantly higher rates of grade 2 nausea and vomiting. Treatment with paclitaxel was associated with significantly higher rates of alopecia and allergic reactions relative to treatment with CAP.

Overall summary of the trial: The results of this trial suggest that, in platinum- sensitive patients who are taxane naïve, re-challenge with a platinum containing regimen at even third-line may be more beneficial than re-challenge with paclitaxel. However, it should also be noted that the quality of the trial from which these results were obtained is variable. The trial was a small, randomised pilot study and therefore the number of participants upon which these results are based is small. This limited sample size makes the point estimate of efficacy imprecise. Furthermore, due to the trial design that allowed cross-over to third line therapy, the interpretation of the

226 survival data is problematic. Due to these points, it is not possible to draw any firm conclusions regarding the relative efficacy of CAP compared to paclitaxel. Further trials assessing CAP as a comparator in platinum-sensitive patients do, however, seem to be warranted.

Paclitaxel in combination with platinum-based chemotherapy versus platinum- based therapy alone (ICON4). This trial was a good quality randomised, multicentre parallel trial. In total 392 participants were randomised to treatment with paclitaxel in combination with platinum-based therapy and 410 to conventional platinum-based therapy alone. Approximately 40% of the participants had received prior chemotherapy with a taxane. In 25% of the participants the treatment free interval was less than 12 months, and in 75% greater than 12 months.

At a median follow up of 42 months, paclitaxel and platinum combination was more effective than platinum-based therapy alone for the outcomes of overall survival and progression free survival. An analysis to assess the effect of the two different treatment regimens, on both overall survival and progression-free survival in sub- groups, indicated that there were no significant differences between the two treatment group in terms of these outcome measures in any of the sub-groups explored. This suggests that treatment with the combination of paclitaxel and platinum is beneficial even across a population of patients who are heterogeneous in terms of the number of previous lines of chemotherapy that have been received, their previous exposure to taxanes and their treatment free interval. However, many of the sub-groups were small and therefore may lack the power to detect any realistic differences in treatment effect in these groups. There was no significant difference between the two treatment regimens in terms of response rate but, again, the response rates appeared to favour treatment with the combination therapy. The response rate to paclitaxel plus platinum therapy was 66% compared to 55% for the conventional platinum-based therapy group. In terms of toxicities experienced during the trial, the incidence of grade 2-4 neurological toxicities and alopecia was significantly higher in the paclitaxel plus platinum chemotherapy regimen group compared to the conventional monotherapy arm. However, the rates of haematological toxicity were significantly higher in the conventional platinum chemotherapy group.

227 Overall summary of the trial. The ICON4/AGO-OVAR 2.2 trial was a good quality randomised multicenter trial. However, the trial can be criticised regarding the heterogeneous nature of the patient population that were enrolled into the trial, in terms of the differences in prior therapy and treatment free interval. A further criticism can be made regarding the treatment received in the platinum-based monotherapy group. Thirty-three percent of participants in this arm were taxane naïve as they had not received a taxane either as part of first-line therapy or within the trial. However, despite these criticisms and the difficulties they present in interpreting the trial results, ICON4 is the first adequately powered trial to evaluate the extent of the benefit of combination paclitaxel and platinum-based therapy relative to platinum- based therapy alone in participants with platinum sensitive disease.

Overall discussion of clinical effectiveness

This review included five trials that assessed the efficacy and safety of PLDH, topotecan and paclitaxel with both comparators used within their licensed indications, in patients with relapsed advanced ovarian cancer. The evidence base from these trials is somewhat heterogeneous in terms of the trial quality, comparators assessed and the patient populations. Whilst it is clearly recognised that response to prior platinum-based chemotherapy is a continuum, the results from the trials will be compared in relation to the two sub-groups of participants, those with platinum- sensitive disease and those with platinum-resistant disease. This is due to the fact that the clinical issues that need to be addressed in these two patient populations are somewhat different due to the differing prognosis in terms of potential tumour chemo- responsiveness and overall survival within each of these groups.

228

Platinum-resistant disease sub-group

Three trials included participants who were classified at baseline as having platinum- resistant disease.(19, 22, 24) However, only overall survival data was available for the trial of PLDH versus paclitaxel.(27) The median survival, median progression- free survival and overall response rate data for the platinum-resistant sub-groups of participants from these three trials is summarised in Table 68..

Table 68 shows that there were no substantial differences in the median survival rates across the three trials with these ranging from 36.7 weeks to 54.3 weeks. The most favourable median survival time was observed for the paclitaxel treatment arm within the trial of PLDH versus paclitaxel, at 54.3 weeks.(27) This trial was conducted in participants who were taxane naïve. In terms of response rate, it can again be observed from the data that there were no substantial differences in terms of response rates across the three trials, with these ranging from 6.7% to 13.3%. The most favourable response rate was observed for topotecan, within the trial of topotecan versus paclitaxel at 13.3%.(24, 25)

Overall from the summary data of the three trials that included participants with platinum-resistant disease it can be observed that there was a low probability of response to treatment with PLDH, topotecan or paclitaxel. Response rates varied from 6.7% to 13.3%. Likewise, there was little difference between the three comparators in relation to overall survival, with median survival times varying from 36.7 weeks to 54.3 weeks. Given the low survival times and response rates observed in these trials, and the different toxicity profiles of PLDH, topotecan and paclitaxel, it would appear that the crucial issues to be addressed when choosing a second-line treatment for patients who have platinum-resistant disease would be the maintenance of quality of life, and the control of symptoms and toxicity. Further factors that would therefore need to be considered would be the cost-effectiveness of the three comparators, the ease of administration, and patient choice in terms of the potential toxicities they were prepared to undergo. It can also be suggested these patients may benefit from being included in clinical trials of new drugs.

229 Table 68. Summary of the median survival, progression-free survival and response rates for participants with platinum-resistant disease Reference Drug n Median Overall response Median Taxane at first-line therapy Treatment free interval survival rate progression-free survival Trial 30-49(44) PLDH 130 38.3 wks 11.5% (95%CI: 9.1 wks 74% received a taxane at first- 54% relapsed within 6 6.0%, 17.0%) line months Topotecan 125 42.1 wks 7.2% (95% CI: 2.7%, 13.6 wks 72% received a taxane at first- 53% relapsed within 6 11.7%) line months Trial 039(24, 25) Topotecan 60 Not available 13.3% Not available Taxane naive 53% relapsed within 6 months Paclitaxel 59 Not available 6.7% Not available Taxane naive 53% relapsed within 6 months Trial 30-57 (27) PLDH 64 36.7 wks Not available Not available Taxane naive Not available Paclitaxel 67 54.3 wks Not available Not available Taxane naive Not available Platinum-sensitive disease sub-group

All the five trials included participants who were classified at baseline as having platinum-sensitive disease.(24, 27-29, 44) Again however, only overall survival data were available for the trial of PLDH versus paclitaxel.(27) These five trials were reasonably heterogeneous in terms of the prior chemotherapy regimens that participants had received and the treatment-free interval of the trial participants. The differences in the treatment-free intervals of the participants between the trials may be particularly pertinent when assessing the differences observed in the median survival times and overall response rates of these trials. A summary of the median survival times, progression-free survival times and the overall response rates for the platinum- sensitive participants that were assessed in the five trials is summarised in Table 69.

As can be observed in table 69 there was a considerable difference in the median survival times observed across the five trials, with these ranging from 65.4 weeks to 156 weeks. It is however problematic to indirectly compare the results of the different trial arms due to the heterogeneity of the prior chemotherapy regimens that participants had received, and also the differences in the treatment free intervals between the trials. The differences in the overall survival rates between the trials do however suggest that there are large potential differences in relation to overall survival in this patient population, and therefore the choice of treatment may make a substantial difference. It can be suggested from the results of the two trials that compared combination chemotherapy to a single agent regimen, that re-challenge with a platinum-based combination therapy is superior to re-challenge with a single agent. However, no inferences can be drawn regarding the relative efficacy of paclitaxel- platinum combination therapy and CAP, as both of these regimens were compared to different mono-therapies, and the treatment free interval of the participants in these trial populations differed considerably.

In terms of re-challenge with a single-agent therapy, again the median survival times varied susbstantially across the trials, ranging from 65.4 weeks for the PLDH treatment arm, in the trial that compared PLDH and paclitaxel(27) to 116 weeks in the single-agent paclitaxel arm, in the trial that compared CAP versus paclitaxel. However, it is difficult to compare the results of these trials as the treatment free interval and first-line therapy that participants had received differed considerably across the trials It can however, be stated that PLDH was significantly more beneficial in terms of overall survival than topotecan, although this is based on a subgroup analysis. In a second trial that directly compared topotecan and paclitaxel there were no significant differences between the two comparators in terms of overall survival. In a further trial, that compared PLDH with paclitaxel in a head-to-head comparison, the limited data available suggested there were no significant differences in overall survival between the two treatment arms.

Within this group of women with advanced ovarian cancer it would appear that prolongation of progression-free and overall survival, as well as the maintenance of quality of life and symptom control can be realistic treatment objectives. Based upon the results of these trials it would appear that combination chemotherapy either with paclitaxel in combination with a platinum based therapy or CAP may be beneficial compared to monotherapy alone. However, combination therapy can also be associated with a higher incidence of toxicity compared to single agent therapy, and therefore may not be a suitable treatment option for some patients. In terms of re- challenge with single agent therapy, there appears from the limited evidence available to be an indication that PLDH is more effective than topotecan. This is however, based upon the data from a sub-group analysis of one trial and therefore the evidence base is very limited. Furthermore, as there are no trials that have directly compared PLDH with a single-agent platinum compound it is not possible to comment on whether re-challenge with a different chemotherapeutic compound is more beneficial than re-challenge with a platinum-based chemotherapeutic regimen in platinum- sensitive patients.

232 Table 69. Summary of the median survival, progression-free survival and response rates for participants with platinum-sensitive disease. Reference Drug n Median Overall response rate Median Taxane at first-line therapy Treatment free interval survival progression-free survival Trial 30-49(44) PLDH 109 107.9 wks 29.4% (95% CI: 27.3 wks 74% received a taxane first- 54% relapsed within 6 20.8%, 37.9%) line months Topotecan 110 70.1 wks 28.2% (95% CI: 22.7 wks 72% received a taxane first- 53% relapsed within 6 19.8%, 36.6%) line months Trial 039(24, 25) Topotecan 52 Not available 28.8% Not available Taxene naive 53% relapsed within 6 months Paclitaxel 55 Not available 20.0% Not available Taxane naive 53% relapsed within 6 months Trial 30-57 PLDH 44 65.4 wks Not available Not available Taxane naive Not available (27) Paclitaxel 41 75.7 wks Not available Not available Taxane naive Not available Cantu et al.(28) Paclitaxel 47 25.8 months 45% 9 months Taxane naïve 30.2 months (136 wks) (116 wks) (40 wks) CAP 47 34.7 months 55% 15.7 months Taxane naïve 38.8 months (175 wks) (156 wks) (70 wks) ICON & AGO Paclitaxel & 392 29 months 66% 12 months 41% received a taxane first- 25% ≤ 12 months; 75% > 12 Collaborators(29) platinum (130 wks) (54 wks) line months combination Platinum 410 24 months 54% 9 months 39% received a taxane first- 25% ≤ 12 months; 75% > 12 alone (108 wks) (40 wks) line months

233

6.2 Summary of cost-effectiveness

A total of 4 published studies met the inclusion criteria for the cost-effectiveness review. In addition, separate submissions were received from Bristol Myers Squibb, GlaxoSmithKline and Schering-Plough Ltd. The published studies and manufacturers’ submissions were assessed, and a new model was developed to address the limitations identified in these sources and to provide a direct comparison of the full range of possible strategies that are relevant to the NHS. The model explored a range of uncertainties and sources of variability that were not fully addressed in existing data sources.

An integral component of the model was the use of Bayesian approaches to synthesise effectiveness data from a series of mixed and indirect treatment comparisons. This approach provides an analytic framework to incorporate evidence in situations where there exists both direct head-to-head evidence and indirect evidence relative to a common comparator. It allows consideration of the complete evidence base and facilitates a direct comparison of the full range of treatment strategies. Clearly when indirect evidence is used to estimate treatment effects it is not possible to rule out the introduction of bias, and the results should be interpreted accordingly. The approach is, however, based on only a few additional assumptions over standard meta-analysis.

The evidence reported from ICON4 (platinum plus paclitaxel and platinum alone) provided no evidence on the relationship between these comparators and any of the other treatments under consideration. In order to incorporate the results of this trial into the model it was necessary to break randomisation in order to obtain estimates of the relative treatment effect. Consequently, although it was possible to make direct comparisons amongst the full range of treatment strategies, this was at the expense of a less robust estimate for the effectiveness data. Furthermore, systematic searches for all possible comparators were not undertaken. Hence there may be additional indirect evidence on the effectiveness of these comparators that could be considered alongside the evidence reviewed in this report.

234 In addition, the model presented here has several potential limitations that need to be considered in conjunction with the main results. First, although the model made adjustments for the quality of life of patients according to different disease states (e.g. progression-free and disease progression), no separate adjustments could be made to reflect the impact of the different treatments in terms of adverse events reported during the treatment period. No utility estimates were identified which could be used to reflect the different types and severity grades of the various adverse events reported in each of the trials. Consequently it is not possible to assess the robustness of the model results to this aspect. The lack of suitable utility data emphasises the need for this data to be an integral part of future trials.

Second, in the absence of patient level data it was not possible to conduct a detailed analysis of the resource use and costs associated with the management of adverse events. Although several patient-level costing studies were identified as part of the review of existing cost-effectiveness evidence, these analyses were restricted to a comparison between topotecan and PLDH. In order to adopt a consistent approach to the full range of comparators considered in the model, it was necessary to use the aggregated data reported in the trials. Furthermore, the analysis was restricted to severity grades III/IV, due to a lack of data reported on the less severe toxicities in some of the trials. While this approach provided a common basis on which to compare the different treatments, this method is likely to underestimate the overall costs (since we cannot account for patients who have had more than one adverse event of the same type and severity) and also may not adequately capture the true differences in costs between the treatments. Indeed, a comparison between our own results and those reported in Smith (50) and the submission by Schering-Plough, highlights that the differences in total costs mainly relate mainly to the costs of managing adverse events. Both Smith and Schering-Plough estimated that the additional costs of managing adverse events for patients receiving topetecan compared to PLDH ranged between £2,593 and £2,909. In contrast, using the aggregate data reported in the trials, our model estimated this difference to be only £1,289. Consequently while Smith and Schering-Plough conclude that the overall costs (including the costs of the drugs, administration, monitoring and management of adverse events) of topotecan are higher than PLDH, our model results in marginally higher costs for PLDH. Clearly if the actual difference in the costs of managing adverse events between topotecan and

235 PLDH is significantly higher than our own estimates imply, then PLDH will dominate topotecan. However, our overall conclusions would not be altered (since topotecan is already ruled out by extended dominance regardless). It is less clear what impact a more accurate costing of adverse events would have on the conclusions arrived at when comparing the full range of treatment comparators.

Third, although the impact of patient heterogeneity was explored in a series of subgroup analyses, it was not possible to undertake separate analyses for platinum refractory and resistant patients. Consequently, the results for these patients were presented by combining data from the two groups. Although the base-case results for the main analysis indicated that PLDH was cost-effective for the overall patient population and for the combined group of resistant and refractory patients, these results may not hold when considering the refractory group in isolation. Since the survival of these patients will be lower than the overall survival estimated by combining this group with the resistant population, it is likely that the ICER for PLDH compared to paclitaxel will exceed £40,000 per QALY in this particular subgroup.

Fourth, in order to estimate mean survival times from the estimated hazards it was necessary to assume that the survival data was approximately exponential (i.e. a constant hazard) in form. In the absence of patient-level data it is difficult to establish the validity of this assumption or to determine whether alternative distributional forms would be more appropriate (e.g. Weibull). However, since overall survival times in this patient population were relatively short, it is unlikely that alternative assumptions would significantly impact on the results presented here.

Finally, the analysis presented here does not directly consider the impact of treatments provided as part of third-line and subsequent therapies. It is possible that the differences observed in the various trials may be partly confounded by the different therapies received after 2nd line drugs. For example, patients receiving PLDH as 2nd line therapy might have received topotecan as 3rd line therapy, while the same pathway may not be possible for patients receiving topotecan as 2nd line drug. In other words, differences in the long-term results could also depend on treatments received after the 2nd line therapies.

236 Although the economic model presented here has a number of potenial weaknesses, its does represent the most comprehensive comparison of topotecan, paclitaxel and PLDH in advanced ovarian cancer. The analysis conducted here also helps to reiterate the need for a formal comparison, via a RCT, of all the treaments considered, in particular the comparison between paclitaxel in combination with platinum therapy, topotecan, PLDH and paclitaxel monotherapy in platinum sensitive patients.

7. CONCLUSIONS

7.1 Clinical effectiveness

Trials including participants with refractory, resistant and platinum-sensitive disease

PLDH versus topotecan

PLDH was marginally more effective than topotecan in terms of overall survival in the total trial population that included both participants with platinum resistant and platinum refractory disease. That is, the point estimate favoured treatment with PLDH but the lower boundary of the 95% confidence intervals suggests that the size of this benefit may be very small.

The overall survival benefit from treatment with PLDH compared to topotecan was most pronounced in the platinum-sensitive sub-group of participants. For participants with platinum-refractory disease there was no statistically significant difference in overall survival between the PLDH and topotecan treatment groups.

There were no statistically significant differences between the PLDH and topotecan groups in terms of progression-free survival, response or quality of life.

The rates of grade 3 stomatitis, PPE, mucous membrane disorder and rash were significantly higher in the PLDH treatment arm.

In the topotecan arm the rates of grade 3 and 4 haematological toxicities, and grade 3 alopecia and fever were significantly higher.

Topotecan versus paclitaxel

There were no statistically significant differences between the two treatment groups in terms of overall survival, time to progression, response rate or response duration. The point estimates for all of these outcomes favoured treatment with topotecan over paclitaxel, but these differences were not statistically significant. However, there was a significant difference between

237 the two treatment groups in terms of time to response. This difference favoured paclitaxel.

After cross-over to third-line therapy there were no statistically significant differences between the two groups on any of the effectiveness outcomes assessed.

Treatment with topotecan was associated with significantly more grade 3 and 4 haematological toxicities compared to paclitaxel. In addition rates of grade 3 and 4 nausea, vomiting, constipation, abdominal pain, asthenia, fatigue, and fever / infection were significantly higher in this group.

Treatment with paclitaxel was associated with significantly more grade 3 and 4 alopecia, arthralgia, myalgia and skeletal pain compared to the topotecan.

PLDH versus paclitaxel

In relation to overall survival there were no significant differences between treatment with PLDH or paclitaxel.

Treatment with PLDH was associated with significantly more grade 3 PPE, ascites, stomatitis and dyspnea compared to treatment with paclitaxel. Treatment with paclitaxel was associated only with a higher incidence of grade 3 alopecia.

Trials including participants with platinum-sensitive disease only

Paclitaxel verus CAP

CAP was more effective than paclitaxel in terms of both overall and progression-free survival. However, there were no significant differences between the two treatment regimens in terms of response.

The incidence of grade 3 and 4 haematological toxicities and grade 2 nausea and vomiting was significantly higher in the CAP treatment arm.

Treatment with paclitaxel was associated with significantly higher rates of alopecia and allergic reactions relative to treatment with CAP.

Paclitaxel in combination with platinum-based chemotherapy versus platinum based therapy alone

Paclitaxel in combination with platinum-based chemotherapy was more effective than platinum monotherapy in relation to both overall survival and progression-free survival.

There were no significant differences between the groups treated with paclitaxel in combination with platinum and platinum monotherapy for the outcomes of response rate or overall quality of life.

238 Treatment with paclitaxel in combination with platinum was associated with significantly higher rates of grades 2 to 4 neurological toxicity and alopecia. Treatment with platinum monotherapy was associated with significantly higher rates of haematological toxicity.

Overall conclusions

Participants with platinum-resistant disease: For participants with platinum- resistant disease there was a low probability of response to treatment with PLDH, topotecan or paclitaxel. Furthermore there was little difference between the three comparators in relation to overall survival. The comparators did however differ considerably in their toxicity profiles. Given the low survival and response rates it appears that the maintenance of quality of life, control of symptoms and toxicity are paramount in this patient group. As the three comparators differed significantly in terms of their toxicity profiles, patient and physician choice is also an important element that should be addressed when decisions are made regarding second-line therapy. It can also be suggested that this group of patients may benefit from being included in further clinical trials of new drugs.

Participants with platinum-sensitive disease: For participants with platinum- sensitive disease there was a considerable range of median survival times observed across the trials. The results of the trials that compared combination therapy to single-agent therapy suggest that re-challenge with combination therapy may be more beneficial than re-challenge with a single-agent chemotherapeutic regimen. In terms of single-agent compounds, the evidence suggests that PLDH is more effective than topotecan. Evidence from a further trial that compared PLDH and paclitaxel suggests that there is no significant difference between these two comparators. The three comparators did however differ significantly in terms of their toxicity profiles across the trials. Although re-challenge with PLDH may therefore be more beneficial than with topotecan, patient and physician choice as to the potential toxicities associated with each of the comparators, and the patients ability and willingness to tolerate these are of importance.

7.2 Cost effectiveness • The model developed by the University of York TAR team sought to assess the cost-effectiveness of intravenous formulations of topotecan monotherapy, PLDH monotherapy, and paclitaxel used alone or in combination with a platinum-based compound for the second-line or subsequent treatment of advanced ovarian cancer. • The model considered two main analyses and several alternative scenarios, due to the difficulties encountered in obtaining robust estimates using a consistent approach in the methods of evidence synthesis. Analysis 1 was restricted to a comparison of topetecan, PLDH and paclitaxel monotherapies. Analyis 2 explored the cost-effectiveness of PLDH, topotecan, paclitaxel monotherapy, CAP, paclitaxel plus platinum and platinum monotherapy in platinum sensitive patients.

239 The following conclusions are possible assuming the NHS is willing to pay up to £20,000 to £40,000 per additional QALY.

• PLDH appears to be cost-effective compared to topotecan and paclitaxel monotherapy in terms of the overall patient population and the main subgroups considered. These conclusions were robust to alternative cost assumptions and the use of subgroup specific treatment effects. • The cost-effectiveness results for the base-case analysis were sensitive to the inclusion of trial 30-57. Incorporating the results of trial 30-57 resulted in less favourable estimates for the ICER for PLDH vs paclitaxel monotherapy compared to the base-case results. Under this scenario the results would suggest that paclitaxel monotherapy was the optimal treatment. The results from Analysis 2 explored the cost-effectiveness of the full range of treatment comparators for platinum sensitive patients. For platinum sensitive patients, the combination of paclitaxel and platinum appears to be cost- effective.

240

8. REFERENCES

1. Cancer Research UK. CancerStats: Ovarian cancer - UK. London: Cancer Research UK; 2004. 2. Office for National Statistics. Registrations of cancer diagnosed in 2000, England. London: Office for National Statistics; 2003. 3. Office for National Statistics web site. In. 4. Cancer Incidence in Wales 1992 -2001. In: Welsh Cancer Intelligence and Surveillance Unit; 2002. 5. Huncharek M, Geshwind J, Kupelnick B. Perineal application of cosmetic talc and risk of invasive epithelial ovarian cancer: a meta-analysis of 11,933 subjects from 16 observational studies. Anticancer Research 2003;23:1955-1960. 6. Ness RB, Cramer D, Goodman M, et al. Infertility, fertility drugs and ovarian cancer: a pooled analysis of case control studies. Am J Epidemiol 2002;155:217-224. 7. Brenton J. Expression Profiling of Advanced Epithelial Ovarian Cancer to Predict Chemotherapy Response [NRR ongoing study]. In: National Research Register 2004 Issue 1. Oxford: Update Software; 2004. 8. Whittemore A, Harris R, Itnyre J, Collaborative Ovarian Cancer Group. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 US case-control studies. Am J Epidemiol 1992;136:1184-1203. 9. Riman T, Dickman PW, Nilsson S, et al. Risk factors for invasive epithelial ovarian cancer: results from a Swedish case-control study. Am J Epidemiol 2002;156:363-373. 10. Dick ML, Bain CJ, Purdie DM, Siskind V, Molloy D, Green AC. Self-reported difficulty in conceiving as a measure of infertility. Human Reproduction 2003;18:2711-2717. 11. WHO Collaborative study of Neoplasia and Steroid Contraceptives. Epithelial ovarian cancer and combined oral contraceptives. Int J Epidemiol 1989;18:538-45. 12. Bosetti C, Negri E, Trichopoulos D, et al. Long-term effects of oral contraceptives on ovarian cancer risk. Int J Cancer 2002;102:262-265. 13. Green A, Purdie D, Bain C, et al. Tubal sterilisation, hysterectomy and decreased risk of ovarian cancer: survey of women's health study group. Int J Cancer 1997;71:948-951. 14. Berek JS, Bertelson K, Du Bois A, Brady MF, Carmichael J, Eisenhauer EA, et al. Advanced epithelial ovarian cancer: 1998 consensus statements. Ann Oncol 1999;10:S87-S92. 15. Berek JS, Bast RC. Ovarian cancer screening. The use of serial complementary tumor markers to improve sensitivity and specificity for early detection. Cancer 1995;76:2092-6. 16. National Institute for Clinical Excellence. Guidance on the use of paclitaxel in the treatment of ovarian cancer. London: National Institute for Clinical Excellence (NICE); 2003. Report No.: Technology Appraisal No. 55. 17. National Institute for Clinical Excellence. Guidance on the use of topotecan for the treatment of advanced ovarian cancer. London: National Institute for Clinical Excellence (NICE); 2001. Report No.: Technology Appraisal No. 28. 18. National Institute for Clinical Excellence. Guidance on the use of pegylated liposomal doxorubicin hydrochloride (PLDH) for the treatment of advanced ovarian cancer. London: National Institute for Clinical Excellence (NICE); 2002. Report No.: Technology Appraisal No. 45. 19. British Medical Association. BNF 45. London: British Medical Association and the Royal Pharmaceutical Society of Great Britain; 2003. 20. Centre for Reviews and Dissemination. Undertaking systematic reviews of research on effectiveness: CRD's guidance for those carrying out or commissioning reviews. 2nd ed. York: University of York; 2001. 21. Drummond M, O'Brien B, Stoddart G, Torrance G. Methods for the economic evaluation of health care programmes. Oxford: Oxford Medical Publications; 1997. 22. Gordon AN, Fleagle JT, Guthrie D, Parkin DE, Gore ME, Lacave AJ. Recurrent epithelial ovarian carcinoma: a randomized phase III study of pegylated liposomal doxorubicin versus topotecan. Journal of Clinical Oncology 2001;19:3312-22. 23. Schering-Plough Ltd. Caelyx (pegylated liposomal doxorubicin hydrocholoride) in the treatment of recurrent ovarian cancer in the United Kingdom. A submission to the National Institute for Clinical Excellence; 2004 6 June.

241 24. ten Bokkel Huinink W, Gore M, Carmichael J, Gordon A, Malfetano J, Hudon I, et al. Topotecan versus palitaxel for the treatment of recurrent epithelial ovarian cancer. J Clin Oncol 1997;15:2183-93. 25. ten Bokkel Huinink W, Lane SR, Ross GA. Long-term survival in a phase III, randomised study of topotecan versus paclitaxel in advanced epithelial ovarian carcinoma. Annals of Oncology 2004;15:100-103. 26. Gore M, ten Bokkel Huinink W, Carmichael J, Gordon A, Davidson N, Coleman R, et al. Clinical evidence for topotecan-paclitaxel non--cross-resistance in ovarian cancer. Journal of Clinical Oncology 2001;19:1893-900. 27. Johnson & Johnson Pharmaceutical Research and Devlopment. L.L.C. Clinical Study Abbreviated Report. A phase III, randomized, open-label, comparative study of caelyx versus paclitaxel HCI in patients with epithelial ovarian carcinoma following failure of first-line, platinum- based chemotherapy, protocol 30-57; 2004. 28. Cantu MG, Buda A, Parma G, Rossi R, Floriani I, Bonazzi C, et al. Randomized controlled trial of single-agent paclitaxel versus cyclophosphamide, doxorubicin, and cisplatin in patients with recurrent ovarian cancer who responded to first-line platinum-based regimens. Journal of Clinical Oncology 2002;20:1232-7. 29. The ICON and AGO Collaborators. Paclitaxel plus platinum-based chemotherapy versus conventional platinum-based chemotherapy in women with relapsed ovarian cancer: the ICON4/AGO- OVAR-2.2 trial. Lancet 2003;361:2099-106. 30. Piccart MJ, Green JA, Lacave AJ, Reed N, Vergote I, Benedetti Panici P, et al. Oxaliplatin or paclitaxel in patients with platinum-pretreated advanced ovarian cancer: a randomised phase II study of the European Organisation for Research and Treatment of Cancer Gynaecolgy. J Clin Oncol 2000;18:1193-1202. 31. Rosenberg P, Andersson H, Boman K, Ridderheim M, Sorbe B, Puistola U, et al. Randomized trial of single agent paclitaxel given weekly versus every three weeks and with peroral versus intravenous steroid premedication to patients with ovarian cancer previously treated with platinum. Acta Oncologica 2002;41:418-24. 32. Omura GA, Brady MF, Look KY, Averette HE, Delmore JE, Long HJ, et al. Phase III trial of paclitaxel at two dose levels, the higher dose accompanied by filgrastim at two dose levels in platinum- pretreated epithelial ovarian cancer: an intergroup study. Journal of Clinical Oncology 2003;21:2843-8. 33. Gore M, Oza A, Rustin G, Malfetano J, Calvert H, Clarke-Pearson D, et al. A randomised trial of oral versus intravenous topotecan in patients with relapsed epithelial ovarian cancer.[see comment]. European Journal of Cancer 2002;38:57-63. 34. Vermorken JB, Gore M, Perren T, Vergote I, Colombo N, Harper P, et al. Multicenter randomized phase II study of oxaliplatin (OXA) or topotecan (TOPO) in platinum-pretreated epithelial ovarian cancer (EOC) patients (pts). Proceedings of ASCO 2001Abs 847. 35. Torri V, Floriani I, Tinazzi A, Conte PF, Ravaioli A, Maria Cantu G, et al. Randomized trial comparing paclitaxel+doxorubicin (AT) versus paclitaxel (T) as second line therapy for advanced ovarian cancer (AOC) patients in early progression after platinum based chemotherapy. Proc Am Soc Clin Oncol 2000;18:2395. 36. Loibl S, Meier W, du Bois A, Kuhn W, Pfisterer J, Kimmig R, et al. Topotecan versus treosulfan in recurrent ovarian cancer after initial chemotherapy with platinum and paclitaxel. A prospective randomised phase III study of the AGO Ovarian Cancer Study Group. EJC Supplements 2003;1:S16. 37. Gonzalez Martin AA, Calvo E, Bover I, Rubio MJ, Arcusa A, Casado A, et al. Randomised phase II study of carboplatin (C) versus paclitaxel-carboplatin (PC) in platinum-sensitive (PS) recurrent advanced ovarian carcinoma (AOC) with assessment of quality of life (QoL): a GEICO Study (Spanish group for investigation on ovarian carcinoma). Proc ASCO: American Society of Clinical Oncology 39th Annual Meeting 31 May- 3 June 2003, Chicago, Illinois 2003Abs 1812. 38. Forbes C, Wilby J, Richardson G, Sculpher M, Mather L, Riemsma R. A systematic review and economic evaluation of pegylated liposomal doxorubicin hydrochloride for ovarian cancer. Health Technology Assessment 2002;6:1-119. 39. Forbes C, Shirran L, Bagnall AM, Duffy S, ter Riet G. A rapid and systematic review of the clinical effectiveness and cost-effectiveness of topotecan for ovarian cancer. Health Technology Assessment 2001;5:1-110. 40. Covens A, Kerr I, Esmail R, Fung MFK, Browman G, Bryson P, et al. Use of topotecan in pre-treated recurrent or relapsed ovarian cancer patients. Current Oncology 2000;7:136-148.

242 41. Fung Kee Fung M, Johnston ME, Eisenhauer EA, Elit L, Hirte HW, Rosen B. Chemotherapy for recurrent epithelial ovarian cancer previously treated with platinum - a systematic review of the evidence from randomized trials. European Journal of Gynaecological Oncology 2002;23:104-110. 42. Gordon A, Carmichael J, Malfetano J, Gore M, Spaczynski M, Clark Pearson D, et al. Final analysis of a phase III, randomized study of topotecan (T) versus paclitaxel (P) in advanced epithelial overian carcinoma (OC): international topotecan study group (meeting abstract). Proceedings of the 34th Annual Meeting of the American Society of Clinical Oncology; 1998 May 15-18; Los Angeles, California, USA. 1998A1374. 43. Gordon A, Teitelbaum A. Overall survival advantage for pegylated liposomal doxorubicin compared to topotecan in recurrent epithelial ovarian cancer. EJC Supplements 2003;1:S51. 44. Schering-Plough Ltd. Data on File. A Phase III randomised, open-label comparative study of Doxil/Caelyx versus Topotecan HCI in patients with epithelial ovarian carcinoma following failure of first-line, platinum-based chemotherapy. Final Report 2004. Kenilworth, NJ: Schering-Plough Research Institute; 2004. 45. Schering-Plough Ltd. Caelyx (pegylated liposomal doxorubicin hydrocholoride) in the treatment of recurrent ovarian cancer in the United Kingdom. A submission to the National Institute for Clinical Excellence; 2001 29 August. 46. Ross G, Lane S, Dane G. Long-term survival in a phase III randomised study of topotecan (T) vs paclitaxel (P) in advanced epithelial ovarian carcinoma. European Journal of Cancer 2001;37:S326. 47. O’Byrne KJ, Bliss P, Graham JD, Gerber J, Vasey PA, Khanna S, et al. A phase III study of doxil/caelyx versus paclitaxel in platinum treated, taxane-naïve relapsed ovarian cancer. Proc ASCO 2002: American Society of Clinical Oncology 38th Annual Meeting 18-21 May 2002, Orlando, Florida 2002;21:Abs 808. 48. Colombo N. Randomised trial of paclitaxel in combination with platinum chemotherapy versus platinum-based chemotherapy in the treatment of relapsed ovarian cancer (ICON4/OVAR 2.2). EJC Supplements 2003;1:S101. 49. Calvert AH, R. ND, Gumbrell LAea. Carboplatin dosage: prospective evaluation of a simple formula based on renal function. J Clin Oncol 1987;7:1748 - 56. 50. Smith DH, Adams JR, Johnston SR, Gordon A, Drummond MF, Bennett CL. A comparative economic analysis of pegylated liposomal doxorubicin versus topotecan in ovarian cancer in the USA and the UK. Annals of Oncology 2002;13:1590-7. 51. Ojeda B, de Sande LM, Casado A, Merino P, Casado MA. Cost-minimisation analysis of pegylated liposomal doxorubicin hydrochloride versus topotecan in the treatment of patients with recurrent epithelial ovarian cancer in Spain. British Journal of Cancer 2003;89:1002-7. 52. Capri S, Cattaneo G. Cost-minimization analysis of pegylated liposomal doxorubicin versus topotecan for the treatment of ovarian cancer in Italy. Clinical Therapeutics 2003;25:1826-45. 53. Prasad M, Ben-Porat L, Hoppe B, Aghajanian C, Sabbatini P, Chi DS, et al. Costs of treatment and outcomes associated with second-line therapy and greater for relapsed ovarian cancer. Gynecologic Oncology 2004;93:223-8. 54. Merck Pharmaceuticals, GlaxoSmithKline. Research Report: An audit of relapsed ovarian cancer (OC) management in 9 UK centres; 2004 July. 55. Rodriguez M, Rose PG. Improved therapeutic index of lower dose topotecan chemotherapy in recurrent ovarian cancer. Gynecologic Oncology 2001;83:257-62. 56. British Medical Association. BNF 47. London: British Medical Association and the Royal Pharmaceutical Society of Great Britain; 2004. 57. Netten A, Curtis L. Unit costs of health and social care. Canterbury: Personal Social Services Research Unit, University of Kent; 2000. 58. Briggs AH. Handling uncertainty in cost-effectiveness models. Pharmacoeconomics 2000;17:479-500. 59. National Institute for Clinical Excellence. Guidance for manufacturers and sponsors. London: National Institute for Clinical Excellence; 2001. 60. Ades AE, Cliffe S. Markov Chain Monte Carlo estimation of a multi-parameter decision model: consistency of evidence and the accurate assessment of uncertainty. Medical Decision Making 2002;22:359-371. 61. Higgins JPT, Whitehead A. Borrowing strength from external trials in a meta-analysis. Statistics in Medicine 1996;15:2733-2749. 62. Bridle C, Palmer S, Bagnall AM, Darba J, Duffy S, Sculpher M, et al. A rapid and systematic review and economic evaluation of the clinical and cost-effectiveness of newer drugs for treatment of mania associated with bipolar affective disorder. Health Technology Assessment 2004;8:1-200.

243 63. Spiegelhalter DJ, Thomas A, Best NG, Gilks WR. BUGS: Bayesian inference using Gibbs sampling, version 0.5, (version ii). Cambridge: MRC Biostatistics Unit; 1996. 64. Salom E, Almeida Z, Mirhashemi R. Management of recurrent ovarian cancer: Evidence- based decisions. Current Opinion in Oncology 2002;14:519-527. 65. Jones L, Hawkins N, Westwood M, Wright K, Richardson G, Riemsma R. Systematic review of the clinical effectiveness and cost-effectiveness of capecitabine (XelodaR) for locally advanced and/or metastatic breast cancer. Health Technology Assessment 2004;8:1-156. 66. National Institute for Clinical Excellence. National Institute for Clinical Excellence methodological guidance: economic evaluations. London: NICE; 2004. 67. Aaronson NK, Ahmedzai S, Bergman B, Bullinger M, Cull A, Duez NJ, et al. The European Organization for Research and Treatment of Cancer QLQ-C30: A quality-of-life instrument for use in international trials in oncology. J Natl Cancer Inst 1993;85:365-376. 68. Tengs T, Wallace A. One-thousand health-related quality of life estimates. Medical Care 2000;38:583-637. 69. Ortega A, Dranitsaris G, Sturgeon J, Sutherland H, Oza A. Cost-utility analysis of paclitaxel in combination with cisplatin for patients with advanced ovarian cancer. Gynecologic Oncology 1997;66:454-63. 70. Bennett CL, Golub RM, Calhoun EA, Weinstein J, Fishman D, Lurain J, et al. Cost-utility assessment of amifostine as first-line therapy for ovarian cancer. International Journal of Gynecological Cancer 1998;8:64-72. 71. Grann VR, Jacobson JS, Thomason D, Hershman D, Heitjan DF, Neugut AI. Effect of prevention strategies on survival and quality-adjusted survival of women with BRCA1/2 mutations: an updated decision analysis. Journal of Clinical Oncology 2002;20:2520-9. 72. Calhoun EA, Fishman DA, Lurain JR, Welshman EE, Bennett CL. A comparison of ovarian cancer treatments: analysis of utility assessments of ovarian cancer patients, at-risk population, general population, and physicians. Gynecologic Oncology 2004;93:164-9. 73. Brown RE, Hutton J. Cost-utility model comparing docetaxel and paclitaxel in advanced breast cancer patients. Anti-Cancer Drugs 1998;9:899-907. 74. Dolan P, Gudex C, Kind P, Williams A. The time-trade-off method: results from a general population study. Health Economics 1996;5:141-154. 75. Bolis G, Scarfone G, Giardina G, Villa A, Mangili G, Melpignano M, et al. Carboplatin alone vs carboplatin plus epidoxorubicin as second-line therapy for cisplatin- or carboplatin-sensitive ovarian cancer. Gynecologic Oncology 2001;81:3-9. 76. GlaxoSmithKline. HycamtinT Topotecan Hydrochloride. National Institute for Clinical Excellence: Health Technology Re-appraisal; 2004 July. 77. Bristol-Myers Squibb. Health technology appraisal of topotecan, pegylated liposomal doxorubicin, and paclitaxel for second-line or subsequent advanced ovarian cancer. A submission to the National Institute for Clinical Excellence; 2004 7 July. 78. Fenwick E, Claxton K, Sculpher M. Representing uncertainty: the role of cost-effectiveness acceptability curves. Health Economics 2001;10:779-789. 79. Paclitaxel - As an adjunct to surgery for advanced ovarian cancer. Prescrire International 2002;11:30. 80. Piccart MJ, Bertelsen K, Stuart G, Cassidy J, Mangioni C, Simonsen E, et al. Long-term follow-up confirms a survival advantage of the paclitaxel-cisplatin regimen over the cyclophosphamide-cisplatin combination in advanced ovarian cancer. International Journal of Gynecological Cancer 2003;2:144-8. 81. Villella JA, Chaudhry T, Pearl ML, Valea F, DiSilvestro PA, Pollack S, et al. Comparison of tolerance of combination carboplatin and paclitaxel chemotherapy by age in women with ovarian cancer. Gynecologic Oncology 2002;86:316-22. 82. Gore M, Clarke-Pearson D, Beckman R, Fields S, Lane S, Dane G, et al. Topotecan (TOPO) in the treatment of first-relapse patients following platinum( P)/paclitaxel (TX) therapy for advanced ovarian cancer: results of a pooled analysis. Annals of Oncology 2000;11:81. 83. Soriano V, Balana C, Izquiredo M, Herrero A, del Campo JM, Cervantes A, et al. Topotecan in platinum-resistant advanced ovarian cancer (AOC) patients: evaluation of activity and prognostic factors of response. Preliminary results: a study of the Spanish Group for Research on Ovarian Cancer (GEICO). Annals of Oncology 2000;11:85. 84. Glimelius B, Bergh J, Brandt L, Brorsson B, Gunnars B, Hafstrom L, et al. The Swedish Council on Technology Assessment in Health Care (SBU) systematic overview of chemotherapy effects in some major tumour types - Summary and conclusions. Acta Oncologica 2001;40:135-154.

244 85. Hogberg T, Glimelius B, Nygren P. A systematic overview of chemotherapy effects in ovarian cancer. Acta Oncologica 2001;40:340-60. 86. Karlsson G, Nygren P, Glimelius B. Economic aspects of chemotherapy. Acta Oncologica 2001;40:412-433. 87. Vermorken JB. Optimal treatment for ovarian cancer: taxoids and beyond. Annals of Oncology 2000;11:131-139. 88. Mobus V. Tolerability and survival associated with extended treatment with topotecan in relapsed ovarian cancer (OC). Gynecologic Oncology Issues in the 8th Igcs Meeting of Buenos Aires 2000185-189. 89. Fracasso PM, Blessing JA, Morgan MA, Sood AK, Hoffman JS. Phase II study of oxaliplatin in platinum-resistant and refractory ovarian cancer: A gynecologic group study. Journal of Clinical Oncology 2003;21:2856-2859. 90. Gronlund B, Hogdall C, Hansen HH, Engelholm SA. Performance status rather than age is the key prognostic factor in second-line treatment of elderly patients with epithelial ovarian carcinoma. Cancer 2002;94:1961-7. 91. Liposomal doxorubicin - No progress in advanced ovarian cancer. Prescrire International 2002;11. 92. Cannistra SA. Is there a "best" choice of second-line agent in the treatment of recurrent, potentially platinum-sensitive ovarian cancer? Journal of Clinical Oncology 2002;20:1158-1160. 93. du Bois A, Luck HJ, Pfisterer J, Schroeder W, Blohmer JU, Kimmig R, et al. Second-line carboplatin and gemcitabine in platinum sensitive ovarian cancer - A dose-finding study by the Arbeitsgemeinschaft Gynakologische Onkologie (AGO) Ovarian Cancer Study Group. Annals of Oncology 2001;12:1115-1120. 94. Chan YM, Ngan HYS, Li BYG, Yip AMW, Ng TY, Lee PWH, et al. A longitudinal study on quality of life after gynecologic cancer treatment. Gynecologic Oncology 2001;83:10-19. 95. Slimane K. [The role of paclitaxel in the first line treatment of cancer of the ovary?]. Bulletin du Cancer 2003;90:202-3. 96. Encouraging results achieved using topotecan as salvage therapy in recurrent ovarian cancer. Oncology 2000;14:828-828. 97. Bernardi M, Mangili G, Peccatori J, Corti C, Ciceri F, Patti C, et al. High-dose topotecan alone or in combination for treatment of advanced ovarian cancer: preliminary results. Bone Marrow Transplantation 2003;31:S4-S5. 98. Markman M, Liu PY, Wilczynski S, Monk BJ, Copeland L, Alberts D. Phase 3 randomized trial of 12 versus 3 months of single-agent paclitaxel in patients with advanced ovarian cancer who attained a clinically defined complete response to platinum/paclitaxel-based chemotherapy. A Southwest Oncology Group and Gynecologic Oncology Group trial. Gynecologic Oncology 2002;84:479. 99. Conte PF, Cianci C, Tanganelli L, Gadducci A. Ovarian cancer: optimal chemotherapy in relapsed disease. Annals of Oncology 2000;11:145-150. 100. Muggia F, Hamilton A. Phase III data on Caelyx in ovarian cancer. European Journal of Cancer 2001;37:S15-8. 101. Gonzalez-Martin A. Is combination chemotherapy superior to single-agent chemotherapy in second-line treatment? International Journal of Gynecological Cancer 2003;13:185-191. 102. Zeimet AG. For a careful risk estimation: Current therapeutic concepts in ovarian cancer with regard to aspects of quality of life in long-term therapy. [German]. Wiener Klinische Wochenschrift 2002;114:16-22. 103. Johnston SR, Kaye S. Caelyx: treatment for relapsing ovarian cancer. Hospital Medicine 2001;62:611-6. 104. Mayer F, Peaud PY, Tigaud JD, Kaminsky MC, Culine S, Walter S, et al. Topotecan (T) and cyclophosphamide (Cy) in second line treatment of advanced ovarian cancer (AOC): A GINECO phase II trial. European Journal of Cancer 2001;37:S324. 105. Markman M. Second-line therapy for potentially platinum-sensitive recurrent ovarian cancer: what is optimal treatment?[comment]. Gynecologic Oncology 2001;81:1-2. 106. Costa SD, du Bois A, Lueck HJ, Meier W, Moebus V, Bauknecht T, et al. Cisplatin/paclitaxel vs. carboplatin/paclitaxel in 798 patients with ovarian cancer FIGO IIB-IV: Randomized phase III study the AGO (arbeitsgemeinschaft gynaekologische onkologie) Study Group (OVAR-3 trial). European Journal of Cancer 2001;37:S325. 107. Sun CC, Bodurka DC, Donato ML, Rubenstein EB, Borden CL, Basen-Engquist K, et al. Patient preferences regarding side effects of chemotherapy for ovarian cancer: do they change over time? Gynecologic Oncology 2002;87:118-128.

245 108. Health Technology Board for Scotland (HTBS). The use of pegylated liposomal doxorubicin hydrochloride (Caelyx) for ovarian cancer. Glasgow: Health Technology Board for Scotland (HTBS) (merged into NHS Quality Improvement Scotland (NHS QIS)); 2002. 109. Harper P. Current clinical practices for ovarian cancers. Seminars in Oncology 2002;29:3-6. 110. Le T, Leis A, Pahwa P, Wright K, Ali K, Reeder B. Quality-of-Life Issues in Patients With Ovarian Cancer and Their Caregivers: A Review. Obstetrical & Gynecological Survey 2003;58:749- 758. 111. Kaye SB, Vasey PA. Docetaxel in ovarian cancer: phase III perspectives and future development. Seminars in Oncology 2002;29:22-7. 112. Bookman MA. Developmental chemotherapy in advanced ovarian cancer: incorporation of newer cytotoxic agents in a phase III randomized trial of the Gynecologic Oncology Group (GOG- 0182). Seminars in Oncology 2002;29:20-31. 113. Phase III study on paclitaxel versus docetaxel for the treatment of ovarian carcinoma. [German]. Deutsche Apotheker Zeitung 2001;141. 114. Ohara N. Ovarian function during chemotherapy with paclitaxel and carboplatin. Journal of Obstetrics & Gynaecology 2002;22:226-7. 115. Lehoczky O, Pulay T. Transverse leukonychia secondary to paclitaxel-carboplatin chemotherapy in a patient with ovarian cancer. Journal of Obstetrics & Gynaecology 2002;22:694. 116. Topuz E, Aydiner A, Eralp Y, Saip P, Tas F, Salihoglu Y, et al. Efficacy of paclitaxel in combination with intraperitoneal cisplatinum in patients with advanced ovarian cancer. European Journal of Cancer 2001;37:S322-S323. 117. Markman M, Elson P, Kulp B, Peterson G, Zanotti K, Webster K, et al. Carboplatin plus paclitaxel combination chemotherapy: impact of sequence of drug administration on treatment-induced neutropenia. Gynecologic Oncology 2003;91:118-122. 118. Clinical and pharmacoeconomic aspects both play an important role in the treatment of ovarian cancer. Drugs & Therapy Perspectives 2001;17:12-15. 119. Oxaliplatin: shows efficacy in metastatic colorectal cancer and has potential for the treatment of other malignancies. Drugs & Therapy Perspectives 2001;17:1-5. 120. Bilgin T, Ozalp S, Yalcin OT, Zorlu G, Vardar MA, Ozerkan K. Efficacy of gemcitabine in heavily pretreated advanced ovarian cancer patients. European Journal of Gynaecological Oncology 2003;24:169-70. 121. Wojciechowska-Lacka A, Markowska J, Skasko E, Kruczek A, Steffen J. Frequent disease progression and early recurrence in patients with familial ovarian cancer primarily treated with paclitaxel and cis- or carboplatin (preliminary report). European Journal of Gynaecological Oncology 2003;24:21-4. 122. Castellano D, Mendiola C, Gonzalez de Sande L, Oramas J, Virizuela Echebaru JA, Domine M, et al. [Phase II study of pegylated liposomal doxorubicin (Caelyx) in advanced ovarian cancer: Analysis of efficacy and toxocity]. Oncologia 2002;25:42-49. 123. Alvarez AA, Clarke-Pearson DL. Platinum-resistant and refractory ovarian cancer: second- line treatment options. American Journal of Cancer 2003;2:1-13. 124. Armstrong DK. Topotecan dosing guidelines in ovarian cancer: reduction and management of hematologic toxicity. Oncologist 2004;9:33-42. 125. Armstrong DK. Novel therapies in ovarian cancer management: an update on the role of topotecan. Oncologist 2002;7:1-2. 126. Lakusta CM, Atkinson MJ, Robinson JW, Nation J, Taenzer PA, Campo MG. Quality of life in ovarian cancer patients receiving chemotherapy. Gynecologic Oncology 2001;81:490-5. 127. Hahn CA, Jones EL, Blivin JL, Sanders LL, Yu D, Dewhirst MW, et al. Prospective quality of life assessment in patients with persistent or recurrent ovarian cancer receiving whole abdomen hyperthermia and doxil. International Journal of Radiation Oncology, Biology, Physics 2003;57:S440. 128. Patel N, Solimando DA, Jr., Waddell JA. Liposomal doxorubicin for ovarian cancer. Hospital Pharmacy 2001;36:610-612. 129. Gold MA, Walker JL, Berek JS, Hallum AV, Garcia DJ, Alberts DS. Amifostine pretreatment for protection against topotecan-induced hematologic toxicity: results of a multicenter phase III trial in patients with advanced gynecologic malignancies. Gynecologic Oncology 2003;90:325-330. 130. Hensley ML, Hoppe B, Leon L, Sabbatini P, Aghajanian C, Chi D, et al. The costs and efficacy of liposomal doxorubicin in platinum-refractory ovarian cancer in heavily pretreated patients. Gynecologic Oncology 2001;82:464-9. 131. Stebbing J, Gaya A. Pegylated liposomal doxorubicin (Caelyx (TM)) in recurrent ovarian cancer. Cancer Treatment Reviews 2002;28:121-5.

246 132. Holzner B, Kemmler G, Greil R, Kopp M, Zeimet A, Raderer M, et al. The impact of hemoglobin levels on fatigue and quality of life in cancer patients. Annals of Oncology 2002;13:965- 973. 133. Toffoli G, Sorio R, Sartor F, Raffin L, Veronesi A, Boiocchi M. Carboplatin and topotecan combination and myelosuppression. Journal of Clinical Oncology 2002;20:3558. 134. Armstrong DK, Blessing JA, Look KY, Schilder R, Nunez ER. A randomized phase II evaluation of bryostatin-1 (NSC #339555) in recurrent or persistent platinum-sensitive ovarian cancer: a Gynecologic Oncology Group study. Investigational New Drugs 2003;21:373-377. 135. Chiara S, Tognoni A, Pastrone I, Tomasello L, Brema F, Di Costanzo G, et al. Topotecan and ifosfamide as salvage treatment in advanced ovarian cancer. Gynecologic Oncology 2004;93:474-8. 136. Latorre A, De Lena M, Catino A, Crucitta E, Sambiasi D, Guida M, et al. Epithelial ovarian cancer: second and third line chemotherapy (review). International Journal of Oncology 2002;21:179- 86. 137. Markman M, Bookman MA. Second-line treatment of ovarian cancer. Oncologist 2000;5:26- 35. 138. Yasuda M, Kobayashi S, Kikuchi Y, Kuramoto H, Ohta Y, Tsunoda H, et al. A phase II study of combination chemotherapy with docetaxel and carboplatin for epithelial ovarian cancer - Interim report. 9th Biennial Meeting of Gynecologic Cancer Society 2002177-181. 139. Gorbunova VA, Khokhlova SV, Komarova BP, Orel NF, Besova NS. [Comparative pharmaco-economic analysis of docetaxel with cisplatin and cyclophosphamide with cisplatin in first- line chemotherapy of advanced ovarian cancer]. Voprosy Onkologii 2002;48:695-9. 140. Piccart M, Bertelsen K, Stuart G, Cassidy J, Vergote I, Simonsen E, et al. Long term follow- up confirms a survival advantage of the paclitaxel-cisplatin (TP) regimen over the cyclophosphamide- cisplatin (CP) combination in advanced ovarian cancer (AOC). Annals of Oncology 2002;13:109. 141. Pignata S, Ballatori E, Favalli G, Scambia G. Quality of life: gynaecological cancers. Annals of Oncology 2001;12:S37-S42. 142. Goldwasser F, Misset JL. [Clinical use of oxaliplatin in solid tumors]. Bulletin du Cancer 2001;88:S40-4. 143. Boruta DM, II, Van Le L, Fowler WC, Jr., Gehrig PA, Boggess JF, Walton LA. Weekly paclitaxel infusion as salvage therapy in ovarian cancer. Gynecologic Oncology 2001;80:314. 144. Ozols RF. Update on the management of ovarian cancer. Cancer Journal 2002;8:S22-30. 145. Mielcarek P, Majdak E, Krasinska L, Kobierski J, Kozaka J, Emerich J. Comparison of quality of life in patients with advanced ovarian cancer treated with intravenous paclitaxel and carboplatin versus cyclophosphamide and cisplatin as first line chemotherapy - A preliminary report. Nowotwory 2002;52:33-36. 146. Breidenbach M, Rein DT, Schondorf T, Schmidt T, Konig E, Valter M, et al. Hematological side-effect profiles of individualized chemotherapy regimen for recurrent ovarian cancer. Anti-Cancer Drugs 2003;14:341-6. 147. Woronoff-Lemsi MH, Menat C, Limat SJ, et al. Cost-effectiveness of paclitaxel in first-line chemotherapy of advanced ovarian cancer. Proceedings of ASCO 2001;20:Abs 844. 148. Favalli G, Conte PF, Katsaros D, Zola P, Nanni O, Gadducci A, et al. Randomized phase III trial of observation versus 6 courses of single agent paclitaxel in patients with advanced ovarian cancer who attained a complete response to platinum-paclitaxel based chemotherapy. An Italian study - (AFTER-6 protocol 1). 9th Biennial Meeting of Gynecologic Cancer Society 200283-88. 149. Beshara N, Fung Kee Fung M, Faught W. The role of topotecan as second-line therapy in patients with recurrent ovarian cancer. European Journal of Gynaecological Oncology 2002;23:287-90. 150. Markman M. The myth of measurable disease in ovarian cancer. Journal of Clinical Oncology 2003;21:3013-3015. 151. Markman M, Liu PY, Wilczynski S, Monk B, Copeland LJ, Alvarez RD, et al. Phase III randomized trial of 12 versus 3 months of maintenance paclitaxel in patients with advanced ovarian cancer after complete response to platinum and paclitaxel-based chemotherapy: a Southwest Oncology Group and Gynecologic Oncology Group trial. Journal of Clinical Oncology 2003;21:2460-5. 152. Bolis G, Scarfone G, Polverino G, Raspagliesi F, Tateo S, Richiardi G, et al. Paclitaxel 175 or 225 mg per meters squared with carboplatin in advanced ovarian cancer: a randomized trial. Journal of Clinical Oncology 2004;22:686-90. 153. Slater S. Non-curative chemotherapy for cancer - is it worth it? Clinical Medicine 2001;1:220- 222. 154. Boehnke Michaud L, Valero V, Hortobagyi GN. Risks and benefits of taxanes in breast and ovarian cancer. Drug Safety 2000;23:401-428. 155. Baron-Hay S. Oxaliplatin (Eloxatin(R)). Cancer Forum 2001;25:154-155.

247 156. Doyle C, Crump M, Pintilie M, Oza AM. Does palliative chemotherapy palliate? Evaluation of expectations, outcomes, and costs in women receiving chemotherapy for advanced ovarian cancer. Journal of Clinical Oncology 2001;19:1266-1274. 157. Wagner KA, Waddell JA, Solimando DA, Jr. Paclitaxel and cisplatin (TC) regimen for advanced ovarian cancer. Hospital Pharmacy 2001;36:723-8, 797. 158. Le T, Leis A, Pahwa P, Wright K, Ali K, Reeder B, et al. Quality of life evaluations in patients with ovarian cancer during chemotherapy treatment. Gynecologic Oncology 2004;92:839-44. 159. Boos J. Off label use - label off use? Annals of Oncology 2003;14:1-5. 160. Fireman BH, Fehrenbacher L, Gruskin EP, Ray GT. Cost of care for patients in cancer clinical trials. Journal of the National Cancer Institute 2000;92:136-142. 161. Lendermann JA, Colombo N, Du Bois A, Torri V, Floriani I, Qian W, et al. Erratum: The ICON and AGO Collaborators. Paclitaxel plus platinum-based chemotherapy versus conventional platinum-based chemotherapy in women with relapsed ovarian cancer: The ICON4/AGO-OVAR-2.2 trial (Lancet (2003) 361 (2099-106)). Lancet 2003;362:1504. 162. Bodurka-Bevers D, Sun CC, Gershenson DM. Pharmacoeconomic considerations in treating ovarian cancer. Pharmacoeconomics 2000;17:133-150. 163. Chan YM, Ng TY, Ngan HY, Wong LC. Quality of life in women treated with neoadjuvant chemotherapy for advanced ovarian cancer: a prospective longitudinal study. Gynecologic Oncology 2003;88:9-16. 164. Donato ML. High-dose chemotherapy in ovarian cancer: still waiting for the right study. Gynecologic Oncology 2003;88:1-2. 165. Flynn PM, Paul J, Cruickshank DJ. Does the interval from primary surgery to chemotherapy influence progression-free survival in ovarian cancer? Gynecologic Oncology 2002;86:354-357. 166. Gadducci A, Conte P, Cianci C, Negri S, Genazzani AR. Treatment options in patients with recurrent ovarian cancer. Anticancer Research 2001;21:3557-64. 167. What is a correct difference in prices? Annals of Oncology 2001;12:1336. 168. Neymark N, Gorlia T, Adriaenssen I, Baron B, Piccart M. Cost effectiveness of paclitaxel/cisplatin compared with cyclophosphamide/cisplatin in the treatment of advanced ovarian cancer in Belgium. Pharmacoeconomics 2002;20:485-97. 169. Gelderblom AJ, Sparreboom A, de Jonge MJA, Loos WJ, Hennis B, Verweij J, et al. Dose and schedule finding study of oral topotecan (T) in combination with weekly cisplatin (C) +/- cremophor EL (CrEl) in patients (pts) with recurrent ovarian cancer. Clinical Cancer Research 2000;6:246. 170. Szucs TD, Wyss P, Dedes KJ. Cost-effectiveness studies in ovarian cancer. International Journal of Gynecological Cancer 2003;13:212-9. 171. Van Den Bosch J, Hoekman K, Verheyen RHM, Pinedo HM. Phase II study of paclitaxel (taxol, gemcitabine, and cisplatin for patients with advanced ovarian cancer). European Journal of Cancer 2001;37:S319. 172. International Collaborative Ovarian Neoplasm (ICON1) Collaborators. International collaborative ovarian neoplasm trial 1: a randomized trial of adjuvant chemotherapy in women with early-stage ovarian cancer. Journal of the National Cancer Institute 2003;95:125-132. 173. Piccart MJ. RESPONSE: re: randomized intergroup trial of cisplatin-paclitaxel versus cisplatin-cyclophosphamide in women with advanced epithelial ovarian cancer: three-year results. Journal of the National Cancer Institute 2000;92:1446-1447. 174. Chan S. Second-line chemotherapy for ovarian cancer. Lancet Oncology 2003;4:333-4. 175. NICE updates paclitaxel guidance. Pharmaceutical Journal 2003;270:104. 176. Ovarian cancer treatment for second-line use endorsed by NICE. Pharmaceutical Journal 2002;269:122. 177. Trial supports first-line carboplatin monotherapy for ovarian cancer. Pharmaceutical Journal 2002;269:208. 178. Young M, Plosker GL. Paclitaxel: a pharmacoeconomic review of its use in the treatment of ovarian cancer. Pharmacoeconomics 2001;19:1227-59. 179. Copeland LJ, Bookman M, Trimble E. Clinical trials of newer regimens for treating ovarian cancer: the rationale for Gynecologic Oncology Group Protocol GOG 182-ICON5. Gynecologic Oncology 2003;90:S1-S7. 180. Wenzel LB, Huang H, Armstrong D, Walker J, Cella D. Quality of life (QOL) results of a randomized study of intravenous (IV) paclitaxel and cisplatin vs IV paclitaxel, intraperitoneal (IP) cisplatin and IP paclitaxel in optimal stage III epithelial ovarian cancer (OC): A Gynecologic Oncology Group trial. 2004 ASCO Annual Meeting 2004Abstract No: 5026.

248 181. Scarfone G, Parazzini F, Sciatta C, Rabaiotti E, Richiardi G, Tateo S, et al. A multicenter Randomized Trial Comparing Two Different Doses of TAXOL (T) Plus a Fixed Dose of Carboplatin (C) in Advanced Ovarian Cancer (AOC). 2001 ASCO Annual Meeting 2001Abstract No: 816. 182. Gennatas C, Mouratidou D, Andreadis C. A Phase III Trial Comparing Taxol and Cisplatin Versus Cyclophosphamide and Cisplatin in Advanced Ovarian Cancer: A Preliminary Report. 2000 ASCO Annual Meeting 2000Abstract No: 1600. 183. Muggia FM, Braly PS, Brady MR, et al. Phase III randomised study of cisplatin versus paclitaxel versus cisplatin and paclitaxel in patients with subopimal stage III or IV ovarian cancer: a Gynaecological Oncology Group Study. Journal of Clinical Oncology 2000;18:106-115. 184. Piccart M, Bertelsen K, James K, Cassidy J, Mangioni C, Simonsen E, et al. Randomised Intergroup trial of cisplatin-paclitaxel versus cisplatin-cyclophosphamide in women with advanced epitehelial ovarian cancer: 3 year results. Journal of National Cancer Institute 2000;92. 185. Winter IW, McBroom JW, Carlson JW, Rose GS, Elkas JC. The utility of gastrojejunostomy in secondary cytoreduction and palliation of proximal intestinal obstruction in recurrent ovarian cancer. Gynecologic Oncology 2003;91:261-264. 186. Nicholls C, Cassidy J, Freemantle N, Harrison M, Carita P. Cost-effectiveness of combination chemotherapy (oxaliplatin or irinotecan in combination with 5-FU/FA) compared with 5-FU/FA alone. Journal of Medical Economics 2001;4:115-125. 187. De Haes JCJM, Raatgever JW, van der Burg MEL, Hamersma E, Neijt JP. Evaluation of the quality of life of patients with advanced ovarian cancer treated with combination chemotherapy. In: Aaronson NK, Beckman J, editors. The quality of life of cancer patients. New York: Raven Press; 1987. p. 215-27. 188. Bolis G, Parazzini F, Scarfone G, Villa A, Amoroso M, Rabaiotti E, et al. Paclitaxel vs epidoxorubicin plus paclitaxel as second-line therapy for platinum-refractory and -resistant ovarian cancer. Gynecol Oncol 1999;72:60-64. 189. Dranitsaris G, Elia-Pacitti J, Cottrell W. Measuring treatment preferences and willingness to pay for docetaxel in advanced ovarian cancer. Pharmacoeconomics 2004;22:375-87. 190. The International Collaborative Ovarian Neoplasm (ICON) Group. Paclitaxel plus carboplatin versus standard chemotherapy with either single-agent carboplatin or cyclophosphamide, doxorubicin, and cisplatin in women with ovarian cancer: the ICON3 randomised trial. Lancet 2002;360:505-515. 191. Fiorica JV. The role of topotecan in the treatment of advanced cervical cancer. Gynecologic Oncology 2003;90:S16-S21. 192. Mutch DG. Gemcitabine combination chemotherapy of ovarian cancer. Gynecologic Oncology 2003;90:S16-S20. 193. Stuart GCE. First-line treatment regimens and the role of consolidation therapy in advanced ovarian cancer. Gynecologic Oncology 2003;90:S8-S15. 194. Fabbro M, Leduc B, Mignot L, Ayela P, Assouline D, Gouttebel MC, et al. Topotecan and paclitaxel in second line treatment of advanced ovarian cancer (AOC): a GINECO phase II trial. European Journal of Cancer 2001;37:S324. 195. Bankhead C. For ovarian cancer, an optimal treatment remains to be found. Journal of the National Cancer Institute 2004;96:96-7. 196. Wang L, Arnold K. Paclitaxel combination improves survival from relapsed ovarian cancer. Journal of the National Cancer Institute 2003;95:1037. 197. Trimbos JB, Parmar M, Vergote I, Guthrie D, Bolis G, Colombo N, et al. International collaborative ovarian neoplasm trial 1 and adjuvant chemotherapy in ovarian neoplasm trial: Two parallel randomized phase III trials of adjuvant chemotherapy in patients with early-stage ovarian carcinoma. Journal of the National Cancer Institute 2003;95:105-112. 198. Calhoun EA, Welshman EE, Chang C, et al. Cost of chemotherapy induced neutropenia, thrombocytopenia and neurotoxicity. Proc Am Soc Clin Oncol 2001;20. 199. Moe M, El-Sharkawi S. Clinical audit on the use of taxanes for ovarian cancer. Journal of Clinical Excellence 2003;4:312-314. 200. Wu CH, Yang CH, Lee JN, Hsu SC, Tsai EM. Weekly and monthly regimens of paclitaxel and carboplatin in the management of advanced ovarian cancer. A preliminary report on side effects. International Journal of Gynecological Cancer 2001;11:295-9. 201. Calhoun EA, Chang C-H, Welshman EE, Cella D. The impact of chemotherapy delays on quality of life in patients with cancer. Blood 2003;102:749a. 202. Biamonte R, Pignata S, Bianco AR, Divagno G, Balestrino M, Rossi A, et al. The MITO (multicenter Italian trials in ovarian cancer) study: topotecan vs. nihil following response to carboplatin + paclitaxel in advanced ovarian cancer (AOC). Initial compliance and toxicity data. International Journal of Gynecological Cancer, special online supplement: www.blackwellscience.com/ijg 2000.

249 203. Moss C, Kaye SB. Ovarian cancer: progress and continuing controversies in management. European Journal of Cancer 2002;38:1701-7. 204. Adams JR, Elting LS, Lyman GH, George JN, Lembersky BC, Armitage JO, et al. Use of erythropoietin in cancer patients: assessment of oncologists' practice patterns in the United States and other countries. American Journal of Medicine 2004;116:28-34. 205. Guastalla JP, 3rd, Dieras V. The taxanes: toxicity and quality of life considerations in advanced ovarian cancer. British Journal of Cancer 2003;89:S16-22. 206. Katsumata N. Docetaxel: an alternative taxane in ovarian cancer. British Journal of Cancer 2003;89:S9-S15. 207. Kose F, Sufliarsky J, Beslija S, Saip P, Krejcy K, Minarik T, et al. Gemcitabine (G) plus carboplatin (C) in patients whose epithelial ovarian carcinomas (EOC) relapsed >=6 months after platinum-containing first-line therapy: preliminary results of a phase II study. EJC Supplements 2003;1:S53. 208. Maenpaa JU. Docetaxel: promising and novel combinations in ovarian cancer. British Journal of Cancer 2003;89:S29-34. 209. Gronlund B, Hogdall C, Hansen HH, Engelholm SA. Re-induction therapy with paclitaxel and carboplatin in recurrent epithelial ovarian cancer (EOC). European Journal of Cancer 2001;37:S318. 210. Calhoun EA, Welshman EE, Chang CH, Lurain JR, Fishman DA, Hunt TL, et al. Psychometric evaluation of the Functional Assessment of Cancer Therapy/Gynecologic Oncology Group - Neurotoxicity (Fact/GOG-Ntx) questionnaire for patients receiving systemic chemotherapy. International Journal of Gynecological Cancer 2003;13:741-748. 211. Piccart MJ, Floquet A, Scarfone G, Willemse PHB, Emerich J, Vergote I, et al. Intraperitoneal cisplatin versus no further treatment: 8-year results of EORTC 55875, a randomized phase III study in ovarian cancer patients with a pathologically complete remission after platinum-based intravenous chemotherapy. International Journal of Gynecological Cancer 2003;13:196-203. 212. Bookman MA, Greer BE, Ozols RF. Optimal therapy of advanced ovarian cancer: carboplatin and paclitaxel versus cisplatin and paclitaxel (GOG158) and an update on GOG0182-ICON5. International Journal of Gynecological Cancer 2003;13:149-55. 213. Malik IA. An open label evaluation of topotecan in patients with relapsed or refractory epithelial ovarian cancer - single institution experience in a developing country. International Journal of Gynecological Cancer 2000;10:443-448. 214. Vermorken JB. The integration of paclitaxel and new platinum compounds in the treatment of advanced ovarian cancer. International Journal of Gynecological Cancer 2001;11:21-30. 215. Bookman M, Malmstrom H, Bolis G, Gordon A, Lissoni A, Krebs JB, et al. Topotecan for the treatment of advanced epithelial ovarian cancer: an open-label phase II study in patients treated after prior chemotherapy that contained cisplatin or carboplatin and paclitaxel. J Clin Oncol 1998;16. 216. Kavanagh JJ, Kudelka A, Spriggs DR, Bookman MA, Lewis L, Maack C, et al. Phase 2 study of TLK286 (GST P1-1 Activated Glutathione Analog) in patients with platinum and paclitaxel refractory/resistant advanced epithelial ovarian cancer. European Journal of Cancer 2002;38:100. 217. Jasas K, Trudeau M, Stuart G, Gillesby BE, Fields N, Oza A. Early results of a phase II study of oral topotecan and intravenous cisplatin in epithelial ovarian cancer recurring more than 6 months following initial platinum therapy. EJC Supplements 2003;1:S59. 218. Sehouli J, Lichtenegger W, Oskay G, Koensgen D, Hindenburg HJ, Klare P, et al. Relapsed ovarian cancer after failure of first-line chemotherapy with platin and paclitaxel: a phase II study. European Journal of Cancer 2001;37:S323-S324. 219. Lalisang R, Erdkamp F, Wils J, Vreeswijk J, Wals J, Stoot J, et al. A phase II study of docetaxel, epirubicin, and cisplatin with G-CSF (lenograstim) support in patients with advanced ovarian cancer. EJC Supplements 2003;1:S54. 220. Wheatley K, Gray R. ICON4/AGO-OVAR-2.2: questions about trial design, cost- effectiveness, and clinical benefit. Lancet 2003;362:1333. 221. Mielke S, Mross K, Gerds TA, Schmidt A, Lange W, Behringer D. A multicenter, randomized phase III study on neurotoxicity, safety and efficacy of weekly paclitaxel infused over 1-h vs. 3-h in patients with advanced solid tumors. EJC Supplements 2003;1:S164. 222. Green JA. ICON4/AGO-OVAR-2.2: questions about trial design, cost-effectiveness, and clinical benefit. Lancet 2003;362:1333-4. 223. Survival in relapsed ovarian cancer improved with paclitaxel. Pharmaceutical Journal 2003;270:882. 224. Vorobiof DA, Rapoport BL, Mahomed R, Uys A, Slabber C, Eek R, et al. A phase II second line study of liposomal doxorubicin and carboplatin in patients with recurrent ovarian cancer with a disease free interval equal or greater than 6 months. EJC Supplements 2003;1:S57.

250 225. Newman G. ICON4/AGO-OVAR-2.2: questions about trial design, cost-effectiveness, and clinical benefit. Lancet 2003;362:1333. 226. Smith DH, Drummond MF, Johnston S, Gordon A. Economic evaluation of liposomal doxorubicin versus toptecan for recurrent ovarian cancer in the UK. Value in Health 2001;4:87. 227. Girre V, Pujade-Lauraine E, Durand-Zaleski I. [Economic assessment of Caelyx(R) versus topotecan in advanced ovarian cancer]. Bulletin du Cancer 2003;90:983-8. 228. Smith DH, Johnston SRD, Gordon AN, Drummond M. Economic evaluation of Doxil/Caelyx versus topotecan for recurrent epithelial ovarian carcinoma: the UK perspective. Proc Am Soc Clin Oncol 2001;20:A808. 229. van Hattum AH, Pinedo HM, Schluper HM, Erkelens CA, Tohgo A, Boven E. The activity profile of the hexacyclic camptothecin derivative DX-8951f in experimental human colon cancer and ovarian cancer. Biochemical Pharmacology 2002;64:1267-77.

251 APPPENDICES Appendix 1. Details about FIGO cancer staging

Stage I: Growth limited to the ovaries Ia One ovary Ib Both ovaries involved Ic Ascites (an accumulation of fluid in the abdominal (peritoneal) cavity) present or positive peritoneal washings

Stage II: Growth limited to pelvis IIa Extension to gynaecological adnexae (on or in a structure associated with the uterus such as on ovary, fallopian tube or uterine ligament) IIb Extension to other pelvic tissues IIc Ascites or positive washings

Stage III: Extra pelvic tumour present – limited to true pelvis but with superficial liver metastases, peritoneal surface seedlings, or histologic-proven malignant extension to the omentum. IIIa Limited to true pelvis with negative nodes, but seeding of abdominal peritoneal surfaces or histologic-proven extension to small bowel or mesentery IIIb Peritoneal metastasis of abdominal peritoneal surfaces, none exceeding 2cm in diameter; nodes negative IIIc Peritoneal metastasis beyond pelvis > 2cm in diameter and/or positive retroperitoneal or inguinal nodes.

Stage IV: Metastases to distant sites (including hepatic parenchymal disease)

252 Appendix 2. Summary of current manufacturers information provided for health professionals

Indications for Warnings Side Dosage Intervention treatment in UK or Cautions Effects in Adults Cost Topotecan is Topotecan is Contra-indications General side effects of The recommended dose of Topotecan (as available in a 4 indicated for the A history of severe hypersensitivity to topotecan or to cytotoxic drugs: topotecan is 1.5 mg/m2 body hydrochloride) net price mg vial of treatment of patients any of the excipients extravasation of surface area/day administered 1-mg vial = £105; 4-mg powder. with metastatic Pregnant or breast feeding intravenous drugs, oral by intravenous infusion over vial = £312.50 carcinoma of the Severe bone marrow depression prior to starting first mucositis, nausea and 30 minutes daily for 5 ovary after failure course (baseline neutrophils < 1.5 x 109/l and/or a vomiting, bone-marrow consecutive days with a 3 of first-line or platelet count of 100 x 109/l). suppression, alopecia, and week interval between the subsequent therapy. reproductive function start of each course. Prior to Special warnings and special indications for use adverse effects. administration of the first Haematological toxicity is dose-related and full blood Specific to topotecan: course of topotecan, patients count including platelets should be monitored regularly. Gastro-intestinal effects, must have a baseline As expected, patients with poor performance status neutrophil count of 1.5 x asthenia, alopecia, and 9 have a lower response rate and an increased incidence anorexia. 10 /l, and a platelet count of of complications such as fever and infection. 100 x 109/l. Routine pre- There is no experience of the use of topotecan in medication for non- patients with severely impaired renal function haematological adverse (creatinine clearance < 20 ml/min) or severely impaired events is not required with hepatic function (serum bilirubin 10 mg/dl) due to topotecan. cirrhosis. Topotecan is not recommended to be used in these patient groups. A small number of hepatically impaired patients (serum bilirubin between 1.5 and 10 mg/dl) were able to tolerate 1.5 mg/m2 for five days every three weeks although a reduction in topotecan clearance was observed. There are insufficient data available to make a dose recommendation for this patient group. Pegylated Pegylated liposomal Contra-indications General side effects of The recommended dose of For intravenous liposomal doxorubicin hypersensitivity to the active substance or to any of the cytotoxic drugs: Caelyx is 50 mg/m2 infusion, doxorubicin doxorubicin hydrochloride is excipients extravasation of administered intravenously hydrochloride 2 mg/mL hydrochloride indicated as a Breast-feeding intravenous drugs, oral once every 4 weeks. for as encapsulated in (Caelyx) treatment for Caelyx must not be used to treat AIDS-KS that may be mucositis, nausea and long as the disease does not liposomes. For dilution treated effectively with local therapy or systemic alfa- Caelyx is advanced ovarian vomiting, bone-marrow progress and the patient before use. Net price 10- cancer in women interferon. suppression, alopecia, and continues to tolerate mL vial = £411.30, available in a 2

253 Indications for Warnings Side Dosage Intervention treatment in UK or Cautions Effects in Adults Cost mg/ml who have failed a Special warnings and special indications for use reproductive function treatment. 25mL-vial = £813.49 concentrate for first-line platinum- Cardiac toxicity: It is recommended that all patients adverse effects. solution for based chemotherapy receiving Caelyx routinely undergo frequent ECG Specific to Caelyx: infusion. regimen. The drug monitoring. Transient ECG changes such as T-wave nausea and vomiting, is also indicated as a flattening, S-T segment depression and benign myelosupression, monotherapy for arrhythmias are not considered mandatory indications alopecia, and mucositis. patients with for the suspension of Caelyx therapy. However, metastatic breast reduction of the QRS complex is considered more cancer, where there indicative of cardiac toxicity. If this change occurs, the is an increased most definitive test for anthracycline myocardial injury, cardiac risk and for i.e., endomyocardial biopsy, must be considered. The treatment of AIDS- evaluation of left ventricular function is considered to related Kaposi’s be mandatory before each additional administration of sarcoma (KS) in Caelyx that exceeds a lifetime cumulative anthracycline patients with low dose of 450 mg/m2. CD4 counts (< 200 CD4 3 Myelosuppression: Many patients treated with Caelyx lymphocytes/mm ) have baseline myelosuppression due to such factors as and extensive their pre-existing HIV disease or numerous mucocutaneous or concomitant or previous medications, or tumours visceral disease. involving bone marrow. In the pivotal trial in patients with ovarian cancer treated at a dose of 50 mg/m2, myelosuppression was generally mild to moderate, reversible, and was not associated with episodes of neutropaenic infection or sepsis. However, because of the potential for bone marrow suppression, periodic blood counts must be performed frequently during the course of Caelyx therapy, and at a minimum, prior to each dose of Caelyx. Persistent severe myelosuppression, may result in superinfection or haemorrhage.

Infusion-associated reactions: Serious and sometimes life-threatening infusion reactions, which are characterised by allergic-like or anaphylactoid-like reactions, with symptoms including asthma, flushing,

254 Indications for Warnings Side Dosage Intervention treatment in UK or Cautions Effects in Adults Cost urticarial rash, chest pain, fever, hypertension, tachycardia, pruritus, sweating, shortness of breath, facial oedema, chills, back pain, tightness in the chest and throat and/or hypotension may occur within minutes of starting the infusion of Caelyx (see 4.8). Temporarily stopping the infusion usually resolves these symptoms without further therapy. However, medications to treat these symptoms (e.g., antihistamines, corticosteroids, and adrenaline), as well as emergency equipment should be available for immediate use. In most patients treatment can be resumed after all symptoms have resolved, without recurrence. Infusion reactions rarely recur after the first treatment cycle.

Diabetic patients: Please note that each vial of Caelyx contains sucrose and the dose is administered in 5 % (50 mg/ml) glucose solution for infusion. Paclitaxel Paclitaxel is Contra-indications General side effects of The recommended dose of Paclitaxel 6mg/mL, net paciltaxel for the secondary Pacilatexel is indicated for Paclitaxel is contra-indicated in patients with severe cytotoxic drugs: price 5-mL vial = second-line hypersensitivity reactions to paclitaxel or any other extravasation of treatment of ovarian and £124.79, 16.7-mL vial = available in vials breast carcinomia is containing chemotherapy of component of the formulation, especially intravenous drugs, oral 2 £374.00, 50mL-vial = ovarian cancer in polyethoxylated castor oil. mucositis, nausea and 175mg/m administered over £1122.00 (hosp. only) 30mg, 100mg or a period of 3 hours with a 3- 300mg paclitaxel women who have Paclitaxel is contra-indicated during pregnancy and vomiting, bone-marrow failed a first-line lactation. suppression, alopecia, and week interval between as a 6mg/ml courses. Subsequent doses of solution. platinum-based Paclitaxel should not be used in patients with baseline reproductive function chemotherapy neutrophils < 1.5 x 109/L. adverse effects. paclitaxel should be regimen. The drug administered according to Specific to paclitaxel: individual patient tolerance. is also indicated for Special warnings and special precautions for use myelosuppression, the primary Patients must be pretreated with corticosteroids, Paclitaxel should not be peripheral neuropathy, readministered until the treatment of ovarian antihistamines and H2 antagonists. and cardiac conduction cancer, in Taxol should be given before cisplatin when used in neutrophil count is ≥ 1.5 x defects with arrhythmias 9 combination with combination. 10 /L and the platelet count is (which are nearly always ≥ 100 x 109/L. Patients who cisplatin, in patients Hypersensitivity reactions: Significant hypersensitivity asymptomatic). It also with advanced reactions characterised by dyspnoea and hypotension experience severe neutropenia causes muscle pain and (neutrophil count < 0.5 x disease or residual requiring treatment, angiodema and generalised alopecia; nausea and

255 Indications for Warnings Side Dosage Intervention treatment in UK or Cautions Effects in Adults Cost disease (>1cm) after urticaria have occurred in < 1% of patients receiving vomiting is mild to 109/L for ≥ 7 days) or severe initial surgery. paclitaxel after adequate premedication. These moderate. peripheral neuropathy should Paclitaxel is also reactions are probably histamine mediated. In the case receive a dose reduction of indicated for the of severe hypersensitivity reactions, paclitaxel should 20% for subsequent courses. treatment of be discontinued immediately, symptomatic therapy All patients must be metastatic should be initiated and the patient should not be premedicated with carcinoma of the rechallenged with the drug. corticosteroids, antihistamines breast in patients Haematological: Bone marrow suppression (primarily and H antagonists prior to who have failed or neutropenia) is the dose limiting toxicity. Frequent 2 paclitaxel. are not candidates monitoring of blood counts should be instituted. for standard Patients should not be retreated until neutrophils anthracycline recover to a level ≥ 1.5 x 109/L and the platelets containing therapy recover to a level ≥ 100 x 109/L. and also for the Cardiovascular: Severe cardiac conduction treatment of non- abnormalities have been reported rarely. If patients small cell lung develop significant conduction abnormalities during cancer. paclitaxel administration, appropriate therapy should be administered and continuous cardiac monitoring should be performed during subsequent therapy with paclitaxel. Nervous system: Although the occurrence of peripheral neuropathy is frequent, the development of severe symptoms is unusual. In severe cases, a dose reduction of 20% is recommended for all subsequent courses of paclitaxel. Patients with liver impairment: There is no evidence that the toxicity of paclitaxel is increased when given as a 3 hour infusion to patients with mildly abnormal liver function. No data are available for patients with severe baseline cholestasis. When paclitaxel is given as a longer infusion, increased myelosuppression may be seen in patients with moderate to severe hepatic impairment. Paclitaxel is not recommended for patients with severely impaired hepatic function. Other: Since paclitaxel contains dehydrated alcohol (396 mg/mL), consideration should be given to possible

256 Indications for Warnings Side Dosage Intervention treatment in UK or Cautions Effects in Adults Cost central nervous system and other effects. Special care should be taken to avoid intra-arterial administration of paclitaxel. In animal trials investigating local tolerance, severe tissue reactions occurred following intra-arterial administration.

257 Appendix 3. Search strategy

Search Strategies

Limits: • Date limits applied to update searches run for previous NICE TARs (by publication date) for topotecan (2000-2004), caelyx (2001-2004) and paclitaxel (2001-2004) • Animal-only studies were excluded where possible • Study design: No limits • Language: No limits

MEDLINE (Ovid host) 2000/2001 - April 2004

1. exp Ovarian Neoplasms/ 2. (ovar$ adj4 (cancer$ or tumo?r$ or malignan$ or oncolog$ or carcinoma$ or neoplas$ or mass$ or growth$ or cyst$)).mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 3. (adenexa$ adj4 mass$).mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 4. or/1-3 5. Topotecan/ 6. topotecan.mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 7. (hycamtin or hycamptamine).mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 8. or/5-7 9. exp Doxorubicin/ 10. doxil.mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 11. (doxorubicin hydrochloride or doxorubicin hcl).mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 12. liposomal doxorubicin.mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 13. (caelyx or adriamycin or rubex).mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 14. liposome encapsulated doxorubicin.mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 15. or/9-14 16. Paclitaxel/ 17. paclitaxel.mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 18. docetaxel.mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 19. taxol.mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 20. taxotere.mp. [mp=title, original title, abstract, name of substance, mesh subject heading] 21. or/16-20 22. 8 23. limit 22 to yr=2000-2004 24. 15 25. limit 24 to yr=2001-2004 26. 21 27. limit 26 to yr=2001-2004 28. 23 or 25 or 27 29. 4 and 28

258 30. animal/ not (animal/ and human/) 31. 29 not 30

PREMEDLINE (Ovid host) 2000/2001 - April 2004

1. (ovar$ adj4 (cancer$ or tumo?r$ or malignan$ or oncolog$ or carcinoma$ or neoplas$ or mass$ or growth$ or cyst$)).mp. [mp=title, abstract] 2. (adenexa$ adj4 mass$).mp. [mp=title, abstract] 3. 1 or 2 4. topotecan.mp. [mp=title, abstract] 5. (hycamtin or hycamptamine).mp. [mp=title, abstract] 6. 4 or 5 7. doxil.mp. [mp=title, abstract] 8. (doxorubicin hydrochloride or doxorubicin hcl).mp. [mp=title, abstract] 9. liposomal doxorubicin.mp. [mp=title, abstract] 10. (caelyx or adriamycin or rubex).mp. [mp=title, abstract] 11. liposome encapsulated doxorubicin.mp. [mp=title, abstract] 12. or/7-11 13. paclitaxel.mp. [mp=title, abstract] 14. docetaxel.mp. [mp=title, abstract] 15. taxol.mp. [mp=title, abstract] 16. taxotere.mp. [mp=title, abstract] 17. or/13-16

EMBASE (Ovid host) 2000/2001 - April 2004

1. exp Ovary Cancer/ 2. (ovar$ adj4 (cancer$ or tumo?r$ or malignan$ or oncolog$ or carcinoma$ or neoplas$ or mass$ or growth$ or cyst$)).mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 3. (adenexa$ adj4 mass$).mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 4. or/1-3 5. Topotecan/ 6. topotecan.mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 7. (hycamtin or hycamptamine).mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 8. or/5-7 9. Doxorubicin/ 10. doxil.mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 11. (doxorubicin hydrochloride or doxorubicin hcl).mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 12. liposomal doxorubicin.mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 13. (caelyx or adriamycin or rubex).mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name]

259 14. liposome encapsulated doxorubicin.mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 15. or/9-14 16. Paclitaxel/ 17. paclitaxel.mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 18. docetaxel.mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 19. taxol.mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 20. taxotere.mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 21. or/16-20 22. 8 23. limit 22 to yr=2000-2004 24. 15 25. limit 24 to yr=2001-2004 26. 21 27. limit 26 to yr=2001-2004 28. 23 or 25 or 27 29. 4 and 28 30. animal/ or non-human/ 31. human/ 32. 30 not (30 and 31) 33. 29 not 32

CINAHL (Ovid host) 2000/2001 - April 2004

260 20. or/15-19 21. 7 22. limit 21 to yr=2000-2004 23. 14 24. limit 23 to yr=2001-2004 25. 20 26. limit 25 to yr=2001-2004 27. 22 or 24 or 26 28. 4 and 27

DARE NHS EED HTA Database (http://www.york.ac.uk/inst/crd/crddatabases.htm) 2000/2001 - April 2004 s ovar$(4w)(cancer$ or tumor$ or tumour$ or malignan$ or oncolog$ or carcinoma$ or neoplas$ or mass$ or growth$ or cyst$) s adenexa$(4w)mass$ s s1 or s2 s topotecan s hycamtin or hycamptamine s s4 or s5 s s6 and ((2000 or 2001 or 2002 or 2003 or 2004)/xyr) s doxil s doxorubicin hydrochloride or doxorubicin(w)hcl s liposomal(w)doxorubicin s caelyx or adriamycin or rubex s liposome(w)encapsulated(w)doxorubicin s s8 or s9 or s10 or s11 or s12 s s13 and ((2001 or 2002 or 2003 or 2004)/xyr) s paclitaxel s docetaxel s taxol s taxotere s s15 or s16 or s17 or s18 s s19 and ((2001 or 2002 or 2003 or 2004)/xyr) s s7 or s14 or s20 s s3 and s21

Cochrane Controlled Trials Register Cochrane Database of Systematic Reviews (The Cochrane Library) 2000/2001 - Issue 2, 2004

261 #10. (ovar* near growth*) #11. (ovar* near cyst*) #12. (adenexa* near mass*) #13. (#1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11 or #12) #14. TOPOTECAN single term (MeSH) #15. (topotecan or hycamtin or hycamptamine) #16. (#14 or #15) #17. (#14 or #15) ( 2000 to current date ) #18. DOXORUBICIN explode tree 1 (MeSH) #19. (doxil or (doxorubicin next hydrochloride) or (doxorubicin next hcl)) #20. (liposomal next doxorubicin) #21. (caelyx or adriamycin or rubex) #22. (liposome next encapsulated next doxorubicin) #23. (#18 or #19 or #20 or #21 or #22) #24. #23 ( 2001 to current date ) #25. PACLITAXEL single term (MeSH) #26. (paclitaxel or docetaxel or taxol or taxotere) #27. (#25 or #26) #28. (#25 or #26) ( 2001 to current date ) #29. (#17 or #24 or #28) #30. (#13 and #29)

Science Citation Index Index to Scientific and Technical Proceedings (Web of Knowledge) 2000/2001 - April 2004

((ovar* same (cancer* or tumo?r* or malignan* or oncolog* or carcinoma* or neoplas* or mass* or growth* or cyst*)) or (adenexa* same mass*)) and (topotecan or hycamtin or hycamptamine) [limit: 2000-2004]

((ovar* same (cancer* or tumo?r* or malignan* or oncolog* or carcinoma* or neoplas* or mass* or growth* or cyst*)) or (adenexa* same mass*)) and (doxil or doxorubicin hydrochloride or doxorubicin hcl or liposomal doxorubicin or caelyx or adriamycin or rubex or liposome encapsulated doxorubicin) [limit: 2001-2004]

((ovar* same (cancer* or tumo?r* or malignan* or oncolog* or carcinoma* or neoplas* or mass* or growth* or cyst*)) or (adenexa* same mass*)) and (paclitaxel or docetaxel or taxol or taxotere) [limit: 2001-2004]

Biosis (Edina host) 2000/2001 - April 2004 ovar* w4 cancer* ovar* w4 tumo*r* ovar* w4 malignan* ovar* w4 oncolog* ovar* w4 carcinoma* ovar* w4 neoplas*

262 ovar* w4 mass* ovar* w4 growth* ovar* w4 cyst* adenexa* w4 mass* and topotecan or hycamtin or hycamptamine (2000-2004) or doxil or "doxorubicin hydrochloride" or "doxorubicin hcl" (2001-2004) or "liposomal doxorubicin" or caelyx or adriamycin (2001-2004) or rubex or "liposome encapsulated doxorubicin" (2001-2004) or paclitaxel or docetaxel or taxol or taxotere (2001-2004)

OHE HEED (CD-ROM) 2000/2001 - April 2004 ovarian or ovary or ovaries and topotecan or hycamtin or hycamptamine or doxil or doxorubicin or caelyx or adriamycin or rubex or paclitaxel or docetaxel or taxol or taxotere

Controlled Trials.com - mRCT / ISRCTN (http://controlled-trials.com/) 2000/2001 - April 2004 ovarian and (topotecan or hycamtin or hycamptamine or doxorubicin or doxil or caelyx or adriamycin or rubex or paclitaxel or docetaxel or taxol or taxotere)

Cancer Portfolio.org (http://www.cancerportfolio.org/) 2000/2001 - April 2004

Year = 2004; 2003; 2002; 2001; 2000 Type of Research = Causes of Cancer/Etiology; Prevention; Early Detection, Diagnosis, and Prognosis; Treatment; Cancer Control, Survivorship and Outcomes Research Type of Cancer = Ovarian Cancer Any of these words = topotecan or hycamtin or hycamptamine or doxorubicin or doxil or caelyx or adriamycin or rubex or paclitaxel or docetaxel or taxol or taxotere

Clinical Trials,gov (http://www.clinicaltrials.gov/) 2000/2001 - April 2004

263 topotecan or hycamtin or hycamptamine or doxorubicin or doxil or caelyx or adriamycin or rubex or paclitaxel or docetaxel or taxol or taxotere [ALL-FIELDS] AND "ovarian cancer" [CONDITION]

264 Appendix 4. Excluded studies

Study details Reason for exclusion Anonymous (2002)(79) Disscusion article Piccart (2003)(80) First-line therapy Villella (2002)(81) Retrospective analysis Gore (2000)(82) Review not SR Sariano (2000)(83) Phase II study, not RCT Glimelius (2001)(84) Background on treatment pathway in ovarian cancer Hogberg (2001)(85) Review, not systematic Karlsson (2001)(86) Included as relavant related SR - pass onto CHE Vermoken (2002)(87) Background discussion on treatment in ovarian cancer Mobus (2000)(88) Retrospective survival analysis of patients treated with topotecan Fracasso (2003)(89) Phase II trial on oxaliplatin Gronlund (2002)(90) Info on prognostic factors Anonymous (2002)(91) Disscusion article Cannistra (2002)(92) Background article on CAQP versus paclitaxel trial du Bois (2001)(93) Dose finding study of carboplatin gemcitabine combination Chan (2001)(94) longitudinal study on QoL after treatment Slimane (2003)(95) Commentary on first-line treatment trials Anonymous (2002)(96) Commentary on GOG111 Bernardi (2003)(97) Phase II study, not RCT Baker (2002)(98) Abstract on the results of GOG/SWOG Conte (2002)(99) Discussion on chemotherapy in second line disease Muggia (2001)(100) Background: commentary on PLDH versus topotecan trial. Gonzalez-Martin (2003)(101) Background: single versus combination therapy Ross (2001)(46) Background: review article on treatment options for second-line therapy Zeimet (2002)(102) Commentary paper in German Covens (2000)(40) Relevant SR on topotecan Johnston (2001)(103) background Mayer (2001)(104) Not an RCT; phase II study Markman (2001)(105) General background on second-line therapy Costa (2001)(106) First-line therapy Sun (2002)(107) Background: Pass onto CHE Health Technology Board for Scotland Same as the NICE Guidance on the use of PLDH (2002)(108) (Cae;yx) for the treatment of advanced ovarian cancer. Harper (2002)(109) Review Le (2003)(110) review of QoL issues Kaye (2002)(111) Review Bookman (2002)(112) Info on considerations in trial design Anon (2001)(113) Not data Ohara (2002)(114) Case report Lehoczky (2002)(115) Case report Topuz (2001)(116) Not an RCT; phase II study. Markman (2003)(117) Study on the impac of the sequence of delivery on

265 neutropenia Anonymous (2001)(118) Background references for economics Anonymous (2001)(119) Background on oxaliplatin Bilgin (2002)(120) Treatment with second-line temcitabline therapy Wojciechowska-Lacka (2002)(121) Study on disease progression after 1st-line paclitaxe and cis-or carboplatin treatment Castellano (2002)(122) Phase II non-RCT Alvarez (2003)(123) Background on second-line treatment options Armstrong (2004)(124) Commentary on use of topotecan dosing Armstrong (2002)(125) Commentary on the use of topotecan Lakusta (2001)(126) Cohort study on QoL measures Hahn (2003)(127) Study on QoL in patients treated with doxil as second-line therapy Patel (2001)(128) Background on the use of PLDH. Gold (2002)(129) Topotecan combined with amifostine as sedond-line therapy Hensley (2001)(130) Background on AE of PLDH. Stebbing (2002)(131) Commentary on trial 30-49 Holzner (2002)(132) Background on QoL Toffoli (2002)(133) Correspondence article Armstrong (2003)(134) Bryostatin1 as the intervention Chiara (2003)(135) Non RCTphase II study Latorre (2002)(136) Background Markman (2000)(137) Background on choices for second-line therapy Yasuda (2002)(138) Phase II study not an RCT Gorbunova (2002)(139) Paper in Polish Piccart (2002)(140) Long-term follow-up results of GOG111 Pignata (2001)(141) Background on QoL issues in gynaecological cancers Goldwasser (2001)(142) Commentary on oxaliplatin in cancers Boruta (2002)(143) Retrospective study; not an RCT Ozols (2002)(144) Commentary on the management of ovarian cancer Mielcarek (2002)(145) QoL of patients treated with paclitaxel as first-line therapy Breidenbach (2003)(146) Discussion article on side-effect profiles of different chemotherapy regimens Woronoff (2001)(147) Background: economics Favalli (2002)(148) Protocol for the After-6 study. Beshara (2002)(149) Retrospective analysis of the use of topotecan as a second-line therapy Markman (2003)(150) Background on disease stage measurement in ovarian cancer Markman (2003)(151) First-line therapy Bolis (2004)(152) First-line therapy Slater (2001)(153) Commentary on non-curative chemotherapy Michaud (2000)(154) Background on paclitaxel Baron-Hay (2001)(155) Commentary on the use of oxiliplatin Doyle (2001)(156) Background: economics Wagner (2001)(157) Bbackground on the use of paclitaxel-cisplatin Le (2003)(158) Background: pass onto CHE Boos (2003)(159) Discussion on off-label use of treatments Fireman (2000)(160) Cost of care for patients in cancer trials (CHE?)

266 Lenderman (2003)(161) Correction on letter to the editor Bodurka-Bevers (2000)(162) Background: economics (CHE) Bolis (2001)(75) Intervention is epidoxorubicin Chan (2003((163) Background: QoL during chemotherpay Donato (2003)(164) Background: editorial Flynn (2002((165) Background on optimum time interval between surgery and chemotherapy Gadducci (2001)(166) Background Anonymous (2001)(167) CHE Neymark (2002)(168) Background: economics (CHE) Gelderblorn (2000)(169) Phase I study Szucs (2003)(170) Background: cost-effectiveness (CHE) Van Den Bosch (2001)(171) Not an RCT and is not second-line therapy. ICON1 Collaborators (2003)(172) Background: ICON 1 Piccart (2000)(173) Commentary Chan (2003)(174) Commentary on the NICE guidance on PLDH Anonymous (2003)(175) Commentary on the NICE£ guidance on paclitaxel Anonymous (2002)(176) Commentary on NICE guidance for PLDH Anonymous (2002)(177) Commentary on single agent carboplatin as first-line treatment choice. Young (2001)(178) Background: economics (CHE) Copeland (2003)(179) Protocol and rational for ICON 5 (USEFUL BACKGROUND) Wenzel (2004)(180) Abstract only: not clear that intervention is after failure of 1st line platinum based therapy. Scarfone (2001)(181) Abstract only: not clear that this is second-line therapy after failure of 1st line platinum based therapy Gennatas (2000)(182) Abstract only: not clear that this is second-line after failure of 1st line platinum-based therapy Muggia (2000)(183) 1st line therapy Piccart (2000)(184) First-line therapy Einhorn (2003)(185) Background on radiation therapy Nicholls (2001)(186) Background: economics (CHE) de Haes (1987)(187) Background: QoL Bolis (1999)(188) Excluded: pre-search up-date date of 2000 Dranitsaris (2004)(189) Background: economics (CHE) ICON Group (2002)(190) Background: ICON 3 (first-line therapy) Fiorica (2003)(191) Background: use of topotecan for advanced treatment (USEFUL SPECIFIC BACKGROUND) Mutch (2003)(192) Backgroudn: gemitabine combination chemotherapy (general) Stuart (2003)(193) Background: first-line treatment regimens (general) Fabbro (2001)(194) Non-randomised phase II study Bankhead (2004)(195) Useful general background article Wang (2003)(196) Commentary on ICON 4 (not useful as background) ICON 1 & EOTTC – ACTION (2003) platinum-based first line therapy (197) Calhoun (2001)(198) Background: economics (CHE) Moe (2002)(199) Background: economics (CHE) Wu (2001)(200) Reports on first-line therapy Calhoun (2003)(201) Impact of chemotherapy delays on QoL in patients

267 undergoing first-line chemotherapy (not specific to ovarian cancer) Biamonte (2000)(202) Topotecan verus Nihil as maintainance therapy after first-line treatment. Moss (2002)(203) Background: management of advanced ovarian caner patients (general) Adams (2004)(204) Article on use of erythropoietin in USA Guastalla (2003)(205) Background on adverse events associated eith paclitaxel Katsumata (2003)(206) Background on docetaxel Sufliarsky (2003)(207) Phase II study of gemcitabine + carboplatin Maenpaa (2003)(208) Backgorund article on docetaxel Gronlund (2001)(209) Retrospective study of results of patients receiving paclitaxel/carboplatin first-line therapy Calhoun (2003)(210) Psychometric properties of scale evaluating neuropathy Piccart (2003)(211) Does not have PLDH, topotecan or paclitaxel as an intervention. Bookman (2003)(212) Background to GOG158 and ICON 5 Malik (2000)(213) commentary on one single institutions experience of participating in topotecan trials Vermorken (2001)(214) Background on drug actions of oxaliplatin and paclitaxel Bookman (1998)(215) Not an RCT; pre up-dated searches Kavanagh (2002)(216) Not an RCT Ojeda (2003)(51) Background: economics (CHE) Jasas (2003)(217) Phase II study of second-line use of oral topotecan + cisplain (i.v) Sehouli (2001)(218) Abstract only: phase IIstudy (not randomised) Lalisang (2003)(219) First-line therapy with decetaxel, epirubicin and cisplatin Wheatley (2003)(220) Letter to the editor: useful commentary on ICON 4 Mielke (2003)(221) Study of different infusion times for paclitaxel in patients with different types of cancer Green (2003)(222) Reply to letter to the editor: useful background on ICON 4 Anonymous (2003)(223) Commentary on the results of ICON 4 Vorobiof (2003)(224) Phase II study of caelyx + carboplatin (not randomised) Newman (2003)(225) Letter to the editor: useful commentary on ICON4

268 Appendix 5. Details of quality assessment A. Clinical effectiveness studies were assessed using the following criteria based on CRD Report 4

1. Was the method used to assign participants to the treatment groups really random? (Computer generated random numbers and random number tables were accepted as adequate, while inadequate approaches will include the use of alternation, case record numbers, birth dates and days of the week) 2. Was the allocation of treatment concealed? (Concealment was deemed adequate where randomisation is centralised or pharmacy-controlled, or where the following are used: serially-numbered identical containers, on-site computer based systems where the randomisation sequence is unreadable until after allocation, other approaches with robust methods to prevent foreknowledge of the allocation sequence to clinicians and patients. Inadequate approaches will include: the use of alternation, case record numbers, days of the week, open random number lists and serially numbered envelopes even if opaque). 3. Was the number of participants who were randomised stated? 4. Were details of baseline comparability presented in terms of myocardial infarction, stroke, heart failure, hypertension, diabetes and current or former smoker? 5. Was baseline comparability achieved in terms of myocardial infarction, stroke, heart failure, hypertension, diabetes and current or former smoker? 6. Were the eligibility criteria for study entry specified? 7. Were any co-interventions identified that may influence the outcomes for each group? 8. Were the outcome assessors blinded to the treatment allocation? 9. Were the individuals who administered the intervention blinded to the treatment allocation? 10. Were the participants who received the intervention blinded to the treatment allocation? 11. Was the success of the blinding procedure assessed? 12. Were at least 80% of the participants originally included in the randomisation process followed up in the final analysis? 13. Were the reasons for withdrawals stated? 14. Was an intention to treat analysis included?

Items were graded in terms of yes (item properly addressed), no (item not properly addressed), partially addressed (item partially addressed), unclear or not enough information, or NA not applicable

269 B. Studies of cost effectiveness were assessed using the following criteria, which is an updated version of the checklist developed by Drummond

Study question 1. Costs and effects examined 2. Alternatives compared 3. The viewpoint(s)/perspective of the analysis is clearly stated (e.g. NHS, society) Selection of alternatives 4. All relevant alternatives are compared (including do-nothing if applicable) 5. The alternatives being compared are clearly described (who did what, to whom, where and how often) 6. The rationale for choosing the alternative programmes or interventions compared is stated

Form of evaluation 7. The choice of form of economic evaluation is justified in relation to the questions addressed. 8. If a cost-minimisation design is chosen, have equivalent outcomes been adequately demonstrated?

Effectiveness data 9. The source(s) of effectiveness estimates used are stated (e.g. single study, selection of studies, systematic review, expert opinion) 10. Effectiveness data from RCT or review of RCTs 11. Potential biases identified (especially if data not from RCTs) 12. Details of the method of synthesis or meta-analysis of estimates are given (if based on an overview of a number of effectiveness studies)

Costs 13. All the important and relevant resource use included 14. All the important and relevant resource use measured accurately (with methodology) 15. Appropriate unit costs estimated (with methodology) 16. Unit costs reported separately from resource use data 17. Productivity costs treated separately from other costs 18. The year and country to which unit costs apply is stated with appropriate adjustments for inflation and/or currency conversion.

Benefit measurement and valuation 19. The primary outcome measure(s) for the economic evaluation are clearly stated (cases detected, life years, QALYs, etc.) 20. Methods to value health states and other benefits are stated (e.g. time trade off) 21. Details of the individuals from whom valuations were obtained are given (patients, members of the public, health care professionals etc.)

Decision modelling 22. Details of any decision model used are given (e.g. decision tree, Markov model) 23. The choice of model used and the key input parameters on which it is based are adequately detailed and justified 24. All model outputs described adequately.

270 Discounting 25. Discount rate used for both costs and benefits 26. Do discount rates accord with NHS guidance (1.5%-2% for benefits; 6% for costs)?

Allowance for uncertainty Stochastic analysis of patient-level data 27. Details of statistical tests and confidence intervals are given for stochastic data 28. Uncertainty around cost-effectiveness expressed (e.g. confidence interval around incremental cost-effectiveness ratio (ICER), cost-effectiveness acceptability curves). 29. Sensitivity analysis used to assess uncertainty in non-stochastic variables (e.g. unit costs, discount rates) and analytic decisions (e.g. methods to handle missing data).

Stochastic analysis of decision models 30. Are all appropriate input parameters included with uncertainty? 31. Is second-order uncertainty (uncertainty in means) included rather than first order (uncertainty between patients)? 32. Are the probability distributions adequately detailed and appropriate? 33. Sensitivity analysis used to assess uncertainty in non-stochastic variables (e.g. unit costs, discount rates) and analytic decisions (e.g. methods to handle missing data).

Deterministic analysis 34. The approach to sensitivity analysis is given (e.g. univariate, threshold analysis etc) 35. The choice of variables for sensitivity analysis is justified 36. The ranges over which the variables are varied are stated

Presentation of results 37. Incremental analysis is reported using appropriate decision rules 38. Major outcomes are presented in a disaggregated as well as aggregated form 39. Applicable to the NHS setting

All items were graded as either yes (item adequately addressed), r no (item not adequately addressed), sunclear or not enough information, NA not applicable or NS not stated.

271 Appendix 6. Data extraction tables

Trial: 30-49(22)

Study Details and Design Participant Details Intervention Details Follow-up/withdrawals and Adverse Conclusion and Comments Events Author: Gordon 2001(22) Number randomised Intervention (I): Withdrawals from intervention (I) Authors’ conclusions 481randomised; 474 in ITT Type: - Pegylated liposomal Overall seven participants were lost to The comparable efficacy, favourable Aim doxorubicin hydrochloride (PDL) follow-up prior to the first dose. The safety profile and convenient dosing To compare the efficacy and safety of Disease type: Epithelial No. randomised:- Not reported. 239 study group assignment and reason for support the role of PLD as a valuable pegylated liposomal doxorubicin Disease stage: Advanced were treated. these withdrawals is not stated. No treatment option in this patient (PLDH) and topotecan in patients with Occurrence of secondary spread:- Route of administration : i.v. other reasons for lost to follow-up are population. epithelial ovarian carcinoma that Yes, in the majority of cases, but not all Dose:- 50 mg/m2 as a 1hr infusion reported. recurred after or didn't respond to No. of cycles - Not stated (protocol up Comments platinum-based, first-line, chemo- Therapy stage: Second-line to 6) Withdrawals from control (C) This was designed as an equivalence therapy. Length per cycle: - 28 days Overall seven participants were lost to study with 80% power to detect Previous treatments follow-up prior to the first dose. The statistical equivalence of the two drugs. Trial ID: 30-49 Chemotherapy (platinum-based first-line Control (C) study group assignment and reason for Phase: Phase III regimen); radiotherapy (but only if Type: Topotecan these withdrawals is not stated. No A sample size calculation was involves less than one third of No. randomised: Not reported. 235 other reasons for lost to follow-up are performed prior to recruitment which Method of randomisation haemopoietic sites). were treated. reported. suggested a total of 360 participants, Centrally randomised and stratified for Route of administration: i.v. 175 in each treatment group, were platinum sensitivity and bulky disease* Disease present after first-line Dose: - 1.5 mg/m2/d as a 30min infusion Adverse events required to ensure with a probability of treatment? for five days 80% than the lower 95% one-sided Length of follow-up Residual : Yes No. of cycles:- Not stated (protocol up PLDH: 43 participants in the group confidence interval of the hazard ratio of Up to one year; or less if disease Refractory - Yes to 8) withdrew from the study due to adverse topotecan to PLD could not fall below progression occurred Length per cycle:- 21 days events, but there were no treatment- 0.757. However, the sample size was Age/age range of participants related deaths in the group. increased to 460 to cover anticipated PLDH: Median 60.0yrs (range 27-87yrs) losses to follow-up. Top: Median 60.0 yrs (range25-85yrs) Comments about PLDH: (n=239) intervention/control: † All grades Karnofsky performance status data is Characteristics In the absence of disease progression, Body as a whole:- not provided in terms of the individual PLDH (I) : treatment with both agents was to be Asthenia: 96 (40.2%) study groups. Mean (SD) 59.2 (11.3) continued for up to 1 year. Treatment Abdominal pain: 80 (33.5%) Median age: 60.0yrs (range: 27-87yrs) could also continue if the patient Fever: 51 (21.3%) The analysis is not an ITT analysis as < 65: 156 (65.3% demonstrated sustained clinical benefit. Pain: 50 (20.9%) reported by the trialists. 481 participants >=65: 83 (34.7%) Patients who discontinued treatment Mucous membrane disorder: 34 (14.2%) were originally randomised but the after 6 months (six cycles of PLDH, Back pain: 28 (11.7%) analysis performed was based on only ± Race (n=239):- eight cycles of topotecan) were Infection: 28 (11.7%) 474 participants. White: 217/239 (90.8%) considered protocol completed. Headache: 25 (10.5%) Black: 14/239 (5.9%) Response to study drug was based on Digestive system:- Hispanic: 2/239 (0.8%) objective tumor assessments. A Nausea: 110 (46.0%) Asian: 3/239 (1.3%) complete response was defined as Stomatitis: 99 (41.4%)

272 Other: 3/239 (1.3%) complete disappearance of all Vomiting: 78 (32.6%) measurable and assessable disease, no Constipation: 72 (30.1%) ± CA-125 at baseline (U/mL) (n=224) new lesions, and no disease-related Diarrhoea: 50 (20.9%) Mean (SD): 900.26 (1,933.342) symptoms. A partial response (PR) was Anorexia: 48 (20.1%) Median: 198.50 documented in patients with >=50% Dyspepia: 29 (12.1%) Range: 3.0 – 1,8801.0 decrease in the sum of the products of Intestinal obstruction: 27 (11.3%) bidimensional perpendicular diameters Hemic & lymphatic system:- Initial FIGO stage of all measurable lesions; progession of Anemia: 96 (40.2%) I 11/239 (5%) assessable disease and new lesions were Leukopenia: 88 (36.8%) II 13/239 (5%) not allowed. Progressive disease was Neutropenia: 84 (35.1%) II 175/239 (73%) classified in patients with a >=50% Thrombocytopenia: 31 (13.0%) IV 40/239 (17%) increase in the sum of products of Metabolic/nutritional disorder:- bidensionally measured lesions over the Peripheral oedema: 27 (11.3%) ± Pre-study chemotherapy regimens: smallest sum obtained at the best Nervous system: Combination platinum & taxane: response, or reappearance of any lesion Paresthesia: 24 (10.0%) 176/239 (73.6%) that had disappeared, or clear worsening Dizziness: 10 (4.2%) Combination platinum & non-taxane: of any assessable disease, or failure to Respiratory system:- 6/239 (2.5%) return for evaluation because of death or Pharyngitis: 38 (15.9%) Platinum alone: 57/239 (23.9%) deteriorating condition, or the Dyspnea: 36 (15.1%) Type of prior platinum therapy:- appearance of any new lesion or site. Cough increased: 23 (9.6%) Carboplatin only: 145/239 (60.7%) Patients were clssified as having stable Skin & appendages:- Cisplatin only: 68/239 (28.5%) disease if they did not qualify for CR, Hand-foot syndrome: 121 (50.6%) Both: 26/239 (10.9%) PR or progressive disease. Rash: 68 (28.5%) Prior anthracycline therapy:- Alopecia: 46 (19.2%) Yes: 11/239 (4.6%) Actual study treatment No: 228/239 (95.4%) administration: Grade 3 PLDH (n=239):- Body as a whole:- Drug-free interval : Total number of dosing cycles: 1,164 Asthenia: 17 (7.1%) Mean (SD): 10.2 (11.98) Cumulative dose, mg/m2:- Abdominal pain: 24 (10.0%) Median: 7.0mths (range: 0.9-82.1mths) median: 200 Fever: 2 (0.8%) range: 47 - 1,301 Pain: 4 (1.7%) Median CA-125:199U/ml (range: 3.0- Mean cycle dose, mg/m2:- Mucous membrane disorder: 9 (3.8%) 18801.0U/ml) median: 50 Back pain: 4 (1.7%) range: 34 - 58 Infection: 5 (2.1%) Sum of lesions at baseline (cm2):- Mean cycle length: Headache: 2 (0.8%) Mean (SD): 38.7 (54) median: 30 Digestive system:- Median: 20cm2 (range: 1-441cm2) range: 27- 56 (n=212) Nausea: 12 (5.0%) Stomatitis: 9 (7.9%) Platininum sensitivity ± Most patients received 4-5 cycles of Vomiting: 17 (7.1%) Sensitive 109/239 (46%) study drug (estimated by dividing Constipation: 6 (2.5%) Refractory 130/239 (54%) cumulative dose by the cycle dose) Diarrhoea: 5 (2.1%) Anorexia: 6 (2.5%) Presence of bulky disease Topotecan (n=235):- Dyspepia: 2 (0.8%) Present 111/239 (46%) Total number of dosing cycles: 1,249 Intestinal obstruction: 19 (7.9%) Absent 128/239 (54%) Cumulative dose, mg/m2:- Hemic & lymphatic system:-

273 median: 36 Anemia: 13 (5.4%) ± Platinum sensitivity/bulky disease:- range: 3-165 Leukopenia: 21 (8.8%) Refractory/present: 64 (26.8%) Mean cycle dose, mg/m2:- Neutropenia:19 (7.9%) Refractory/absent: 66 (27.6%) median: 7 Thrombocytopenia: 3 (1.3%) Sensitive/present: 44 (18.4%) range: 3-10 Metabolic/nutritional disorder:- Sensitive/absent: 65 (27.2%) Mean cycle length, days:- Peripheral oedema: 5 (2.1%) median: 24 Nervous system: ± KPS at baseline:- range: 20-38 (n=211) Paresthesia: 0 50: 1 (0.4%) Dizziness: 0 60: 7 (2.9%) ± Most patients received 4-5 cycles of Respiratory system:- 65: 1 (0.4%) study drug (estimated by dividing Pharyngitis: 0 70: 30 (12.6%) cumulative dose by the cycle dose) Dyspnea: 8 (3.3%) 75: 0 Cough increased: 0 80: 53 (22.2%) Protocol deviations: Skin & appendages:- 85: 0 Protocol deviations included (1) failure Hand-foot syndrome: 55 (23.0%) 90: 96 (40.2%) to meet entry criteria (7 receiving Rash: 10 (4.2%) 100: 50 (20.9%) PLDH, two receiving topotecan; (2) Alopecia: 3 (1.3%) Total <80: 39 (16.3%) patients who continued in study after Total >=: 80: 199 (83.3%) first clinically significant change in Grade 4 Not available: 1 (0.4%) LVEF (13 receiving PLDH); (3) patients Body as a whole:- who continued treatment after Asthenia: 0 ± Histologic tumour type: documented disease progression (40 Abdominal pain: 1 (0.4%) Serous papillary: 145 (60.7%) receiving PLDH, 42 receiving Fever: 0 Mucinous: 6 (2.5%) topotecan); (4) patients who completed Pain: 1 (0.4%) Endometrioid: 11 (4.6%) fewer than eight cycles of treatment but Mucous membrane disorder: 0 Clear cell: 9 (3.8%) were deemed protocol-completed by the Back pain: 0 Transitional cell: 1 (0.4%) investigator (20 receiving topotecan). Infection: 0 Undiffernentiated: 55 (23.0%) Headache: 0 Unspecified adenocarcinoma: 11 (4.6%) Digestive system:- Not specified: 1 (0.4%) Nausea: 1 (0.4%) Histologic tumor grade:- Stomatitis: 1 (0.4%) Well differentiated: 4 (1.7%) Vomiting: 2 (0.8%) Moderately differentiated: 16 (6.7%) Constipation: 0 Poorly differentiated: 53 (22.2%) Diarrhoea: 1 (0.4%) Unspecified differentiated: 125 (52.3%) Anorexia: 0 Not specified: 41 (17.2%) Dyspepia: 0 Intestinal obstruction: 4 (1.7%) ± Ascites at baseline: Hemic & lymphatic system:- Absent: 162 (67.8%) Anemia: 1 (0.4%) Present: 77 (32.2%) Leukopenia: 3 (1.3%) Not available: 0 Neutropenia: 10 (4.2%) Thrombocytopenia: 0 Topotecan (C): Metabolic/nutritional disorder:- Median age: 60.0yrs (range: 25-85yrs) Peripheral oedema: 0 <65: 138 (58.7%) Nervous system:

274 >=: 97 (41.3% Paresthesia: 0 Dizziness:0 ± Race (n=235) Respiratory system:- White: 226/235 (96.2%) Pharyngitis: 0 Black:7/235 (3.0%) Dyspnea: 2 (0.8%) Hispanic: 2/235 (0.9%) Cough increased: 0 Asian: 0 Skin & appendages:- Other: 0 Hand-foot syndrome: 2 (0.8%) Rash: 0 Initial FIGO stage Alopecia: 0 I 15/235 (6%) II 8/235 (3%) Topotecan (n=235) II 164/235 (70%) Topotecan:- 37 participants withdrew IV 48/235 (20%) due to treatment related adverse events and three participants died (all from ± Prestudy chemotherapy regimens complications of neutropenia and (n=235) sepsis). Combination platinum & taxane: 170/235 (72.3%) † All grades Combination platinum & non-taxane: Body as a whole:- 13/235 (5.5%) Asthenia: 121 (51.5%) Platinum alone: 52/235 (22.1%) Abdominal pain: 89 (37.9%) Type of prior platinum therapy:- Fever: 72 (30.6%) Carboplatin alone: 145/235 (61.7%) Pain: 40 (17.0%) Cisplatin alone: 61/235 (26.0%) Mucous membrane disorder: 8 (3.4%) Both: 29/235 (12.3%) Back pain: 24 (10.2%) Infection: 15 (6.4%) Prior anthracycline therapy:- Headache: 35 (14.9%) Yes: 19/235 (8.1%) Digestive system:- No: 216/235 (91.9%) Nausea: 148 (63.0%) Stomatitis: 36 (15.3%) Drug free interval:_ Vomiting: 103 (43.8%) Mean (SD): 10.1 (13.81) Constipation: 107 (45.5%) Median: 6.7mths (range: 0.5-09.6mths) Diarrhoea: 82 (34.9%) Anorexia: 51 (21.7%) ± CA-125 at baseline (U/mL) Dyspepia: 33 (14.0%) Mean (SD): 931.80 (2,454.587) Intestinal obstruction: 26 (11.1%) Median: 178.00 (range: 3.0- Hemic & lymphatic system:- 29330.0U/ml) Anemia: 177 (75.3%) Leukopenia: 151 (64.3%) , Sum of lesions at baseline:- Neutropenia: 193 (82.1%) Mean (SD) : 33.5 (38.67) Thrombocytopenia: 153 (65.1%) Median 20cm2 (range: 1-296cm2) Metabolic/nutritional disorder:- Peripheral oedema: 41 (17.4%) Platininum sensitivity Nervous system:

275 Sensitive 111/235 (47%) Paresthesia: 21 (8.9%) Refractory 124/235 (53%) Dizziness: 24 (10.2%) Respiratory system:- Presence of bulky disease Pharyngitis: 42 (17.9%) Present 111/235 (47%) Dyspnea: 55 (23.4%) Absent 124/235 (53%) Cough increased: 27 (11.5%) Skin & appendages:- ± Platinum sensitivity/bulky disease:- Hand-foot syndrome: 2 (0.9%) Refractory/present: 60 (25.5%) Rash: 29 (12.3%) Refracory/absent: 65 (27.7%) Alopecia: 123(52.3%) Sensitive/Present: 45 (19.1%) Sensitive/Absent: 65 (27.7%) Grade 3

± KPS at baseline: Body as a whole:- 50: 0 Asthenia: 19 (8.1%) 60: 7 (3.0%) Abdominal pain: 19 (8.1%) 65: 0 Fever: 8 (3.4%) 70: 29 (12.3%) Pain: 4 (1.7%) 75: 1 (0.4%) Mucous membrane disorder: 0 80: 56 (23.8%) Back pain: 2 (0.9%) 85: 1 (0.4%) Infection: 2 (0.9%) 90: 85 (36.2%) Headache: 0 100: 53 (22.6%) Digestive system:- Total ,80: 37 (15.7%) Nausea: 16 (6.8%) Total >=: 195 (83.0%) Stomatitis: 1 (0.4%) Not available: 3 (1.3%) Vomiting: 18 (7.7%) Constipation:11 (4.7%) ± Histologic tumor type:- Diarrhoea: 9 (3.8%) Serous papillary: 130 (58.3%) Anorexia: 3 (1.3%) Mucinous: 9 (3.8%) Dyspepia:0 Endometrioid: 15 (6.4%) Intestinal obstruction: 14 (6.0%) Clear cell: 10 (4.3%) Hemic & lymphatic system:- Transitional cell: 0 Anemia: 59 (25.1%) Undifferentiated: 53 (22.6%) Leukopenia: 83 (35.3%) Unspecified adenocarcinoma: 18 (7.7%) Neutropenia: 33 (14.0%) Not specified: 0 Thrombocytopenia: 40 (17.0%) Metabolic/nutritional disorder:- ± Histologic tumour grade: Peripheral oedema: 6 (2.6%) Well differentiated: 3 (1.3%) Nervous system: Moderately differentiated: 13 (5.5%) Paresthesia: 0 Poorly differentiated: 72 (30.6%) Dizziness: 0 Unspecified differentiated: 110 (46.8%) Respiratory system:- Not specified: 37 (15.7%) Pharyngitis: 1 (0.4%) Dyspnea: 7 (3.0%) ± Ascites at baseline:- Cough increased: 0

276 Absent: 168 (71.5%) Skin & appendages:- Present: 65 (27.7%) Hand-foot syndrome: 0 Not available: 2 (0.9%) Rash: 1 (0.4%) Alopecia: 15 (6.4%) Inclusion/exclusion criteria Women <= 18 years of age with Grade 4 measurable, or measureable and Body as a whole:- assessable diease that had recurred or Asthenia: 0 failed first-line, platinum-based Abdominal pain: 4 (1.7%) chemotherapy. Measurable disease was Fever: 5 (2.1%) defined as bidimensionally measurable Pain: 0 lesions(s) with clearly defined margings Mucous membrane disorder: 0 by plain x-ray with at least one diameter Back pain: 0 >= 0.5 cm (excluding bone lesions) or Infection: 0 by computed tomography, magnetic Headache: 0 resonance imaging, or another imaging Digestive system:- scan with both diameters greater than Nausea: 3 (1.3%) the distance between cuts of the imaging Stomatitis: 0 study or palpation with both diameters Vomiting: 5 (2.1%) >= 2cm. Assessable disease included Constipation: 2 (0.9%) unidimensionally measurable lesions(s), Diarrhoea: 1 (0.4%) mass (es) with margins not clearly Anorexia: 0 defined, lesion(s) with both diameters Dyspepia: 0 <=0.5cm, lesion(s) on scan with either Intestinal obstruction: 7 (3.0%) diameters <=2 cm, and malignant ascites Hemic & lymphatic system:- or pleural effusion in conjunction with Anemia: 10 (4.3%) serum CA-125 levles more than 100 Leukopenia: 36 (15.3%) U/mL in the absence of cirrhosis. Neutropenia: 146 (62.1% Additional criteria included adequate Thrombocytopenia: 40 (17.0% bone marrow function (platelets Metabolic/nutritional disorder:- >=100,000/mm3, hemoglobin >= 9 Peripheral oedema: 0 g/dL, absolute neutrophil count >= Nervous system: 1,500 cells/mm3) renal function (serum Paresthesia: 0 creatinine <=2.5 mg/dL), liver function Dizziness: 0 (AST<= two times the upper limit of Respiratory system:- normal, alkaline phosphate <= two times Pharyngitis: 0 the upper limits of normal, bilirubin <= Dyspnea: 3 (1.3%) upper limit of normal), cardiac function Cough increased: 0 (leftventricular ejection fraction (LVEF) Skin & appendages:- >= 50% or the instiutional normal), Hand-foot syndrome: 0 Karnofsky performance status >= 60%, Rash: 0 and a disease-free period of more than 5 Alopecia: 0 years from prior malignancies (except curatively treated basal cell carcinoma, Dosing adjustments:- squamous cell carcinoma of the skin, or

277 carcinoma in situ of the cervix). PLDH: Patients were excluded if they were Total dosing adjustments:137 (57.3 %) pregnant or breast0-feeding, expected to 25.2% doses were reduced due to PPE live <= 3 months, had received prior Delayed doses: 124 (51.9%) radiation therapy to greater than one Interrupted doses: 23 (9.6%) third of hematopoietic sites, had a Reduced doses: 65 (27.2%) history of cardiac disease that net the ± Reason for delay, interruption or New York State Heart Association reduction:- Classification class 2 or greater, had an Hand-foot syndrome: 105/415 (25.3% uncontrolled systemic infection, had Stomatitis: 105/415 (25.3% received an investigational agent within Hematological toxicity: 46/415 (11.1%) 30 days of the first dose of study drug, Other laboratory toxicity: 3/415 (0.7%) had received prior pegylated liposomal Infusion reaction: 17/415 (4.1%) doxorubicin or topotecan therapy, had Intercurrent illness: 20/415 (4.8%) received chemotherapy within 29 days Scheduling problem: 36/415 (8.7%) of the first dose of study drug (or within Other: 204/415 (49.2%) 42 days if patient had received a nitosourea or mitomycin). Additionally Granulocyte/granulocyte-macrophage no concurrent investigational or colony-stimulating factor: 11 (4.6%) antineoplastic agents were permitted during the study. Topotecan:- Total dosing adjustments: 184 (78.3%) Comments about participants: 43.3% doses reduced due to Patients were prospectively stratified haematological toxicity. based on platinum sensitivity and the Delayed doses: 151 (64.3%) presence or absence of bulky disease. Interrupted doses: 34 (14.5%) Patients were categorised as platinum- Reduced doses: 122 (51.9%) sensitive if they had a progression-free ± Reason for delay, interruption or survival (PFS) interval of greater than 6 reduction:- months after first-line, platinum-based Hand-foot syndrome: 0 chemotherapy. Patients were considered Stomatitis: 1/684 (0.1%) platinum-refractory if they progressed Hematologic toxicity: 295/235 (43.1%) during initial platinum-based Other laboratory toxicity: 1/684 (0.1%) chemotherapy, demonstrated stable Infusion reaction: 3/684 (0.4%) disease, or relapsed within 6 months Intercurrent illness: 29/684 (4.2%) after completing platinum-based Scheduling problem: 72/684 (10.5%) chemotherapy. Bulky disease was Other: 397/684 (58.0%) defined as the presence of a tumor mass larger than 5cm. Granulocyte/granulocyte-macrophage colony-stimulating factor: 68 (29.1%)

Discontinuations:

± 48% of PLD group and 47% of topotecan discontinued the study due to

278 disease progression.

± Reasons for discontinuation PLD (n=239) Disease progression: 114/239 (48%) Adverse event: 43/239 (18%) Death: 15/239 (6%) Non-compliance: 1/239 (,1%) In appropriate enrolment: 0 Other/unknown: 31/239 (13%) Protocol completed: 34/239 Ongoing: 5/239 (2%)

± Topotcan (n=235) Disease progression: 110/235 (47%) Adverse events: 37/245 (16%) Death: 18/235 (8%) Non-compliance: 1/235 (<1%) In appropriate enrolment: 1/235 (<1%) Other/unknown: 35/235 (15%) Protocol completed: 39/235 (17%) Ongoing: 2/235 (1%) Results Outcome 1 Outcome 2 Outcome 3 Outcome: Progression free survival (time form the start of study Outcome: Response Outcome: Quality of life drug dosing to the documented disease progression or death due to any cause) PLD group (I) (n = 239):- Baseline data: for both the PLD and topotecan groups: – Not Overall response: 47/239 (19.7%) (p=.390) reported; questionnaire only completed by 82% of participants. PLD group(I) :- Complete response: 9/239 (3.8%) However, function and symptom scale scores were similar between All participants (n=239) Partial response: 38/239 (15.9%) the two treatment groups. Median progression free survival = 16.1 weeks (p=.095) Stable disease: 77/239 (32.2%) Median survival = 60 weeks (p=.341) Follow-up data: For both the PLD and topotecan groups: No ± Time to progression: HR = 1.176 (90% CI: 1.002, 1.381) p=0.095 Platinum-sensitive participants (n=109) significant differences between the intervention and control groups at Overall response: 31/109 (28.4%) (p=.964) 12 weeks follow-up. Platinum-sensitive participants (n=109) Complete response: 8/109 (7.3%) Median progression free survival = 28.9 weeks (p=0.37) Partial response: 23/109 (21.1%) Comments:- Median survival = 108 weeks Stable disease: 41/109 (37.6%) QoL (assessed by European Organization for Research and Treatment ± Progression free survival: HR=1.72 (95% CI: 1.145, 2.585) of Cancer Quality of Life Questionnaire (QIQ-C30). 82% of Test for interaction between platinum sensitivity/resistance and Platinum-resistant participants (n=130) participants completed QoL measure at baseline, at 12 weeks less survival: ∆=0.256 (95% CI: 0.151, 1.155) p=0.011 Overall response: 16/130 (12.3%) (n=130) than 50% of participants in either treatment arm were still in the Complete response: 1/130 (0.8%) HRQoL study. Platinum-resistant participants (n=130) Partial response: 15/130 (11.5%) Median progression free survival = 9.1 weeks (p=733) Stable disease: 36/130 (27.7%) Median survival = 35.6 weeks (p=.455)

279 Topotecan group (C) (n=235):- Topotecan group (C):- Overall response: 40/235 (17%) All participants (n=235) Complete response: 11/235 (4.7%) Median progression free survival = 17.0 weeks (p = .095) Partial response: 29/235 (12.3%) Median survival = 56.7 weeks (p = .341) Stable disease: 95/235 (40.4%) ± Time to progression: HR = 1.176 (90% CI: 1.002, 1.381) p=0.095 Platinum-sensitive participants (n=111): Platinum-sensitive participants (n=111) Overall response: 32/111 (28.8%) (p=.964) Median progression free survival = 23.3 weeks (p = .37) Complete response: 10/111 (9.0%) Median survival= 71.1 weeks (p=.008) Partial response: 22/111 (19.8%) Stable disease: 42/111 (37.8%) Platinum-resistant participants (n=124) Median progression free survival = 13.6 weeks (p = .733) Platinum-resistant participants (n=124) Median survival = 41.3 weeks (p = .455) Overall response: 8/124 (6.5%) (p=0.118) Complete response: 1/124 (0.8%) Partial response: 7/124 (5.6%) Stable disease: 53/1214 (42.7%)-

280 ± Time to progression for subgroups based on baseline characteristics Variable Subgroup Treatment N Median time to Hazard 95% CI for HR* progression (days) ratio (HR) Overall response ITT Caelyx 239 113 1.176 0.972, 1.423 Topotecan 235 119 Age < 65 years Caelyx 156 121 1.190 0.932, 1.520 Topotecan 138 113

>= 65 years Caelyx 83 103 1.147 0.835, 1.575 Topotecan 97 128 KPS at baseline <80 Caelyx 39 53 0.867 0.523, 1.438 Topotecan 37 71

>=80 Caelyx 200 131 1.157 0.939, 1.426 Topotecan 194 134

Drug-free interval after <= 6 months Caelyx 102 57 1.095 0.815, 1.470 last dose of first-line Topotecan 109 94 therapy >6 to <=18 months Caelyx 107 148 1.170 0.874, 1.566 Topotecan 94 131 >18 months

Caelyx 30 290 1.530 0.832, 2.812 Topotecan 32 228 Bulky disease Present Caelyx 111 92 1.143 0.863, 1.151 Topotecan 111 110

Absent Caelyx 128 131 1.206 0.929, 1.565 Topotecan 124 128 Platinum sensitivity Platinum Caelyx 109 202 1.349 1.018, 1.788 sensitive Topotecan 111 163

Platinum Caelyx 130 66 1.046 0.807, 1.356 refractory Topotecan 124 95 Ascites Present Caelyx 77 63 0.930 0.653, 1.325 Topotecan 65 102

Absent Caelyx 162 157 1.295 1.026, 1.635 Topotecan 168 134

* 95% CI were estimated from the original 90% CI reported using the formulae: In 95% CI lower CI = In HR-[1.96x(In of HR – In of lower 90% CI)/1.645] In 95% upper CI = In HR+[1.96x(In of HR – In of upper 90% CI)/1.645]

281 † Long-term follow-up efficacy analysis † denotes: extracted from Data on File. A Phase III randomised, open-label comparative study of Doxil/Caelyx versus Topotecan HCI in patients with epithelial ovarian carcinoma following failure of first- line, platinum-based chemotherapy. Kenilworth, NJ: Schering-Plough Research Institute; Final Report 2004.

Objective: To compare the efficacy of PLD versus topotecan in terms of survival and progression free survival when 90% of participants had died or were lost to follow-up

Length of follow-up: 3 years

† Outcome 1: overall survival

PLD (n=239) % censored: 16.7 Median survival (weeks): 62.7 (range: 1.7 – 258.3)

Topotecan (n=235) % censored: 8.9 Median survival (weeks): 59.7 (range: 1.6, 247.1) HR = 1.216 (95% CI: 1.00, 1.478) p = 0.050.

Overall Survival Rates

PLD Topotecan Overall Alive Dead Censored Rate 95% CI Alive Dead Censored Rate (%) 95% CI Survival (%) 1 year 132 104 3 56.3 50.0, 62.6 126 108 1 54.0 47.6, 60.3 2 years 80 154 5 34.7 28.6, 40.8 54 178 3 23.6 18.1, 29.1 3 years 38 185 16 20.2 14.9, 25.5 24 201 10 13.2 8.8, 17.7 1 year = 365 days; 2 years = 731 days; 3 years = 1,096 days

† Platinum sensitive disease PLD (n=109) % censored: 22.0 Median survival (weeks): 107.9 (range: 6.9 – 258.3)

Topotecan (n=110) % censored: 10.9 Median survival (weeks): 70.1 (range: 1.6 – 247.1+) HR = 1.432 (95% CI: 1.066, 1.923); p = 0.017

† Platinum refractory disease PLD (n=130) % censored: 12.3 Median survival (weeks): 38.3 (range: 1.7 – 253.9+)

282 Topotecan (n=125) % censored: 7.2 Median survival (weeks): 42.1 (range: 1.6 – 239.3) HR = 1.069 (95% CI: 0.823, 1.387); p = 0.618

† Overall Survival Rates – platinum refractory disease

PLD Topotecan Overall Alive Dead Censored Rate 95% CI Alive Dead Censored Rate (%) 95% CI Survival (%) 1 year 53 76 1 41.5 32.8, 50.1 54 71 0 43.2 34.5, 51.9 2 years 27 102 1 21.1 14.1, 28.2 21 103 1 17.2 10.5, 23.8 3 years 12 110 8 13.8 7.6, 20.0 8 112 5 9.5 4.2, 14.7 1 year = 365 days; 2 years = 731 days; 3 years = 1,096 days

† Overall survival for sub-groups according to baseline disease characteristics

Variable Group N HR 95% CI for HR Manufacturers ITT 474 1.216 1.00, 1.478 Age <65 year 294 1.322 1.022, 1.710 => 65 years 180 1.077 0.786, 1.477 Baseline KPS <80 76 0.871 0.531,k 1.427 =>80 394 1.242 0.999, 1.543 Drug-free interval =<6 months 211 1.103 0.826, 1.474 >6 to =<18 months 201 1.284 0.945, 1.744 >18 months 62 1.191 0.663, 2.137 Bulky disease Present 213 1.l131 0.849, 1.506 Absent 261 1.294 0.991, 1.691 Platinum sensitivity Sensitive 219 1.432 1.066, 1.923 Refractory 255 1.069 0.823, 1.387 Baseline ascites Present 142 0.978 0.689, 1.389 Absent v 330 1.387 1.088, 1.768

† Outcome 2: progression free survival

PLD (n=239) % censored: 2.9 Median progression free survival (weeks): 16.1 (range) 1.3 – 163.4+)

Topotecan (n=235) % censored: 3.0 Median progression free survival (weeks): 16.9 (range: 0.4 - 178.6) HR = 1.118 (95% CI: 0.928, 1.347); p = 0.241

283

Platinum sensitive disease PLD (n=109) % censored: 2.8 Median progression free survival (weeks): 27.3 (range: 2.4 – 151.9)

Topotecan (n=110) % censored: 1.8 Median progression free survival (weeks): 22.7 (range: 0.4 = 155.9+) HR = 1.287 (95% CI: 0.l977, 1.694); p = 0.072

Platinum refractory disease PLD (n=130) % censored: 3.1 Median progression free survival (weeks): 9.1 (range: 1.3 – 162.4)

Topotecan (n=125) % censored: 4.0 Median progression free survival (weeks): 13.6 (range: 1.4 – 178.6) HR = 0.992 (95% CI: 0.770, 1.279); p = 0.952

† Overall survival for sub-groups according to baseline disease characteristics Variable Group N HR 95% CI for HR Manufacturers ITT 474 1.118 0.928, 1.347 Age <65 year 294 1.207 0.949, 1.534 => 65 years 180 1.027 0.754, 1.398 Baseline KPS <80 76 0.869 0.519, 1.454 =>80 394 1.104 0.900, 1.354 Drug-free interval =<6 months 211 0.994 0.747, 1.322 >6 to =<18 months 201 1.093 0.817, 1.461 >18 months 62 1.371 0.789, 2.382

284 ± Numbers and percentages of patients with maintenance of or improvement in QoL scores at 12 weeks

Scales No of patients and % Caelyx Topotecan Physical functioning 66/118 56% 60/107 56% Role functioning 77/118 65% 63/109 58% Emotional functioning 80/119 67% 80/108 74% Cognitive functioning 87/119 73% 79/108 73% Social functioning 82/119 69% 69/108 64% Global QoL 68/117 58% 54/104 52% Fatigue 67/118 57% 61/109 56% Nausea / vomiting 86/119 72% 77/109 71% Pain 76/119 64% 88/109 81%

± Number and percentages of patients with maintenance of or improvement in QoL scores at 12 weeks for platinum-sensitive and platinum-refractory sub-groups

Scales No of patients and % Platinum sensitive Platinum-refractory Caelyx Topotecan Caelyx Topotecan Physical functioning 38/65 58% 30/57 53% 28/53 53% 30/50 60% Role functioning 36/65 55% 30/59 51% 41/53 77% 33/50 66% Emotional functioning 38/65 58% 40/58 69% 42/54 78% 40/50 80% Cognitive functioning 48/65 74% 42/58 72% 39/54 72% 37/50 74% Social functioning 39/65 60% 34/58 59% 43/54 80% 35/50 70% Global QoL 36/64 56% 25/56 45% 32/53 60% 29/48 60% Fatigue 29/65 45% 30/59 51% 38/53 72% 31/50 62% Nausea / vomiting 44/65 68% 42/59 71% 42/54 78% 35/50 70% Pain 35/65 54% 49/59 83% 41/54 76% 39/50 78%

± Quality-adjusted time without symptoms or toxicity survival analysis (Q-TwiST)

Time (in months) spent in the three health states and their differences

Health states Caelyx Topotecan Difference (95% CI)F TOX (grade 3 or higher) 0.84 1.54 -0.70 (-1.04, -0.36) TWIST 4.65 3.51 1.14 (0.46, 1.82) PROG (until death or 15 months post 5.07 5.53 -0.46 (-1.31, 0.39) randomisation)

285 Key: ± denotes extracted from : Schering-Plough Ltd, Caelyx (pegylated liposomal doxorubicin hydrocholoride) in the treatment of recurrent ovarian cancer in the United Kingdom – A submission to the National Institute for Clinical Excellence, August 29, 2001.

† Outcome 2: progression free survival

PLD (n=239) % censored: 2.9 Median progression free survival (weeks): 16.1 (range) 1.3 – 163.4+)

Topotecan (n=235) % censored: 3.0 Median progression free survival (weeks): 16.9 (range: 0.4 - 178.6) HR = 1.118 (95% CI: 0.928, 1.347); p = 0.241

Platinum sensitive disease PLD (n=109) % censored: 2.8 Median progression free survival (weeks): 27.3 (range: 2.4 – 151.9)

Topotecan (n=110) % censored: 1.8 Median progression free survival (weeks): 22.7 (range: 0.4 = 155.9+) HR = 1.287 (95% CI: 0.l977, 1.694); p = 0.072

Platinum refractory disease PLD (n=130) % censored: 3.1 Median progression free survival (weeks): 9.1 (range: 1.3 – 162.4)

Platinum sensitive disease

PLD (n=109) Total (CR + PR): 32/109 (29.4%) (95% CI: 20.8%, 37.9%) Complete: 8/109 (7.3%) Partial: 24/109 (22.0%)

Topotecan (n=110) Total (CR + PR): 31/110 (28.2%) (95% CI: 19.8%, 36.6%) Complete: 9/110 (8.2%) Partial: 22/110 (20.0%)

PLD (n=130) Total (CR + PR): 15/130 (11.5%) (95% CI: 6.0%, 17.0%) Complete: 1/130 (0.8%)

286 Partial: 14/130 (10.8%)

Topotecan (n=125) Total (CR + PR): 9/125 (7.2%) (95% CI: 2.7%, 11.7%) Complete: 2/125 (1.6%) Partial: 7/125 (5.6%)

Overall survival using All randomised participants population (true ITT analysis) (extracted from Schering-Plough Ltd, Caelyx (pegylated liposomal doxorubicin hydrocholoride) in the treatment of recurrent ovarian cancer in the United Kingdom – A submission to the National Institute for Clinical Excellence, June 6th , 2004

Outcome 1: Overall survival (nonstratifed log-rank test)

PLD (n=240) % censored: 16.7 Median overall survival (weeks); 63.6 (range: 1.7, 258.3)

Topotecan (n=241) % censored: 9.1 Median overall survival (weeks): 57.0 (range: 0.1 – 247.1+) HR = 1.246 (95% CI: 1.028, 1.511); p = 0.025

287

Trial: 039(24-26)

Study Details and Design Participant Details Intervention Details Follow-up/withdrawals and Adverse Conclusion and Comments Events Author: ten Bokkel Huinink (1997) Number randomised Intervention (I): Withdrawals from intervention (I) Authors’ conclusions (24) 235 randomised topotcan n=117; Type: - Topotecan Five patients randomised to topotecan Topotecan has efficacy at least paclitaxel n=118) No. randomised: 117 did not receive treatment and therefore equivalent to paclitaxel manifested by Objective: Route of administration : i.v. were not included in the analysis. A the higher response rate and To compare the efficacy and toxicity of Disease type: Epithelial Dose:- 1.5mg/m2 as a 30 min infusion further 16 were not evaluated for significantly longer time to progression. topotecan and paclitaxel in patients with Disease stage: Advanced stage III/IV on five consecutive days response: withdrawal for adverse events advanced epithelial ovarian cancer who Occurrence of secondary spread:- Not No. of cycles: Dependent on response (n=7), lost to follow-up (n=2), patient Comments have progressed during or after reported (see comments). 555 cycles in total for refusal (n=2), protocol violation - no treatment with one platinum-based the whole group. measurable disease at baseline (n=2), all The methods section of the report states chemotherapy regimen. Therapy stage: Second-line Length per cycle: - 21 days lesions noted at screening were not that hazard ratios with 95% CI were assessed throughout study (n=2) and calculated. Survival curves were Trial ID: 039 Previous treatments Control (C) other (pulmonary embolism (n=1). presented for the duration of response, Phase: Phase III Radiotherapy:- Type: Paclitaxel time to progression, and survival, but topotecan: n=3 No. randomised: 118. Withdrawals from control (C) hazard ratios were not reported. HRs are Lengrth of follow-up: 60 wks? paclitaxel: n=4 Route of administration: i.v. Four patients randomised to paclitaxel the most appropriate representation of Dose: 175mg/m2 as a 3 hour infusion - did not receive treatment and therefore survival or time to event data. It was Immunotherapy:- No. of cycles: Dependent upon response were not included in the analysis. A also not clear from the data presented topotecan: n=2 to treatment (see comments). 550 cycles further 8 were not evaluated for whether the median times quoted were Paclitxel: n=1 in total for the whole group response: withdrawal for adverse events based on Kaplan-Meier estimates. Length per cycle:- 21 days (n=3), patient refusal (n=1), protocol Hormonal therapy:- violation - no measurable disease at Topotecan: n=0 Comments about baseline (n=1), indicator lesions (n=1), Paclitxel: n=6 intervention/control: baseline performance status of 3 (n=1), Patients in the paclitaxel group were patient entered study in renal failure Prior chemotherapy:- premedicated with dexamethasone, and (n=1). Topotecan: (n=112) (%) both H1 and H2 receptor antagoinists to Cyclophosphamide: 66.0% prevent hypersensitivity reactions. Adverse events Carboplatin:55.0% Premedication was not given to the 7% of patients in topotecan grup and 4% Cisplatin: 54.0% topotecan group unless nausea or of patients in paclitaxel group were Epirubicin: 8.0% vomiting occurred, in which case it was withdrawn due to adverse effects. The Doxorubicin hydrochloride: 4.5% permitted in subsequent cycles. primary reasons for withdrawal in Doxorubicin: 3.6% However, prophylactic recombinant topotecan group were febrile Etoposide: 1.8% granulocyte colong-stimulating factor neutropenia, infection and sepsis. Mitoxantrone: 1.8% (G-CSF) was allowed after the first Neurotoxicity was the primary reason Ifosfamide: 1.8% course of therapy to maintain dose for withdrawal in the paclitaxel group. Epirubicin hydrochloride: 0.9% intensity, on day 6 of the topotecan and Chlorambucil: 0.9% day 2 of the paclitaxel group, if Suspected or documented infection Prednimustine: 0.9% participants had experienced any of the occurred within two days of grade 4

288 Fluorouracil: 0.0% following: grade 4 neutropenia with neutropenia in a total of 25% of patients Pirarubicin: 0.0% fever or infection, grade 4 neutorpenia and 7% of courses in the topotecan lasting more than 7 days, or grade 3 group patients and in 4% of patients and Paclitaxel: (n=114) (%) neutropenia that required a delay in 1% of courses in the paclitaxel group. Cyclophosphamide: 69.0% treatment. In addition, 5% of patients and 1% of Carboplatin: 61.0% topotecan courses and 2% of patients Cisplatin: 51.0% Dependent on toxicity the doses of the and 0.4% of paclitaxel courses were Epirubicin: 5.3% two drugs could vary from 1.0 mg.m2/d associated with sepsis. Two patients in Doxorubicin hydrochloride: 6.1% to 2.0 mg/m2/d for topotecan, and 135 topotecan group died due to topotecan- Doxorubicin: 3.5% mg/m2 to 175 mg/m2 for paclitaxel. induced sepsis. There were no deaths Etoposide: 0.9% Patients were withdrawn from treatment attributed to myelosuppression in the Mitoxantrone: 0.9% if there was a greater than 2-week delay paclitaxel group. Prophylactic G-CSF Ifosfamide: 0.0% in treatment at the minimum dose of was administered to maintain dose Epirubicin hydrochloride: 1.8% either medication because of toxicity. intensity in 23% of topotecan courses Chlorambucil: 0.9% and 3% of paclitaxel courses; platelet Prednimustine: 0.0% The number of cycles of both the transfusions were given in 3% and red Fluorouracil: 0.9% topotecan and paclitaxel interventions blood cell transfusions in 27% of Pirarubicin: 0.9% were determined by the patients' topotecan courses. Treatment G-CSF response. Patients with a CR/PR was given in 7% topotecan and 1% Disease present after first-line continued until progression or for 6 paclitaxel courses. treatment? months after the maximal response. Residual : at least 4 weeks after Patients who progressed during Hematological toxicity:- previous treatment treatment were removed from the study. topotecan (n=112) Refractory - Yes Those whose best response was stable leukopenia (grade 3): 50.9% disease after 6 courses were removed or leukopenia (grade 4): 33.6% Age/age range of participants switched to the other treatment. neutropenia (grade 3): 15.3% topotecan: mean=59.2 yrs (range 29-85 neutropenia (grade 4): 79.3% yrs) Complete response (CR) was defined as thrombocytopenia (grade 3): 24.3% paclitaxel: mean = 58.3 yrs (range 29 - the complete disappearance of all known thrombocytopenia (grade 4): 25.2% 79 yrs) measurable and assessable disease on anemia (grade 3): 36.9% two separate measurements at least 4 anemia (grade 4): 3.6% Characteristics wks apart. Partial response (PR) was Topotecan (n=112): defined as a 50% reduction in the sum paclitaxel (n=114) Mean age: 59.2 years of products of the perpendicular leukopenia (grade 3): 17.9% Age range: 29-85 diametres of all measurable lesions for leukopenia (grade 4): 2.7% Performance status: at least 4 wks and with no new lesion or neutropenia (grade 3): 28.6% 0; 41/112 (36.6%) progression of assessable disease. neutropenia (grade 4): 23.3% 1: 51/112 (45.5%) Progressive disease was defined as a thrombocytopenia (grade 3): 0.9% 2: 20/112 (17.9%) 25% increase in a single measurable thrombocytopenia (grade 4): 1.8% 3: 0/112 (0%) lesion, reappearance of measurable anemia (grade 3): 3.6% Weight (kg):- disease, clear worsening of assessable anemia (grade 4): 2.7% Mean: 65.0 disease, or the development of new Range: 41-95 metastatic disease. Stable disease was Number of courses with hematological Body surface (m2):- defined as any measurement not toxicity:- Mean: 1.7 fulfilling the criteria for response or

289 Range: 1.3-2.3 progression, and lasting more than 8 topotecan (n=555) Tumor diameter (cm):- wks. Non-assessable disease was leukopenia (grade 3): 40.1% <5: 54/112 (48.2%) defined as non-measurable lesions with leukopenia (grade 4): 10.4% >/= 5: 56 (50/0%) an elevated CA-125 tumour marker. neutropenia (grade 3): 27.9% ND*: 2/112 (1.8%) The duration of response was measured neutropenia (grade 4): 37.4% Tumor histology:- from the time of initial documented thrombocytopenia (grade 3): 15.3% Malignant serous: 58/112 (51.8%) response to the first sign of disease thrombocytopenia (grade 4): 9.6% Malignant mucinous: 6/112 (5.4%) progression. The time to progression anemia (grade 3): 14.9% Malignant endometriod: 10/112 (8.9%) was measured from the time of first anemia (grade 4): 1.1% Undifferntiated carcinoma: 18/112 study drug administration to (16.1%) documented progressive disease or paclitaxel (n=550) Other: 20/112 (17.9%) initiation of third-line therapy. The leukopenia (grade 3): 8.8% Histologic grade:- criteria for efficacy, time to reponse and leukopenia (grade 4): 0.5% 0-1: 6/112 (5.0%) survival were measured from the time of neutropenia (grade 3): 21.1% 2: 23/112 (20.5%) initial drug administration to initial neutropenia (grade 4): 9.0% 3: 56/112 (50%) reponse and death, respectively. thrombocytopenia (grade 3): 0.2% 4: 10/112 (8.9%) thrombocytopenia (grade 4): 0.4% ND*: 17/112 (15.2%) All responses were subject to anemia (grade 3): 1.5% Race: independent review and confirmation of anemia (grade 4): 0.5% White: 112 (100%) scans by a radiololgist who was blind to the treatment assignment. Non-hematological toxicities occurrng Paclitaxel: (n=114) in more than 10% of treated Mean age: 58.3 Actual study treatment patients:- Age range: 29-79 administration: topotecan: (n = not reported) Performance status: alopecia (grade 1/2): 75.9% 0: 42/114 (36.8%) Protocol deviations: alopecia (grade 3/4): 0.0% 1: 53/114 (46.5%) Topotecan: 16/112 (14.3%) nausea (grade 1/2): 67.9% 2: 17/114 (14.9%) Not assessed beyond 5 days: 11 (9.8%) nausea (grade 3/4): 9.8% 3: 2/114 (1.8%) Did not have required measurable vomiting (grade 1/2): 53.6% Weight (kg):- disease: 2 (1.8%) vomiting (grade 3/4): 9.9% Mean: 67.6 Indicator lesion in field of prior fatigue (grade 1/2): 33.1% Range: 46-136 radiation; 1 (09%) fatigue (grade 3/4): 8.0% Body surface (m2):- Entered with second primary cancer: 1 constipation (grade 1/2): 37.5% Mean: 1.7 (0.9%) constipation (grade 3/4): 5.4% Range: 1.4-2.4 >1 prior chemotherapy regimen: 1 diarrhea (grade 1/2): 33.9% Tumor diameter (cm):- (0.9%) diarrhea (grade 3/4): 6.3% <5: 53/114 (46.5%) Did not have required performance abdominal pain (grade 1/2): 21.5% >/= 5: 59/114 (51.8%) status: 0 abdominal pain (grade 3/4): 5.4% ND*: 2/114 (1.8%) fever* (grade 1/2): 27.7% Tumor histology:- Paclitaxel: 9* fever* (grade 3/4): 0.9% Malignant serous: 59/114 (51.8%) Not assessed beyond 5 days: 6 (5.3%0 stomatitis (grade 1/2): 23.2% Malignant mucinous: 6/114 (5.3%) Did not have required measurable stomatitis (grade 3/4): 0.9% Malignant endometriod: 15/114 (13.2%) disease: 1 (0.9%) dyspnea (grade 1/2): 17.8% Undifferentiated carcinoma: 8/114 Indicator lesion in field of prior dyspnea (grade 3/4): 6.3% (7.0%) asthenia (grade 1/2): 17.0%

290 Other: 26/114 (22.8%) radiation; 1 (09%) asthenia (grade 3/4): 5.4% Histologic grade:- Entered with second primary cancer: 0 arthralgia (grade 1/2): 5.5% 0-1: 8/114 (7.0%) >1 prior chemotherapy regimen: 0 arthralgia (grade 3/4): 0.9% 2: 29/114 (25.4%) Did not have required performance myalgia (grade 1/2): 3.6% 3: 50/114 (43.9%) status: 2 (1.8%) myalgia (grade 3/4): 0.0% 4: 12/114 (10.5%) neuropathy (grade 1/2): 0.9% ND*: 15/114 (13.2%) * 1 patient had 2 protocol violations: neuropathy (grade 3/4): 0.0% Race: performance status and lack of skeletal pain (grade 1/2): 4.5% White: 100 assessment skeletal pai8n (grade 3/4): 0.0% Black: 5 flushing (grade 1/2): 4.5% Hispanic: 3 flushing (grade 3/4): 0.0% paresthesia (grade 1/2): 0.9% * ND = not determined paresthesia (grade 3/4): 0 (0%)

Inclusion/exclusion criteria paclitaxel (n = not reported) Women with the following were alopecia (grade 1/2):92.1% included: stage III/IV disease, histologic alopecia (grade 3/4): 0.9% diagnosis of epithelial ovarian nausea (grade 1/2): 43.0% carcinoma, failed first line threapy with nausea (grade 3/4): 1.8% a platinum-based chemotherapy vomiting (grade 1/2): 28.1% regimen; at least one bidimensionally vomiting (grade 3/4): 2.7% measurable lesion as evidence by fatigue (grade 1/2): 25.4% computed tomography (CT) or magnetic fatigue (grade 3/4): 6.1% resonance imaging (MRI) scan, constipation (grade 1/2): 30.7% ultrasound, or physical examination; at constipation (grade 3/4): 0.0% least a 4 week period between prior diarrhea (grade 1/2): 37.8% surgery, hormonal therapy, diarrhea (grade 3/4): 0.9% radiotherapy, or chemotherapy and abdominal pain (grade 1/2): 36.0% treatment in the trial; an Eastern abdominal pain (grade 3/4): 3.5% Cooperative Oncology Group (ECOG) fever* (grade 1/2): 17.7% performance status of /= 3500, neutrophil count >/= 1500 uL, stomatitis (grade 3/4): 0.9% and platelet count >/+ 100000/uL); dyspnea (grade 1/2): 13.2% normal liver function (bilirubin level dyspnea (grade 3/4): 5.3% /= asthenia (grade 3/4): 3.5% 1.5mg/dL or creatinine clearance > 60 arthralgia (grade 1/2): 28.9% mL/min). Patients who had received arthralgia (grade 3/4): 2.6% more than one previous chemotherapy myalgia (grade 1/2): 25.4% regimen or who had received topotecan myalgia (grade 3/4): 2.6% or paclitaxel previously were excluded. neuropathy (grade 1/2): 15.8% neuropathy (grade 3/4): 0.0% Comments about participants: skeletal pain (grade 1/2): 11.4% Demographic and baseline disease skeletal pai8n (grade 3/4): 5.3%

291 characteristics in the ITT population flushing (grade 1/2): 14.1% were comparable between the two flushing (grade 3/4): 0.0% treatment groups. paresthesia (grade 1/2): 29.0% paresthesia (grade 3/4): 0.0% Patients were stratified by age (<65yrs or >/= 65 yrs), ascites (present or * excludes febrile neturopenia absent) and response to prior platinumj- based therapy (resistant, early, interim Dosing adjustments:- or late relapse). Topotecan Patients receiving > 1 course: 96/112 The response rates for the ITT Delays after course 1: 58/112 (60.4%) population will be conservative as the patients; 100/597 courses (20.6%) denominator includes women who Reductions after course 1: 15/112 withdrew from the trial and were not patients (15.6%); 20/597 courses evaluated for response. However, a true (4.1%). ITT analysis was not performed as only Escalations after course 1: 3/112 226 (not the 235 randomised) were patients (3.1%); 3/597 (0.6%) included in the analysis. Patients who progressed or whose best Paclitaxel response was stable disease after 6 Patients receiving > 1 course:108/114 courses on one regimen were eligible to Delays after course 1: 20/114 (18.5%) be switched to the other (alternate) patients; 26/589 courses (5.5%) regimen, or were removed. 90 patients Reductions after course 1: 6/114 patients in total were entered into the alternative (5.6%); 7/589 courses (1.5%). regimen, 53 received topotecan after Escalations after course 1: 0/114 failing to respond to paclitaxel and 37 patients (0%); 0/589 (0%) received paclitaxel having failed to respond to topotecan. Reasons for Treatment Delay* in patients receiving > 1 course: no courses (%)

Topotecan(n=485) Hematologic: 56 (11.5%) Non-Hematologic: 11 (2.3%) Both hematologic & non-henatologic: 1 (0.2%) Other: 31 (6.4%) Unassigned: 53: 10.9% Total: 152: (31.3%)

Paclitaxel (n=475) Hematologic: 9 (1.9%) Non-Hematologic: 2 (0.4%0 Both hematologic & non-henatologic: 0 (0%)

292 Other: 19 (4.0%) Unassigned: 84 (17.7%) Total: 114 (24.0%)

*Delays are not asociated with course 1 of treatment.

Reasons for Dose reductions Topotecan(n=485) Hematologic: 21 (4.3%) Non-Hematologic: 1 (0.2%) Both hematologic & non-henatologic: 0 (0%) Other: 0 (0%) Unassigned: 1(0.2%) Total: 23 (4.7%)

Paclitaxel (n=475) Hematologic: 7 (1.5%) Non-Hematologic: 0 (0%) Both hematologic & non-henatologic: 0 (0%) Other: 0 (0%) Unassigned: 0 (0%) Total: 7 (1.5%)

Withdrawals Topotecan (n = 112) Adverse experience: 13 (11.6%) Protocol violation: 1 (0.9%) Lost to follow up: 2 (1.8%) Other*: 11 (9.8%) Total: 27 (24.1%)

Paclitaxel (n = 114) Adverse experience: 8 (7.0%) Protocol violation: 0 (0%) Lost to follow up: 2 (1.8%) Other*: 5 (4.4%) Total: 15 (13.2%)

*comprised primarily of patient request or refusal.

293 Results Outcome 1 Outcome 2 Outcome 3 Outcome: Response rate Outcome: Duration of response (measured from the time of initial Outcome: Time to progression (measured from the time of first documented response to the first sign of disease progression). study drug administration to documented progressive disease or Topotecan:- (n=112) initiation of third-line therapy) CR: 5/112 (4.5%) Topotecan PR: 18/112 (16.1%) Median = 32.1 (n=23) Topotecan (n=112) Total (CR+PR): 23/112 (20.5%); p=0.138 (95%CI: 13, 28.3) Range: 5.4 - 53.1 Median = 23.1 Risk ratio=0.416; p=0.2218 Range: 0.7 - 62.1 ITT (n = 117) response rate = 18.8% (95% CI: 11.7, 25.9%) Risk ratio: 0.578; p=0.021 Paclitaxel:- Sub-group analysis according to platinum sensitivity Median = 19.71 (n=15) Paclitaxel (n=114) Resistant disease:- (n=34) Range: 6.3 - 24.3 Median = 14.0 CR: 0/34 (0%) Risk ratio=0.416; p=0.2218 Range: 0.1 - 30.9 PR: 3/34 (8.8%) Risk ratio: 0.578; p=0.021 Total (CR=PR): 3/34 (8.8%)

Early relapse:- (n=6) CR: 0/6 (0.0%) PR: 1/6 (16.7%) Total (CR+PR) 1/6 (16.7%)

Interim relapse:- (n=20) CR: 1/20 (5.0%) PR: 3/20 (15.0%) Total (CR=PR): 4/20 (20.0%)

Late relapse:- (n=52) C: 4/52 (7.7%) PR: 11/52 (21.2%) Total (CR=PR) 15/52 (28.8%)

Non-responders:- (n=112) Stable disease: 33/112 (29.5%) Progressive disease: 39/112 (34.8%) Not assessable 17/112 (15.2%) Total 89/112 (79.5%)

Responders: Response in relation to baseline disease status: Age: /= 65 yrs: 23.7%

294 Ascites:- Present: 18.9% Absent: 21.3% Performance status:- 0: 22.0% 1: 25.5% 2: 5.0% Tumour burden:- < 5cm: 33.3% 5 -

Paclitaxel (n=114) Responders:- CR: 3/114 (2.6%) PR: 12/114 (10.5%) Total (CR+PR) 15/114 (13.2%); p=0.138 (95%CI: 7, 19.4)

ITT (N = 118) response rate = 13.6% (95% CI: 7.4, 19.7%)

Sub-group analysis according to platinum-sensitivity:- Resistant disease: (n=33) CR: 0/33 (0.0%) PR: 1/33 (3.0%) Total (CR+PR) 1/33 (3.0%)

Early relapse:(n=10) CR%: 0/10 (0.0%) PR: 1/10 (10%) Total (CR+PR) 1/10 (10%)

Interim relapse:(n=16) CR: 0/16 (0.0%) PR: 2/26 (12.5%) Total (CR + PR) 2/16 (12.5%)

Late relapse:- (n=55) CR: 3/55 (5.5%) PR: 8/55 (14.5%) Total (CR + PR): 11/55 (20.0%)

Non-responsers:- Stable disease: 38/114 (33.3%)

295 Progressive disease: 56/114 (49.1%) Not assessable: 5/114 (4.4%) Total: 99/114 (86.8%)

Responders: Response in relation to baseline disease status: Age:- /= 65 yrs: 16.7% Ascites:- Present: 7.5% Absent: 16.2% Performance status:- 0: 14.3% 1: 13.2% 2: 11.8% Tumour burden:- < 5cm: 18.0% 5 -

Outcome 4 Outcome 5 Outcome 6 Outcome: Time to response (measured from the time of initial Outcome: Survial (measured from the time of initial drug Outcome: Response to alternative regimen drug administration to initial response). administration to death) Fifty-three patients received topotecan after failing to respond to Topotecan:- Topotecan:- paclitaxel and five achieved PRs (9.4%), while 37 patients received Median = 9.0 (n=23) Median = 61.3 (n=112) paclitaxel after topotecan and one achieved a CR (2.7%) Range = 3.1 = 19.0 Range= 0.7 - 62.1 Risk ratio=0.476; p=0.0409. Risk ratio = 1.210; p=0.5153

Paclitaxel:- 11 patients died within 30 days of treatment, 31 after longer than 30 Median = 6.0 (n=15) days of completing therapy. Death following alternate paclitaxel Range= 2.4 - 12.3 therapy occurred in 33 patients. Deaths were not recorded for 37 Risk ratio=0.476; p=0.0409. patients.

Paclitaxel: Median=42.6 (n=114) Range=0.1 - 75.3 Risk ratio = 1.210; p=0.5153

296 3 patients died within 30 days of treatment, 28 after longer than 30 days of completing therapy. Death following alternate topotecan therapy occurred in 46 patients. Deaths were not recorded for 37 patients.

Long-term follow-up from trial 039 Participants: same as reported in trial 039.

Author: ten Bokkel Huinink (2004) (25) Interventions: same as reported in trial 039.

Objective: To monitor progression-free survival and overall survival of patients randomised in a previously reported multicentre phase III study of topotecan versus paclitaxel in patients with advanced epithelial ovarian cancer who had failed one prior platinum-based regimen.

Length of follow-up: median 4 years

Outcome 1 Outcome 2 Outcome 3 Outcome: Time to progression (definition as before) Outcome: Survival (definition as before) Outcome: Quality of life ( assessed by EORTC-QOL-C30 questionnaire. Questionnaire completed at baseline to end of best Topotecan:- Topotecan:- response. Between 75% and 85% of patients who enrolled in the 18.9 weeks (range <1 to 92.6 weeks; 25% censored) 63.0 weeks (range <1 to 238.4+ weeks; 20.5% censored) study had evaluable quality of life data) Paclitaxel:- Paclitaxel:- 14.7 weeks (range <1 to 137.3+ weeks; 12.3% censored); P=0.08. 53.0 weeks (range <1 to 226.3+ weeks; 12.3% censored); p=0.44 The EORTC-QOL-C30) questionnaire results for pain, anorexia, diarrhoea, fatigue, nausea and vomiting, dyspnea, constipation and insommnia were similar for the two groups, and treatment with either topotecan or paclitaxel was not associated with any compromise of QoL for the patients.

Changes from baseline to end of best response: median (range) Topotecan Appetite loss: (n = 88): 0.0 (-100 – 67) Cognitive function (n = 85): 0.0 (- 67 – 50) Constipation (n = 89): 0.0 (-100 – 67) Diarrhea (n = 88): 0.0 (-67 – 100) Dyspnoea (n = 89): 0.0 (-67 – 67) Emotional function (n = 85): 8.0 (-83 – 75) Fatigue (n = 90) 0.0 (-78 – 89) Financial impact (n = 89): 0.0 (-67 – 100) Nausea/Vomiting (n = 90): 0.0 (100 – 100) Pain (n = 91): 0.0 (-100 – 67) Physical function (n = 93): 0.0 (-100 – 80) Quality of Life (n = 89): -8.0 (-58 – 83) Rloe function (n = 93): 0.0 (-100 – 100)

297 Socal function (n = 89): 0.0 (-83 – 67) Sleep disturbance (n = 89): 0.0 (-100 – 100)

Paclitaxel Appetite loss: (n = 89): 0.0 (-67 – 100) Cognitive function (n = 90): 0.0 (-67 – 50) Constipation (n = 91): 0.0 (-100 – 67) Diarrhea (n = 94): 0.0 (-100 – 33) Dyspnoea (n = 94): 0.0 (-67 – 67) Emotional function (n = 91): 8.0 (-100 – 75) Fatigue (n = 94): 0.0 (-67 – 67) Financial impact (n = 94): 0.0 (-100 – 67) Nausea/Vomiting (n = 94): 0.0 (-83 – 50) Pain (n = 93): 0.0 (-100 – 67) Physical function (n = 98): 0.0 (-100 – 80) Quality of Life (n = 95): 0.0 (-67 – 50) Rloe function (n = 97): 0.0 (-100 – 100) Socal function (n = 95): 0.0 (-100 – 100) Sleep disturbance (n = 93): 0.0 (-100 –100)

Changes from baseline to end of best response for patients who responded to treatment: median (range) Topotecan Appetite loss: (n = 22): 0.0 (-100 – 67) Cognitive function (n = 21): 0.0 (- 66 – 50) Constipation (n = 22): 0.0 (-67 – 67) Diarrhea (n = 22): 0.0 (-67 – 67) Dyspnoea (n = 22): 0.0 (-67 – 67) Emotional function (n = 21): 9.0 (-42 – 75) Fatigue (n = 22) 0.0 (-78 – 56) Financial impact (n = 21): 0.0 (-33 – 34) Nausea/Vomiting (n = 22): 0.0 (100 – 50) Pain (n = 22): 0.0 (-100 – 67) Physical function (n = 23): 0.0 (-80 – 80) Quality of Life (n = 21): -8.0 (-58 – 83) Role function (n = 23): 0.0 (-100 – 100) Socal function (n = 21): 0.0 (-67 – 50) Sleep disturbance (n = 22): 0.0 (-100 – 67)

Paclitaxel Appetite loss: (n = 12): 0.0 (-34 - 0) Cognitive function (n = 13): 0.0 (-17 – 33) Constipation (n = 12): 0.0 (-33 – 0) Diarrhea (n = 13): 0.0 (-67 – 0) Dyspnoea (n = 13): 0.0 (-34 – 33)

298 Emotional function (n = 13): 9.0 (0 – 50) Fatigue (n = 13): -11.0 (-45 – 12) Financial impact (n = 13): 0.0 (0 – 34) Nausea/Vomiting (n = 13): 0.0 (-33 – 0) Pain (n = 12): -16.5 (-67 – 33) Physical function (n = 13): 0.0 (-30 – 20) Quality of Life (n = 14):16.0 (-50 – 34) Role function (n = 13):0.0 (0 – 50) Socal function (n = 13): 0.0 (-50 – 33) Sleep disturbance (n = 12): -33.0 (-100 –33)

Outcome 4 Outcome 5 Outcome 6 Outcome: Long-term survival (topotecan median follow-up = Outcome: Survial stratified by platinum sensitivity Outcome: Alternative regimen 20.8 weeks, range 0-86 weeks; paclitaxel median follow-up = 23.6 weeks, range = 0-117 weeks). Topotecan:- Demographic characteristics of patients crossing over: Refractory patients (n=34, censored=8): median survival = 28.4 Topotecan (N = 61) Topotecan:- weeks Age Median survival = 63 weels Early/interim relapse (n=26, censored=4): median survival relapse = < 40: 1 (1.6%) 71.9 weeks 41 – 64: 46 (75.4%) Paclitaxel:- Late relapse (nh=52, censored=11): median survival = 63.4 weeks >/= 65: 14 (23.0%) Median survival = 53 weeks (log0rank p=0.44) Difference between the groups: p=0.02 Mean: 56.9 Range: 29 –77 Paclitaxel:- Race Refractory patients (n=33, censored=4): median survival = 39.7 Caucasian: 57 (93.4%) weeks Black: 3 (4.9%) Early/interim relapse patients(n=26, censored=2): median survival = Other: 1 (1.6%) 35.0 weeks Weight (kgs) Late relapse patients (n=54, censored=8): median survival = 85.1 Mean: 66.4 weeks Range: 46 –100 Difference between the groups (p=0.0004) Body surface area (m2) Mean: 1.71 Range: 1.45 – 2.10

Paclitaxel (n = 49) Age < 40: 1 (2.0%) 41 – 64: 28 (57.1%) >/= 65: 20 (40.8%) Mean: 60.3 Range: 30 – 75 Race Caucasian: 49 (100%) Black: 0

299 Other: 0 Weight (kgs) Mean: 66.2 Range (42 – 90) Body surface area (m2) Mean: 1.70 Range: 1.33 – 2.04

Baseline disease characteristics Topotecan (N = 61) Primary tumour histology: 28 (25.6%) Malignent serous: 4 (3.6%) Malignant endometrioid:9 (8.2%) Undifferentiated: 7 (6.4%0 Other: 13 (11.8%) Histologic grade 1: 5 (8.2%) 2: 14 (23.0%) 3: 24 (39.3%) 4: 8 (13.1%) Missing: 10 (16.4%) Unknown: 0 Prior therapy: Radiotherapy: 2 (3.3%) Hormonal therapy: 3 (4.9%) Immunotherapy: 1 (1.6%) Performance status at end of randomised treatment 0: 15 (24.6%) 1: 37 (60.7%) 2: 6 (9.8%) 3: 3 (4.9%) Platinum sensitivity at baseline Refractory: 18 (29.5%) Early relapse: 7 (11.5%) Interim relapse: 9 (14.8%) Late relapse: 26 (42.6%) Missing: 1 (1.6%) Largest measurable lesions (end of randomised treatment) < 2cm: 0 2 -< 5cm: 26 (42.6%) 5 –10cm: 25 (41.0%) >10cm: 8 (13.1%) data missing: 2 (3.3%) number of courses of randomised treatment median courses (range) 5 (1 –13)

300 Best response to randomised treatment Partial response: 2 (3.3%) Stable disease: 23 (37.7%) Progressive disease: 35 (57.4%) Not evaluated: 1 (1.6%) Time to progression while on randomised therapy Median (weeks): 11.9 Minimum: 2.4 Maximum: 49 (censored event)

Paclitaxel (N = 49) Primary tumour histology:27 (24.5%) Malignent serous: 3 (2.7%) Malignant endometrioid:4 (3.6%) Undifferentiated: 7 (6.4%0 Other: 8 (7.3%) Histologic grade 1: 2 (4.1%) 2: 9 (18.4%) 3: 23 (47.0%) 4: 9 (18.4%) Missing: 5 (10.2%) Unknown; 1 (2.0%) Prior therapy: Radiotherapy: 2 (4.1%) Hormonal therapy: 0 Immunotherapy: 0 Performance status at end of randomised treatment 0: 17 (34.7%) 1: 23 (46.9%) 2: 7 (14.3%) 3: 2 (4.1%) Platinum sensitivity at baseline Refractory: 15 (30.6%) Early relapse: 2 (4.1%) Interim relapse: 14 (28.6%) Late relapse: 18 (36.7%) Missing: 0 Largest measurable lesions (end of randomised treatment) < 2cm: 6 (12.2%) 2 -< 5cm: 21 (42.9%) 5 –10cm: 18 (36.7%) >10cm: 3 (6.1%) data missing: 1 (2.0%)

301 number of courses of randomised treatment median courses (range): 6 (2 –13) Best response to randomised treatment Partial response; 7 (14.3%) Stable disease: 17 (34.7%) Progressive disease: 22 (44.9%) Not evaluated: 3 (6.1%) Time to progression while on randomised therapy Median (weeks): 18.0 Minimum: 3.1 Maximum : 53.0 (censored event)

Demographic characteristics were similar betweent the two groups who-crossed over to third-line therapy. A total of 110 patients received the alternate drug (61 topotecan, 49 paclitaxel). Response rates of 13.1% for topotecan (eight of 61) and 10.2% for paclitaxel (five of 49) were demonstrated. Seven patients who failed to respond to second-line treatment with paclitaxel responded to topotecan third- line, as did four to paclitaxel third-line.

Topotecan All 8 (13.1%) responses were partial responses. (95% CI: 4 –21) Median duration of response was 29 weeks; median time to progression was 9 weeks, median survival was 40 weeks.

Completed study: 50 (82%) Reasons for withdrawal: Adverse experience; 3 (4.9%) Other: 7 (11.5%) Protocol violation: 1 (1.6%)

Paclitaxel 2 complete responses (4%) and 3 partial responses (6%) occurred. Total: 5 (10.2%) (95% CI: 0.5 –14). Median duration of response was 27 weeks; medina time to progression was 9 weeks, median survival was 48 weeks.

Completed study: 44 (89.8%) Reasons for withdrawal: Adverse experience: 3 (6.1%) Other: 2 (4.1%) Protocol violation: 0 (0%)

302

Study Details and Design Participant Details Intervention Details Adverse Events Conclusion and Comments Author: Gore (2001) (26) Topotecan (n=61) Patients were permitted to cross over Hematologic toxicity during cross- Authors’ conclusions Age: mean= 57 (range 29-77) and receive the alternative drug as third- over (third-line) treatment Objective Histology grade*:- line therapy, i.e. patients who were Topotecan (n=61) Topotecan and paclitaxel have similar In a large RCT comparing the efficacy 0/1: 5/61 (8%) randomised to topotecan crossed over to Neutropenia:- activity as second-line therapies with and safety of topotecan versus 2: 14/61 (23%) receive paclitaxel and those who were Grade 1: 0% regard to response rates and paclitaxel, a number of patients crossed 3: 24/61 (39%) treated with paclitaxel crossed over to Grade 2: 3% progression-free and overall survival. over to the alternative drug as third-line 4: 8/61 (13%) receive topotecan. This cross-over Grade 3: 14% The results of the long-term follow-up therapy (i.e) from paclitaxel to Unknown: 10/61 (16%) occurred at the investigator's discretion Grade 4: 81% of patients who crossed over between topotecan and vice versa. We therefore but was usually for one of three reasons Thrombocytopenia:- the two regimens demonstrated that the aimed to assess the degree of non-cross- ECOG performance status** (1) failure to respond to second-line Grade 1: 47% two drugs have a degree of non-cross- resistance between these two 0: 15/61 (25%) therapy (2) relapse after an initial Grade 2: 12% resistance. Thus, there is a good compounds in these patient groups. 1: 37/61 (61%) response to second-line therapy or (3) Grade 3: 8% rationale for incorporating these drugs 2: 6/61 (10%) toxicity. Grade 4: 22% into future first-line regimens. 3: 3/61 (5%) Anemia:- Platinum sensitivity at baseline*: A total of 110 patients crossed over to Grade 1: 8% Refractory: 18/61 (30%) the alternative drug as third-line therapy: Grade 2: 62% Early relapse: 7/61 (11%) 61 patients crossed over from paclitaxel Grade 3: 27% Interim relapse: 9/61 (15%) to topotecan, and 49 patients crossed Grade 4: 2% Late relapse: 26/61 (43%) over from topotecan to paclitaxel. Missing: 1/61 (2%) Paclitaxel (n=49) The 61 patients treated with topotecan Neutropenia:- Tumour size**:- received 270 courses of treatment, and Grade 1: 15% <5cm: 26/61 (43%) the 49 patients treated with paclitaxel Grade 2: 15% 5-10 cm: 25/61 (41%) received 223 courses. Patients in the Grade 3: 31% >10cm: 8/61 (13%) topotecan group received a median of 3 Grade 4: 23% Missing: 2/61 (3%) courses of treatment (range 1 - 23) and Thrombocytopenia:- those in the paclitaxel group received a Grade 1: 17% Paclitaxel (n=49) median of 4 courses (range 1 - 12). The Grade 2: 0% Age: mean= 60 (range 30-75) planned dose was maintained in 95% of Grade 3: 2% Histology grade*:- courses of topotecan and 94% of courses Grade 4: 0% 0/1: 2/49 (4%) of paclitaxel. Haematological toxicity Anemia:- 2: 9/49 (18%) caused dose delays of more than 7 days Grade 1: 52% 3: 23/49 (47%) in only 1.4% of topotecan courses and in Grade 2: 42% 4: 9/49 (18%) none of the paclitaxel courses. Grade 3: 2% Unknown: 6/49 (12%) Grade 4: 0% Prophylactic G-CSF was administered ECOG performance status** to 23% of patients (22% courses) in the Nonhematologic toxicity to cross-over 0: 17/49 (35%) topotecan group and 12% of patients (third-line) treatment 1: 23/49 (47%) (6% courses) in the paclitaxel group. 2: 7/49 (14%) Treatment with G-CSF was Topotecan (n=61) 3: 2/49 (4%) administered to 8% and 2% of patients Nausea (grade 1/2): 59% Platinum sensitivity at baseline*: (3% and 0% courses) in the topotecan Nausea (grade 3/4): 13%

303 Refractory: 15/49 (31%) and paclitaxel groups, respectively. Vomiting (grade 1/2): 36% Early relapse: 2/49 (4%) Intravenous antibiotics were Vomiting (grade 3/4): 12% Interim relapse: 14/49 (29%) administered to 26% of patients in the Diarrhoea (grade 1/2): 33% Late relapse: 18/49 (37%) topotecan group and 12% of patients in Diarrhoea (grade 3/4): 3% Missing: 0 (0%) the paclitaxel group. Thirteen patients Abdominal pain (grade 1/2): 25% (21%) who received topotecan and one Abdominal pain (grade 3/4): 8% Tumour size**:- who received paclitaxel (2%) were Constipation (grade 1/2): 26% <5cm: 27/49 (55%) hospitalized because of infection. Constipation (grade 3/4): 5% 5-10 cm: 18/49 (37%) Fatigue (grade 1/2): 23% >10cm: 3/49 (6%) Fatigue (grade 3/4): 8% Missing: 1/49 (2%) Alopecia (grade 1/2): 21% Alopecia (grade 3/4): 0% *at entry to the randomized phase of the Myalgia (grade 1/2):16% study. Myalgia (grade 3/4): 0% ** Eastern Cooperative Oncology Arthralgia (grade 1/2): 13% Group; assessed separately at the start of Arthralgia (grade 3/4): 2% each phase. Paresthesia (grade 1/2): 8% Paresthesia (grade 3/4): 0% At third-line therapy there were differences in the two groups' Paclitaxel (n=49) potential chemo-responsiveness. Nausea (grade 1/2): 39% Nausea (grade 3/4):10% Response to second-line treatment Vomiting (grade 1/2): 27% Topotecan (n=61) Vomiting (grade 3/4): 8% Partial response: 2/61 (3%) Diarrhoea (grade 1/2):35% Stable disease: 23/61 (38%) Diarrhoea (grade 3/4): 2% Progressive disease:35/61 (57%) Abdominal pain (grade 1/2):20% Non-assessable: 1/61 (2%) Abdominal pain (grade 3/4): 8% Constipation (grade 1/2): 35% Paclitaxel (n=49) Constipation (grade 3/4): 0% Partial response: 7/49 (14%) Fatigue (grade 1/2):16% Stable disease: 17/49 (35%) Fatigue (grade 3/4): 0% Progressive disease: 22/49 (45%) Alopecia (grade 1/2): 41% Non-assessable: 3/49 (6%) Alopecia (grade 3/4): 0% Myalgia (grade 1/2): 31% Platinum sensitivity to initial Myalgia (grade 3/4): 2% chemotherapy Arthralgia (grade 1/2): 37% Topotecan (n=61) Arthralgia (grade 3/4): 0% Refractory n=18 Paresthesia (grade 1/2): 31% Early relapse (within 3 months) n=7 Paresthesia (grade 3/4): 0% Interim relapse (within 3-6 months) n=9 Late relapse (> 6 months) n=26 At the time of analysis, 46 (75%) of the Missing: n=1 patients who were treated with topotecan as third-line therapy had died, Paclitaxel (n=49) 45 from progressive disease. One

304 Refractory n=15 patient died as a result of paralytic ileus Early relapse (within 3 months) n=2 and septic shock, which was not thought Interim relapse (within 6 months) n=14 to be drug related. Thirty-three (67%) Late relapse (> 6 months) n=18 patients in the paclitaxel group were reported to have died, all as a result of progressive disease. Results Outcome 1 Outcome 2 Outcome: Response rate Outcome: Response stratified by sensitivity to first-line platinum therapy.

Topotecan (n=61) Only those patients who had a treatment-free interval of more than 6 months Median response duration = 29 weeks (range 18 - 70) after completion of first-line platinum therapy responded to third-line topotecan, Median time to progression = 9 weeks although four patients who relapses within 6 months of their first-line platinum Median survival = 40 weeks (range 1 - 123) therapy responded to paclitaxel at the third-line phase. There were no p=NS responders to any third-line treatment in patients who had platinum-refractory Complete response: 0 (0%) disease (relapse within 4 months or no response or progressive disease to first- Partial response: 8/61 (13%) line therapy) Stable disease: 8/61 (13%) Progressive disease: 35/61 (57%) Nonevaluable: 10/61 (16%) Overall response: 13% (95% CI: 4%, 21%)

Paclitaxel (n=49) Median response duration: 27 weeks (range 13 - 66) Median time to progression: 9 weeks Median survival: 48 weeks (range 2 - 86) p=NS complete response: 2/49 (4%) Partial response: 3/49 (6%) Stable disease: 12/49 (25%) Progressive disease: 25/49 (51%) Nonevaluable: 7/49 (14%) Overall response: 10% (95% CI: 0.5%, 14%)

Outcome 3:

Response to Topotecan (third-line) treatment according to response to paclitaxel (second-line) treatment

Patient response to paclitaxel (third-line) treatment (no.) CR PR SD PD NE Response to topotecan (second-line) treatment PR, n= 7 1 0 2 4 0 SD, n=27 0 2 4 10 1

305 PD, n=22 1 1 6 10 4 NE, n=3 0 0 0 1 2

Response to Paclitaxel (third-line) treatment according to response to topotecan (second-line) treatment

Patient response to topotecan (third-line) treatment (no.) CR PR SD PD NE Response to paclitaxel (second-line) treatment PR, n= 2 0 0 1 1 0 SD, n=17 0 4 4 10 5 PD, n=35 0 3 3 24 5 NE, n=1 0 1 0 0 0

306 Trial: ICON4 /AGO-OVAR 2.2(29)

Study Details and Design Participant Details Intervention Details Follow-up/withdrawals and Adverse Conclusion and Comments Events Author: The ICON and AGO Number randomised Intervention (I): Withdrawals from intervention (I) Authors’ conclusions Collaborators (2003)(29) 802 (Paclitaxel plus platinum Type: - Paclitaxel in combination with 8 in total: 1 treatment never began, 7 Paclitaxel plus platinum chemotherapy chemotherapy n=392; platinum based carboplatin or cisplatin missing details (no further details seems to improve survival and Objective: To compare the efficacy and monotherapy n=410) No. randomised:- 392 reported) progression-free survival among patients safety of paclitaxel in combination with Route of administration : i.v. Withdrawals from control (C) with relapsed platinum-sensitive ovarian platinum-based chemotherapy compared Disease type: Epithelial Dose:- ICON4 (MRC CTU & IRFMN) 18 in total: 2 treatment never began, 16 cancer compared with conventional to platinum-based chemotherapy alone Disease stage: Advanced paclitaxel 175mg/m2 given as a 3 hr missing details (no further details platinum-based chemotherapy. in patients with epi8thelial ovarian Occurrence of secondary spread:- Not infusion, followed by either carboplatin reported) carcinoma who had relapsed after 6 reported. at 5(GFR+25) mg (where GFR was the months following first-line platinum- GFR determined by radioisotope Adverse events based chemotherapy. Therapy stage: Second-line with some method or 24 hr urine collection) or Paclitaxel/platinum (n=392) 3rd and 4th line permitted in the MRC 6(GFR+25) mg (where the glomerular Neurological (grade 2-4): 76/392 (20%) Trial ID: ICON4/AGO-OVAR 2.2 CTU protocol. filtration rate was assessed by the Not yet known: 15/392 Phase: Phase III Cockcroft method) or a minimum dose Haematological*: 111/392 (29%) Previous treatments of 50 mg/m2 cisplatin as a 3 hr Not yet known: 8/392 Length of follow-up: median 42 Paclitaxel plus carboplatin/cisplatin infusiion.. Infection*: 64/392 (17%) months (intervention):- Not yet known: 15/392 Last chemotherapy received: AGO – paclitaxel 185mg/m2 given as a Renal*: 31/392 (8%) Paclitaxel and carboplatin: 133/392 3 hr infusion, followed by carboplatin at Not yet known: 8/392 (34%) 5(GFR+25) mg (where GFR was the Mucositis (grade 2/3): 26/392 (7%) Carboplatin: 119/392 (30%) GFR determined by radioisotope Not yet known: 15/392 CAP: 62/392 (16%) method or 24 hr urine collection) or Nausea and vomiting (grade 2-4): Paclitaxel and cisplatin: 27/392 (7%) 6(GFR+25) mg (where the glomerular 131/392 (35%) Docetaxel and carboplatin: 7/392 (2%) filtration rate was assessed by the Not yet known: 15/392 Other platinum-based: 34/392 (9%) Cockcroft method) as a 3 hr infusion. Alopecia (grade 2-4): 322/392 (86%) Other non-platinum based: 10/392 (3%) Not yet known: 19/392 No. of cycles – MRC CTU protocol: at Number of previous chemotherapy least 6 cycles. Platinum monotherapy (n=410) lines: IRFMN: at least 3 cycles and a further Neurological (grade 2-4): 4/410 (1%) 1: 354/392 (90%) 3 cycles being administered according to Not yet known: 31/410 2*: 22/392 (6%) the results of response assessment. Haematological*: 182/410 (46%) >2*: 15/392 (4%) AGO protocol: a minimum of 6 cycles Not yet known: 16/410 Not yet known: 1/392 (<1%) and a maximum of 8 cycles, with Infection*: 53/410 (14%) response assessment done after 2nd and Not yet known: 24/410 Time since completion of last 4th cycles. Renal*: 37/410 (9%) chemotherapy (months): Not yet known: 16/410 <=12: 92/392 (23%) Length per cycle: - 21 days Mucositis (grade 2/3): 21/410 (6%) >12: 300/392 (77%) Not yet known: 31/410 * MRC CTU patients only. Control (C) Nausea and vomiting (grade 2-4): Type: Carboplatin or cisplatin 153/410 (40%)

307 Carboplatin/cisplatin (control) monotherapy Not yet known: 29/410 Last chemotherapy received: No. randomised: 410. Alopecia (grade 2-4): 95/410 (25%) Paclitaxel and carboplatin: 141/410 Route of administration: i.v. Not yet known: 28/410 (34%) Dose: - ICON4 (MRC CTU & IRFMN) Carboplatin: 128 /410 (31%) carboplatin at 5(GFR+25) mg (where * Toxic effect leading to treatment CAP: 72/410 (18%) GFR was the GFR determined by modification or interruption Paclitaxel and cisplatin: 20/410 (7%) radioisotope method or 24 hr urine reported. Docetaxel and carboplatin: 14/410 (3%) collection) or 6(GFR+25) mg (where the Other platinum-based: 30/410 (7%) glomerular filtration rate was assessed Other non-platinum based: 5/410 (1%) by the Cockcroft method) or a minimum dose of 75 mg/m2 cisplatin as a 3 hr Number of previous chemotherapy infusion. . lines: 1: 380/410 (93%) AGO: carboplatin at 5(GFR+25) mg 2*: 24/410 (6%) (where GFR was the GFR determined >2*: 6/410 (1%) by radioisotope method or 24 hr urine Not yet known: 0/410 (0%) collection) or 6(GFR+25) mg (where the glomerular filtration rate was assessed Time since completion of last by the Cockcroft method) as a 3 hr chemotherapy (months): infusion. <=12: 111/410 (27%) >12: 299/410 (73%) No. of cycles:- MRC CTU protocol: at least 6 cycles. Disease present after first-line IRFMN: at least 3 cycles and a further treatment? 3 cycles being administered according to Residual : In IRFMN protocol, but not the results of response assessment. in MRC-CTU or AGO AGO protocol: a minimum of 6 cycles Refractory - No and a maximum of 8 cycles, with response assessment done after 2nd and Age/age range of participants 4th cycles. Paclitaxel/platinum: Median 60.0 yrs Platinum monotherpay : Meddian: 59.2 Length per cycle:- 21 days yrs Comments about Characteristics intervention/control: Pacliaxel/platinum (n=392) Protocol treatments received during Age:-_ the trial <55: 127/392 (32%) 55-65: 151/392 (39%) Paclitaxel plus platinum >65: 114/392 (29%) chemotherapy (intervention) n=392) Median 60.0 yrs Carboplatin alone: 1/392 (<1%) CAP: 0 Accural by randomisation group:- Carboplatin and cisplatin*: 0 ICON MRC CTU: 266/392 (68%) Cisplatin and doxoruicin: 0 ICON IRFMN: 100/392 (26%) Cisplatin alone: 0

308 AGO: 26/392 (7%) Carboplatin in combination (non- taxane): 0 WHO performance status:- Paclitaxel and carboplatin: 314/392 0: 246/392 (63%) (80%) 1: 121/392 (31%) Paclitaxel and cisplatin: 38/392 (10%) 2-3: 25/392 (6%) Paclitaxel, carboplain and cisplatin*: 19/392 (5%) Intended platinum agent:- Paclitaxel alone: 17/392 (4%) Carboplatin: 332/392 (85%) Other: 1/392 (<1%) Cisplatin: 60/392 (15%) Treatment not received or not yet known: 2/392 (<1%) Platinum monotherapy (n=410) * Switches between platinum during Age:- treatment <55: 123/410 (23%) 55-65: 162/410 (40%) Carboplatin/cisplatin monotherapy >65: 125/410 (30%) (control ) (n=410) Median: 59.2 yrs Carboplatin alone: 292/410 (71%) CAP: 68/410 (17%) Accural by randomisation group:- Carboplatin and cisplatin*: 18/410 (4%) ICON MRC CTU: 270/410 (66%) Cisplatin and doxoruicin: 12/410 (3%) ICON IRFMN: 113/410 (28%) Cisplatin alone: 8/410 (2%) AGO: 27/410 (7%) Carboplatin in combination (non- taxane): 3/410 (<1%) WHO performance status:- Paclitaxel and carboplatin: 1/410 (<1%) 0: 262/410 (64%) Paclitaxel and cisplatin: 0 1: 122/410 (30%) Paclitaxel, carboplain and cisplatin*: 0 2-3: 26/410 6%) Paclitaxel alone: 4/410 (1%) Other: 3/410 (<1%) Intended platinum agent:- Treatment not received or not yet Carboplatin: 341/410 (83%) known: 1/410 (<1%) Cisplatin: 69/410 (17%) * Switches between platinum during treatment Inclusion/exclusion criteria: Eligible patients had: relapsed epithelial Actual study treatment ovarian cancer requiring chemotherapy; administration: previously received platinum-based 580 (72%) patients in the two treatment chemotherapy; been treatment-free for groups received a minimum of 6 cycles more than 6 months (>12 months in the of chemotherapy. Reasons for not ICON IRFMN group) and no completing the 6 cycles of treatment concomitant or previous malignant were: disease progression or death (109 disease likely to interfere with treatment [56%]), toxic effects during treatment or outcomes. (77 [39%]), and patient’s preference (9 [5%]). Only 3 patients received no Eligibility criteria differed slightly assigned chemotherapy. 123 (41%) of across the three protocols. Patients in 300 patients in the control group and

309 the MRC CTU protocol trial were 129 (40%) of 322 in the paclitaxel and permitted to have had more than one carboplatin group received more than line of previous chemotherapy, whereas 90% of the planned total dose, and 54 those randomised in the IRFMN and (18%) of 300 and 76 (23%) of 330 AGO protocols must have had only one respectively, received more than 90% of previous line. Measurable disease was the dose intensity of carboplatin. The required for patients randomised in the relevant values for paclitaxel in the IFRMN protocol, but not in the MRC paclitaxel and carboplatin group were CTU or AGO protocols. The diagnosis 266 (71%) of 376 for total dose and 160 of relapsed disease at entry into the trial (43%) of 376 for total dose intensity. in 18 patients in the MRC CTU protocol was based on raised CA125 Protocol deviations: concentrations. Patients randomised None reported. into the AGO protocol must have previously received cisplatin plus paclitaxel or carboplatin plus paclitaxel; all patients in ICON 4 (MRC CTU & IRFMN) were required to have had previous platinum-based chemotherapy, with or without paclitaxel.

Comments about participants: Randomisation was undertaken independently in the 3 protocols. The randomisation ratio between treatment groups was 1:1. In the 2 ICON protocols (MRC CTU & IRFMN) stratification was undertaken by centre, age, last chemotherapy received, time since completion of last chemotherapy and intended platinum treatment. The platinum treatment to be used had to be specified prior to randomisation. In the AGO protocol, stratification was done by time since completion of last chemotherapy and whether the patient had undergone secondary debulking surgery. Pretreatment data was collected at the time of randomisation. Treatment data was collected at each cycle of treatment. Further follow-up data was collected every 3 months for the first 2 years and 6 months thereafter. Quality of life data were collected for all patients in the

310 MRC CTU and AGO protocols using the European Organization for Research and Treatment of Cancer quality of life questionnaire, EORTC QLQ-C30 (version 2.0). Patients in the MRC CTU protocol complete the questionnaire at baseline, before each cycle of protocol treatment, and at follow-up visits at 6,9 and 12 months. Patients in the AGO protocol completed the questionnaire at baseline, before the second and third cycles of protocol treatment, at the end of protocol treatment, and every 3 months for 1 year after the end of protocol treatment..

Results Outcome 1 Outcome 2 Outcome 3 Outcome: Survival (the time from randomisation to death from Progression free survival (time from randomisation to first Outcome: Tumour response (not defined) any cause) (patients known to still be alive at the time of the appearance of progressive disease or death from any cause analysis were censored at the time of their last follow-up) (patients known to be alived and without progressive disease at Complete or partial response:- the time of analysis were censored at their last follow-up). Paclitaxel/platinum:: 78/119 (66%) 530/802 patients died. HR = 0.82(95% CI: 0.69, 0.97); p=0.02. Platinum monotherapy: 69/128 (54%) 717/802 developed progressive disease or died. HR = 0.76 (95% CI: Difference: 12% (95% CI for difference: -0.1%, 24%); p=0.06. Absolute difference in 2-year survival:- 0.66, 0.89); p=0.0004. Paclitaxel/platinum:: 57% Absolute difference in 1-year progression free survival:- Platinum monotherapy: 50% Paclitaxel/platinum,: 50% (95% CI for difference: 1, 12) Platinum monntherapy:: 40% (95% CI for difference: 4,15) Median survival (months):- Paclitaxel/platinum: 29 months Absolute difference in median progression-free survival:- Platinum monotherapy : 24 months Paclitaxel/platinum: 12 months (95% CI for difference: 1, 9). Platinum monntherapy:: 9 months (95% CI for difference: 1, 4).

Outcome 4 Outcome 5 Outcome: QoL assessment (EORTC QoL-C30) Outcome: Overall survival in sub-groups

482/536 (90%) of patients completed a questionnaire at baseline, Paclitaxel/platinum (n=392) before treatment. Most patients had little or no functional difficulties Randomised group:- and few moderate or severe sysmptoms at baseline. In the first 6 MRC-CTU: 243/266 months after randomisation the worst scores and the areas under the IRFMN: 80/100 curve did not differ betwwen treatment groups for all five functional AGO: 23/26 scales (physical, role, emotional, cognitive and social), for eight of the (p=0.93)

311 nine symptom scales (fatigue, pain, dyspnoea, insomnia, appetite loss, constipation, diarrhoea and financial difficulties) or global health. Age (years) The exception was nausea and vomiting, which was worse for the 55: 114/127 patients receiving conventional chemotherapy (p=0.0014 for worst 55 - 65: 135/151 score, p=0.005 for area under the curve). >65: 97/114 (p=0.08)

WHO performance:- WHO 0: 212/246 WHO >0: 134/146 (p=0.66)

Intended platinum agent:- Carboplatin: 294/332 Cisplatin: 52/60 (p=0.17)

Previous lines of chemotherapy:- 1: 310/354 2: 22/22 >2: 14/15 (p=0.19)

Time since completion of last chemotherapy cycle (months):- =/< 12: 90/92 > 12: 256/300 (p=0.87)

Previous exposure to taxanes:_ No: 195/223 Yes: 151/169 (p=0.62)

Platinum monotherapy (n=410) Randomised group:- MRC-CTU: 253/270 IRFMN: 94/113 AGO: 24/27 (p=0.93)

Age (years) 55: 111/123 55 - 65: 146/162 >65: 114/125 (p=0.08)

312

WHO performance:- WHO 0: 232/262 WHO >0: 139/148 (p=0.66)

Intended platinum agent:- Carboplatin: 303/341 Cisplatin: 68/69 (p=0.17)

Previous lines of chemotherapy:- 1: 343/380 2: 22/24 >2: 6/6 (p=0.19)

Time since completion of last chemotherapy cycle (months):- =/< 12: 109/111 > 12: 262/299 (p=0.87)

Previous exposure to taxanes:_ No: 214/235 Yes: 157/175 (p=0.62)

313 Trial: Cantu et al.(28)

Study Details and Design Participant Details Intervention Details Follow-up/withdrawals and Adverse Conclusion and Comments Events Author: Cantu 2002 (28) Number randomised Intervention (I): Withdrawals from intervention (I) Authors’ conclusions 97 (paclitaxel n=50; CAP n=47) Type: - Paclitaxel 3 lost to follow up Rechallenge with either single-agent Objective: To assess the activity, No. randomised:- 50. paclitaxel or platinum-based efficacy and tolerability of single-agent Disease type: Epithelial Route of administration : i.v. Withdrawals from control (C) chemotherapy is effective in this patient paclitaxel and a platinum-containing- Disease stage: Advanced Dose:- 175mg/m2 over 3 hours. Pre- None. population. Preliminary results suggest regimen in previously treated patients Occurrence of secondary spread:- In medication: dexamethasone 40mg in that single-agent paclitaxel may not be with recurrent ovarian cancer. some patients, most had regional spread. 20mg doses 12 & 6 hours before Adverse events as active as platinum-based paclitaxel, cimetidine 300mg IV and chemotherapy in recurrent ovarian Trial ID: N/A Therapy stage: Second-line chlorpheniramine 10mg IV both 30min Paclitaxel (n = 47) cancer. Larger randomised trials are Phase: Phase II before treatment. Leukopenia (grade 3-4): 2 (4%): P = needed. Previous treatments No. of cycles - 6 cycles planned 0.001 Chemotherapy - platinum-based 1st line. Length per cycle: - 21 days Neutropenia (grade 3-4): 6 (13%): P = Comments All patients had previously been treated 0.009 The inclusion of a cross-over for non- with 2 or 3 chemotherapy regimens. Control (C) Thrombocytopenia (grade 3-4): 0 (0%): responders compromises study design. It Type: Cyclophosphamide, doxorubicin P = 0.012 is not stated whether assessors were Disease present after first-line & cisplatin (CAP)) Nausea & vomiting (grade 2-3): 8 blinded to allocation and it is not clear treatment? No. randomised: 47. (17%): P = 0.004 whether randomisation was appropriate Residual : Yes Route of administration: i.v. Alopeica: 41 (87%): P = 0.010 or allocation concealment adequate. Refractory - Yes Dose: - cyclophosphamide 500mg/m2, Allergic reactions: 7 (15%): P = 0.085 doxorubicin 50mg/m2, cisplatin Sensory neuropathy >/= grade 2: 5 Age/age range of participants 50mg/m2. (11%): P = 0.002 Paclitaxel No of cycles:- 6 cycles Myalgia (grade 2): 9 (19%): P = 0.025 Median: 54.8 years (range: 35-72) Length per cycle:- 21 days Other (cardiac toxicity, renal toxicity, CAP stomatitis): 4 (8%): P > 0.05 Median: 53 years (range: 27-69) Comments about No. cycles delivered: median (main intervention/control: reason for stopping treatment was Characteristics Treatment administration regulated by disease progression): 6 evaluation of blood cell count before Paclitaxel (n = 47) start of treatment cycle. Treatment given CAP (n = 47) Age if granulocyte count =/> 2000uL and Leukopenia (grade 3-4): 16 (34%): P = Median: 54.8 platelet count was =/> 100,000uL. 0.001 Range: 35-72 Otherwise treatment was delayed 1 Neutropenia (grade 3-4): 17 (36%): P = < 45: 7/47 (15%) week. 0.009 46-55: 23/47 (49%) When chemotherpay failed patients Thrombocytopenia (grade 3-4): 6 >/= 56: 17/47 (36%) were crossed over to the alternative (13%): P = 0.012 intervention. 23 crossed from CAP to Nausea & vomiting (grade 2-3): 24 Serous histology: 31/47 (66%) Paclitaxel, 30 from paclitaxel to CAP. It (51%): P = 0.004 Other histology: 16/47(34%) is not stated at what point cross-over Alopeica: 28 (60%): P = 0.010 took place or how treatment failure was Allergic reactions: 1 (2%): P = 0.085 Poor differentiation: 35/47 (74%) defined. Sensory neuropathy >/= grade 2: 3

314 Actual study treatment (6%): P = 0.002 Disease site administration: Myalgia (grade 2): 2 (4%): P = 0.025 Pelvis: 17/47 (36%) Not reported. Other (cardiac toxicity, renal toxicity, Abdomen: 5/47 (11%) stomatitis): 1 (2%): P > 0.05 No. cycles delivered: median (main Retroperitoneal nodes: 5/47 (11%) reason for stopping treatment was Multiple: 16/47 (34%) disease progression): 6 Other: 3/47 (6%)

Residual bulk > 2cm: 27 (57%)

Median time since last chemo (months): 30.2 Time since last chemo > 2yrs: 23/47 (49%)

Prior therapy: Cisplatin alone: 6/47(13%) Carboplatin alone: 7/47 (15%) Carboplatin combination: 34/47 (72%) Cisplatin + Doxorubicin: 31/47 (66%) Dose of previously administered carboplatin/cisplatin: median (range) mg/m2 Cisplatin: 553mg/m2 (240-1000) Carboplatin: 2130mg/m2 (600-2400)

CAP (n = 47) Age Median: 53 Range: 27-69 < 45: 10/47 (21%) 46-55: 15/47 (32%) >/= 56: 22/47 (47%)

Serous histology: 31/47 (66%) Other histology: 16/47 (34%)

Poor differentiation: 28/47 (60%)

Disease site: Pelvis: 14/47 (30%) Abdomen: 5/47 (11%)

Retroperitoneal nodes: 10/47 (21%)

315 Multiple: 14/47 (30%) Other: 2/47 (4%)

Residual bulk > 2cm: 16/47 (34%)

Median time since last chemo (months): 38.8 Time since last chemo > 2yrs: 21/47 (45%)

Prior therapy: Cisplatin alone: 9/47 (19%) Carboplatin alone: 5/47 (11%) Carboplatin combination: 30/47 (64%) Cisplatin + Doxorubicin: 26/47 (55%) Dose of previously administered carboplatin/cisplatin: median (range) mg/m2 Cisplatin: 530mg/m2 (300-1000) Carboplatin: 2400mg/m2 (2400-2400)

Inclusion/exclusion criteria 1.Histologically proven ovarian carcinoma. 2.Disease recurred or progressed after 12 months since end of 1st line therapy including cisplatin or carboplatin. 3.Measurable disease (not defined) 4.WHO performance status /= 2000/uL, platelet count >/= 100,000/uL) 6.Adequate renal, hepatic and cardiac function (not defined).

Results Outcome 1 Outcome 2 Outcome 3 Outcome: Overall Survival. Defined as time from randomisation Outcome: Tumour responses (ECOG criteria) to death by any cause. Progression-free survival defined as interval from randomisation to 1st appearance of progressive Paclitaxel (n = 47) disease (not defined) CR: 8/47 (17%) PR: 13/47 (28%) Times of follow-up: median=49 months (range: 40-54) Total: 21/47 (45%)

316 Paclitaxel (n=47) SD: 10/47 (21%) Survival PD: 16/47 (34%) Median: 25.8 months At 4 year follow up: Responses after cross-over following no response to paclitaxel Deaths: 34/47 (with progression: 31, without progression: 3) treatment: i.e. response to CAP Still alive: 13 Total number crossed over: 30 HR = 0.70 (95% CI: 0.42, 1.15, P = 0.160). Adjusted for residual CR: 7/30 (23%) tumour, time since end last therapy & age HR = 0.58, (95% CI: 0.34, PR: 7/30 (23%) 0.98, P = 0.043) Total: 14/30 (46%) SD: 7/30 (23%) Multivariate analysis: mortality risk (calculated using Cox PD: 9/30 (30%) proportional hazards regression model) Age: CAP (n=47) 50: HR = 1.70 (95% CI: 0.90, 3.19, P = 0.10) PR: 12/47 (25%) Total: 26/47 (55%) Residual disease: SD: 14/47 (30%) 2cm: HR = 0.49 (95% CI: 0.28, 0.86, P = 0.013) Responses after cross-over following no response to CAP Time since end last chemo treatment: i.e. response to paclitaxel 2yrs: HR = 0.62 (95% CI: 0.36,1.08, P = 0.091) CR: 3/23 (13%) Arm: PR: 2/23 (9%) Paclitaxel: HR = 1 Total: 5/23 (22%) CAP: HR = 0.58 (95% CI: 0.35, 0.98, P = 0.043) SD: 7/23 (30%) PR: 11/23 (48%) Progression-free survival: Median: 9 months Alive at time of follow up with progression: 10 Alive no progression (extrapolated): 3 Died 34 Died without progression 4

HR = 0.65 (95% CI: 0.41, 1.02, P = 0.08) Adjusted for residual tumour, time since end last therapy & age HR = 0.60 (95% CI: 0.37, 0.97, P = 0.038) Multivariate analysis: progression risk Age: 50: HR = 1.26 (95% CI: 0.72, 2.20, P = 0.412)

Residual disease: 2cm: HR = 0.81 (95% CI: 0.50, 1.33, P = 0.420)

317

Time since end last chemo 2yrs: HR = 0.76 (95% CI: 0.47, 1.23, P = 0.270)

Arm Paclitaxel: HR = 1 CAP: HR = 0.60 (95% CI: 0.37, 0.97, P = 0.038)

CAP (n=47) Survival:- Median: 34.7 months At 4 year follow up: Deaths: 27/47 (with progression: 23, without progression: 4) Still alive: 20 HR = 0.70 (95% CI: 0.42, 1.15, P = 0.160). Adjusted for residual tumour, time since end last therapy & age HR = 0.58, (95% CI: 0.34, 0.98, P = 0.043)

Multivariate analysis: mortality risk (calculated using Cox proportional hazards regression model) Age: 50: HR = 1.70 (95% CI: 0.90, 3.19, P = 0.10)

Residual disease: 2cm: HR = 0.49 (95% CI: 0.28, 0.86, P = 0.013)

Time since end last chemo 2yrs: HR = 0.62 (95% CI: 0.36,1.08, P = 0.091)

Arm: Paclitaxel: HR = 1 CAP: HR = 0.58 (95% CI: 0.34, 0.98, P = 0.043) Progression-free survival: Median: 15.7 months Alive at time of follow up with progression: 14 Alive no progression (extrapolated): 6 Died 27 Died without progression 3

HR = 0.65 (95% CI: 0.41, 1.02, P = 0.08) Adjusted for residual tumour, time since end last therapy & age HR =

318 0.60 (95% CI: 0.37, 0.97, P = 0.038)

Multivariate analysis: progression risk Age: 50: HR = 1.26 (95% CI: 0.72, 2.20, P = 0.412)

Residual disease: 2cm: HR = 0.81 (95% CI: 0.50, 1.33, P = 0.420)

Time since end last chemo 2yrs: HR = 0.76 (95% CI: 0.47, 1.23, P = 0.270)

Arm Paclitaxel: HR = 1 CAP: HR = 0.60 (95% CI: 0.37, 0.97, P = 0.038)

319 Trial: Piccart et al.(30)

Study Details and Design Participant Details Intervention Details Follow-up/withdrawals and Adverse Conclusion and Comments Events Author: Piccart 2000(30)] Number randomised Intervention (I): Withdrawals from intervention: Authors’ conclusions 86 randomised (paclitaxel n=41; Type: - Paclitaxel 2: both were ineligible. Of the total of 4 Single-agent oxaliplatin at 130mg/m2 Objective oxaliplatin n=45) No. randomised:- 41. patients who were ineligible, 1 had every 3 weeks is active with moderate To evaluate the efficacy of oxaliplatin in Route of administration : i.v. relapsed disease > 1 year after last toxicity in patients with relapsing progressive ovarian cancer Disease type: Epithelial Dose:-175mg/m2/3 weeks over 3 hours platinum-based chemotherapy regiomen cisplatin/carboplatin-pretreated patients and to analyze the safety profile Disease stage: Advanced No. of cycles - median: 6 (range: 1 –8) and 3 had no measurable lesions at study advanced ovarian cancer. and impact of single-agent paclitaxel Occurrence of secondary spread:- Yes Length per cycle: - 21 days entry, according to independent and oxaliplatin on QoL, time to in all patients. radiologic experts Comments progression and survival. Control (C) It is not clear what the length of follow- Therapy stage: Second-line Type: Oxaliplatin Withdrawals from control up is and this may have implications for Trial ID: N/A No. randomised: 45 3: 2 were ineligible and 1 died of the usefulness of survival data (see Phase: Phase II Previous treatments Route of administration: i.v. massive pulmonary thromboembolism outcomes). At least 1 and not more than 2 platinum- Dose: 130mg/m2/3 weeks over 2 hours- 6 days after 1st cycle (not related to Length of follow-up: unclear baed chemotherapy regimens with last No. of cycles:- median: 4 (range 1-8) treatment). Of the total of 4 patients who Comments on the results: regimen featuring carboplatin or Length per cycle:- 21 days were ineligible, 1 had relapsed disease > The overall survival results are based on cisplatin at therapeutically adequate and 1 year after last platinum-based estimates, without further explanation of potentially active doses chemotherapy regiomen and 3 had no how these estimatew were derived. Comments about measurable lesions at study entry, Given that it is unclear what the length Disease present after first-line intervention/control: according to independent radiologic of follow-up was it is difficult to treatment? Paclitaxel: Recommended experts ascertain how many patietns were Residual : Yes premedication: oral dexamethasone censored at follow-up and what the Refractory - Yes (20mg), 12 and 6 hours before infusion, Adverse events impact of this on survival figures may diphenhydramine 50mg IV, cimetidine Severe (Grade 3/4 NCIC) hematologic have been. Age/age range of participants 300mg or ranitidine 50mg IV 30 and nonhematologic toxicity by patient Paclitaxel: Median 62.0yrs (range 37- minutes before PXL Paclitaxel (n = 41) 81yrs) Oxaliplatin: Recommended Hematologic Oxaliplatin: Median 59.0 yrs (range28- premedication 5HT3 with a single dose Neutropenia 71yrs) of corticosteroid (eg dexamethasone Grade 3: 6/41 (15%) 20mg) Grade 4: 3/41 (7%) Characteristics Paclitaxel (n =41): Regime to be continued until disease Anemia Age: progression, unacceptable toxicity, or Grade 3: 0/41 median: 62 patient refusal. Initial doses could be Grade 4: 1/41 (2%) range: 37-81 reduced in subsequent cycles or cycles could be delayed by 1 or 2 weeks Thrombocytopenia WHO performance status depending on toxicity. When dose Grade 3: 0/41 0-1: 35/41 (85%) reduction was required no subsequent Grade 4: 0/41 2: 6/41 (15%) dose escalation was allowed. Dose reduction could be cumulative from Liver function FIGO initial stage cycle to cycle if required by AST

320 I: 5/41 (12%) hematological or nonhematological Grade 3: 0/41 II: 2/41 (5%) toxicities but doses could not fall below Grade 4: 0/41 III: 26/41 (63%) established minimum dose per cycle IV: 8/41 (20%) (90mg/m2 PXL & 75mg/m2 OXA), ALT patients who required lower doses than Grade 3: 2/41 (5%) Histologic subtype after central the minimum went off-study. Grade 4: 0/41 review serous: 17/41 (41%) CR defined as disappearance of all Gastrointestinal other: 24/41 (59%) known disease, without appearance of Nausea new lesions, lasting more than 4 weeks. Grade 3: 1/41 (2%) Prior chemotherapy An elevated CA125 serum level had to Grade 4: n/a 1 regimen: 30/41 (73%) regain normal levels for a response to 2 regimens: 11/41 (27%) qualify as complete. Vomiting PR was defined as a decrease of at least Grade 3: 1/41 (2%) Time since last chemotherapy 50% of the sum of the products of of the Grade 4: 0/41 0-6 months (platinum-refractory): 31/41 largest perpendicular diameters of all (76%) measurable lesions being confirmed by a Diarrhea 6-12 months (platinum-sensitive): 10/41 further observation no less than 4 weeks Grade 3: 0/41 (24%) later without any new lesions. Grade 4: 0/41 No change was defined in the case of No. of involved sites bidimensional lesions as a decrease of at Neurosensory 0: 0/41 less than 50% and an increase of less Grade 3: 3/41 (7%) 1: 22/41 (54%) than 25% in the sum of the products of Grade 4: n/a 2: 17/41 (41%) the largest perpendicular dimensions of 3: 2/41 (5%) all measurable lesions. Other >3: 0/41 PD was defined as an increase of at least Lethargy 25% in the sum of the products of the Grade 3: 3/41 (7%) Involved site largest perpendicular diameters of Grade 4: n/a pelvi-peritoneum: 30/41 (73%) measurable lesions or the appearance of lymph nodes: 15/41 (37%) a new lesion. The occurrence of positive Pain lung: 2/41 (5%) cytology pleural effusion or ascites was Grade 3: 5/41 (12%) liver: 7/41 (17%) also considered as PD. Grade 4: 0/41 other: 7/41 (17%) Quality of Life was assessed using the European Organisation for Research and Oxaliplatin (n = 45) CA125 level Treatmetn of Cancer (QLQ-C30) Hematologic Median, IU/mL: 222 questionnaire Neutropenia range, IU/mL: 8 - 8,165 Grade 3: 0/45 < 20 IU/mL: 2/41 (5%) Dose modifications Grade 4: 0/45 >/= 20 IU/mL: 29/41 (71%) Exposure to treatment Paclitaxel (n = 41) Anemia Regimens containing Grade 3: 1/45 (2%) Cisplatin: 16/41 (39%) No of cylces administered: Grade 4: 2/45 (4%) Median dose, mg: 442 median: 197 Range, mg: 223-456 No per patient: 6 Thrombocytopenia Carboplatin: 21/41 (51%) range: 1 -8 Grade 3: 2/45 (4%)

321 Median dose, mg: 1,970 Grade 4: 0/45 Range, mg: 652 - 3,600 Cumulative dose per patients, mg/m2 Both cisplatin & carboplatin: 4/41 median: 1028 Liver function (10%) range; 173 - 1453 AST Grade 3: 0/45 Oxaliplatin (n = 45) dose intensity mg/m2/week Grade 4: 0/45 Age: median: 57 median: 59 median at cycle 3: 58 ALT range: 28-71 Grade 3: 0/45 No patients given relative dose Grade 4: 0/45 WHO performance status intensity 0-1: 38/45 (84%) < 80%: 0 Gastrointestinal 2: 7/45 (16%) 80 - 95%: 11 Nausea 95-100%: 30 Grade 3: 2/45 (4%) FIGO initial stage Grade 4: n/a I: 7/45 (16%) Exposure to treatment II: 1/45 (2%) Oxaliplatin (n = 45) Vomiting III: 29/45 (64%) Grade 3: 3/45 (7%) IV: 8/45 (18%) No of cylces administered: Grade 4: 0/45 median: 181 Histologic subtype after central No per patient: 4 Diarrhea review range: 1 - 8 Grade 3: 2/45 (4%) serous: 33/45 (73%) Grade 4: 0/45 other: 12/45 (27%) Cumulative dose per patients, mg/m2 median: 504 Neurosensory Prior chemotherapy range: 132 - 993 Grade 3: 4/45 (9%) 1 regimen: 29/45 (64%) Grade 4: n/a 2 regimens: 16/45 (36%) dose intensity mg/m2/week median: 42 Other Time since last chemotherapy median at cycle 3: 42 Lethargy 0-6 months (platinum-refractory): 32/45 Grade 3: 3/45 (7%) (71%) No patients given relative dose Grade 4: n/a 6-12 months (platinum-sensitive): 13/45 intensity (29%) < 80%: 4 Pain 80 - 95%: 15 Grade 3: 2/45 (4%) No. of involved sites 95-100%: 26 Grade 4: 0/45 0: 1/45 (patient with no target lesion at study entry) 1: 30/45 (67%) 2: 11/45 (25%) 3: 2/45 (4%) >3: 1/45 (2%)

Involved site pelvi-peritoneum: 25/45 (56%)

322 lymph nodes: 13/45 (29%) lung: 2/45 (4%) liver: 15/45 (33%) other: 7/45 (16%)

CA125 level Median, IU/mL: 316 range, IU/mL: 7 - 14,350 < 20 IU/mL: 5/45 (11%) >/= 20 IU/mL: 32/45 (71%)

Regimens containing Cisplatin: 19/45 (42%) Median dose, mg: 440 Range, mg: 223 - 478 Carboplatin: 21/45 (47%) Median dose, mg: 1,888 Range, mg: 1,402 - 3,568 Both cisplatin & carboplatin: 5/45 (11%)

Inclusion/exclusion criteria

1. Histologically or cytologically proven metastatic ovarian carcinoma excluding Brenner or borderline tumour, low potential (grade 0 tumours), squamous cell carcinoma, and granulosa theca cell tumours. 2. Progressed or stabilised after prior treatment with relapse observed within 1 year of last platinum-based chemotherapy. 3. Received at least 1 and no more than 2 chemotherapeutic regimens, with the last regimen featuring carboplatin or cisplatin at therapeutically adequate and potentially active doses. 4. Those with progressive or stable disease must have received at least 2 or 4 consecutive cycles, respectively and must have 1 bidimensionally measurable lesion by CT scan or MRI with at least 1 diameter of greater than or equal to 2 cm.

323 5. 18 years old or over 6. WHO performance status of 0 to 2 7. Estimated life expectancy of >/= 12 weeks 8. Baseline blood lab criteria: neutrophil count>/= 1.5 x 10 to the 9 cells/L; platelet count>/= 100 x 10 to the 9 platelts/L; creatinine level /= grade 2 (NCIC criteria) or any other serious illness

Comments about participants: patients were stratified by centre, performance status (0, 1 or 2), platinum free interval (0-6 months v 6 - 12 months) & number of previous platinum-based Results Outcome 1 Outcome 2 Outcome 3 Outcome: Survival (not defined) Outcome: Tumour response Outcome: Quality of life (EORTC QoL-C30)

Paclitaxel group (n=41):- Paclitaxel group (n=41) Completion and follow-up of QoL data: median duration of response; 33 weeks 66 patients completed at baseline

324 median time to progression: 14 weeks CR: 0 ( 1 patient had a CR in a pelvic mass) 47 at end of 2nd cycle estimated median progression-free survival: 14 weeks PR : 7 (17%) (95% CI: 7%, 32%) 5/7 had a sustained CA125 serum 31 at end of 4th cycle median time to treatment failure: 13 weeks level decrease, 1 did not show a significant decrease of CA125 and 1 Number of deaths: 25 was not assessable by this method Results: Median overall survival estimated: 37 weeks Platinum-sensitive: 2/10 (20%) Paclitaxel: mean QoL scores increased by 10 points between baseline Platinum-refractory: 5/31 (16%) and cycle 4, regardless of time of study withdrawal. Oxaliplatin group (n=45):- 2 prior chemotherapy regimens: 3/ 7 median duration of response; 31 weeks NC: 18 (44%) Oxaliplatin: mean QoL scores decreased by < 10 points through median time to progression: 12 weeks PD: 14 (34%) cycle 2 then tended to return to baseline values for the majority of estimated median progression-free survival: 12 weeks Not assessable: 2 (5%) patients after 2 cycles. median time to treatmetn failure: 12 weeks Total: 41 (100%) Number of deaths: 20 Median overall survival estimated: 42 weeks Oxaliplatin group (n=45) CR: 0 PR : 7 (16%) (95% CI: 7%, 29%) 4/7 had a sustained Ca125 serum level decrease, 2 did not show a significant decrease in Ca125 amd 1 was not assessable by this method. Platinum-sensitive: 5/13 (38%) Platinum-refractory: 2/32 (6%) 2 prior chemotherapy regimens: 4/16 NC: 20 (44%) PD: 15 (33%) Not assessable: 3 (7%) Total: 45 (100%)

325 Trial: Omura et al.(32)

Study Details and Design Participant Details Intervention Details Follow-up/withdrawals and Conclusion and Comments Adverse Events Author: Omura 2003 (32) Number randomised Intervention (I): Withdrawals from intervention (I) Authors’ conclusions 477 (250 mg/m2 paclitaxel n=188; 175 Type: - Paclitaxel 22 patients from this group randomised Paclitaxel exhibits a dose effect with Objective mg/m2 paclitaxel n=184) No. randomised:- 188 but did not receive treatment as they regard to response rate, but there is more To determine whether increasing the Route of administration : i.v. were not deemed eligible. Total toxicity and no survival benefit to justify dose of paclitaxel increases the Disease type: Epithelial Dose:- 250mg/m2 given as a 24hr i/v exclusions before trial: 42: : paclitaxel 250mg/m2 plus filgrastim. probability of response, progression-free Disease stage: Advanced infusion. Patients were also assigned to inappropriate primary disease site (n = Doubling the dose of filgrastim from 5 survival, or overall survival in women Occurrence of secondary spread:- Not 5 or 10ug/kg of filgrastim/day subcut. 34), improper prior treatment (n = 7), to 10ug/kg did not reduce the who have persistent or recurrent ovarian reported. G-CSF was started on 3rd after inadequately documented histology (n = probability of neutropenic fever after cancer, and whether doubling the dose paclitaxel administration and continued 3), 2nd primary cancer (n = 3), high dose paclitaxel. of prophylactic filgrastim accompanying Therapy stage: Second-line through the neutrophil nadir until the inadequate documentation of recurrence the higher paclitaxel dose decreases the absolute neutrophil count was at leasr (n = 2), borderline tumour histology (n = Comments frequency of neutropenic fever. Previous treatments 10,000/uL 1) and wrong stage (n = 1) Not stated Chemotherapy; no more than 1 platinum No. of cycles - Protocol specified 6 but which group patients had been The inclusion criteria were altered in 2 Trial ID: N/A containing 1st line regimen. treatment could continue indefinitely if randomised to. critical respects during recruitment, Phase: Phase III clinical progression or excess toxicity Withdrawals from control (C) including patients without measurable Disease present after first-line were not experienced. 20 patients from this group randomised disease who were subsequently Length of follow-up: 36 months treatment? Length per cycle: - 21 days but did not receive treatment as they excluded from tumour response analysis Residual : Yes were not deemed eligible. Total and including patients who were Refractory - Yes Control (C) exclusions before trial: 42: : platinum sensitive. This raises questions Type: Paclitaxek inappropriate primary disease site (n = about the screening of patients for the Age/age range of participants No. randomised: 184 34), improper prior treatment (n = 7), trial and application and appropriateness 250 mg paclitaxel: Route of administration: i.v. inadequately documented histology (n = of the inclusion criteria. Median: 62 years (range: 24-80) Dose: -175mg/m2 given as a 24hr i/v 3), 2nd primary cancer (n = 3), . 175 mg paclitaxel: infusion inadequate documentation of recurrence Median: 60 years (range: 23-88) No. of cycles:- Protocol specified 6 but (n = 2), borderline tumour histology (n = treatment could continue indefinitely if 1) and wrong stage (n = 1). Not stated Characteristics clinical progression or excess toxicity which group patients had been 250 mg paclitaxel (n=166) were not experienced. randomised to. Age: Length per cycle:- 21 days median: 62 Adverse events Range: 24-80 Comments about < 40: N (%): 9 (5) intervention/control: 250mg/m2 (also given granulocyte 40-49: N (%): 30 (18) Patients were initially (1st year of colony-stimulating factor) 50-59: N (%): 31 (19) recruitment) randomised to 3 groups. 60-69: N (%): 64 (39) Assignment to 135mg/m2 group ceased Neureopenic fever (defined as grade 4 >/= 70: N (%): 32 (19) after the accession of 77 patients (88% neutropenia (absolute neutrophil count < deemed eligible) who are only partially 500/uL) and grade 2 (>/= 38.1C) or histology (cell type) reported. above fever): 19% (n = not reported) serous: N (%): 100 (60) (19% on 5ug/kg filgrastim, 18% on

326 endometrioid: N (%): 22 (13) Actual study treatment 10mg/kg filgrastim). (95% CI for mucinous: N (%): 7 (4) administration: differences between 2 doses: -11%, clear cell: N (%): 11 (7) Paclitaxel 250 mg/m2: 13%) other: N (%): 26 (16) 55% of patients received 6 or more cycles of therapy. Over the 1st 6 courses All other adverse events were grade 3/4 Measurable disease: N (%): 134 (81) of treatments approx. 70% of the (N = not reported) planned ideal dose was delivered to Anemia: 15%, P = 0.102 Platinum Sensitivity patients. Although dose reductions Thrombocytopenia: 15%, P = 0.009 Resistant: N (%): 132 (79) occurred in both arms a difference in the Nausea & Vomiting: 10%, P = 0.211 Sensitive: N (%): 34 (21) total dose and dose intensity was Neuropathy: 16%,P = 0.024 maintained during the 1st 6 courses of Myalgia/arthralgia: 10%, P = 0.022 GOG performance status therapy. 0: N (%): 88 (53) 175mg/kg 1: N (%): 63 (38) Paclitaxel 175 mg/m2: Neureopenic fever (defined as grade 4 2: N (%): 15 (9) 58% of patients received 6 or more neutropenia (absolute neutrophil count < cycles of therapy. Over the 1st 6 courses 500/uL) and grade 2 (>/= 38.1C) or Race: of treatments approx. 76% of the above fever): 22% (n = not reported) Black: N (%): 7 (4) planned ideal dose was delivered to Hispanic: N (%): 6 (4) patients. Although dose reductions All other adverse events were grade 3/4 White: N (%): 146 (88) occurred in both arms a difference in the (n = not reported) Other/not specified: N (%): 7 (4) total dose and dose intensity was Anemia: 7%, P = 0.102 maintained during the 1st 6 courses of Thrombocytopenia: 5%, P = 0.009 Paclitaxel 175mg/m2 (n = 164) therapy. Nausea & Vomiting: 5%, P = 0.211 Age: Neuropathy: 7%,P = 0.024 median: 60 % planned ideal dose delivered over 1st Myalgia/arthralgia: 3%, P = 0.022 Range: 23-88 6 cycles: < 40: N (%): 12 (7) 250mg/m2: 70% 40-49: N (%): 25 (15) 175mg/m2: 76% 50-59: N (%): 46 (28) 60-69: N (%): 53 (32) Outcome definitions: >/= 70: N (%): 28 (17) Tumour responses: Complete response (defined as histology disappearance of all gross evidence of serous: N (%): 105 (63) disease for at least 4 weeks. endometrioid: N (%): 17 (10) Partial response (defined as a reduction mucinous: N (%): 0 (0) =/> 50% in the product of perpendicular clear cell: N (%): 8 (5) measurements of each lesion for at least other: N (%): 34 (21) 4 weeks.

Measurable disease: N (%): 131 (80)

Platinum Sensitivity Resistant: N (%): 125 (76) Sensitive: N (%): 39 (24)

327 GOG performance status 0: N (%): 89 (54) 1: N (%):65 (40) 2: N (%): 10 (6)

Race: Black: N (%): 4 (2) Hispanic: N (%): 5 (3) White: N (%): 149 (91) Other/not specified: N (%): 6 (4)

Inclusion/exclusion criteria 1. Histologically confirmed epithelial ovarian cancer 2. No more than 1 prior platinum-based regimen 3. No prior taxane therapy 4. Platinum-resistant disease defined as progression during 1st line platinum treatment or within 6 months of completing therapy, a best response of stable disease after 6 courses of platinum, or stable disease with a rising CA-125 while on platium or recurrence, were included.* see comments on participants. 5. Clinically measurable disease.* see comments about participants 6. Performance status of 0, 1 or 2. 7. Adequate marrow, renal and hepatic function Ineligible patients included those with: A diagnosis of borderline carcinoma (grade 0) or neoplasm termed probably malignant Patients who had received prior paclitaxel, irradiation or more than one prior chemotherapy regimen Patients with septicaemia, other active infection, acute hepatitis, severe gastrointestinal bleeding or other serious medical conditions likely to limit a patient’s ability to tolerate treatment; patients with a history congestive heart failure, unstable angina, or a myocardial

328 infarction within the last 6 months, or a history of cardiac arrhythmia requiring antiarrythmic medication Patients with unclassified ovarian cancer (cancers thought to be of ovarian origin but not explored, or where it was not possible to verify tumour arising from ovarian tissue; patients with a past or concomitant malignancy other than skin (excluding melanoma), or known hypersensitivity to a Escherichia coli- derived drug preparation were also ineligible.

Comments about participants: Inclusion criteria were altered during selection to allow patients without measurable disease or who were defined as platinum sensitive (patients with an initial response to platinum therapy lasting at least 6 months) to participate

Results Outcome 1 Outcome 2 Outcome 3 Outcome: survival (months) (measured from date registered on to Outcome: Tumour response in patients with measurable disease. Outcome: Quality of life study to date of 1st progression/death from any cause Covariates adjusted for:- (progression-free survival) and death or last contact if date of 1. Initial performance status, 2. Cell type, 3. Response to prior Baseline data: for both the PLD and topotecan groups: – Not death unknown (overall survival) platinum, 4. Co-operative group, 5. Measurable disease status reported; questionnaire only completed by 82% of participants. However, function and symptom scale scores were similar between Paclitaxel 250 mg/m2 (n=166) Paclitaxek 250 mg/m2 (n= 134) the two treatment groups. CR: 17/134 Progression-free survival: Platinum resistant: 13/109 (12%) Follow-up data: For both the PLD and topotecan groups: No Median: 5.5 months Platinum-sensitive: 4/25 (16%) significant differences between the intervention and control groups at Without progression at 36 months: 5/166 (3%) 12 weeks follow-up. PR: 32/134 Survival: Platinum resistant: 27/109 (25%) Comments:- Median: 12.3 months Platinum-sensitive: 5/25 (20%) QoL (assessed by European Organization for Research and Treatment Alive at 36 months: 9/166 of Cancer Quality of Life Questionnaire (QIQ-C30). 82% of Total: 49 (36%) participants completed QoL measure at baseline, at 12 weeks less Paclitaxel 175 mg/m2 (n=164) Platinum-resistant: 40/109 than 50% of participants in either treatment arm were still in the Platinum-sensitive: 9/25 HRQoL study. Progression-free survival: Median: 4.8 months Filgrastim dosage:

329 Without progression at 36 months: 2/164 (1%) 5ug/kg: (N = 68) 24/68 (35%) (95% CI: 24%, 48%) 10ug/kg: (N = 66) 25/66 (37.9%) (95% CI: 26%, 51%) Survival: Median: 13.1 months Paclitaxel 125 mg/m2 (n=131) Alive at 36 months: 8/164 (5%) CR: 9/131 Platinum resistant: 5/104 (5%) From both groups: Survival (median) Platinum-sensitive: 4/27 (15%) Platinum-resistant: 11.0 months Platinum-sensitive: 19.6 months PR: 27/131 (P < 0.001) Platinum resistant: 18/104 (17%) Platinum-sensitive: 9/27 (33%) Multivariate analysis: survival (relative hazard ratio) Paclitaxel regimen Total: 36/131 (27%) 175mg: 1.0 Platinum-resistant: 23/104 250mg: 0.972 (95% CI: 0.772, 1.22) Platinum-sensitive: 13/27

Platinum sensitive (relative HR): For both groups Resistant: 1.0 Sensitive: 0.576 (95% CI: 0.434, 0.764) OR responding to 250mg compared to 175mg = 1.89 (95% CI: 1.07, 3.31, P = 0.027) after adjusting for other factors (see above). Clinically measurable disease (relative HR): Platinum-resistant: (N = 213) OR = 2.59 (95% CI: 1.36, 4.95) No: 1.0 Platinum-sensitive: (N = 52) OR = 0.630 (95% CI: 0.191, 2.07) Yes: 1.97 (95% CI: 1.44, 2.69) Serous histology: 66/165 (40%) (95% CI: 32%, 48%) Histologic cell type (relative HR): Mucinous/clear cell histology: 1/21 (5%) (95% CI: 0.1%, 24%) Papillary serous adeno: 1.0 Other histology: 18/79 (23%) (95% CI: 14%, 34%) Clear cell or mucinous: 2.13 (95% CI: 1.38, 3.27) Other: 1.04 (95% CI: 0.810, 1.34) No response: 85 Platinum-resistant 69/109 (63%) Initial performance status (relative HR): Platinum-sensitive: 16/25 (64%) 0: 1.0 Not assessed due to death, toxicity or withdrawal: 7/134 1: 1.17 (95% CI: 0.915, 1.49) 2: 5.23 (95% CI: 3.36, 8.15)

330 Trial: Rosenberg et al. (31)

Study Details and Design Participant Details Intervention Details Follow-up/withdrawals and Adverse Conclusion and Comments Events Author: Rosenberg 2002(31) Number randomised Intervention (I): Withdrawals from intervention (I) Authors’ conclusions 208 randomised (paclitaxel weekly Type: - Paclitaxel weekly 1: ineligible – had non-measurable Weekly paclitaxel has a better safety Objective: To evaluate the efficacy and n=105; paclitaxel 3 weekly n=103) No. randomised:- 105 disease at inclusion profile and was as effective as the toxicity of paclitaxel given at the same Route of administration : i.v. equivalently dosed schedule every 3 dose intensity and administered weekly Disease type: Epithelial Dose:- 67mg/m2/wk. Group randomised Withdrawals from control (C) weeks. I/v steroids are a safe alternative or every 3 weeks, and to assess the Disease stage: Advanced to oral steroids 12 and 6 hrs before In total 2: 1: ineligible – was aged over to oral steroids. safety of intravenous steroids versus Occurrence of secondary spread:- Not paclitaxel OR parenteral steroids 30 min 75, 1 not stated – standard peroral permedication. stated before paclitaxel. All patients received clemastine 2mg and cimetidine 300mg Adverse events Trial ID: N/A Therapy stage: Second-line or ranitidine 50mg i.v. 30 min before Paclitaxel 67mg/m2/week (n = 104) Phase: Not reported. paclitaxel. Hematological toxicity (lowest Previous treatments No. of cycles - Protocol allowed value/patient) Length of follow-up: median 27 One prior platinum-containing regimen indefinite treatment (see below) Hemoglobin: 81/104 (78%) P = 0.04 months (range: 7-47+) of chemotherapy not containing a Length per cycle: - 21 days Grade 3-4: 4/104 (4%) P = 1.0 taxane. White blood cell: 74/104 (71%)P = 0.27 Control (C) Grade 3-4: 17/104 (16%) P = 1.0 Disease present after first-line Type: Paclitaxel 3 weekly Neutrophils: 63/104 (61%) P < 0.01 treatment? No. randomised: 103 Grade 3-4: 19/104 (18%) P < 0.001 Residual : Yes Route of administration: i.v. Platelets: 1/104 (1%) P = 0.12 Refractory - Yes Dose: - 200mg/m2/3 wks Group Grade 3-4: 0/104 (0%) P = 0.49 randomised to oral steroids 12 and 6 hrs Age/age range of participants before paclitaxel OR parenteral steroids Non-hematological toxicity (worst Weekly: median: 59 (range: 37-74) 30 min before paclitaxel. All patients value/patient) 3 weekly: median: 60 (range: 40 -76) received clemastine 2mg and cimetidine Neuropathy: 84/104 (81%) P = 0.72 300mg or ranitidine 50mg i.v. 30 min Grade 3: 11/104 (11%) P < 0.001 Characteristics before paclitaxel. Alopecia: 85/104 (82%) P = 0.11 Paclitaxel 67mg/m2 weekly (N = 105) No. of cycles:- protocol allowed Grade 3: 48/104 (46%) P < 0.001 Age: indefinite treatment (see below) Arthralgia/Myalgia: 61/104 (59%) P < median: 59 Length per cycle:- 21 days 0.001 range: 37-74 Grade 3: 5/104 (5%) P = 0.40 Comments about intervention/control: Nausea/Vomiting : 48/104 (46%) P = WHO performance status If no haematological toxicity occurred 0.57 0: 57/105 the dose was escalated maximally by 2 Grade 3: 4/104 (4%) P = 1.0 1: 40/105 steps. Dose reduction was performed in Nails: 37/104 (36%) P < 0.001 2: 8/105 case of severe cytopenia. Patients who Grade 3: 9/104 (9%) P < 0.01 could not tolerate the lowest dose level Platinum-resistant: 57/105 were taken off the study treatment. No Withdrawals at 9 weeks Platinum-sensitive: 48/105 dose escalation was allowed once a dose Total: 32 reduction had been made. If infusion Disease progression: 14 Tumour size at inclusion was interrupted due to a hypersensitivity Prolongation of treatment-free interval:

331 /= 10cm: 33/105 was made on basis of tumour Study drug toxicity: 1 unknown: 1/105 assessments every 6 weeks. Patients Other reasons: 3 with PD were taken off the study. Death: 0 Paresthesia at inclusion (WHO) Patients with SD received treatment Grade 0: 92/105 until either progression or unacceptable Dosage Grade 1: 13/105 toxicity occurred. Patients who achieved Median dose intensity (mg/m2/wk): 77.6 a CR or a PR continued study treatment mg/m2/week, P < 0.0001 Paclitaxel 200mg/m2 3weekly (n = 103) for a minimum of 6 weeks and thereafter % doses increased by +2: 61% Age: at the investigator’s discretion to tumour [Dose alterations] median: 60 progression/relapse or unacceptable -2: 48 range: 40-76 toxicity whichever came first. -1: 57 Cycles were to be given as planned – 0: 67 WHO performance status not permissible to prolong treatment free +1: 75 0: 56/103 interval. +2: 81 1: 33/103 2: 14/103 Paclitaxel 200mg/m2/3 weeks (n = 101) Platinum-resistant: 51/103 Hematological toxicity (lowest Platinum-sensitive: 52/103 value/patient) Hemoglobin: 65/101 (64%) P = 0.04 Tumour size at inclusion Grade 3-4: 4/101 (4%) P = 1.0 /= 10cm: 40/103 Grade 3-4: 45/101 (45%) P < 0.001 unknown: 0/103 Platelets: 5/101 (5%) P = 0.12 Grade 3-4: 1/101 (1%)P = 0.49 Paresthesia at inclusion (WHO) Grade 0: 91/103 Non-nematological toxicity (worst Grade 1: 12/103 value/patient) Neuropathy: 86/101 (85%) P = 0.72 Inclusion/exclusion criteria Grade 3: 29/101 (29%) P < 0.001 1. Histologically proven diagnosis of Alopecia: 91/101 (90%) P = 0.11 epithelial ovarian carcinoma Grade 3: 80/101 (79%) P < 0.001 2. 18 – 75 yrs old Arthralgia/Myalgia: 85/101 (84%) P < 3. Received no more than 1 prior 0.001 platinum-based regimen Grade 3: 8/101 (8%) P = 0.40 4. Evidence of progression during/ Nausea/Vomiting: 42/101 (42%) P = relapse after administration of last 0.57 platinum course documented before Grade 3: 3/101 (3%) P = 1.0 entering study Nails: 2/101 (2%) P < 0.001 5 .Measurable disease documented Grade 3: 0/101 (0%) P < 0.01

332 clinically and/or radiologically 6. Adequate physiologic function and Withdrawals at 9 weeks status (absolute neutrophil count >/= 2.0 Total: 20 x10 to the 9, platelet count >/= 100 x10 Disease progression: 9 to the 9, serum creatine /= 60) Study drug toxicity: 4 7. Anticipated survival >/= 12 weeks Other reasons: 1 8. No atrial or ventricular arrythmias, or Death: 1 congestive heart failure, even if medically controlled, or documented Dosage myocardial infarction within 6 months, Median dose intensity (mg/m2/wk): or a history of 2nd or 3rd degree heart 72.7, P < 0.0001 block. % doses increased by +2: 28% 9. No pre-existing motor or sensory Dose alterations: neurotoxicty > grade 2 of WHO criteria -2: 145 -1: 170 Comments about participants: 0: 200 platinum resistant defined as relapse < 6 +1: 225 months after 1st line therapy +2: 242 patients were stratified prior to entry according to platinum resistance. Oral steroids 12 and 6 hours prior to paclitaxel (n = 106) Survival data: Hypersensitivty reactions (no. of events) Response duration: Complete response Skin: 17/106 (16.0%) Ns was calculated from day of 1st Grade 3-4: 3/106 (2.8%) Ns observation of CR to day of documented Generalised urticaria: 1/106 (0.9%) Ns PD or censored observation; PR Grade 3-4: 1/106 (0.9%) Ns calculated from 1st day of study Dyspnea: 4/106 (3.8%) Ns treatment to day of documented PD or Grade 3-4: 0/106 (0%) Ns censored observation. Respiratory distress requiring treatment: Tumour responses: If 2 observations 1/106 (0.9%) Ns were missing not less than 4 weeks apart Grade 3-4: 1/106 (0.9%) Ns tumour response was judged as not Hypotension: 1/106 (0.9%) Ns confirmed Time to progression was calculated from Parenteral steroids 30 minutes prior to 1st day of study treatment to day of paclitaxel (n = 99) documented PD or censored Hypersensitivty reactions (no. of events) observation. Skin: 18/99 (18.2%) Ns Overall survival was calculated from Grade 3-4: 4/99 (4.0%) Ns date of randomisation to death or Generalised urticaria: 0/99 (0%) Ns censored observation Grade 3-4: 0/99 (0%) Ns Dyspnea: 4/99 (4.0%) Ns Grade 3-4: 1/99 (1.0%) Ns

333 Respiratory distress requiring treatment: 1/99 (1.0%) Ns Grade 3-4: 0/99 (0%) Ns Hypotension: 1/99 (1.0%) Ns

Results Outcome 1 Outcome 2 Outcome 3 Outcome: Survival Outcome: Time to progression Outcome: Tumour responses (WHO criteria)

Paclitaxel 67mg/m2/week (n = 105) Paclitaxel 67mg/m2/week (n = 105) Paclitaxel 67mg/m2/week (n = 105) Median 13.6 months (95% CI: 10.5, 18.7). P = 0.98 Median 6.1 months (95% CI: 5.0, 8.0) P = 0.85 ITT analysis CR: 13/105 (12.4%). 3 patients had unconfirmed CRs Paclitaxel 200mg/3 weeks (n = 103) Paclitaxel 200mg/3 weeks (n = 103) PR: 24/105 (22.8%). 7 patients had unconfirmed PRs Median 14.7 months (95% CI: 12.3, 19.1). P = 0.98 Median: 8.1 months (95% CI: 6.4, 9.7). P = 0.85 Total response: 37/105 (35.2%) SD: 43/105 (41.0%) PD: 15/105 (14.3) Not evaluable: 9/105 (8.6%) Not treated 1/105 (0.9%) Response duration: 9.4 months (95% CI: 6.2, 13.9) CR duration: 4.5 months (95% CI: 3.6, 10.7; P = 0.84)

per protocol analysis for patients treated for >/= 9 weeks N = 87 CR: 12/87 (13.8%) 3 unconfirmed CRs PR: 24/87 (27.6%) 7 unconfirmed PRs Total response: 36/87 (41.4%) SD: 36/87 (41.4%) PD: 15/87 (17.2%)

Paclitaxel 200mg/3 weeks (N = 103) ITT analysis CR: 17/103 (16.5%). 3 patients had unconfirmed CRs PR: 21/103 (20.4%). 6 patients had unconfirmed PRs Total response: 38/103 (36.9%) SD: 33/103 (32.0%) PD: 19/103 (18.5) Not evaluable: 11/103 (10.7%) Not treated 2/103 (1.9%) Response duration: 12.4 months (95% CI: 9.1, 14.3) CR duration: 7.8 months (95% CI: 4.2, 10.2; P = 0.84)

per protocol analysis for patients treated for >/= 9 weeks N = 90 CR: 16/90 (17.8%) 2 unconfirmed CRs

334 PR: 21/90 (23.3%) 6 unconfirmed PRs Total response: 37/90 (41.1%) SD: 34/90 (37.8%) PD: 19/90 (21.1%)

335

Trial: Gore et al. (33)

Author: Gore 2002(33) Number randomised Intervention (I): Withdrawals from intervention (I) Authors’ conclusions 266 randomised; 266 inITT Type: - i.v. topotecan None. Oral topotecan is active as 2nd line Objective: To compare the efficacy, No. randomised:- 131 therapy and is well tolerated. Although safety and tolerability of oral topotecan Disease type: Epithelial Route of administration : i.v. Withdrawals from control (C) the response rate was lower (non- compared with standard i.v. regiments Disease stage: stage III and IV Dose:- 1.5mg/m2/day for 5 days None significant) and there was a small but in a similar group of patients with Occurrence of secondary spread:- Not No. of cycles - Discretion of investigator significant survival benefit for I.v. relapsed ovarian cancer. stated. (see below) Median No of cycles: 6 Adverse events topotecan, oral topotecan is well (range 1-26) i.v. topotecan (n = 131) tolerated & convenient, important Trial ID: N/A Therapy stage: Second-line Length per cycle: - 21 days Grade 3/4 hematological toxicity per considerations where treatment is Phase: Not stated patient palliative. Previous treatments: One platinum- Control (C) based chemotherapy regimen which Type: oral topotecan Death from haematological toxicity: . could have included a taxane No. randomised: 135 5/131 Route of administration: oral. Neutropenia Disease present after first-line Dose: - 2.3mg/m2/day Grade 3: 15/131 (11%) treatment? No. of cycles:- Protocol: Discretion of Grade 4: 110/131 (84%) Residual : Yes investigator (see below) Median No of Mean neutrophil nadir 0.7 x 10 to the 9/l Refractory - Yes cycles: 4 (range 1-23) Leucopenia Length per cycle:- 21 days Grade 3: 8/131 (60%) Age/age range of participants Grade 4: 40/131 (31%) i.v. topotecan: median 60 years (range: Thrombocytopenia range: 27-80) Comments about Grade 3: /131 (1%) oral topotecan: median 60 years (range: intervention/control: Grade 4: 23/131 (18%) 23-80) Response to study drug was based on Anemia objective tumour assessments and Grade 3: /131 (33%) Characteristics evaluated by abdominal/pelvic CT or Grade 4: 10/131 (8%) MRI, chest X-ray or photography. All Per cycle: 778 cycles i.v. topotecan: (n = 131) claimed responses were subject to Neutropenia Age: median: 60 independent blinded radiological Grade 3: 249/778 (32%) range: 27-80 review. Complete response was defined Grade 4: 393/778 (51%) as complete disappearance of all known Leucopenia FIGO stage III: 82/131 63%) measurable & evaluable disease Grade 3: 371/778 (48%) FIGO stage IV: 42/131 (32%) determined by 2 measurements not less Grade 4: 68/778 (9%) Missing data: 7/131 (5%) than 4 weeks apart. Thrombocytopenia Partial response was defined as > 50% Grade 3:90/778 (12%) Performance status decrease in measurable lesion size for Grade 4: 29/778 (4%) 0: 47/131 (35%) >/= 4 weeks with no simultaneous Anemia 1: 77/131 (57%) increase in a known lesion or Grade 3: 78/778 (10%) 2: 11/131 (8%) appearance of new lesions or increase in Grade 4: 10/778 (1%) evaluable disease Infectious complications needing i.v.. Tumour size Response rate was also evaluated by antibiotics: 8%

336 < 5cm: 65/131 (50%) serial CA125 value measurement. Sepsis or infection/fever grade 2: 3% 5-10cm: 50/131 (38%) Response was defined as a 50% Platelet transfusions: 3% >10cm: 11/131 (8%) decrease in 2 samples, confirmed by a Red blood cell transfusions: 20% Missing data: 5/131 (4%) further sample, or serial decrease over 3 Non-hematological toxicity samples of > 75%. The final sample had Nausea: 8/131 (16%) Classification of relapse to be at least 28 days after the previous Grades 3/4: 6/131 (5%) Platinum-refractory: 39/131 (30%) sample. Diarrhoea: 40/131 (31%) Platinum-resistant: 36/131 (28-27%) Dose reductions in increments of Grades 3/4: 6/131 (5%) Platinum sensitive: 56/131 (43%) 0.4mg/m2/day p.o. or 0.25mg/m2/day Vomiting: 52/131 (40%) i.v. Were made for grade 4 neutropenia Grades 3/4: 4/131 (3%) Oral topotecan: (N = 135) lasting beyond day 21, grade 3-4 Alopecia: 68/131 (52%) Age: thrombocytopenia and any non- Grades 3/4: 8/131 (6%) Median: 60 hematological toxicity of grade 3-4 Fatigue: 50/131 (38%) Range: 23-80 (excluding grade 3-4 nausea and Grades 3/4: 5/131 (4%) vomiting). The daily dose could be Abdominal pain: 39/131 (30%) FIGO stage: increased in similar increments if, Grades 3/4: 9/131 (7%) Stage III: 84/135 (62%) during the previous course there had Constipation: 42/131 (32%) Stage IV: 43/135 (32%) been no toxicity greater than grade 2. grades 3/4: 2/131 (2%) Missing data: 8/135 (6%) Fever: 31/131 (24%) Grades 3/4: 7/131 (5%) Performance status: Dosing adjustments: (N =778) 0: 59/135 (45%) Courses at starting dose: 58% 1: 60/135 (46%) Courses above starting dose: 4% 2: 12/135 (9%) Dose reductions: 11%

Tumour size: Oral Topotecan (n = 135) < 5cm: 66/135 (49%) Grade ¾ hematological toxicity per 5-10cm: 58/135 (43%) patient > 10cm: 10/135 (7%) Death from haematological toxicity: Missing data: 1 (/135 1%) 2/135 Neutropenia Classification of relapse Grade 3: 40/135 (30%) Platinum-refractory: 40/135 (30%) Grade 4: 67/135 (50%) Platinum-resistant: 37/135 (28%) Mean neutrophil nadir 1.5 x 10 to the 9/l Platinum sensitive: 58/135 (43%) Leucopenia Grade 3: 59/135 (44%) Inclusion criteria: Grade 4: 28/135 (21%) 1 Histological diagnosis of epithelial Thrombocytopenia ovarian cancer Grade 3: 30/135 (22%) 2. Originally FIGO stage III or IV Grade 4: 27/135 (20%) disease Anemia 3. Either progressed on 1st line Grade 3: 51/135 (38%) chemotherapy or relapsed within 12 Grade 4: 5/135 (4%) months of completing initial treatment Per cycle: 729 cycles 4. 1st line chemotherapy included a Neutropenia

337 platinum compound which may have Grade 3: 190/729 (26%) been combined with a taxane Grade 4: 106/729 (15%) 5. Only 1 previous line of chemotherapy Leucopenia 6. Aged >/= 18 Grade 3: 163/729 (22%) 7.Measurable disease with 1 lesion >/= Grade 4: 31/729 (4%) 2cm in diameter or >/= 1cm for skin Thrombocytopenia lesions Grade 3: 70/729 (10%) 8.ECOG performance status/= 3 months Anemia 10.No surgery, radiotherapy, Grade 3: 85/729 (12%) chemotherapy or hormone therapy for 4 Grade 4: 7/729 (1%) weeks prior to study entry or 60 days in Infectious complications needing i.v. case of prior immunotherapy antibiotics: 6% 11.Adequate bone marrow, renal and Sepsis or infection/fever grade 2: 5% hepatic function (haemoglobin >/= Platelet transfusions: 3% 90g/l, WBC >/= 3.5 x 10 to the 9/l, Red blood cell transfusions: 21% neutrophils >/= 1.5 x 10 to the 9/l, platelets >/=100 x 10 to the 9/l, Non-hematological toxicity (n = 135) creatinine 1ml/s), serum bilirubin

Patients were classified for platinum- - sensitivity as follows: Refractory: progressive or stable disease during 1st line chemotherapy Resistant: responded to 1st line chemotherapy but relapsed within 6

338 months of treatment ending Sensitive: responded to initial therapy but relapsed after > 6 months

Results Outcome 1 Outcome 2 Outcome 3 Outcome: Survival (not defined). Patients for whom an event had Outcome: Tumour response not occurred at the cut-off date for inclusion in the analysis were censored at time of last contract. i.v. topotecan CR: 4/131 (3%) i.v. Topotecan PR: 22/131 (17%) n = 131 Overall response: 26/131 (20%) (95% CI: 13.0-26.7%) Survival: Median = 58 weeks (range 0.3-120.0) p = 0.033 Platinum-sensitive patients: (n = 56) 20/56 (36%) Time to progression (weeks): Median = 17 weeks (range (0.1-91.6) Platinum-resistant patients: (n = 36) 4/36 (11%) n = 26 Platinum refractory patients: (n = 39) Time to response: Median = 8 weeks (range5.1-25.4) 2/39 (5%)

(n = 26) Tumour size < 5 cm: N = 65 Duration of response: Median = 26 weeks (6.6 - 52.7) 15/65 (23%) >/= 5cm: (n=61) 10/61 (16%) oral topotecan (n = 135) 1st line treatment: Platinum/paclitaxel: N = 54 Survival: median = 51 weeks (range 1.6-109.0) 12/54 (22%)

Time to progression (weeks): median = 13 weeks (range: 1.6-76.6) SD: 35/131 (27% PD: 59/131 (45%) (n = 17) Not evaluated 11/131 (8%) Time to response: Median = 12 weeks (range: 5.6-18.1) response by CA125: N = 99 (n = 17) 28/99 (28%) Response duration: Median = 34 (range: 13.1-62.3) ITT analysis: 28/131 (21%)

Oral topotecan

CR: 2/135 (1%) PR: 15/135 (11%) Overall response: 17/135 (13%) (95% CI: 7.6-19.1%)

Platinum-sensitive patients: (n = 58) 11/58 (19%)

339 Platinum-resistant patients: (n = 37) 3/37 (8%) Platinum refractory patients: (n=40) 3/40 (8%)

Tumour size < 5 cm: (n = 66) 10/66 (15%) >/= 5cm: (n = 68) 7/68 (10%)

1st line treatment: Platinum/paclitaxel: (n = 53) 8/53 (15%)

SD: 39/135 (29%) PD: 65/135 (48%) Not evaluated: 14/135 (10%) Response by CA125: (n= 105) 21/105 (20%) ITT analysis: 21/135 (16%)

340 Trial: 30-57(27)

Study Details and Design Participant Details Intervention Details Follow-up/withdrawals and Adverse Conclusion and Comments Events Author: Johnson & Johnson Number randomised Intervention (I): Withdrawals from intervention (I) Authors’ conclusions Pharmaceutical Research & 438 planned enrolment; 216 randomised Type: - Pegylated liposomal PLD:- In this population of women with Development (PLD n=108, paclitaxel n=108) doxorubicin hydrochloride (PDL) Disease progression: 42/108 (38.9%) epithelial ovarian cancer whose disease No. randomised:- 108 Death: 12/108 (11.1%) did not respond to or relapsed after first- Objective: To compare the efficacy and Disease type: Epithelial Route of administration : i.v. Adverse event: 18/108 (16.7%) line platinum-based therapies, overall safety of pegylated liposomal Disease stage: Advanced Dose:- 50 mg/m2 as a 1hr infusion Lost-to-follow up: 0 (0%) survival appears to be similar for doxorubicin (PLD) and paclitaxel in Therapy stage: Second-line No. of cycles –at least 6 for participants Other/unknown: 10/108 (9.3%) participants treated with PLD or patients with epithelial ovarian who responded Completed protocol treatment: 26/108 paclitaxel. PLD treatment was less carcinoma following failure of first-line Previous treatments Length per cycle: - 28 days (24.1%) often associated with alopecia and platinum based chemotherapy. Chemotherapy (platinum-based first-line Grade 3-4 haematological adverse monotherapy regimen) Control (C) Withdrawals from control (C) events than paclitaxel treatment. Grade Trial ID: 30-57 Type: Paclitaxel Paclitaxel:- 3-4 hand-foot syndrome (PPE) and Phase: Phase III non-inferiority trial. PLD (n=108) No. randomised: 108 Disease progression: 30/108 (27.8%) stomatitis were more frequently Carboplatin alone: 63/108 (58.3%) Route of administration: i.v. Death: 5/108 (4.6%) associated with PLD treatment. Length of follow-up: Cisplatin alone: 35/108 (32.4%) Dose: - 175 mg/m2 as a 3 hour infusion Adverse event: 7/108 (6.5%) Treatment should have continued for up Both: 10/108 (9.3%) No. of cycles:- at least 6 for participants Lost-to-follow up: 1/108 (0.1%) Comments to one year in the absence of disease Neither: 0 (0%) who responded Other/unknown: 12/108 (11.1%) The study was terminated early due to progression. Prior anthracycline therapy:- Length per cycle:- 21 days Completed protocol treatment: 53/108 poor patient accrual after paclitaxel was Yes: 10/108 (9.3%) All paclitaxel treated participants were (49.1%) approved for use in combination with No: 98/108 (90.7%) to be premedicated with corticosteroids, platinum-based therapy for first-line antiphistamines, and H2 antagonists The majority of participants who treatment by the European Agency for Paclitaxel (n=108) prior to paclitaxel administration. discontinued from treatment for ‘other’ the Evaluation of Medicinal Products. Carboplatin alone: 58/108 (53.7%) reasons did so because of the Cisplatin alone: 39/108 (36.1%) Comments about investigator’s choice, patient choice or Both: 10/108 (9.3%) intervention/control: both. Neither: 1/108 (0.9%) Participants with ongoing clinical Prior anthracycline therapy:- benefit were to have continued on the Adverse events Yes: 15/108 (13.9%) study drug as long as it was in their PLD (n=108) No: 93/108 (86.1%) interest and in the absence of severe toxicity. Participants exhibiting a Body as a whole Disease present after first-line complete response (CR) were to have Asthenia:- treatment? received 2 subsequent cycles of All grades: 42/108 (38.9%) Residual : yes treatment. Participants exhibiting partial Grade 3: 4/108 (3.7%) Refractory yes– response (PR) were to have continued to Grade4: 0 (0%)) received study drug as long as Abdominal pain:- Characteristics therapeutic benefit was derived, with it All grades: 34/108 (31.5%) PLD (n=108) recommended they received at least 3 Grade3: 12/108 (11.1%) Age: subsequent cycles of study drug. Grade 4: 0/ (0%) Mean (SD): 58.4 years Fever:- Median: 60.5 (range 27-80) No concurrent investigational agents or All grades: 28/108 (25.9%)

341 < 65 years: 74/108 (68.5%) antineoplastic agents were permitted. Grade 3: 7/108 (6.5%) > 65 years: 34/108 (31.5%) The prophylactic use of hematopoietic Grade 4: 0 (0%) cytokines was discouraged in Pain:- Race:- conjunction with the first dose of study All grades: 24/108 (22.2%) White: 105/108 (97.2%) drug. Their use was recommended in Grade 3: 1/108 (0.9%) Black: 0 (0%) subsequent cycles under specific Grade 4: 0 (0%) Hispanic: 2/108 (1.9%) circumstances: in participants with Infection:- Asian: 1/108 (0.9%) prolonged neutropenia (Grade 4 All grades: 23/108 (21.3%) neutropenia lasting >7days, or failure of Grade 3: 2/108 (1.9%) CA-125 at baseline (U/mL): ANC to recover within 22 days), or the Grade 4: 1/108 (0.9%) (n=105) occurrence of febrile neutropenia in a Headache:- Mean (SD): 1165.69 (2117.529) prior cycle of treatment. Pyridoxine All grades: 12/108 (11.1%) Median: 470.00 (range 7.0 – 14513.0) (B6) was recommended for the Grade 3: 1/108 (0.9%) treatment of hand-foot syndrome Grade 4: 0 (0%) Histologic tumour type: symptoms. Ascites:- Serious papillary: 29/105 (26.9%) All grades: 11/108 (10.2%) Mucinous: 0 (0%) Actual study treatment Grade 3: 6/108 (5.6%) Unspecified adenocarcinoma: 12/105 administration: Grade 4: 0 (0%) (11.1%) Backpain:- Not specified: 67/105 (62%) PLD (n=108) All grades: 11/108 (10.2%) Cumulative dose (mg/m2) Grade 3: 1/108 (0.9%) Histologic tumour grade:- Mean (SD): 205.56 (121.998) Grade 4: 0 (0%) Moderate differentiated: 1/108 (0.9%) Median: 199.99 (range 49-800) Poorly differentiated: 12/108 (11.1%) Cardiovascular system Unspecified differentiated: 28/108 Mean dose intensity (mg/m2 per Vasodilitation:- (25.9%) week) (n=92) All grades: 5/108 (4.6%) Not specified: 67/108 (62%) Mean (SD): 11.09 (1.261) Grade 3: 1/108 (0.9%) Median: 85.0 (range 1-448) Grade 4: 0 (0%) Ascites at baseline: Absent: 70/108 (64.8%) Days on drug (n=92) Digestive system Present: 38/108 (35.2%) Mean (SD) 98.7 (77.05) Nausea:- Not available: 0 (0%) Median: 85.0 (1-448) All grades: 56/108 (51.9%) Grade 3: 6/108 (5.6%) Platinum sensitivity: Paclitaxel (n=108) Grade 4: 1/108 (0.9%) Platinum sensitive: Cumulative dose (mg/m2)(n=108) Stomatitis:- Platinum refractory: Mean (SD): 999.89 (392.622) All grades: 52/108 (48.1%) Median: 1049.8 (range 175-2221.1) Grade 3: 11/108 (10.2%) Bulky disease: Grade 4: 0 (0%) Present: 61/108 (56.5%) Mean dose intensity (mg/m2 per Vomiting:- Absent: 47/108 (43.5%) week)(n=104) All grades: 37/108 (34.3%) Mean (SD): 52.74 (4.787) Grade 3: 10/108 (9.3%) Platinum sensitivity/bulky disease: Median: 54.69 Grade 4: 2/108 (1.9%) Refractory/present: 37/108 (34.3%) Constipation:- Refractory/absent: 27/108 (25.0%) Days on drugs (n=104) All grades: 30/108 (27.8%) Sensitive/present: 24/108 (22.2%) Mean (SD): 106.2 (50.13) Grade 3: 4/108 (3.7%)

342 Sensitive/absent: 20/108 (18.5%) Median: 106.0 (1-260) Grade 4: 0 (0%) Diarrhoea:- KPS at baseline: Protocol deviations: One participant All grades: 23/108 (21.3%) <80: 11/108 (10.2%) randomised to paclitaxel was classified Grade 3: 3/108 (2.8%) >80: 95/108 (88%) as having platinum sensitive disease. It Grade 4: 0 (0%) Not available: 2/108 (1.9%) was later discovered that the participant Anorexia:- had never received prior platinum All grades: 18/108 (16.7%) FIGO staging at initial diagnosis: therapy. Grade 3: 1/108 (0.9%) 1: 10/108 (9.3%) Grade 4: 0 (0%) 2: 11/108 (10.2%) Definition of response Dyspepsia:- 3: 64/108 (59.3%) All grades: 14/108 (13.0%) 4: 22/108 (20.4%) Complete response (CR): Complete Grade 3: 1/108 (0.9%) Not available: 1/108 (0.9%) disappearance of all measurable and Grade 4: 0 (0%) evaluable disease. No new lesions and Treatment free interval (months) no disese related symptoms. Hemic and lymphatic system Mean (SD): 9.0 (9.98) Partial response (PR): A >50% Neutropenia:- Median: 6.6 (1.0 – 69.4) decrease in the sum of the products of All grades: 18/108 (16.7%) bidimensional perpendicular diameters Grade 3: 6/108 (5.6%) Sum of lesions at baseline (cm2) of all measurable lesions. No Grade 4: 1/108 (0.9%) (n=107) : progression of evaluable disease. No Anemia:- Mean (SD): 68.5 (111.72) new lesions. All grades: 17/108 (15.7%) Median: 30.0 (range 0.7- 900) Stable disease: Does not ualify for CR, Grade 3: 3/108 (2.8%) PR or progression. Grade 4: 0 (0%) Paclitaxel (n=108) Progressive disease (PD): A >50% Leukopenia:- Age:- increase in the sum of the products of All grades: 15/108 (13.9%) Mean (SD): 59.5 ((10.79) bidimensionally measured lesions over Grade 3: 5/108 (4.6%) Median: 61.0 (range 20-78) the smallest sum observed (over Grade 4: 1/108 (0.9%) < 65 years: 68/108 (63.0%) baseline, if no increase), or reappearance > 65 years: 40/108 (37.0%) of any lesion that had disappeared, or Metabolic/nutritional disorder clear worsening of any evaluable Peripheral oedema:- Race (n=108): disease, or failure to return for All grades: 14/108 (13.0%) White: 105/108 (97.2%) evaluation due to death or deteriorating Grade 3: 0 (0%) Black: 1/108 (0.9%) condition (unless clearly unrelated to Grade 4: 0 (0%) Hispanic: 0 (0%) this cancer), or appearance of any new Asian: 2/108 (1.9%) lesion/site. For bone metastases, Musculoskeletal system increased uptake on the scan does not Myalgia:- CA-125 at baseline (U/mL): constitute clear worsening or disease. All grades: 4/108 (3.7%) (n=102) Worsening of existing non-evaluable Grade 3: 1/108 (0.9%) Mean (SD): 1543.77 (3111.981) disease does not constitute progression. Grade 4: 0 (0%) Median: 338.50 (range 2.0-17374.0) Arthralgia:- All grades: 2/108 ((1.9%) Histologic tumour type (n=102) Grade 3: 0 (0%) Serious papillary: 24/102 (22.2%) Grade 4: 0 (0%) Mucinous: 1/102 (0.9%) Unspecified adenocarcinoma: 18/102 Nervous system

343 (16.7%) Paresthesia:- Not specified: 65/102 (60.2%) All grades: 15/108 (13.9%) Grade 3: 0 (0%) Histologic tumour grade (n=108) Grade 4: 0 (0%) Moderate differentiated: 6.108 (5.6%) Somnolence:- Poorly differentiated: 13/108 (12.0%) All grades: 11/108 (10.2%) Unspecified differentiated: 24/108 Grade 3: 3/108 (2.8%) (22.2%) Grade 4: 0 (0%) Not specified: 65/108 (60.2%) Respiratory system Ascites at baseline Dyspnea:- Absent: 69/108 (63.9%) All grades: 18/108 (16.7%) Present: 38/108 (35.2%) . Grade 3: 6/108 (5.6%) Not available: 1/108 (0.9%) Grade 4: 1/108 (0.9%) Pharyngitis:- Platinum sensitivity All grades: 8/108 (7.4%) Platinum sensitive: 41/108 (38%) Grade 3: 0 (0%) Platinum refractory: 67/108 (62%) Grade 4: 0 (0%)

Bulky disease: Skin and appendages Present: 56/108 (51.9%) Hand-foot syndrome:- Absent: 52/108 (48.1%) All grades: 55/108 (50.9%) Grade 3: 16/108 (14.8%) Platinum sensitivity/bulky disease Grade 4: 1/108 (0.9%) Refractory/present: 36/108 (33.3%) Alopecia:- Refractory/absent: 31/108 (28.7%) All grades: 47/108 (43.5%) Sensitive/present: 20/108 (18.5%) Grade 3: 3/108 (2.8%) Sensitive/absent: 21/108 (19.4%) Grade 4: 0 (0%) Rash:- KPS at baseline All grades: 15/108 (13.9%) <80: 12/108 (11.1%) Grade 3: 2/108 (1.9%) >80: 90/108 (83.3%) Grade 4: 0 (0%) Not available: 6/108 (5.6%) Paclitaxel (n=108) FIGO staging at initial diagnosis 1: 10/108 (9.3%) Body as a whole 2: 8/108 (7.4%) Asthenia:- 3: 77/108 (71.3%) All grades: 36/108 (33.3%) 4: 13/108 (12.0%) Grade 3: 6 /108 (5.6%) Not available: 0 (0%) Grade4: 1/108 (0.9%) Abdominal pain:- Treatment free interval (months) All grades: 35/108 (32.4%) Mean (SD): 11.1 months (17.34) Grade3: 7/108 (6.5%) Median: 6.7 (range 0.9-109.1) Grade 4: 0 (0%) Fever:-

344 Sum of lesions at baseline (cm2) All grades: 8/108 (7.4%) (n=107) Grade 3: 3/108 (2.8%) Mean (SD): 50.7 (60.95) Grade 4: 0 (0%) Median: 32.5 (range 1.1-420.0) Pain:- All grades: 24/108 (22.2%) Inclusion/exclusion criteria Grade 3: 3/108 (2.8%) Participants with histologically proven Grade 4: 0 epithelial ovarian carcinoma with Infection:- measurable disease; a recurrence of All grades: 10/108 (9.3%) disease or disease progression indicative Grade 3: 1/108 (0.9%) of failure of first-line platinum based Grade 4: 0 (0%) chemotherapy; Karnofsky performance Headache:- status (KPS) >60%; age >18 years; All grades: 13/108 (12.0%) adequate bone marrow function: Grade 3: 2/108 (1.9%) platelets >100,000 / mm3, haemoglobin Grade 4: 0 (0%) >9g/dL, and absolute neutrophil count Ascites:- (ANC) >1500 cells / mm3; adequate All grades: 8/108 (7.4%) renal function: creatinine <2.5 mg/dL Grade 3: 1/108 (0.9%) (<220 umol/L); adequate liver function: Grade 4: 0 (0%) aspartate amino transferase (AST) and Backpain:- alanine amino transferase (ALT) <2 All grades: 14/108 (13%) times upper limit or normal, alkaline Grade 3: 1/108 (0.9%) phosphatase <2.0 times upper limit of Grade 4: 0 (0%) normal, except if attributed to tumor, and bilirubin 5 years with exception of curatively treated basal call Digestive system or squamous cell carcinoma of the skin Nausea:- or carcinoma in situ of the cervix were All grades: 47/108 (43.5%) included. Grade 3: 2/108 (1.9%) Participants who were pregnant or Grade 4: 0 (0%) breast-feeding; life expectance < 3 Stomatitis:- months; prior radiation therapy to more All grades: 12/108 (11.1%) than one-third of hematopoietic sites Grade 3: 1/108 (0.9%) within 30 days prior to first dose of Grade 4: 0 (0%) study drug; history of cardiac disease, Vomiting:- with New York Heart Association Class All grades: 34/108 (31.5%) II or greater with congestive heart Grade 3: 4/108 (3.7%) failure; uncontrolled systemic infection; Grade 4: 0 (0%) any investigational agent within 30 days Constipation:- of first dose of study drug; prior therapy All grades: 41/108 (38.0%)

345 with PLD or paclitaxel; prior Grade 3: 5/108 (4.6%) chemotherapy within 28 days of first Grade 4: 0 (0%) dose of study drug (or 42 days if Diarrhoea:- participanht had receive4d a nitrosourea All grades: 24/108 (22.2%) or mitomycin); treated with high dose Grade 3: 3/108 (2.8%) therapy supported by bone marrow or Grade 4: 0 (0%) peripheral stem cell transplantation at Anorexia:- any time were excluded. All grades: 11/108 (10.2%) Grade 3: 0 (0%) Comments about participants: Grade 4: 0 (0%) Patients were prospectively stratified Dyspepsia:- based on platinum sensitivity and the All grades: 11/108 (10.2%) presence or absence of bulky disease Grade 3: 0 (0%) and were to have been analysed Grade 4: 0 (0%) retrospectively for prior anthracycline therapy. Participants who had initially Hemic and lymphatic system responded to platinum-based therapy Neutropenia:- and who had a progression-free interval All grades: 23/108 (21.3%) of more than 6 months off treatment Grade 3: 10/108 (9.3%) were classified as platinum-sensitive. Grade 4: 3 /108 (2.8%) Participants who progressed during Anemia:- treatment, or who had stable disease in All grades: 23/108 (21.3%) response to initial platinum-based Grade 3: 5/108 (4.6%) therapy, or whose disease relapsed Grade 4: 0 (0%) within 6 months of cessation of therapy Leukopenia:- were classified as having platinum- All grades: 21/108 (19.4%) refractory disease. Bulky disease was Grade 3: 9/108 (8.3%) defined as the presence of a tumour Grade 4: 0 (0%) mass that was > 5cm in size. Metabolic/nutritional disorder Peripheral oedema:- All grades: 15/108 (13.9%) Grade 3: 1/108 (0.9%) Grade 4: 0 (0%)

Musculoskeletal system Myalgia:- All grades: 31/108 (28.7%) Grade 3: 7/108 (6.5%) Grade 4: 0 (0%) Arthralgia:- All grades: 23/108 (21.3%) Grade 3: 2/108 (1.9%) Grade 4: 0 (0%)

346 Nervous system Paresthesia:- All grades:47/108 (43.5%) Grade 3: 4/108 (3.7%) Grade 4: 0 (0%) Somnolence:- All grades: 17/108 (15.7%) Grade 3: 2/108 (1.9%) Grade 4: 0 (0%)

Respiratory system Dyspnea:- All grades: 15/108 (13.9%) Grade 3: 1/108 (0.9%) Grade 4: 0 (0%) Pharyngitis:- All grades: 18/108 (16.7%) Grade 3: 0 (0%) Grade 4: 0 (0%)

Skin and appendages Hand-foot syndrome:- All grades: 13/108 (12.0%) Grade 3: 0 (0%) Grade 4: 0 (0%) Alopecia:- All grades: 94/108 (87.0%) Grade 3: 20/108 (18.5%) Grade 4: 1/108 (0.9%) Rash:- All grades: 19/108 (17.6%) Grade 3: 1/108 (0.9%) Grade 4: 0 (0%)

Dosing adjustments:- PLD (n=108) Participants with delayed, interrupted or reduced doses: 56/108 (51.9%) Doses delayed, interrupted or reduced: 109

Reason for delay or interruption: Hand-foot syndrome: 21/108 (19.4%) Stomatitis/mucositis: 12/108 (11.1%) Haematologic toxicity: 6/108 (5.6%)

347 Other laboratory toxicity: 0 (0%) Infusion reaction: 10/108 (9.3%) Intercurrent illness: 8/108 (7.4%) Scheduling problem: 2/108 (1.9%) Other/unknown: 22/108 (20.4%)

Paclitaxel (n=108) Participants with delayed, interrupted or reduced doses: 51 (47.2%) Doses delayed, interrupted or reduced: 76

Reason for delay or interruption: Hand-foot syndrome: 1/108 (0.9%) Stomatitis/mucositis: 1/108 (0.9%) Haematologic toxicity: 7/108 (6.5%) Other laboratory toxicity: 1/108 (0.9%) Infusion reaction: 3/108 (2.8%) Intercurrent illness: 9/108 (8.3%) Scheduling problem: 17/108 (15.7%) Other/unknown: 27/108 (25.0%)

Resultrsr Outcome 1 Outcome: Overall survival (not defined)

PLD group(I) :- Median overall survival: 46.6 weeks (range: 2.3 – 263.7+) Number censored: 7.4

Paclitaxel group (C):- Median overall survival: 56.3 weeks (range: 1.4 – 211.4) Number censored: 6.5 HR =0.931 (95% CI: 0.702, 1.234) (p=0.618)

Platinum-sensitive disease PLD (n=44) Median overall survival: 65.4 weeks (range: 3.9-263.7+) Number censored: 13.6

Paclitaxel (n=41) Median overall survival: 57.0 weeks (range: 1.4 – 172.3) Number censored: 7.3

348 HR=1.051 (95% CI: 0.663, 1.667) (p=0.833)

Platinum refractory disease PLD (n=64) Median overall survival: 36.7 weeks (range:2.3 – 242.1+) Number censored: 3.1

Paclitaxel (n=67) Median overall survival: 54.3 weeks (range:1.7 – 211.4+) Number censored: 6.0 HR= 0.865 (95% CI: 0.605, 1.237) (p=0.427)

349 Appendix 7. Trials only available in abstract form

Meeting: 2003 ASCO Annual Meeting Category: Gynecologic Cancer SubCategory: Gynecologic Cancer

Randomised phase II study of carboplatin (C) versus paclitaxel-carboplatin (PC) in platinum-sensitive (PS) recurrent advanced ovarian carcinoma (AOC) with assessment of quality of life (QoL): A GEICO study (Spanish Group for Investigation on Ovarian Carcinoma)

Abstract: No: 1812 Citation: Proc Am Soc Clin Oncol 22: page 451, 2003 (abstr 1812)

Author(s): A. A. Gonzalez Martin, E. Calvo, I. Bover, M. J. Rubio, A. Arcusa, A. Casado, B. Ojeda, C. Balana, M. C. Gonzalez, A. Herrero, Amadeo Pelegri, Andres Cervantes; Hospital Universitario Ramon y Cajal, Madrid, Spain; Hospital Virgen del Rocio, Sevilla, Spain; Hospital Sant Joan, Reus, Spain; Hospital Reina Sofia, Cordoba, Spain; Hospital de Terrasa, Barcelona, Spain; Hospital Clinico San Carlos, Madrid, Spain; Hospital Sant Creu i Sant Pau, Barcelona, Spain; Hospital Trias i Pujol de Badalona, Barcelona, Spain; Hospital Virgen de la Luz, Cuenca, Spain; Hospital Miguel Servet, Zaragoza, Spain

Abstract: The role of platinum (P) based poli-chemotherapy (CT) versus monotherapy with P in PS recurrent AOC is not well known because of the small number of randomised clinical trials in this context. Our objective was to determine if PC is superior to C in terms of response rate measured by CT scan and /or CA-125 criteria. Methods: Pts with recurrent AOC, < 3 previous CT lines, P-based CT in the last treatment, and a P-free interval (PFI) of more than 6 months (m), were randomised to C AUC=5 (A) or P 175 mg/m2 plus C AUC=5 (B), both every 21 days for 6 cycles. Stratification between 1 or 2 previous lines and 6-12 or > 12 m PFI was made. Results: 80 pts have been included from may 2000 and data of the first 77 pts have been analysed (38 A/39 B). 74 pts are valuable for toxicity (38/36). 3 pts randomised to PC did not received any cycle but had been included in the intent to treat analysis. No differences in significant prognostic factors except for more ECOG 2 pts in A (18,4% vs 2,6%; p=0,056). Median PFI was (A/B) 14 m and 13 m. 379 cycles have been administered with a median of 3 in both arms. NCI-CTC haematological toxicities included (% of pts C/PC): anemia (15,8/5,6), neutropenia (10,5/19,4) and thrombocytopenia (13,1/2,8). NCI-CTC non- haematological toxicity grade 2-4 (% pts C/PC): Carboplatin allergy (5,3/16,6), vomiting (26,3/13,9), mucositis (0/16,7; p=0,02), neurosensory (2,6/19,4; p=0,051), arthralgia-myalgia (7,9/36,1; p=0,003), asthenia (15,7/25). The intent to treat overall response rate was 52,6 % with C (21% CR+ 31,6% PR) and 74,4% with PC (23,1% CR + 51,3% PR), p=0,047. PD was observed in 31,6% with C and 5,1% with PC (p=0,002). Median time to progression has not been reached with PC. QoL analysis is ongoing. Conclusions: PC polychemotherapy seems superior to C monotherapy in terms of response rate, with an acceptable toxicity profile. Mature survival data and QoL analysis are necessary to consider PC as the treatment of choice in PS recurrent AOC

350 Meeting: 2000 ASCO Annual Meeting Category: Gynecologic Cancer SubCategory: Gynecologic Cancer

Randomized Trial Comparing Paclitaxel+Doxorubicin (AT) Versus Paclitaxel (T) as Second Line Therapy for Advanced Ovarian Cancer (AOC) Patients in Early Progression After Platinum Based Chemotherapy.

Abstract No: 1506

Author(s): Valter Torri, Irene Floriani, Angelo Tinazzi, Pier Franco Conte, Alberto Ravaioli, Maria Grazia Cantù, Rita Rossi, Laura Grassi, Gabriella Parma, Nicoletta Colombo

Abstract: From October 1994 to June 1999, 234 patients from 35 Italian hospitals have been randomised into a study aimed at evaluating the role of AT (doxorubicin (A): 80 mg/m2 iv q 21 days x 4-6 cycles; paclitaxel (T): 175 mg/ m2, 3 hours infusion q 21 days x 4-6 cycles) vs. T (paclitaxel (T): 175 mg/ m2, 3 hours infusion q 21 days x 4-6 cycles), as second line therapy for AOC patients. Patients in progressive disease within 12 months from the end of first line platinum based therapy, were eligible for the study. At the time of this analysis, 96 out of 118 (81%) patients in AT arm and 90 out of 116 in T arm(78%) are evaluable for the analysis. Platinum based monochemotherapy was the first line treatment in 42% patients, while polichemotherapy containing A was the preferred first line therapy in 26% patients. Median duration of first line therapy was 4 months, median time from the end of first line therapy to randomisation was 5 months (range 3-12). Seventyeight percent in T arm and 72% of patients in AT arm received at least 4 cycles of therapy. Treatment was completed without modification in 48% and 28% of T and AT patients respectively. Haemathological toxicity was significantly more common in AT patients (grade 3-4 reduction in WBC count: 24% in AT vs. 7% in T arm). Neuropathies were similar in both arms (Sensory: 15% WHO grade 2 or 3, Motor: 4% grade 2 or 3). Objective response was achieved in 54% of patients in T and in 52% of patients in AT arm. At a median follow-up time of 24 months, a total of 150 patients have progressed and 124 have died. Survival analysis showed no evidence of a difference between AT and T (median time to progression: 6.6 and 7.5 months, median survival: 12 and 14 months for AT and T respectively; hazard ratio for mortality of AT vs. T: 1.17 [95% CI 0.82-1.66] p=0.39). AT is not more effective than T in AOC patients refractory to platinum based chemotherapy.

351 Meeting: 2001 ASCO Annual Meeting Category: Gynecologic Cancer SubCategory: Gynecologic Cancer

Multicenter Randomized Phase II Study of Oxaliplatin (OXA) or Topotecan (TOPO) in Platinum-Pretreated Epithelial Ovarian Cancer (EOC) Patients (pts).

Abstract No: 847 Author(s): Jan Baptist Vermorken, Martin Gore, Tim Perren, Ignace Vergote, Nicoletta Colombo, Peter Harper, Gordon Rustin, Andrea Bonetti, Paolo Zola, Nathalie Le Bail, Silvano Brienza, U.Z. Antwerpen, EDEGEM, Belgium; Royal Marsden Hospital, London, UK; Saint JamesÆs University Hospital, Leeds, UK; U.Z. Gasthuisberg, Leuven, Belgium; Instituto Europeo di Oncologia, Milano, Italy; Guy's Hospital, London, UK; Mount Vernon Hospital, Northwood, UK; Azienda Ospedaliera di Verona, Verona, Italy; Universita di Torino, Torino, Italy; Sanofi-Synthélabo Recherche, Paris, France; Debioclinic S.A., Charenton le Pont, France.

OXA, a new DACH platinum (Pt) with preclinical evidence of partial non cross-resistance with Cisplatin/Carboplatin, has shown in a previous randomized phase II study (JCO. 2000 Mar;18(6):1193-202) similar activity and favorable safety profile when compared to taxol (TAX) in heavily pretreated EOC pts . We conducted a randomized phase II trial to evaluate OXA 130 mg/m¦ day (D)1/q 3 wks vs TOPO 1,5 mg/m¦ D1 to D5/ q 3 wks. Eligibility criteria included a relapsing progressive EOC, pretreatment with 1 or 2 prior Pt chemotherapy lines (Pt free-interval(PFI) <12 months), bidimensionally measurable disease, WHO performance status (PS) 0-2, adequate hematologic and renal functions. Stratification factors were: institution, PS 0-1 vs 2, PFI 0-6 mos ("refractory" pts) or 6-12 mos ("potentially sensitive" pts), number of prior Pt based regimens (1 vs 2), prior TAX (yes vs no). The primary endpoint was response rate (RR), with time to progression, overall survival, safety and quality of life as secondary endpoints. Efficacy was evaluated by radiological assessment every 6 wks and externally reviewed; responses were confirmed at least 4 wks later. 158 pts were randomized, 156 treated and evaluable for toxicity (78 pts in each arm). Pts characteristics were well balanced between the two arms. Results : the confirmed RR were 11.4% (9/79) for OXA vs 8.9% (7/79) for TOPO. The confirmed RR in Pt-refractory pts were 3.9% (2/51) for OXA and 5.7% (3/53) for TOPO. Discontinuations due to treatment-related adverse events (AE) occurred in 4 pts in each arm. There were no treatment-related deaths with OXA vs 2 with TOPO. Main AEs are shown in the table below. Conclusion: the favorable safety profile combined with the similar activity observed in this study supports the role of OXA in patients failing Pt-based chemotherapy. Final results will be presented at the meeting. Sponsors : Sanofi-Synthelabo Recherche / Debiopharm SA.

Toxicities (NCI- OXA(n=78) % TOPO(n=78) % CTC grade 3-4) pts pts Neutropenia 3.9 94 Febrile neutropenia 0 23 Thrombocytopenia 5.2 43.6 Alopecia 0 24.4 Diarrhea 11,6 3,6 Vomiting 19.2 6.4

352 Neurosensory 23 3.9

353 APPENDIX 8 – Economics included studies

Table 1: Included/Excluded studies

Study Details Include/Exclude Ojeda, 2003 (51) Cost-minimisation in 2nd line Include Frequency and cost of adverse events Neymark, 2002 (168) Cost-effectiveness in 1st line Exclude

Smith, 2001(226) Cost-minimisation 2nd line Exclude Abstract of 385 Fireman, 2000 (160) Review Exclude Szucs, 2003 (170) Review Exclude Capri, 2003 (52) Cost-minimisation in 2nd line Include Frequency and cost of adverse events Bodurka-bevers, 2000 Review Exclude (162) Some information about appropriate treatments Young, 2001 (178) Review + model of 1st line Exclude Useful info on how to model for 2nd line Smith, 2002 (50) Cost-minimisation (reports Include results of survival analysis) in 2nd line UK costs and resource use Girre, 2003 (227) Cost-minimisation in 2nd line Exclude Karlsson, 2001 Review Exclude (86) Prasad, 2003 (53) Cost-analysis in 2nd line Include Doyle, 2001 (156) Cost-analysis in 2nd/3rd line Exclude Anon, 2001 (118) Review Exclude Woronoff, 2001 Cost-minimisation in 1st line Exclude (147) abstract Sun, 2002 (107) Review QoL Exclude Le, 2003 (158) Review QoL Exclude Anon, 2001(167) Review of drug prices Exclude Calhoun, 2001 (198) Review of toxicity costs Exclude abstract Smith, 2001 (228) Cost effectiveness of 2nd line Exclude Abstract of 385 Ortega, 1997 (229) Cost-utility of 1st line Exclude

354 Table 2: Data extraction - Smith Endnote reference 385

Primary Source Database Author Smith DH, Adams JR, Johnston SRD, Gordon A, Drummond MF, Bennett CL. Date 2002 Type of economic Cost-minimisation evaluation Currency used $US (UK costs also reports in $US) Year to which costs 2000 apply Perspective used Payer in USA and UK Timeframe In absence of disease progression, treatment continued for up to 1-year Comparators Peglated liposomal doxorubicin hydrochloride (PLD) and topotecan Source(s) of Gordon, et al. See previous effectiveness data Source(s) of resource As above + expert opinion (USA and UK) to estimate the use data resource use associated with adverse events – method not described. Inpatients stays due to adverse events were only counted for those experiencing events graded 3 or 4, to account for likely excess of protocol admissions. Source(s) of unit cost UK: Drug costs from British National Formulary (British data Medical Association. British National Formulary 37. London: BMJ Books; 1999) Blood products from the National Blood Authority, 2000 Tariff (reference not given) Inpatient stays from a national costing database of hospital trusts (NHS Executive. The New NHS 1999 Reference Costs (CD-ROM). Catalogue number 10756 Leeds: NHS Executive; 1999) Intensive care unit cost from UK study (Edbrooke D, Hibbert C, Kingsley J, et al. “The patient-related costs of care for sepsis patients in a United Kingdom adult general intensive care” Critical Care Med 1999; 27: 1760-1767) Costs of outpatient visits and chemotherapy administration taken from tariffs at a UK cancer centre (reference not given)

USA: Costs were taken from prior studies looking at costs of cancer care (Stinson TJ, Calhoun E, Yang T, et al. “Cost analysis of second-line therapies for platinum-refractory ovarian cancer: reimbursement dilemmas for Medicare patients” Cancer Invest 1999; 17: 559-565) Modelling approach Resource use extrapolated between settings (Europe to used USA and vice versa). OLS regression used to predict patient specific resource use for specific items.

355 Resource use for the 35 patients with missing data was imputed as a zero cost from time of censoring. Summary of Gordon, et al. See above effectiveness results Summary of cost Total cost per person in topotecan was $12,325 higher in results North American group and $2909 higher in European group than PLD.

This cost difference was mainly due to increased cost of hospitalisations, management of neutropenia and blood transfusions for topotecan. Summary of cost- Costs and outcomes not combined effectiveness results Sensitivity analysis Some variation in resource use is lost by not using patient- specific estimates, therefore a sensitivity analysis was undertaken using only local patients to estimate resource use. In the North American analysis PLD was cost saving in all of the bootstrap replicates. In the European analysis PLD was cost saving in 89% of the bootstrap replicates.

Given that PLD was less costly a sensitivity analysis was undertaken such that all assumptions were in favour of topotecan. In the North American sensitivity analysis PLD is always cost saving. In the European sensitivity analysis PLD is cost saving 93% of the time. Main Conclusions In this clinical trial of second-line ovarian cancer treatments, PLD in comparison to topotecan was associated with similar clinical outcomes, different toxicity profiles, and higher acquisition costs, but lower toxicity management cost, leading to lower overall cost of care.

Setting also appears to play an important part in assessing costs.

Table 3: Data extraction - Ojeda et al.

Endnote reference 186 Primary Source Database

Author Ojeda B, de Sande LM, Casado A, Merino P, Casado MA. Date 2003 Type of economic Cost minimisation evaluation Currency used Euros Year to which costs 2001 apply Perspective used Health Service Timeframe In absence of disease progression treatment continued for

356 up to 1-year Comparators Peglated liposomal doxorubicin hydrochloride (PLD) and topotecan Source(s) of Gordon AN, Fleagle JT, Guthrie D, Parkin DE, Gore ME, effectiveness data Lacave AJ. “Recurrent epithelial overain carcinoma: a randomized phase III study of pegylated liposomal doxorubicin versus topotecan” Jour Clin Oncol, 2004; 19(14): 3312-3322. Trial included 474 ovarian cancer patients who had failed 1st line therapy with a platinum base agent (cisplatin or carboplatin) This trial was used to support that fact that there was no significant different in survival between the 2 treatments. and provide the resource use and adverse events associated with them. Source(s) of resource As above + expert opinion to estimate the resource use use data associated with adverse events – methods to derive opinion not described. Source(s) of unit cost Costs of drugs from Spanish catalogue of medicinal data products Unit costs of procedures and tests from Spanish database of sanitary costs (SOIKOS, 2001 “Base de datos de costes sanitarios”) and published literature (not referenced) Modelling approach No modelling undertaken used Summary of Overall progression-free survival was 113 for PLD and effectiveness results 117 for topotecan. Overall response was 19.7% for PLD and 17% for topotecan. The difference between PLD and topotecan was more marked in platinum sensitive patients median progression-free survival was 28.9 and 23.3 weeks for PLD and topotecan respectively.

Summary of cost Total costs do differ significantly between PLD and results topotecan. PLD = Euro9614 Topotecan = Euro11, 824 This cost difference is mainly due to the increase in adverse events associated with topotecan Summary of cost- Not undertaken effectiveness results Sensitivity analysis Sensitivity analysis undertaken on length of hospitalisation, total number of cycles given, cost of antiemetics, lower dosage of study drug and assuming application of all changes in favour of topotecan. Sensitivity analysis showed results were robust. Main Conclusions Total costs of drug administration per patient do not differ substantially between the 2 study drugs. However the total costs do show that costs are lower for PLD. Sensitivity analysis shows that these results are robust. Are some limitations, namely, data for analysis taken from an efficacy study not designed to specifically look at

357 resource use, adverse event resource use derived from expert opinion, quality of life and patients preferences not considered and indirect costs associated with the regimes have not been included.

Table 4: Data extraction - Capri et al.

Endnote reference 214

Primary Source Database Author Capri S, Cattaneo G Date 2003 Type of economic Cost-minimisation evaluation Currency used Euros Year to which costs 2002 apply Perspective used Italian NHS Timeframe In absence of disease progression treatment continued for up to 1-year Comparators Peglated liposomal doxorubicin hydrochloride (PLD) and topotecan Source(s) of Gordon, et al. See previous effectiveness data Source(s) of resource As above + expert opinion to estimate the resource use use data associated with adverse events – Delphi method used to derive estimates from 5 Italian oncologists Source(s) of unit cost Drug costs from the drug formulary (Drug formulary, July data 2002. Milan, Italy: Masson; 2002) Medical visits and laboratory tests from national charges (Italian Ministry of Health. “National tariffs for tests and ambulatory services”. Rome, Italy: Italian Ministry of Health; 2002) Hospitalizations from the Italian DRG’s reimbursement rates (Italian Ministry of Health. “National tariffs for hospital stays - DRGs”. Rome, Italy: Italian Ministry of Health; 2002 Modelling approach No modelling undertaken used Summary of Gordon, et al. See above effectiveness results Summary of cost Mean total costs per patient were Euro8,812 for PLD and results Euro15,788 for topotecan. This cost difference is mainly due to the increase in adverse events associated with topotecan Summary of cost- Not undertaken effectiveness results Sensitivity analysis Sensitivity analysis undertaken on resource use associated with adverse events, using minimum and maximum values (source not specified). Sensitivity analysis showed a broad

358 variation in mean costs per patient for both treatments, however the conclusion that PLS was less costly did not change. Main Conclusions Present analysis suggests the net cost-saving capability of PLD, which required fewer ambulatory care-centre visits for treatment administration and was associated with fewer severe adverse events than topotecan. The results of this analysis are consistent with those of a previous study using data from the United Stares and United Kingdom

Table 5: Data extraction - Prasad et al.

Endnote reference 2112

Primary Source Database Author Prasad M, Ben-Porat L, Hoppe B, Aghajanian C, Sabbatini P, Chi DS, Hensley ML. Date 2004 Type of economic Cost-effectiveness (?) evaluation Currency used $US Year to which costs Not stated apply Perspective used Not stated Timeframe 1 course of chemotherapy. Timeframe not stated Comparators Topotecan and gemcitabine Source(s) of Economic evaluation conducted alongside a cohort-study effectiveness data (not randomised) looking at single agent chemotherapy with topotecan or gemacitabine as 2nd line therapy in patients with recurrent (progression or recurrence within or during 6 months of treatment) platinum and paclitaxel resistant ovarian cancer. 56 patients received gemacitabine and 51 patients received topotecan. Topotecan was administered for 5-days every 21days and gemacitabine administered once a week for 3-weeks (1 week rest) Data on response available for 50/51 topotecan patients Source(s) of resource Resource use collected alongside the trial use data 5 patents receiving topotecan were not included in the cost- analysis Source(s) of unit cost Direct medical charges for the trial hospital were collected. data These were then converted to true costs using institutional, departmental or resource use, category-specific ratios of charges to costs. Modelling approach Kaplan Meier method used to estimate progression-free and used overall survival. Potential predictors of progression-free survival and overall survival were investigated using the

359 Cox proportional hazards model. Multivariate logistic regression used to analyse the covariates associated with clinical benefit. Summary of 13 patients receiving gemacitabine displayed clinical effectiveness results benefit compared to 28 receiving topotecan. Median time to progression was 1.8 months for gemacitabine and 3.6 months for topotecan. Median overall survival was 8.2 months for gemacitabine compared with 16.8 months for topotecan.

Potential predictors of response (age, number of prior salvage regimes, time from last prior treatment for recurrent/persistent disease, duration of clinical remission) were evaluated for possible associations with clinical benefit, time to progression and overall survival. There were no significant associations Summary of cost Topotecan was significantly more expensive then results gemacitabine. The mean patient cost per course was $28,098 for topotecan and $13,937 for gemacitabine.

This was mostly due to the chemotherapy drug cost and delivery. Summary of cost- Costs and outcomes were not combined effectiveness results Sensitivity analysis Sensitivity analysis not undertaken Main Conclusions The data suggests that topotecan and gemcitabine are both active as single agents and associated with moderate rates of objective clinical response and clinical benefit in heavily pre-treated platinum and paclitaxel resistant patients.

360 Table 6: Data extraction - GSK submission Author GlaxoSmithKline Date 2004 Type of economic Cost minimisation evaluation Currency used UK Pounds sterling Year to which costs A unique price year was not given apply Perspective used UK NHS Timeframe Treatment continued for up to 1-year Comparators Topotecan and PLD Source(s) of Gordon et al.2 effectiveness data Trial included 474 ovarian cancer patients who had failed 1st line therapy with a platinum base agent (cisplatin or carboplatin) This trial was used to support that fact that there was no significant different in survival between the 2 treatments and to provide resource use associated with adverse events (except neutropenia) Source(s) of resource An audit of current practice for relapsed ovarian cancer in use data UK was used to collect information on resource use. This audit included 9 UK centres and 92 patients receiving topotecan versus 60 patients receiving PLD.

Mean drug dosage (for topotecan and PLD), number of treatment cycles, resource use associated to drug administration and management of neutropenia were directly taken from the audit. Resource utilisation associated with other adverse events was taken from the Gordon et al. study and adjusted by audit data, assuming that lower drug doses would lead to a reduction in the rate of adverse events. Source(s) of unit cost Drug costs, blood products and inpatients stay were taken data from the Smith. study3. In this study the sources of unit costs were the following: Drug costs from British National Formulary (British Medical Association. British National Formulary 37. London: BMJ Books; 1999) Blood products from the National Blood Authority, 2000 Tariff (reference not given) Inpatient stays from a national costing database of hospital trusts (NHS Executive. The New NHS 1999 Reference Costs (CD-ROM). Catalogue number 10756 Leeds: NHS Executive; 1999). Outpatients visits, consultations and monitoring tests were instead taken from NHS Reference costs 2003 and National Tariffs 2004 (reference not given) Modelling approach No modelling undertaken used

361 Summary of The overall response rate was 19.7% in the PLD group effectiveness results versus 17% in the topotecan group (p=0.39), while median PFS was 16.1 weeks versus 17.0 weeks respectively (p=0.095) and overall survival 60 weeks versus 56.7 weeks respectively (p=0.341). However, the difference between PLD and topotecan was more marked in platinum sensitive patients where the median progression-free survival was 28.9 and 23.3 weeks for PLD and topotecan respectively (p=0.037).

Summary of cost Total cost per patient was estimated as £7,773.03 for results topotecan and £8,080.43 for PLD. The higher drug costs for PLD were only partly offset by lower administration costs and adverse event costs. Summary of cost- Costs and outcomes not combined effectiveness results Sensitivity analysis No sensitivity analyses were performed Main Conclusions Topotecan and PLD resulted similar in terms of both costs and effectiveness data, although some differences in drug toxicity were found. It is inappropriate to recommend PLD as preferred option and both treatments should be recommended as equal choices for 2nd line therapies of epithelial ovarian cancer.

Table 7: Data extraction - Schering submission Author Schering-plough Ltd Date 2004 Type of economic Cost minimisation evaluation Currency used UK Pounds sterling Year to which costs A unique price year was not given apply Perspective used UK NHS Timeframe Economic analysis performed on 1-year effectiveness data. Long-term follow-up (3 years) for health outcomes was also conducted. Comparators Topotecan and PLD Source(s) of Gordon et al.-see previous effectiveness data Long-term follow-up efficacy analysis based on: Gordon AN, Tonda M, Sun S, Rackoff W. Long-term survival advantage for women treated with pegylated liposomal doxorubicin compared with topotecan in a phase 3 randomised study of current and refractory epithelian ovarian cancer. Ginecologic Oncology 2004, article in press

Source(s) of resource Drug mean dosage, number of treatment cycles and use data resource use for drug administration were directly taken from trial data. Expert opinion was used to estimate the

362 resource use associated with adverse events. Source(s) of unit cost Drug costs from British National Formulary (British data Medical Association. British National Formulary 37. London: BMJ Books; 1999) Blood products from the National Blood Authority, 2000 Tariff (reference not given) In-patient stays from the 1999 CIPFA database (Institute- of-Public-Finance. CIPFA/HFM Health Database; 1999) ICU based on literature (Edbrooke D, Hibbert C, Kingsley J, Smith S, Bright N, Quinn J. The patient-related costs of care for sepsis patients in a United Kingdom adult general intensive care. Crit Care Med 1999;27(9):1760-7) Outpatient clinical visit from a tariffs at a UK cancer centre (no reference given) Modelling approach No modelling undertaken used Summary of Short-term results for topotecan and PLD- see previous effectiveness results table The median survival weeks in a 3-year follow-up were 62.7 for PLD versus 59.7 for topotecan with a hazard rate that almost reach statistical significance (p=0.05). The trend in favour of PLD was more evident when platinum- sensitive patients were considered. In this group of patients, a statistically significant difference in median survival was found and a higher percentage of patients were alive at 2 and 3-year follow-up. Summary of cost Total costs per patient were £9,957 for PLD versus results £12,610 for topotecan. This difference was statistically significant. The higher drug cost for PLD was more than compensated by lower administration costs and lower costs associated with adverse events (in particular for neutropenia). Summary of cost- Costs and outcomes not combined effectiveness results Sensitivity analysis A extreme analysis favouring topotecan was conducted. PLD remained statistically less costly than topotecan showing the robustness of the basecase results Main Conclusions PLD was associated with similar clinical outcomes compared to topotecan in the short term but in a 3-year follow-up analysis it demonstrated survival advantage over topotecan. Although PLD was associated with higher acquisition costs, the lower toxicity management cost and administration costs led to a statistically significant lower total per patient costs for PLD compared with topotecan. It was concluded that PLD should be considered as the dominant option, being less costly and more effective.

Table 8: Data extraction - BMS submission

363

Author Bristol-Myers Squibb Pharmaceutical Ltd Date 2004 Type of economic Cost-effectiveness evaluation Currency used UK Pounds sterling Year to which costs A unique price year was not given apply Perspective used UK NHS Timeframe Three years Comparators Paclitaxel combined with a platinum agent, paclitaxel alone, topotecan and PLD Source(s) of Evidence for paclitaxel/platinum was taken from the ICON effectiveness data 4 study8. This was a multicentre, randomised clinical trial including 802 platinum-sensitive ovarian cancer patients (relapsing after 6 months). Evidence for paclitaxel monotherapy and topotecan were taken from a study by tenBokkel Huinink et al.7. This was a multicentre randomised trial including 226 ovarian cancer women who had failed a previous platinum-based regimen. Evidence for PLD was taken from Forbes et al.9, a previous Schering-Plough submission to NICE. All these studies were used to collect data on drug effectiveness, drug dosage and administration costs. Source(s) of resource Study drug dosage and drug administration methods were use data taken from the studies mentioned above. The number of treatment cycles for each therapy appears to have been based on authors’ assumptions. The dose of carboplatin was based on the literature (Calvert AH et al. Carboplatin dosage: Prospective evaluation of a simple formula based on renal function. J Clin Oncol;7:1748-1756) For paclitaxel, pre-medication therapies were based on product indications. Source(s) of unit cost Drug costs from British National Formulary, vol 47, 2004 data Outpatient visit from PSSRU (Netten A. and Curtis L. Unit costs of Health & Social Care 2000, PSSRU, University of Kent, Canterbury.)

Modelling approach A model was constructed in order to calculate the life used years gained over three years from clinical trials survival curves. Life years were estimated averaging the proportion of patients alive between two time periods. Summary of The proportion of patients alive at years 1, 2 and 3 were effectiveness results 78%, 43% and 24% respectively for paclitaxel/platinum, 50%, 22% and 15% for paclitaxel alone, 56%, 22% and 18% for topotecan and 56%, 30% and 18% for PLD. These survival rates led to 1.83 LYG with

364 paclitaxel/platinum, 1.295 LYG with paclitaxel alone, 1.37 LYG with topotecan and 1.45 LYG with PLD. Summary of cost Total per patient cost over 3 years (at 6% discount rate) results was estimated at £25,287.79 for paclitaxel/platinum, £18,998.01 for paclitaxel alone, £30,297.51 for topotecan and £26,992.73 for PLD. Summary of cost- Paclitaxel/platinum resulted dominant compared to effectiveness results topotecan and PLD, being less costly and associated with more LYG over 3 years. The combined therapy was relatively cost-effective compared to paclitaxel alone with an ICER of £12,120.07 per LYG. Sensitivity analysis One-way sensitivity analyses were conducted varying the discount rate for health benefits (from 0% to 6%). The basecase results were confirmed in these analyses. LYG were also varied by 15% but its impact was only assessed on ACERs and not on ICERs Main Conclusions It was concluded that paclitaxel/platinum should be considered as the preferred option for treatment of platinum-sensitive ovarian cancer patients. It resulted dominant compared with topotecan and PLD and cost- effective compared with paclitaxel monotherapy.

365 Table 9: Quality assessment economics published studies

Capri Prasad Smith Ojeda Study question 1. Costs and effects x Assumption of 9 Cost-effectiveness x Assumption of x Assumption of examined equivalence between results not reported equivalence between equivalence between treatments based on treatments based on treatments based on data from a previously data from a previously data from a previously reported trial. Main reported trial. Main reported trial. Main clinical results were clinical results were clinical results were reported reported reported 2. Alternatives compared 9 9 9 PLD ‘v’ topotecan 9 3. The 9 x 9 9 viewpoint(s)/perspective of the analysis is clearly stated (e.g. NHS, society) Selection of alternatives 4. All relevant alternatives x Do nothing not x Topotecan not X Do nothing not x Do nothing not are compared (including do- appropriate. compared to multiple appropriate. appropriate. nothing if applicable) However trial did not agents regimes However trial did not However trial did not include palliative care include palliative care include palliative care as an option. as an option. as an option. 5. The alternatives being 9 x Doses not given 9 Refers back to trial for 9 compared are clearly full details described (who did what, to whom, where and how often) 6. The rationale for choosing 9 x 9 As in trial evidence 9 Based on trial data the alternative programmes or interventions compared is stated

366 Form of evaluation 7. The choice of form of 9 9 9 9 economic evaluation is justified in relation to the questions addressed. 8. If a cost-minimisation 9 In terms of overall N/A 9 In terms of overall 9 In terms of overall design is chosen, have survival, however there survival, however there survival, however there equivalent outcomes been were differences in sub- were differences in sub- were differences in sub- adequately demonstrated? groups of patients groups of patients groups of patients Effectiveness data 9. The source(s) of 9 9 9 9 effectiveness estimates used are stated (e.g. single study, selection of studies, systematic review, expert opinion) 10. Effectiveness data from 9 x Non randomised study 9 9 RCT or review of RCTs 11. Potential biases 9 x 9 9 identified (especially if data not from RCTs) 12. Details of the method of N/A N/A N/A N/A synthesis or meta-analysis of estimates are given (if based on an overview of a number of effectiveness studies) Costs 13. All the important and x Indirect costs not x Indirect costs not X Indirect costs not x Indirect costs not relevant resource use included included considered included included 14. All the important and 9 9 X Expert panel used to x Expert panel used to relevant resource use estimate resource use estimate resource use measured accurately (with associated with adverse associated with adverse

367 methodology) events. Methods used events. Methods used were not described. were not described. 15. Appropriate unit costs 9 9 9 9 estimated (with methodology) 16. Unit costs reported 9 x 9 9 separately from resource use data 17. Productivity costs treated N/A N/A N/A Not included N/A separately from other costs 18. The year and country to 9 x 9 9 which unit costs apply is stated with appropriate adjustments for inflation and/or currency conversion. Benefit measurement and valuation 19. The primary outcome N/A 9 9 9 measure(s) for the economic evaluation are clearly stated (cases detected, life years, QALYs, etc.) 20. Methods to value health N/A N/A N/A N/A states and other benefits are stated (e.g. time trade off) 21. Details of the individuals N/A N/A N/A N/A from whom valuations were obtained are given (patients, members of the public, health care professionals etc.) Decision modelling 22. Details of any decision N/A N/A N/A N/A model used are given (e.g. decision tree, Markov model) 23. The choice of model N/A N/A N/A N/A

368 used and the key input parameters on which it is based are adequately detailed and justified 24. All model outputs N/A N/A N/A N/A described adequately. Discounting 25. Discount rate used for N/A N/A Resource use N/A Resource use N/A Resource use both costs and benefits over 1 cycle of up to 1-year collected for chemotherapy up to 1-year 26. Do discount rates accord N/A N/A N/A N/A with NHS guidance (1.5%- 2% for benefits; 6% for costs)? Allowance for uncertainty Stochastic analysis of patient- level data 27. Details of statistical tests x Not x Not x Not x Not and confidence intervals are undertaken undertaken undertaken undertaken given for stochastic data 28. Uncertainty around cost- N/A N/A N/A N/A effectiveness expressed (e.g. confidence interval around incremental cost- effectiveness ratio (ICER), cost-effectiveness acceptability curves). 29. Sensitivity analysis used N/A N/A N/A N/A to assess uncertainty in non- stochastic variables (e.g. unit costs, discount rates) and analytic decisions (e.g. methods to handle missing data).

369 Stochastic analysis of decision models

30. Are all appropriate input N/A N/A N/A N/A parameters included with uncertainty? 31. Is second-order N/A N/A N/A N/A uncertainty (uncertainty in means) included rather than first order (uncertainty between patients)? 32. Are the probability N/A N/A N/A N/A distributions adequately detailed and appropriate? 33. Sensitivity analysis used N/A N/A N/A N/A to assess uncertainty in non- stochastic variables (e.g. unit costs, discount rates) and analytic decisions (e.g. methods to handle missing data). Deterministic analysis 34. The approach to 9 N/A Sensitivity 9 9 sensitivity analysis is given analysis not (e.g. univariate, threshold undertaken analysis etc) 35. The choice of variables 9 N/A 9 9 for sensitivity analysis is justified 36. The ranges over which x Not specified N/A 9 9 the variables are varied are stated Presentation of results 37. Incremental analysis is x N/A x x

370 reported using appropriate decision rules 38. Major outcomes are 9 9 9 9 presented in a disaggregated as well as aggregated form 39. Applicable to the NHS x Trial in US. x Not directly x Costs in £UK x Trial in US. setting Costs from applicable with not given Costs from Italy out adjustment Spain for differences in clinical practice

Key: 9 = Item properly addressed x = Item not properly addressed N/A = Not clear

371 Table 10: Quality assessment industry submitted studies

GSK Schering BMS Study question 1. Costs and effects examined x It was assumed equivalence of r It was assumed equivalence efficacy between treatments of efficacy between based on results from a previous treatments based on results clinical trial. Few data from the from a previous clinical trial. trial were reported Survival benefit for PLD in the long-term were reported 2. Alternatives compared PLD and topotecan Palcitaxel, PLD and topotecan paclitaxel/carboplatin, PLD and topotecan NHS NHS NHS 3. The viewpoint(s)/perspective of the analysis is clearly stated (e.g. NHS, society) Selection of alternatives 4. All relevant alternatives are compared Do nothing is not an appropriate Do nothing is not an Do nothing is not an (including do-nothing if applicable) alternative. Placlitaxel was not appropriate alternative. appropriate alternative. considered because lack of new Placlitaxel was not data since last NICE guidance considered because of lack of new data 5. The alternatives being compared are s However, for some details the However, for some details clearly described (who did what, to whom, reader is referred to the original the reader is referred to the where and how often) trial or the audit. original trial or the audit. 6. The rationale for choosing the alternative programmes or interventions compared is stated Form of evaluation 7. The choice of form of economic r evaluation is justified in relation to the questions addressed. 8. If a cost-minimisation design is chosen, r In terms of overall survival N/A s In terms of overall survival

372 have equivalent outcomes been adequately differences were found in sub- differences were found in demonstrated? group analyses; also different sub-group analyses; also toxicity profiles were found different toxicity profiles between the two treatments were found between the two treatments. Survival benefits for PLD were found in the long-term follow-up- Effectiveness data 9. The source(s) of effectiveness estimates r used are stated (e.g. single study, selection of studies, systematic review, expert opinion)

10. Effectiveness data from RCT or review The majority of effectiveness Effectiveness data are derived Effectiveness data are of RCTs data were taken from a trial. from one clinical trial derived from one clinical However, adverse events rate trial was obtained from an audit 11. Potential biases identified (especially if r The potential biases of the audit r Differences in patient data not from RCTs) were not identified and populations in the trial used addressed for obtaining benefit measure were not stated 12. Details of the method of synthesis or N/A N/A meta-analysis of estimates are given (if based on an overview of a number of effectiveness studies) Costs 13. All the important and relevant resource N/A All NHS costs included r Resource use associated to use included adverse events were not included 14. All the important and relevant resource s Some resource use associated r The number of treatment s Resource utilisation derived use measured accurately (with methodology) with adverse events derived cycles used was not clearly from expert opinion for from expert opinion and the justified management of adverse methods used were not described events (Smith.). Also, the reduction in

373 the rate of adverse events equal to the reduction in cycles of treatment could be questionable. 15. Appropriate unit costs estimated (with methodology) 16. Unit costs reported separately from However, resource use for some resource use data adverse events were not reported 17. Productivity costs treated separately N/A Productivity costs not NA Productivity costs not NA Productivity costs not from other costs considered. considered. considered. 18. The year and country to which unit costs N/A Unit costs from different price s Unit costs from different price s Unit costs from different apply is stated with appropriate adjustments years were used years were used price years were used for inflation and/or currency conversion. Benefit measurement and valuation 19. The primary outcome measure(s) for the N/A Life years gained economic evaluation are clearly stated (cases detected, life years, QALYs, etc.) 20. Methods to value health states and other N/A N/A Area under survival curve benefits are stated (e.g. time trade off) 21. Details of the individuals from whom N/A N/A N/A valuations were obtained are given (patients, members of the public, health care professionals etc.)

Decision modelling 22. Details of any decision model used are N/A N/A N/A given (e.g. decision tree, Markov model) 23. The choice of model used and the key N/A N/A N/A input parameters on which it is based are adequately detailed and justified 24. All model outputs described adequately. N/A N/A N/A Discounting 25. Discount rate used for both costs and r Resource use up to 1-year N/A Resource use up to 1 year

374 benefits 26. Do discount rates accord with NHS N/A N/A Discount rate as guidance (1.5%-2% for benefits; 6% for recommended in the costs)? previous NICE guidelines (6% for costs; 1.5% for benefits) Allowance for uncertainty Stochastic analysis of patient-level data N/A 27. Details of statistical tests and confidence N/A Deterministic analysis. N/A Deterministic analysis. N/A Deterministic analysis. intervals are given for stochastic data 28. Uncertainty around cost-effectiveness N/A N/A N/A expressed (e.g. confidence interval around incremental cost-effectiveness ratio (ICER), cost-effectiveness acceptability curves). 29. Sensitivity analysis used to assess N/A N/A N/A uncertainty in non-stochastic variables (e.g. unit costs, discount rates) and analytic decisions (e.g. methods to handle missing data).

Stochastic analysis of decision models

30. Are all appropriate input parameters r N/A N/A included with uncertainty? 31. Is second-order uncertainty (uncertainty r N/A N/A in means) included rather than first order (uncertainty between patients)? 32. Are the probability distributions r N/A N/A adequately detailed and appropriate? 33. Sensitivity analysis used to assess r N/A N/A uncertainty in non-stochastic variables (e.g. unit costs, discount rates) and analytic decisions (e.g. methods to handle missing data).

375 Deterministic analysis 34. The approach to sensitivity analysis is r No sensitivity analyses were Univariate given (e.g. univariate, threshold analysis etc) performed 35. The choice of variables for sensitivity r However, sensitivity analyses r Sensitivity analyses were analysis is justified were not conducted for all not conducted for relevant relevant parameters parameters 36. The ranges over which the variables are r varied are stated Presentation of results 37. Incremental analysis is reported using r No sensitivity analyses were r Cost-minimisation analysis appropriate decision rules performed 38. Major outcomes are presented in a r disaggregated as well as aggregated form 39. Applicable to the NHS setting r

376 Table 17: Mean per-patient cost of adverse events, as estimated by Capri, et al

Adverse event Grade Cost (euro) Nausea/vomiting 1 10 2 71 3 81 4 1001 PPE 1 0 2 0 3 37 4 39 Stomatitis 1 14 2 28 3 908 4 1385 Anaemia 1 21 2 1370 3 2351 4 2943 Neutropenia 1 0 2 0 3 284 4 1554 Diarrhoea 1 10 2 22 3 921 4 1219 Thrombocytopenia 1 0 2 0 3 0 4 1870 Fever 1 3 2 5 3 1235 4 1343 Sepsis 1 0 2 23 3 1440 4 1449

Table 20a: Results from the GSK audit for patients receiving topotecan

Topotecan as line Total number of Mean dose per cycle No. cycles of therapy patients mg/m2 (SD) (Mean) 2nd Line 42 1.47 (0.38) 4.19 3rd Line 28 1.50 (0.23) 3.57 4th Line 14 1.57 (0.46) 4.31

377 5th 1.43 (0.12) 2.75 (and subsequent 8 lines) Over all lines 92 1.49 3.88

Table 20b. Results from the GSK audit for patients receiving PLD

PLD as line of Total Number of Mean dose per cycle No. cycles chemotherapy Patients mg/m2 (SD) (Mean) 2nd Line 6 50 (0) 3 3rd Line 22 48.4 (3.7) 3.6 4th Line 14 50 (16.3) 3.5 5th 48.9 (7.6) 3.8 (and subsequent 18 lines) Over all lines 60 49.1(11.0) 3.5

Table 24: Numbers and percentages of patients with maintenance of or improvement in HQL scores at 12 weeks (Study 30-49)

Scales No. Patients and % Caelyx Topotecan Physical Functioning 66/118 56% 60/107 56% Role Functioning 77/118 65% 63/109 58% Emotional Functioning 80/119 67% 80/108 74% Cognitive Functioning 87/119 73% 79/108 73% Social Functioning 82/119 69% 69/108 64% Global QOL 68/117 58% 54/104 52% Fatigue 67/118 57% 61/109 56% Nausea / Vomiting 86/119 72% 77/109 71% Pain 76/119 64% 88/109 81%

Table 27. Total annual costs per patients for the four chemotherapeutic agents (Includes administration costs (£68) plus pre-medication costs (£2.96))

Treatment Cost of study Administration Number of Total costs regimen drug (per costs (per cycle) cycles (per year) cycle)

PLD 1,626.98 68 6 10,169.88

Topotecan 1,562.50 340 6 11,415.00

378 Paclitaxel 1,122.00 70.96* 6 7,157.76 monotherapy

Paclitaxel plus 1,516.96 70.96* 6 9,527.52 carboplatin

379 APPENDIX 9: Utility search strategies

These literature searches aimed to identify all studies that include quality of life data in ovarian cancer. The search strategies did not include any drug-specific terms.

The literature searches were conducted in May 2004, were limited to studies published since 1984. There were no limits applied by study design or language of publication. Full details of all databases searched and strategies used are listed below.

This stage of the literature searching retrieved 1,321 unique references after deduplication. References were managed using Endnote software version 6.

Databases

Searches were run on the following databases:

MEDLINE PREMEDLINE EMBASE CINAHL Database of Abstracts of Reviews of Effectiveness (DARE) NHS Economic Evaluation database (NHS EED) HTA database Cochrane Controlled Trials Register Cochrane Database of Systematic Reviews Science Citation Index Index to Scientific and Technical Proceedings Office of Health Economics Health Economic Evaluations Database (OHE HEED)

Search Strategies

Limits: • Date limits: 1984 - date • Animal-only studies were excluded where possible • Study design: No limits • Language: No limits • Publication type: Letters, comments and editorials were excluded where possible

MEDLINE (Ovid host) 1984 - April 2004

380 mesh subject heading] 4. or/1-3 5. animal/ not (animal/ and human/) 6. 4 not 5 7. exp Life Tables/ 8. exp "Quality of Life"/ 9. Health Status/ 10. exp Health Status Indicators/ 11. (utilit$ approach$ or health gain or hui or hui2 or hui 2 or hui3 or hui 3).ti,ab. 12. (health measurement$ scale$ or health measurement$ questionnaire$).ti,ab. 13. (standard gamble$ or categor$ scal$ or linear scal$ or linear analog$ or visual scal$ or magnitude estmat$).ti,ab. 14. (time trade off$ or rosser$ classif$ or rosser$ matrix or rosser$ distress$ or hrqol).ti,ab. 15. (index of wellbeing or quality of wellbeing or qwb).ti,ab. 16. (rating scale$ or multiattribute$ health ind$ or multi attribute$ health ind$).ti,ab. 17. (health utilit$ index or health utilit$ indices).ti,ab. 18. (multiattribute$ theor$ or multi attribute$ theor$ or multiattribute$ analys$ or multi attribute$ analys$).ti,ab. 19. (health utilit$ scale$ or classification of illness state$ or 15d or 15 d or 15 dimension).ti,ab. 20. (health state$ utilit$ or 12d or 12 d or 12 dimension).ti,ab. 21. well year$.ti,ab. 22. (multiattribute$ utilit$ or multi attribute$ utilit$).ti,ab. 23. health utilit$ scale$.ti,ab. 24. (qol or 5d or 5-d or 5 dimension or quality of life or eq-5d or eq5d or eq 5d or euroqol).ti,ab. 25. (qualy or qaly or qualys or qalys or quality adjusted life year$).ti,ab. 26. life year$ gain$.ti,ab. 27. willingness to pay.ti,ab. 28. (hye or hyes or health$ year$ equivalent$).ti,ab. 29. (person trade off$ or person tradeoff$ or time tradeoff$ or time trade off$).ti,ab. 30. theory utilit$.ti,ab. 31. life table$.ti,ab. 32. health state$.ti,ab. 33. (sf36 or sf 36).ti,ab. 34. (short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirtysix or short form thirty six).ti,ab. 35. (6d or 6-d or 6 dimension).ti,ab. 36. or/7-35 37. 6 and 36 38. letter.pt. 39. editorial.pt. 40. comment.pt. 41. or/38-40 42. 37 not 41 43. limit 42 to yr=1984 - 2004

PREMEDLINE

381 (Ovid host) 1984 - May 2004

1. (ovar$ adj4 (cancer$ or tumo?r$ or malignan$ or oncolog$ or carcinoma$ or neoplas$ or mass$ or growth$ or cyst$)).mp. [mp=title, abstract] 2. (adenexa$ adj4 mass$).mp. [mp=title, abstract] 3. (utilit$ approach$ or health gain or hui or hui2 or hui 2 or hui3 or hui 3).ti,ab. 4. (health measurement$ scale$ or health measurement$ questionnaire$).ti,ab. 5. (standard gamble$ or categor$ scal$ or linear scal$ or linear analog$ or visual scal$ or magnitude estmat$).ti,ab. 6. (time trade off$ or rosser$ classif$ or rosser$ matrix or rosser$ distress$ or hrqol).ti,ab. 7. (index of wellbeing or quality of wellbeing or qwb).ti,ab. 8. (rating scale$ or multiattribute$ health ind$ or multi attribute$ health ind$).ti,ab. 9. (health utilit$ index or health utilit$ indices).ti,ab. 10. (multiattribute$ theor$ or multi attribute$ theor$ or multiattribute$ analys$ or multi attribute$ analys$).ti,ab. 11. (health utilit$ scale$ or classification of illness state$ or 15d or 15 d or 15 dimension).ti,ab. 12. (health state$ utilit$ or 12d or 12 d or 12 dimension).ti,ab. 13. well year$.ti,ab. 14. (multiattribute$ utilit$ or multi attribute$ utilit$).ti,ab. 15. health utilit$ scale$.ti,ab. 16. (qol or 5d or 5-d or 5 dimension or quality of life or eq-5d or eq5d or eq 5d or euroqol).ti,ab. 17. (qualy or qaly or qualys or qalys or quality adjusted life year$).ti,ab. 18. life year$ gain$.ti,ab. 19. willingness to pay.ti,ab. 20. (hye or hyes or health$ year$ equivalent$).ti,ab. 21. (person trade off$ or person tradeoff$ or time tradeoff$ or time trade off$).ti,ab. 22. theory utilit$.ti,ab. 23. life table$.ti,ab. 24. health state$.ti,ab. 25. (sf36 or sf 36).ti,ab. 26. (short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirtysix or short form thirty six).ti,ab. 27. (6d or 6-d or 6 dimension).ti,ab. 28. or/1-2 29. or/3-27 30. 28 and 29 31. letter.pt. 32. comment.pt. 33. editorial.pt. 34. or/31-33 35. 30 not 34

EMBASE (Ovid host) 1984 - May 2004

382 1. exp Ovary Cancer/ 2. (ovar$ adj4 (cancer$ or tumo?r$ or malignan$ or oncolog$ or carcinoma$ or neoplas$ or mass$ or growth$ or cyst$)).mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 3. (adenexa$ adj4 mass$).mp. [mp=title, abstract, subject headings, drug trade name, original title, device manufacturer, drug manufacturer name] 4. exp "Quality of Life"/ 5. Health Status/ 6. (utilit$ approach$ or health gain or hui or hui2 or hui 2 or hui3 or hui 3).ti,ab. 7. (health measurement$ scale$ or health measurement$ questionnaire$).ti,ab. 8. (standard gamble$ or categor$ scal$ or linear scal$ or linear analog$ or visual scal$ or magnitude estmat$).ti,ab. 9. (time trade off$ or rosser$ classif$ or rosser$ matrix or rosser$ distress$ or hrqol).ti,ab. 10. (index of wellbeing or quality of wellbeing or qwb).ti,ab. 11. (rating scale$ or multiattribute$ health ind$ or multi attribute$ health ind$).ti,ab. 12. (health utilit$ index or health utilit$ indices).ti,ab. 13. (multiattribute$ theor$ or multi attribute$ theor$ or multiattribute$ analys$ or multi attribute$ analys$).ti,ab. 14. (health utilit$ scale$ or classification of illness state$ or 15d or 15 d or 15 dimension).ti,ab. 15. (health state$ utilit$ or 12d or 12 d or 12 dimension).ti,ab. 16. well year$.ti,ab. 17. (multiattribute$ utilit$ or multi attribute$ utilit$).ti,ab. 18. health utilit$ scale$.ti,ab. 19. (qol or 5d or 5-d or 5 dimension or quality of life or eq-5d or eq5d or eq 5d or euroqol).ti,ab. 20. (qualy or qaly or qualys or qalys or quality adjusted life year$).ti,ab. 21. life year$ gain$.ti,ab. 22. willingness to pay.ti,ab. 23. (hye or hyes or health$ year$ equivalent$).ti,ab. 24. (person trade off$ or person tradeoff$ or time tradeoff$ or time trade off$).ti,ab. 25. theory utilit$.ti,ab. 26. life table$.ti,ab. 27. health state$.ti,ab. 28. (sf36 or sf 36).ti,ab. 29. (short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirtysix or short form thirty six).ti,ab. 30. (6d or 6-d or 6 dimension).ti,ab. 31. life table/ 32. health survey/ 33. or/1-3 34. or/4-32 35. 33 and 34 36. animal/ or nonhuman/ 37. human/ 38. 36 not (36 and 37) 39. 35 not 38 40. (editorial or letter).pt. 41. 39 not 40

383 42. 41 43. limit 42 to yr=1984 - 2004

CINAHL (Ovid host) 1984 - April 2004

1. exp Ovarian Neoplasms/ 2. (ovar$ adj4 (cancer$ or tumo?r$ or malignan$ or oncolog$ or carcinoma$ or neoplas$ or mass$ or growth$ or cyst$)).mp. [mp=title, cinahl subject headings, abstract, instrumentation] 3. (adenexa$ adj4 mass$).mp. [mp=title, cinahl subject headings, abstract, instrumentation] 4. or/1-3 5. exp "Quality of Life"/ 6. Health Status/ 7. exp Health Status Indicators/ 8. (utilit$ approach$ or health gain or hui or hui2 or hui 2 or hui3 or hui 3).ti,ab. 9. (health measurement$ scale$ or health measurement$ questionnaire$).ti,ab. 10. (standard gamble$ or categor$ scal$ or linear scal$ or linear analog$ or visual scal$ or magnitude estmat$).ti,ab. 11. (time trade off$ or rosser$ classif$ or rosser$ matrix or rosser$ distress$ or hrqol).ti,ab. 12. (index of wellbeing or quality of wellbeing or qwb).ti,ab. 13. (rating scale$ or multiattribute$ health ind$ or multi attribute$ health ind$).ti,ab. 14. (health utilit$ index or health utilit$ indices).ti,ab. 15. (multiattribute$ theor$ or multi attribute$ theor$ or multiattribute$ analys$ or multi attribute$ analys$).ti,ab. 16. (health utilit$ scale$ or classification of illness state$ or 15d or 15 d or 15 dimension).ti,ab. 17. (health state$ utilit$ or 12d or 12 d or 12 dimension).ti,ab. 18. well year$.ti,ab. 19. (multiattribute$ utilit$ or multi attribute$ utilit$).ti,ab. 20. health utilit$ scale$.ti,ab. 21. (qol or 5d or 5-d or 5 dimension or quality of life or eq-5d or eq5d or eq 5d or euroqol).ti,ab. 22. (qualy or qaly or qualys or qalys or quality adjusted life year$).ti,ab. 23. life year$ gain$.ti,ab. 24. willingness to pay.ti,ab. 25. (hye or hyes or health$ year$ equivalent$).ti,ab. 26. (person trade off$ or person tradeoff$ or time tradeoff$ or time trade off$).ti,ab. 27. theory utilit$.ti,ab. 28. life table$.ti,ab. 29. health state$.ti,ab. 30. (sf36 or sf 36).ti,ab. 31. (short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirtysix or short form thirty six).ti,ab. 32. (6d or 6-d or 6 dimension).ti,ab. 33. letter.pt. 34. editorial.pt.

384 35. comment.pt. 36. or/33-35 37. Life Table Method/ 38. or/5-37 39. 4 and 38 40. letter.pt. 41. editorial.pt. 42. or/40-41 43. 39 not 42

DARE NHS EED HTA Database (http://www.york.ac.uk/inst/crd/crddatabases.htm) 1984 - April 2004 s ovar$(4w)(cancer$ or tumor$ or tumour$ or malignan$ or oncolog$ or carcinoma$ or neoplas$ or mass$ or growth$ or cyst$) s adenexa$(4w)mass$ s s1 or s2 s health(w)status s utilit$(w)approach$ or health(w)gain or hui or hui2 or hui(w)2 or hui3 or hui(w)3 s health(w)measurement$(w)scale$ or health(w)measurement$(w)questionnaire$ s standard(w)gamble$ or categor$(w)scal$ or linear(w)scal$ or linear(w)analog$ or visual(w)scal$ or magnitude(w)estmat$ s time(w)trade(w)off$ or rosser$(w)classif$ or rosser$(w)matrix or rosser$(w)distress$ or hrqol s index(2w)wellbeing or quality(2w)wellbeing or qwb s rating(w)scale$ or multiattribute$(w)health(w)ind$ or multi(w)attribute$(w)health(w)ind$ s health(w)utilit$(w)index or health(w)utilit$(w)indices s multiattribute$(w)theor$ or multi(w)attribute$(w)theor$ or multiattribute$(w)analys$ or multi(w)attribute$(w)analys$ s health(w)utilit$(w)scale$ or classification(2w)illness(w)state$ or 15d or 15(w)d or 15(w)dimension s health(w)state$(w)utilit$ or 12d or 12(w)d or 12(w)dimension s well(w)year$ s multiattribute$(w)utilit$ or multi(w)attribute$(w)utilit$ s health(w)utilit$(w)scale$ s qol or 5d or 5(w)d or 5(w)dimension or quality(2w)life or eq(w)5d or eq5d or euroqol s qualy or qaly or qualys or qalys or quality(w)adjusted(w)life(w)year$ s life(w)year$(w)gain$ s willingness(w)to(w)pay s hye or hyes or health$(w)year$(w)equivalent$ s person(w)trade(w)off$ or person(w)tradeoff$ or time(w)tradeoff$ or time(w)trade(w)off$ s theory(w)utilit$ s life(w)table$ s health(w)state$

385 s sf36 or sf(w)36 s short(w)form(w)36 or shortform(w)36 or sf(w)thirtysix or sf(w)thirty(w)six or shortform(w)thirtysix or shortform(w)thirty(w)six or short(w)form(w)thirtysix or short(w)form(w)thirty(w)six s 6d or 6(w)d or 6(w)dimension s s4 or s5 or s6 or s7 or s8 or s9 or s10 or s11 or s12 or s13 or s14 or s15 or s16 or s17 or s18 or s19 or s20 or s21 or s22 or s23 or s24 or s25 or s26 or s27 or s28 or s29 s s3 and s30

Cochrane Controlled Trials Register Cochrane Database of Systematic Reviews (The Cochrane Library) 1984 - Issue 2, 2004

#1. OVARIAN NEOPLASMS explode all trees (MeSH) #2. (ovar* near cancer*) #3. (ovar* near tumor*) #4. (ovar* near tumour*) #5. (ovar* near malignan*) #6. (ovar* near oncolog*) #7. (ovar* near carcinoma) #8. (ovar* near neoplas*) #9. (ovar* near mass*) #10. (ovar* near growth*) #11. (ovar* near cyst*) #12. (adenexa* near mass*) #13. (#1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11 or #12) #14. LIFE TABLES explode all trees (MeSH) #15. QUALITY OF LIFE explode all trees (MeSH) #16. ((utilit* next approach*) or (health next gain) or hui or hui2 or hui-2 or hui3 or hui-3) #17. ((health next measurement* next scale*) or (health next measurement* next questionnaire*)) #18. ((standard next gamble*) or (categor* next scal*) or (linear next scal*) or (linear next analog*) or (visual next scal*) or (magnitude next estmat*)) #19. ((time next trade next off*) or (rosser* next classif*) or (rosser* next matrix) or (rosser* next distress*) or hrqol) #20. ((index next wellbeing) or (quality next wellbeing) or qwb) #21. ((rating next scale*) or (multiattribute* next health next ind*) or (multi next attribute* next health next ind*)) #22. ((health next utilit* next index) or (health next utilit* next indices)) #23. ((multiattribute* next theor*) or (multi next attribute* next theor*) or (multiattribute* next analys*) or (multi next attribute* next analys*)) #24. ((health next utilit* next scale*) or (classification next illness next state*) or 15d or 15-d or 15-dimension) #25. ((health next state* next utilit*) or 12d or 12-d or 12-dimension) #26. (well next year*) #27. ((multiattribute* next utilit*) or (multi next attribute* next utilit*)) #28. (health next utilit* next scale*)

386 #29. (qol or 5d or 5-d or 5-dimension or (quality next life) or eq-5d or eq5d or eq-5d or euroqol) #30. (qualy or qaly or qualys or qalys or (quality next adjusted next life next year*)) #31. (life next year* next gain*) #32. (willingness next pay) #33. (hye or hyes or (health* next year* next equivalent*)) #34. ((person next trade next off*) or (person next tradeoff*) or (time next tradeoff*) or (time next trade next off*)) #35. (theory next utilit*) #36. (life next table*) #37. (sf36 or sf-36) #38. ((short next form-36) or shortform-36 or (sf next thirtysix) or (sf next thirty next six) or (shortform next thirtysix) or (shortform next thirty next six) or (short next form next thirtysix) or (short next form next thirty next six)) #39. (6d or 6-d or 6-dimension) #40. (health next status) #41. (#14 or #15 or #16 or #17 or #18 or #19 or #20 or #21 or #22 or #23 or #24 or #25 or #26 or #27 or #28 or #29 or #30 or #31 or #32 or #33 or #34 or #35 or #36 or #37 or #38 or #39 or #40) #42. (#13 and #41) #43. #42 (1984 to current date )

Science Citation Index Index to Scientific and Technical Proceedings (Web of Knowledge) 1984 - May 2004 (Science Citation Index) 1992 - May 2004 (Index to Scientific and Technical Proceedings)

(ovar* same (cancer* or tumo?r* or malignan* or oncolog* or carcinoma* or neoplas* or mass* or growth* or cyst*)) or (adenexa* same mass*) and

Health Status or utilit* approach* or health gain or hui or hui2 or hui 2 or hui3 or hui 3 or health measurement* scale* or health measurement* questionnaire* or standard gamble* or categor* scal* or linear scal* or linear analog* or visual scal* or magnitude estmat* or time trade off* or rosser* classif* or rosser* matrix or rosser* distress* or hrqol or index of wellbeing or quality of wellbeing or qwb or rating scale* or multiattribute* health ind* or multi attribute* health ind* or health utilit* index or health utilit* indices or multiattribute* theor* or multi attribute* theor* or multiattribute* analys* or multi attribute* analys* or health utilit* scale* or classification of illness state* or 15d or 15 d or 15 dimension or health state* utilit* or 12d or 12 d or 12 dimension or well year* or multiattribute* utilit* or multi attribute* utilit* or health utilit* scale* or

387 qol or 5d or 5 d or 5 dimension or quality of life or eq 5d or eq5d or euroqol or qualy or qaly or qualys or qalys or quality adjusted life year* or life year* gain* or willingness to pay or hye or hyes or health* year* equivalent* or person trade off* or person tradeoff* or time tradeoff* or time trade off* or theory utilit* or life table* or health state* or sf36 or sf 36 or short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirtysix or short form thirty six or 6d or 6 d or 6 dimension

OHE HEED (CD-ROM) 1984 - May 2004 ovarian or ovary or ovaries and cancer* or tumor* or tumour* or malignan* or oncolog* or carcinoma* or neoplas* or mass* or growth* or cyst*

388 APPENDIX 10 – Winbugs code

Analysis 1

Below is the WinBUGS code used to estimate PFS and overall survival for the base case analysis. #Estimates overall survival for topotecan, paclitaxel and caelyx, and uses topotecan baseline from 30-49 #Topotecan=a, paclitaxel=b, PLD=c model { #priors for basic parameters dab ~ dnorm(0,.001) #LHR a vs B dac ~ dnorm(0,.001) # LHR A vs C la ~ dnorm(0,.001) # Log hazard rate for A (overall baseline) psla ~ dnorm(0,.001) # Log hazard for a (platinum sensitive baseline) prla ~ dnorm(0,.001) # Log hazard for a (platinum resistant/refractory baseline)

#OVERALL POPULATION # define absolute hazards on log scale ltopo ~ dnorm(la,ptopo) lb <- la - dab lc <- la - dac

# define absolute hazards PER WEEK on natural scale log(a) <- la log(b) <- lb log(c) <- lc

# convert to mean survival OSa <- 1/a OSb <- 1/b OSc <- 1/c

#PLATINUM SENSITIVE POPULATION # define platinum sensitive baseline psltopo ~ dnorm(psla,psptopo) # define absolute hazards for platinum sensitive pslb <- psla - dab pslc <- psla - dac

# define absolute hazards PER WEEK on natural scale log(psa) <- psla log(psb) <- pslb log(psc) <- pslc

# convert to mean survival psOSa <- 1/psa psOSb <- 1/psb psOSc <- 1/psc

#PLATINUM RESISTANT/REFRACTORY POPULATION # define platinum resistant/refractory baseline prltopo ~ dnorm(prla,prptopo)

389 # define absolute hazards for platinum resistant/refractory prlb <- prla - dab prlc <- prla - dac

# define absolute hazards PER WEEK on natural scale log(pra) <- prla log(prb) <- prlb log(prc) <- prlc

# convert to mean survival prOSa <- 1/pra prOSb <- 1/prb prOSc <- 1/prc

#likelihood yab ~ dnorm(dab,pab) yac ~ dnorm(dac,pac) data list list(yab=-0.0899,yac=0.1956, # from 039, 30-49 pab=44.4520,pac=100.6718, # precision=1/variance. SD = 95CI/3.92 ltopo= -4.4558, ptopo=114.9304, #from 30-49 psltopo=-4.6164, psptopo=52.9206, prltopo=-4.1066, prptopo=59.4225) the log baseline (topo) rate is log(-log(0.5) / 59.7) = -4.455845 precision (=1/variance), variance=hazard^2/#events initial values list(dac=.01, dab=.01,la=.01,psla=.01,prla=.01)

#Estimates progression-free survival for topotecan, paclitaxel and caelyx, using only 039 and 30-49, and uses topotecan baseline from 30-49# model { #priors for basic parameters dab ~ dnorm(0,.001) #LHR a vs B dac ~ dnorm(0,.001) # LHR A vs C la ~ dnorm(0,.001) # Log hazard rate for A (overall baseline) psla ~ dnorm(0,.001) # Log hazard rate for A (plat sens baseline) prla ~ dnorm(0,.001) # Log hazard rate for A (plat ref/resis baseline)

#OVERALL POPULATION # define absolute hazards on log scale ltopo ~ dnorm(la,ptopo) lb <- la - dab lc <- la - dac

# define absolute hazards PER WEEK on natural scale log(a) <- la log(b) <- lb log(c) <- lc

# convert to mean survival PFSa <- 1/a PFSb <- 1/b PFSc <- 1/c

390 #PLATINUM SENSITIVE POPULATION # define platinum sensitive baseline psltopo ~ dnorm(psla,psptopo) # define absolute hazards for platinum sensitive pslb <- psla - dab pslc <- psla - dac

# define absolute hazards PER WEEK on natural scale log(psa) <- psla log(psb) <- pslb log(psc) <- pslc

# convert to mean survival psPFSa <- 1/psa psPFSb <- 1/psb psPFSc <- 1/psc

#PLATINUM RESISTANT/REFRACTORY POPULATION # define platinum resistant/refractory baseline prltopo ~ dnorm(prla,prptopo) # define absolute hazards for platinum resistant/refractory prlb <- prla - dab prlc <- prla - dac

# define absolute hazards PER WEEK on natural scale log(pra) <- prla log(prb) <- prlb log(prc) <- prlc

# convert to mean survival prPFSa <- 1/pra prPFSb <- 1/prb prPFSc <- 1/prc

#likelihood yab ~ dnorm(dab,pab) yac ~ dnorm(dac,pac) data list list(yab=-0.2095,yac=0.1115, # from 039, 30-49 pab=43.5733,pac=110.6824, # precision=1/variance. SD = 95CI/3.92 ltopo= -3.1938, ptopo=114.9304, #from 30-49 psltopo=-3.4889, psptopo=52.9206, prltopo=-2.9766, prptopo=59.4225) initial values list(dac=.01, dab=.01,la=.01,psla=.01,prla=.01)

391 Analysis 2

Below is the WinBUGS code used to estimate PFS and overall survival for analysis 2. Estimates overall survival for topotecan, paclitaxel, caelyx, paclitaxel+pt and pt, and uses topotecan baseline from 30-49. This is so PFS can be estimated on same population as OS (no data is available on PFS in platinum sensitive from 039).#

#Topotecan=a, paclitaxel=b, PLD=c, paclitaxel combination=d, platinum=e, CAP=f model #priors for basic parameters dab ~ dnorm(0,.001) # LHR A vs B dac ~ dnorm(0,.001) # LHR A vs C dbf ~ dnorm(0,.001) # LHR B vs F la ~ dnorm(0,.001) # Log hazard rate for A (baseline) ld ~ dnorm(0,.001) # Log hazard rate for D le ~ dnorm(0,.001) # Log hazard rate for E

#define functional parameters daf <- dab + dbf dad <- la-ld dae <- la-le

# define absolute hazards on log scale topotecan ltopo ~ dnorm(la,ptopo) # absolute hazard on log scale pacpt lpacpt ~ dnorm(ld,ppacpt) # absolute hazard on log scale pt lpt ~ dnorm(le,ppt) lb <- la - dab lc <- la - dac lf <- la - daf

# define absolute hazards PER WEEK on natural scale log(a) <- la log(b) <- lb log(c) <- lc log(d) <- ld log(e) <- le log(f) <- lf

# convert to mean survival OSa_2 <- 1/a OSb_2 <- 1/b OSc_2 <- 1/c OSd_2 <- 1/d OSe_2 <- 1/e OSf_2 <- 1/f

#likelihood yab ~ dnorm(dab,pab) yac ~ dnorm(dac,pac) ybf ~ dnorm(dbf,pbf) data list list(yab=0.0100, yac=0.3591,ybf=.5447, #039, 30-49, cantu

392 pab=21.6019,pac=44.1477,pbf=13.7121, # precision=1/variance. SD = 95CI/3.92 ltopo= -4.6164, ptopo=52.9206, #from 30-49 lpacpt= -5.2028, ppacpt=193.4339, #from ICON4 lpt=-5.0205, ppt=202.4341) #from ICON 4

initial values list(dab=.01,dac=.01,dbf=.01,la=.01,ld=.01,le=.01)

#Estimates progression-free survival for topotecan, paclitaxel, caelyx, paclitaxel+pt and pt, and uses topotecan baseline from 30-49. This is so PFS can be estimated on same population as OS (no data is available on PFS in platinum sensitive from 039). model { #priors for basic parameters dab ~ dnorm(0,.001) # LHR A vs B dac ~ dnorm(0,.001) # LHR A vs C dbf ~ dnorm(0,.001) # LHR B vs F la ~ dnorm(0,.001) # Log hazard rate for A (baseline) ld ~ dnorm(0,.001) # Log hazard rate for D le ~ dnorm(0,.001) # Log hazard rate for E

#define functional parameters daf <- dab + dbf dad <- la-ld dae <- la-le

# define absolute hazards PER WEEK on natural scale log(a) <- la log(b) <- lb log(c) <- lc log(d) <- ld log(e) <- le log(f) <- lf

# convert into mean survival PFSa_2 <- 1/a PFSb_2 <- 1/b PFSc_2 <- 1/c PFSd_2 <- 1/d PFSe_2 <- 1/e PFSf_2 <- 1/f #likelihood yab ~ dnorm(dab,pab) yac ~ dnorm(dac,pac) ybf ~ dnorm(dbf,pbf) data list

393 list(yab=-0.1948,yac=0.2523,ybf=.5108, #039, 30-49, Cantu pab=21.1785,pac=50.7314,pbf=16.5426, # precision=1/variance. I got sd as 95CI/3.92 ltopo= -3.4889, ptopo=52.9206, #from 30-49 lpacpt= -4.4096, ppacpt=193.4339, #from ICON4 lpt=-4.1507, ppt=202.4341) #from ICON 4

initial values list(dab=.01,dac=.01,dbf=.01,la=.01,ld=.01,le=.01)

394 Adverse events (Analysis 1)

Below is the WinBUGS code used to estimate the probabilities of adverse events for Analysis 1.

#Probability of Adverse events (model 1, including 039, 30-49 and 30-57) model{

######neutropenia grade 3(t.n3) for (j in 1:2) { delta.n3[j] ~ dnorm(0.0,0.0001) } m.rn3~dnorm(0.0,0.001) t.rn3~dgamma(0.01,0.01) for (j in 1:3){ mu.rn3[j]~dnorm(m.rn3,t.rn3) } for (i in 1:6){ logit(p.n3[i])<-mu.rn3[study.n3[i]] + equals(treat.n3[i],2) * delta.n3[1] + equals(treat.n3[i],3) * delta.n3[2]} for (i in 1:6){ r.n3[i]~dbin(p.n3[i],n.n3[i]) } logit(t.n3[1]) <- m.rn3 for (j in 2: 3) { logit(t.n3[j]) <- m.rn3 + delta.n3[j-1] }

######neutropenia grade 4 (t.n4) for (j in 1:2) { delta.n4[j] ~ dnorm(0.0,0.0001) } m.rn4~dnorm(0.0,0.001) t.rn4~dgamma(0.01,0.01) for (j in 1:3){ mu.rn4[j]~dnorm(m.rn4,t.rn4) } for (i in 1:6){ logit(p.n4[i])<-mu.rn4[study.n4[i]] + equals(treat.n4[i],2) * delta.n4[1] + equals(treat.n4[i],3) * delta.n4[2]} for (i in 1:6){ r.n4[i]~dbin(p.n4[i],n.n4[i]) } logit(t.n4[1]) <- m.rn4 for (j in 2: 3) { logit(t.n4[j]) <- m.rn4 + delta.n4[j-1] }

######thrombocytopenia grade 3 (t.t3) for (j in 1:2) { delta.t3[j] ~ dnorm(0.0,0.0001) } m.rt3~dnorm(0.0,0.001) t.rt3~dgamma(0.01,0.01) for (j in 1:3){ mu.rt3[j]~dnorm(m.rt3,t.rt3) } for (i in 1:6){ logit(p.t3[i])<-mu.rt3[study.t3[i]] + equals(treat.t3[i],2) * delta.t3[1] + equals(treat.t3[i],3) * delta.t3[2]} for (i in 1:6){ r.t3[i]~dbin(p.t3[i],n.t3[i]) } logit(t.t3[1]) <- m.rt3 for (j in 2: 3) { logit(t.t3[j]) <- m.rt3 + delta.t3[j-1] }

######thrombocytopenia grade 4 (t.t4) for (j in 1:2) { delta.t4[j] ~ dnorm(0.0,0.0001) } m.rt4~dnorm(0.0,0.001)

395 t.rt4~dgamma(0.01,0.01) for (j in 1:3){ mu.rt4[j]~dnorm(m.rt4,t.rt4) } for (i in 1:6){ logit(p.t4[i])<-mu.rt4[study.t4[i]] + equals(treat.t4[i],2) * delta.t4[1] + equals(treat.t4[i],3) * delta.t4[2]} for (i in 1:6){ r.t4[i]~dbin(p.t4[i],n.t4[i]) } logit(t.t4[1]) <- m.rt4 for (j in 2: 3) { logit(t.t4[j]) <- m.rt4 + delta.t4[j-1] }

######anemia grade 3 (t.a3) for (j in 1:2) { delta.a3[j] ~ dnorm(0.0,0.0001) } m.ra3~dnorm(0.0,0.001) t.ra3~dgamma(0.01,0.01) for (j in 1:3){ mu.ra3[j]~dnorm(m.ra3,t.ra3) } for (i in 1:6){ logit(p.a3[i])<-mu.ra3[study.a3[i]] + equals(treat.a3[i],2) * delta.a3[1] + equals(treat.a3[i],3) * delta.a3[2]} for (i in 1:6){ r.a3[i]~dbin(p.a3[i],n.a3[i]) } logit(t.a3[1]) <- m.ra3 for (j in 2: 3) { logit(t.a3[j]) <- m.ra3 + delta.a3[j-1] }

######anemia grade 4 (t.a4) for (j in 1:2) { delta.a4[j] ~ dnorm(0.0,0.0001) } m.ra4~dnorm(0.0,0.001) t.ra4~dgamma(0.01,0.01) for (j in 1:3){ mu.ra4[j]~dnorm(m.ra4,t.ra4) } for (i in 1:6){ logit(p.a4[i])<-mu.ra4[study.a4[i]] + equals(treat.a4[i],2) * delta.a4[1] + equals(treat.a4[i],3) * delta.a4[2]} for (i in 1:6){ r.a4[i]~dbin(p.a4[i],n.a4[i]) } logit(t.a4[1]) <- m.ra4 for (j in 2: 3) { logit(t.a4[j]) <- m.ra4 + delta.a4[j-1] }

######PPE grade 3 (t.p3) for (j in 1:2) { delta.p3[j] ~ dnorm(0.0,0.0001) } m.rp3~dnorm(0.0,0.001) t.rp3~dgamma(0.01,0.01) for (j in 1:3){ mu.rp3[j]~dnorm(m.rp3,t.rp3) } for (i in 1:6){ logit(p.p3[i])<-mu.rp3[study.p3[i]] + equals(treat.p3[i],2) * delta.p3[1] + equals(treat.p3[i],3) * delta.p3[2]} for (i in 1:6){ r.p3[i]~dbin(p.p3[i],n.p3[i]) } logit(t.p3[1]) <- m.rp3 for (j in 2: 3) { logit(t.p3[j]) <- m.rp3 + delta.p3[j-1] }

######PPE grade 4 (t.p4)

396 for (j in 1:2) { delta.p4[j] ~ dnorm(0.0,0.0001) } m.rp4~dnorm(0.0,0.001) t.rp4~dgamma(0.01,0.01) for (j in 1:3){ mu.rp4[j]~dnorm(m.rp4,t.rp4) } for (i in 1:6){ logit(p.p4[i])<-mu.rp4[study.p4[i]] + equals(treat.p4[i],2) * delta.p4[1] + equals(treat.p4[i],3) * delta.p4[2]} for (i in 1:6){ r.p4[i]~dbin(p.p4[i],n.p4[i]) } logit(t.p4[1]) <- m.rp4 for (j in 2: 3) { logit(t.p4[j]) <- m.rp4 + delta.p4[j-1] }

######stomatitis/pharyngitis grade 3 (t.s3) for (j in 1:2) { delta.s3[j] ~ dnorm(0.0,0.0001) } m.rs3~dnorm(0.0,0.001) t.rs3~dgamma(0.01,0.01) for (j in 1:3){ mu.rs3[j]~dnorm(m.rs3,t.rs3) } for (i in 1:6){ logit(p.s3[i])<-mu.rs3[study.s3[i]] + equals(treat.s3[i],2) * delta.s3[1] + equals(treat.s3[i],3) * delta.s3[2]} for (i in 1:6){ r.s3[i]~dbin(p.s3[i],n.s3[i]) } logit(t.s3[1]) <- m.rs3 for (j in 2: 3) { logit(t.s3[j]) <- m.rs3 + delta.s3[j-1] }

######stomatitis/pharyngitis grade 4 (t.s4) for (j in 1:2) { delta.s4[j] ~ dnorm(0.0,0.0001) } m.rs4~dnorm(0.0,0.001) t.rs4~dgamma(0.01,0.01) for (j in 1:3){ mu.rs4[j]~dnorm(m.rs4,t.rs4) } for (i in 1:6){ logit(p.s4[i])<-mu.rs4[study.s4[i]] + equals(treat.s4[i],2) * delta.s4[1] + equals(treat.s4[i],3) * delta.s4[2]} for (i in 1:6){ r.s4[i]~dbin(p.s4[i],n.s4[i]) } logit(t.s4[1]) <- m.rs4 for (j in 2: 3) { logit(t.s4[j]) <- m.rs4 + delta.s4[j-1] }

######diarrhoea grade 3 (t.d3) for (j in 1:2) { delta.d3[j] ~ dnorm(0.0,0.0001) } m.rd3~dnorm(0.0,0.001) t.rd3~dgamma(0.01,0.01) for (j in 1:3){ mu.rd3[j]~dnorm(m.rd3,t.rd3) } for (i in 1:6){ logit(p.d3[i])<-mu.rd3[study.d3[i]] + equals(treat.d3[i],2) * delta.d3[1] + equals(treat.d3[i],3) * delta.d3[2]} for (i in 1:6){ r.d3[i]~dbin(p.d3[i],n.d3[i]) } logit(t.d3[1]) <- m.rd3 for (j in 2: 3) { logit(t.d3[j]) <- m.rd3 + delta.d3[j-1] }

######diarrhoea grade 4 (t.d4) for (j in 1:2) { delta.d4[j] ~ dnorm(0.0,0.0001) } m.rd4~dnorm(0.0,0.001)

397 t.rd4~dgamma(0.01,0.01) for (j in 1:3){ mu.rd4[j]~dnorm(m.rd4,t.rd4) } for (i in 1:6){ logit(p.d4[i])<-mu.rd4[study.d4[i]] + equals(treat.d4[i],2) * delta.d4[1] + equals(treat.d4[i],3) * delta.d4[2]} for (i in 1:6){ r.d4[i]~dbin(p.d4[i],n.d4[i]) } logit(t.d4[1]) <- m.rd4 for (j in 2: 3) { logit(t.d4[j]) <- m.rd4 + delta.d4[j-1] }

######nausea/vomiting grade 3 (t.nv3) for (j in 1:2) { delta.nv3[j] ~ dnorm(0.0,0.0001) } m.rnv3~dnorm(0.0,0.001) t.rnv3~dgamma(0.01,0.01) for (j in 1:3){ mu.rnv3[j]~dnorm(m.rnv3,t.rnv3) } for (i in 1:6){ logit(p.nv3[i])<-mu.rnv3[study.nv3[i]] + equals(treat.nv3[i],2) * delta.nv3[1] + equals(treat.nv3[i],3) * delta.nv3[2]} for (i in 1:6){ r.nv3[i]~dbin(p.nv3[i],n.nv3[i]) } logit(t.nv3[1]) <- m.rnv3 for (j in 2: 3) { logit(t.nv3[j]) <- m.rnv3 + delta.nv3[j-1] }

######nausea/vomiting grade 4 (t.nv4) for (j in 1:2) { delta.nv4[j] ~ dnorm(0.0,0.0001) } m.rnv4~dnorm(0.0,0.001) t.rnv4~dgamma(0.01,0.01) for (j in 1:3){ mu.rnv4[j]~dnorm(m.rnv4,t.rnv4) } for (i in 1:6){ logit(p.nv4[i])<-mu.rnv4[study.nv4[i]] + equals(treat.nv4[i],2) * delta.nv4[1] + equals(treat.nv4[i],3) * delta.nv4[2]} for (i in 1:6){ r.nv4[i]~dbin(p.nv4[i],n.nv4[i]) } logit(t.nv4[1]) <- m.rnv4 for (j in 2: 3) { logit(t.nv4[j]) <- m.rnv4 + delta.nv4[j-1] }

######sepsis grade 3 (t.se3) for (j in 1:2) { delta.se3[j] ~ dnorm(0.0,0.0001) } m.rse3~dnorm(0.0,0.001) t.rse3~dgamma(0.01,0.01) for (j in 1:3){ mu.rse3[j]~dnorm(m.rse3,t.rse3) } for (i in 1:6){ logit(p.se3[i])<-mu.rse3[study.se3[i]] + equals(treat.se3[i],2) * delta.se3[1] + equals(treat.se3[i],3) * delta.se3[2]} for (i in 1:6){ r.se3[i]~dbin(p.se3[i],n.se3[i]) } logit(t.se3[1]) <- m.rse3 for (j in 2: 3) { logit(t.se3[j]) <- m.rse3 + delta.se3[j-1] }

######sepsis grade 4 (t.se4) for (j in 1:2) { delta.se4[j] ~ dnorm(0.0,0.0001) } m.rse4~dnorm(0.0,0.001) t.rse4~dgamma(0.01,0.01) for (j in 1:3){ mu.rse4[j]~dnorm(m.rse4,t.rse4) }

398 for (i in 1:6){ logit(p.se4[i])<-mu.rse4[study.se4[i]] + equals(treat.se4[i],2) * delta.se4[1] + equals(treat.se4[i],3) * delta.se4[2]} for (i in 1:6){ r.se4[i]~dbin(p.se4[i],n.se4[i]) } logit(t.se4[1]) <- m.rse4 for (j in 2: 3) { logit(t.se4[j]) <- m.rse4 + delta.se4[j-1] }

#########fever grade 3 (t.f3) for (j in 1:2) { delta.f3[j] ~ dnorm(0.0,0.0001) } m.rf3~dnorm(0.0,0.001) t.rf3~dgamma(0.01,0.01) for (j in 1:3){ mu.rf3[j]~dnorm(m.rf3,t.rf3) } for (i in 1:6){ logit(p.f3[i])<-mu.rf3[study.f3[i]] + equals(treat.f3[i],2) * delta.f3[1] + equals(treat.f3[i],3) * delta.f3[2]} for (i in 1:6){ r.f3[i]~dbin(p.f3[i],n.f3[i]) } logit(t.f3[1]) <- m.rf3 for (j in 2: 3) { logit(t.f3[j]) <- m.rf3 + delta.f3[j-1] }

##########fever grade 4 (t.f4) for (j in 1:2) { delta.f4[j] ~ dnorm(0.0,0.0001) } m.rf4~dnorm(0.0,0.001) t.rf4~dgamma(0.01,0.01) for (j in 1:3){ mu.rf4[j]~dnorm(m.rf4,t.rf4) } for (i in 1:6){ logit(p.f4[i])<-mu.rf4[study.f4[i]] + equals(treat.f4[i],2) * delta.f4[1] + equals(treat.f4[i],3) * delta.f4[2]} for (i in 1:6){ r.f4[i]~dbin(p.f4[i],n.f4[i]) } logit(t.f4[1]) <- m.rf4 for (j in 2: 3) { logit(t.f4[j]) <- m.rf4 + delta.f4[j-1] }

}

# data list(r.n3 =c(33,19,17,34,10,6),(r.n4 = c(146,10,89,24,3,1), (r.t3=c(40,3,25,2,0.5,0.5), (r.t4=c(40,0.5,30,3,0.5,0.5), (r.a3=c(59,13,42,4,5,3), (r.a4=c(10,1,4,3,0.5,0.5), (r.p3=c(0.5,55,0.5,0.5,0.5,16), (r.p4=c(0.5,2,0.5,0.5,0.5,1), (r.s3=c(2,19,1,1,1,11), (r.s4=c(0.5,1,0.5,0.5,0.5,0.5), (r.d3=c(9,5,6,1,3,3), (r.d4=c(1,1,1,0.5,0.5,0.5), (r.nv3=c(34,29,17,4,6,16), (r.nv4=c(8,3,5,1,0.5,3), (r.se3=c(1,0.5,2,2,0.5,0.5), (r.se4=c(1,0.5,2,0.5,0.5,0.5), (r.f3=c(8,2,1,1,3,7), (r.f4=c(5,0.5,0.5,0.5,0.5,0.5), n.n3=c(235,239,112,114,108,108), n.n4=c(235,239,112,114,108,108), n.t3=c(235,239,112,114,108.5,108.5), n.t4=c(235,239.5,112,114,108.5,108.5), n.a3=c(235,239,112,114,108,108), n.a4=c(235,239,112,114,108.5,108.5), n.p3=c(235.5,239,112.5,114.5,108.5,108), n.p4=c(235.5,239,112.5,114.5,108.5,108), n.s3=c(235,239,112,114,108,108), n.s4=c(235.5,239,112.5,114.5,108.5,108.5), n.d3=c(235,239,112,114,108,108),

399 n.d4=c(235,239,112,114.5,108.5,108.5), n.nv3=c(235,239,112,114,108,108), n.nv4=c(235,239,112,114,108.5,108), n.se3=c(235,239.5,112,114,108.5,108.5), n.se4=c(235,239.5,112,114.5,108.5,108.5), n.f3=c(235,239,112,114,108,108), n.f4=c(235,239.5,112.5,114.5,108.5,108.5), study.n3=c(1,1,2,2,3,3), study.n4=c(1,1,2,2,3,3), study.t3=c(1,1,2,2,3,3), study.t4=c(1,1,2,2,3,3), study.a3=c(1,1,2,2,3,3), study.a4=c(1,1,2,2,3,3), study.p3=c(1,1,2,2,3,3), study.p4=c(1,1,2,2,3,3), study.s3=c(1,1,2,2,3,3), study.s4=c(1,1,2,2,3,3), study.d3=c(1,1,2,2,3,3), study.d4=c(1,1,2,2,3,3), study.nv3=c(1,1,2,2,3,3), study.nv4=c(1,1,2,2,3,3), study.se3=c(1,1,2,2,3,3), study.se4=c(1,1,2,2,3,3), study.f3=c(1,1,2,2,3,3), study.f4=c(1,1,2,2,3,3), treat.n3=c(1,3,1,2,2,3), treat.n4=c(1,3,1,2,2,3), treat.t3=c(1,3,1,2,2,3), treat.t4=c(1,3,1,2,2,3), treat.a3=c(1,3,1,2,2,3), treat.a4=c(1,3,1,2,2,3), treat.p3=c(1,3,1,2,2,3), treat.p4=c(1,3,1,2,2,3), treat.s3=c(1,3,1,2,2,3), treat.s4=c(1,3,1,2,2,3), treat.d3=c(1,3,1,2,2,3), treat.d4=c(1,3,1,2,2,3), treat.nv3=c(1,3,1,2,2,3), treat.nv4=c(1,3,1,2,2,3), treat.se3=c(1,3,1,2,2,3), treat.se4=c(1,3,1,2,2,3), treat.f3=c(1,3,1,2,2,3), treat.f4=c(1,3,1,2,2,3))

# initial list(m.rn3=0, t.rn3=1, m.rn4=0,t.rn4=1,m.rt3=0, t.rt3=1, m.rt4=0, t.rt4=1,m.ra3=0,t.ra3=1, m.ra4=0,t.ra4=1, m.rp3=0,t.rp3=1, m.rp4=0,t.rp4=1, m.rs3=0,t.rs3=1, m.rs4=0,t.rs4=1, m.rd3=0,t.rd3=1, m.rd4=0,t.rd4=1, m.rnv3=0, t.rnv3=1, m.rnv4=0,t.rnv4=1, m.rse3=0, t.rse3=1, m.rse4=0, t.rse4=1, m.rf3=0, t.rf3=1, m.rf4=0, t.rf4=1)

Adverse events (Analysis 2)

Below is the Winbugs code used to estimate the probabilities of adverse events, using data from the 30-49, 039, 30-57, Cantu, ICON4 and Bolis trials

400

#Probability of Adverse events (model 1, including 039, 30-49 and 30-57)

#If all 6 trials included then: treatment 1 = topotecan, treatment 2 = paclitaxel, treatment 3 = PLD, treatment 4 = CAP, treatment 5= paclitaxel combination, treatment 6 = carboplatin/cisplatin, treatment 7 = carboplatin+enoxiparin. Trial 1 = 039, trial 2 = 30-49, trial 3 = 30-57, trial 4 = cantu, trial 5 = ICON4 and trial 6 = Bolis model{

######neutropenia grade 3(t.n3) #incorporates Cantu data as well as 3 original trials for (j in 1:3) { delta.n3[j] ~ dnorm(0.0,0.0001) } m.rn3~dnorm(0.0,0.001) t.rn3~dgamma(0.01,0.01) for (j in 1:4){ mu.rn3[j]~dnorm(m.rn3,t.rn3) } for (i in 1:8){ logit(p.n3[i])<-mu.rn3[study.n3[i]] + equals(treat.n3[i],2) * delta.n3[1] + equals(treat.n3[i],3) * delta.n3[2] + equals(treat.n3[i],4) * delta.n3[3]} for (i in 1:8){ r.n3[i]~dbin(p.n3[i],n.n3[i]) } logit(t.n3[1]) <- m.rn3 for (j in 2: 4) { logit(t.n3[j]) <- m.rn3 + delta.n3[j-1] }

######neutropenia grade 4 (t.n4) #incorporates Cantu data as well as 3 original trials for (j in 1:3) { delta.n4[j] ~ dnorm(0.0,0.0001) } m.rn4~dnorm(0.0,0.001) t.rn4~dgamma(0.01,0.01) for (j in 1:4){ mu.rn4[j]~dnorm(m.rn4,t.rn4) } for (i in 1:8){ logit(p.n4[i])<-mu.rn4[study.n4[i]] + equals(treat.n4[i],2) * delta.n4[1] + equals(treat.n4[i],3) * delta.n4[2] + equals(treat.n4[i],4) * delta.n4[3]} for (i in 1:8){ r.n4[i]~dbin(p.n4[i],n.n4[i]) } logit(t.n4[1]) <- m.rn4 for (j in 2: 4) { logit(t.n4[j]) <- m.rn4 + delta.n4[j-1] }

######thrombocytopenia grade 3 (t.t3) #Treatment 5 = carboplatin/cisplatin, treatment 6= carboplatin+enoxiparin #incorporates Cantu and Bolis data for (j in 1:5) { delta.t3[j] ~ dnorm(0.0,0.0001) } m.rt3~dnorm(0.0,0.001) t.rt3~dgamma(0.01,0.01) for (j in 1:6){ mu.rt3[j]~dnorm(m.rt3,t.rt3) } for (i in 1:10){ logit(p.t3[i])<-mu.rt3[study.t3[i]] + equals(treat.t3[i],2) * delta.t3[1] + equals(treat.t3[i],3) * delta.t3[2] + equals(treat.t3[i],4) * delta.t3[3] + equals(treat.t3[i],5) * delta.t3[4] + equals(treat.t3[i],6) * delta.t3[5]} for (i in 1:10){ r.t3[i]~dbin(p.t3[i],n.t3[i]) } logit(t.t3[1]) <- m.rt3 for (j in 2: 6) { logit(t.t3[j]) <- m.rt3 + delta.t3[j-1] }

401 ######thrombocytopenia grade 4 (t.t4) #Treatment 5 = carboplatin/cisplatin, treatment 6= carboplatin+enoxiparin #incorporates Cantu and Bolis data for (j in 1:5) { delta.t4[j] ~ dnorm(0.0,0.0001) } m.rt4~dnorm(0.0,0.001) t.rt4~dgamma(0.01,0.01) for (j in 1:6){ mu.rt4[j]~dnorm(m.rt4,t.rt4) } for (i in 1:10){ logit(p.t4[i])<-mu.rt4[study.t4[i]] + equals(treat.t4[i],2) * delta.t4[1] + equals(treat.t4[i],3) * delta.t4[2] + equals(treat.t4[i],4) * delta.t4[3] + equals(treat.t4[i],5) * delta.t4[4] + equals(treat.t4[i],6) * delta.t4[5]} for (i in 1:10){ r.t4[i]~dbin(p.t4[i],n.t4[i]) } logit(t.t4[1]) <- m.rt4 for (j in 2: 6) { logit(t.t4[j]) <- m.rt4 + delta.t4[j-1] }

######anemia grade 3 (t.a3) #treatment 4 = carboplatin/cisplatin, treatment 5 = carboplatin+enoxaparin #incorporates Bolis data for (j in 1:4) { delta.a3[j] ~ dnorm(0.0,0.0001) } m.ra3~dnorm(0.0,0.001) t.ra3~dgamma(0.01,0.01) for (j in 1:5){ mu.ra3[j]~dnorm(m.ra3,t.ra3) } for (i in 1:8){ logit(p.a3[i])<-mu.ra3[study.a3[i]] + equals(treat.a3[i],2) * delta.a3[1] + equals(treat.a3[i],3) * delta.a3[2] + equals(treat.a3[i],4) * delta.a3[3] + equals(treat.a3[i],5) * delta.a3[4]} for (i in 1:8){ r.a3[i]~dbin(p.a3[i],n.a3[i]) } logit(t.a3[1]) <- m.ra3 for (j in 2: 5) { logit(t.a3[j]) <- m.ra3 + delta.a3[j-1] }

######anemia grade 4 (t.a4) #treatment 4 = carboplatin/cisplatin, treatment 5 = carboplatin+enoxaparin #incorporates Bolis data for (j in 1:4) { delta.a4[j] ~ dnorm(0.0,0.0001) } m.ra4~dnorm(0.0,0.001) t.ra4~dgamma(0.01,0.01) for (j in 1:5){ mu.ra4[j]~dnorm(m.ra4,t.ra4) } for (i in 1:8){ logit(p.a4[i])<-mu.ra4[study.a4[i]] + equals(treat.a4[i],2) * delta.a4[1] + equals(treat.a4[i],3) * delta.a4[2] + equals(treat.a4[i],4) * delta.a4[3] + equals(treat.a4[i],5) * delta.a4[4]} for (i in 1:8){ r.a4[i]~dbin(p.a4[i],n.a4[i]) } logit(t.a4[1]) <- m.ra4 for (j in 2: 5) { logit(t.a4[j]) <- m.ra4 + delta.a4[j-1] }

######PPE grade 3 (t.p3) for (j in 1:2) { delta.p3[j] ~ dnorm(0.0,0.0001) } m.rp3~dnorm(0.0,0.001) t.rp3~dgamma(0.01,0.01) for (j in 1:3){ mu.rp3[j]~dnorm(m.rp3,t.rp3) }

402 for (i in 1:6){ logit(p.p3[i])<-mu.rp3[study.p3[i]] + equals(treat.p3[i],2) * delta.p3[1] + equals(treat.p3[i],3) * delta.p3[2]} for (i in 1:6){ r.p3[i]~dbin(p.p3[i],n.p3[i]) } logit(t.p3[1]) <- m.rp3 for (j in 2: 3) { logit(t.p3[j]) <- m.rp3 + delta.p3[j-1] }

######PPE grade 4 (t.p4) for (j in 1:2) { delta.p4[j] ~ dnorm(0.0,0.0001) } m.rp4~dnorm(0.0,0.001) t.rp4~dgamma(0.01,0.01) for (j in 1:3){ mu.rp4[j]~dnorm(m.rp4,t.rp4) } for (i in 1:6){ logit(p.p4[i])<-mu.rp4[study.p4[i]] + equals(treat.p4[i],2) * delta.p4[1] + equals(treat.p4[i],3) * delta.p4[2]} for (i in 1:6){ r.p4[i]~dbin(p.p4[i],n.p4[i]) } logit(t.p4[1]) <- m.rp4 for (j in 2: 3) { logit(t.p4[j]) <- m.rp4 + delta.p4[j-1] }

######stomatitis/pharyngitis grade 3 (t.s3) #treatment 5 = carboplatin/cisplatin, treatment 6= carboplatin+enoxiparin # incorporates Bolis data and assumed number of events from Cantu for (j in 1:5) { delta.s3[j] ~ dnorm(0.0,0.0001) } m.rs3~dnorm(0.0,0.001) t.rs3~dgamma(0.01,0.01) for (j in 1:6){ mu.rs3[j]~dnorm(m.rs3,t.rs3) } for (i in 1:10){ logit(p.s3[i])<-mu.rs3[study.s3[i]] + equals(treat.s3[i],2) * delta.s3[1] + equals(treat.s3[i],3) * delta.s3[2] + equals(treat.s3[i],4) * delta.s3[3] + equals(treat.s3[i],5) * delta.s3[4] + equals(treat.s3[i],6) * delta.s3[5]} for (i in 1:10){ r.s3[i]~dbin(p.s3[i],n.s3[i]) } logit(t.s3[1]) <- m.rs3 for (j in 2: 6) { logit(t.s3[j]) <- m.rs3 + delta.s3[j-1] }

######stomatitis/pharyngitis grade 4 (t.s4) # incorporates Bolis data and assumed number of events from Cantu for (j in 1:5) { delta.s4[j] ~ dnorm(0.0,0.0001) } m.rs4~dnorm(0.0,0.001) t.rs4~dgamma(0.01,0.01) for (j in 1:6){ mu.rs4[j]~dnorm(m.rs4,t.rs4) } for (i in 1:10){ logit(p.s4[i])<-mu.rs4[study.s4[i]] + equals(treat.s4[i],2) * delta.s4[1] + equals(treat.s4[i],3) * delta.s4[2] + equals(treat.s4[i],4) * delta.s4[3] + equals(treat.s4[i],5) * delta.s4[4] + equals(treat.s4[i],6) * delta.s4[5]} for (i in 1:10){ r.s4[i]~dbin(p.s4[i],n.s4[i]) } logit(t.s4[1]) <- m.rs4 for (j in 2: 6) { logit(t.s4[j]) <- m.rs4 + delta.s4[j-1] }

######diarrhoea grade 3 (t.d3) for (j in 1:2) { delta.d3[j] ~ dnorm(0.0,0.0001) }

403 m.rd3~dnorm(0.0,0.001) t.rd3~dgamma(0.01,0.01) for (j in 1:3){ mu.rd3[j]~dnorm(m.rd3,t.rd3) } for (i in 1:6){ logit(p.d3[i])<-mu.rd3[study.d3[i]] + equals(treat.d3[i],2) * delta.d3[1] + equals(treat.d3[i],3) * delta.d3[2]} for (i in 1:6){ r.d3[i]~dbin(p.d3[i],n.d3[i]) } logit(t.d3[1]) <- m.rd3 for (j in 2: 3) { logit(t.d3[j]) <- m.rd3 + delta.d3[j-1] }

######diarrhoea grade 4 (t.d4) for (j in 1:2) { delta.d4[j] ~ dnorm(0.0,0.0001) } m.rd4~dnorm(0.0,0.001) t.rd4~dgamma(0.01,0.01) for (j in 1:3){ mu.rd4[j]~dnorm(m.rd4,t.rd4) } for (i in 1:6){ logit(p.d4[i])<-mu.rd4[study.d4[i]] + equals(treat.d4[i],2) * delta.d4[1] + equals(treat.d4[i],3) * delta.d4[2]} for (i in 1:6){ r.d4[i]~dbin(p.d4[i],n.d4[i]) } logit(t.d4[1]) <- m.rd4 for (j in 2: 3) { logit(t.d4[j]) <- m.rd4 + delta.d4[j-1] }

######nausea/vomiting grade 3 (t.nv3) #incorporates ICON4, Cantu and Bolis data for (j in 1:6) { delta.nv3[j] ~ dnorm(0.0,0.0001) } m.rnv3~dnorm(0.0,0.001) t.rnv3~dgamma(0.01,0.01) for (j in 1:7){ mu.rnv3[j]~dnorm(m.rnv3,t.rnv3) } for (i in 1:12){ logit(p.nv3[i])<-mu.rnv3[study.nv3[i]] + equals(treat.nv3[i],2) * delta.nv3[1] + equals(treat.nv3[i],3) * delta.nv3[2] + equals(treat.nv3[i],4) * delta.nv3[3] + equals(treat.nv3[i],5) * delta.nv3[4] + equals(treat.nv3[i],6) * delta.nv3[5] + equals(treat.nv3[i],7) * delta.nv3[6]} for (i in 1:12){ r.nv3[i]~dbin(p.nv3[i],n.nv3[i]) } logit(t.nv3[1]) <- m.rnv3 for (j in 2: 7) { logit(t.nv3[j]) <- m.rnv3 + delta.nv3[j-1] }

######nausea/vomiting grade 4 (t.nv4) ##no info from Cantu available for grade 4, therefore T[4] is paclitaxel combination, T[5] is carboplatin/cisplatin # no grade 4 info from Bolis for (j in 1:4) { delta.nv4[j] ~ dnorm(0.0,0.0001) } m.rnv4~dnorm(0.0,0.001) t.rnv4~dgamma(0.01,0.01) for (j in 1:5){ mu.rnv4[j]~dnorm(m.rnv4,t.rnv4) } for (i in 1:8){ logit(p.nv4[i])<-mu.rnv4[study.nv4[i]] + equals(treat.nv4[i],2) * delta.nv4[1] + equals(treat.nv4[i],3) * delta.nv4[2] + equals(treat.nv4[i],4) * delta.nv4[3] + equals(treat.nv4[i],5) * delta.nv4[4] } for (i in 1:8){ r.nv4[i]~dbin(p.nv4[i],n.nv4[i]) } logit(t.nv4[1]) <- m.rnv4 for (j in 2: 5) { logit(t.nv4[j]) <- m.rnv4 + delta.nv4[j-1] }

404 ######sepsis grade 3 (t.se3) #T[4] is carboplatin and T[5] is carboplatin + enoxiparin #incorporates data from Bolis for (j in 1:4) { delta.se3[j] ~ dnorm(0.0,0.0001) } m.rse3~dnorm(0.0,0.001) t.rse3~dgamma(0.01,0.01) for (j in 1:5){ mu.rse3[j]~dnorm(m.rse3,t.rse3) } for (i in 1:8){ logit(p.se3[i])<-mu.rse3[study.se3[i]] + equals(treat.se3[i],2) * delta.se3[1] + equals(treat.se3[i],3) * delta.se3[2] + equals(treat.se3[i],4) * delta.se3[3] + equals(treat.se3[i],5) * delta.se3[4]} for (i in 1:8){ r.se3[i]~dbin(p.se3[i],n.se3[i]) } logit(t.se3[1]) <- m.rse3 for (j in 2: 5) { logit(t.se3[j]) <- m.rse3 + delta.se3[j-1] }

######sepsis grade 4 (t.se4) #T[4] is carboplatin and T[5] is carboplatin + enoxiparin #incorporates data from Bolis for (j in 1:4) { delta.se4[j] ~ dnorm(0.0,0.0001) } m.rse4~dnorm(0.0,0.001) t.rse4~dgamma(0.01,0.01) for (j in 1:5){ mu.rse4[j]~dnorm(m.rse4,t.rse4) } for (i in 1:8){ logit(p.se4[i])<-mu.rse4[study.se4[i]] + equals(treat.se4[i],2) * delta.se4[1] + equals(treat.se4[i],3) * delta.se4[2] + equals(treat.se4[i],4) * delta.se4[3] + equals(treat.se4[i],5) * delta.se4[4]} for (i in 1:8){ r.se4[i]~dbin(p.se4[i],n.se4[i]) } logit(t.se4[1]) <- m.rse4 for (j in 2: 5) { logit(t.se4[j]) <- m.rse4 + delta.se4[j-1] }

#########fever grade 3 (t.f3) #T[4] is carboplatin and T[5] is carboplatin + enoxiparin #incorporates data from Bolis for (j in 1:4) { delta.f3[j] ~ dnorm(0.0,0.0001) } m.rf3~dnorm(0.0,0.001) t.rf3~dgamma(0.01,0.01) for (j in 1:5){ mu.rf3[j]~dnorm(m.rf3,t.rf3) } for (i in 1:8){ logit(p.f3[i])<-mu.rf3[study.f3[i]] + equals(treat.f3[i],2) * delta.f3[1] + equals(treat.f3[i],3) * delta.f3[2] + equals(treat.f3[i],4) * delta.f3[3] + equals(treat.f3[i],5) * delta.f3[4]} for (i in 1:8){ r.f3[i]~dbin(p.f3[i],n.f3[i]) } logit(t.f3[1]) <- m.rf3 for (j in 2: 5) { logit(t.f3[j]) <- m.rf3 + delta.f3[j-1] }

##########fever grade 4 (t.f4) #T[4] is carboplatin and T[5] is carboplatin + enoxiparin #incorporates data from Bolis for (j in 1:4) { delta.f4[j] ~ dnorm(0.0,0.0001) } m.rf4~dnorm(0.0,0.001) t.rf4~dgamma(0.01,0.01) for (j in 1:5){ mu.rf4[j]~dnorm(m.rf4,t.rf4) }

405 for (i in 1:8){ logit(p.f4[i])<-mu.rf4[study.f4[i]] + equals(treat.f4[i],2) * delta.f4[1] + equals(treat.f4[i],3) * delta.f4[2] + equals(treat.f4[i],4) * delta.f4[3] + equals(treat.f4[i],5) * delta.f4[4] } for (i in 1:8){ r.f4[i]~dbin(p.f4[i],n.f4[i]) } logit(t.f4[1]) <- m.rf4 for (j in 2: 5) { logit(t.f4[j]) <- m.rf4 + delta.f4[j-1] }

}

# data list(r.n3 =c(33,19,17,34,10,6,5,8,3,12),(r.n4 = c(146,10,89,24,3,1,1,9), (r.t3=c(40,3,25,2,0.5,0.5,0.5,5,11,31), (r.t4=c(40,0.5,30,3,0.5,0.5,0.5,1,8,29), (r.a3=c(59,13,42,4,5,3,5,18), (r.a4=c(10,1,4,3,0.5,0.5,4,5), (r.p3=c(0.5,55,0.5,0.5,0.5,16), (r.p4=c(0.5,2,0.5,0.5,0.5,1), (r.s3=c(2,19,1,1,1,11,0.5,0.5,0.5,3), (r.s4=c(0.5,1,0.5,0.5,0.5,0.5,0.5,1,0.5,1), (r.d3=c(9,5,6,1,3,3), (r.d4=c(1,1,1,0.5,0.5,0.5), (r.nv3=c(34,29,17,4,6,16,0.5,2,37,47,3,12), (r.nv4=c(8,3,5,1,0.5,3,2,6), (r.se3=c(1,0.5,2,2,0.5,0.5,0.5,0.5), (r.se4=c(1,0.5,2,0.5,0.5,0.5,0.5,0.5), (r.f3=c(8,2,1,1,3,7,0.5,0.5), (r.f4=c(5,0.5,0.5,0.5,0.5,0.5,0.5,0.5), n.n3=c(235,239,112,114,108,108,47,47), n.n4=c(235,239,112,114,108,108,47,47), n.t3=c(235,239,112,114,108.5,108.5,47.5,47,95,95), n.t4=c(235,239.5,112,114,108.5,108.5,47.5,47,95,95), n.a3=c(235,239,112,114,108,108,95,95), n.a4=c(235,239,112,114,108.5,108.5,95,95), n.p3=c(235.5,239,112.5,114.5,108.5,108), n.p4=c(235.5,239,112.5,114.5,108.5,108), n.s3=c(235,239,112,114,108,108,47.5,47.5,95.5,95), n.s4=c(235.5,239,112.5,114.5,108.5,108.5,47.5,47,95.5,95), n.d3=c(235,239,112,114,108,108), n.d4=c(235,239,112,114.5,108.5,108.5),, n.nv3=c(235,239,112,114,108,108,47.5,47,392,410,95,95), n.nv4=c(235,239,112,114,108.5,108,392,410), n.se3=c(235,239.5,112,114,108.5,108.5,95.5,95.5), n.se4=c(235,239.5,112,114.5,108.5,108.5,95.5,95.5), n.f3=c(235,239,112,114,108,108,95.5,95.5), n.f4=c(235,239.5,112.5,114.5,108.5,108.5,95.5,95.5), study.n3=c(1,1,2,2,3,3,4,4), study.n4=c(1,1,2,2,3,3,4,4), study.t3=c(1,1,2,2,3,3,4,4,5,5), study.t4=c(1,1,2,2,3,3,4,4,5,5), study.a3=c(1,1,2,2,3,3,4,4), study.a4=c(1,1,2,2,3,3,4,4), study.p3=c(1,1,2,2,3,3), study.p4=c(1,1,2,2,3,3), study.s3=c(1,1,2,2,3,3,4,4,5,5), study.s4=c(1,1,2,2,3,3,4,4,5,5), study.d3=c(1,1,2,2,3,3), study.d4=c(1,1,2,2,3,3), study.nv3=c(1,1,2,2,3,3,4,4,5,5,6,6), study.nv4=c(1,1,2,2,3,3,4,4), study.se3=c(1,1,2,2,3,3,4,4), study.se4=c(1,1,2,2,3,3,4,4), study.f3=c(1,1,2,2,3,3,4,4), study.f4=c(1,1,2,2,3,3,4,4),

406 treat.n3=c(1,3,1,2,2,3,2,4), treat.n4=c(1,3,1,2,2,3,2,4), treat.t3=c(1,3,1,2,2,3,2,4,5,6), treat.t4=c(1,3,1,2,2,3,2,4,5,6), treat.a3=c(1,3,1,2,2,3,4,5), treat.a4=c(1,3,1,2,2,3,4,5), treat.p3=c(1,3,1,2,2,3), treat.p4=c(1,3,1,2,2,3), treat.s3=c(1,3,1,2,2,3,2,4,5,6), treat.s4=c(1,3,1,2,2,3,2,4,5,6), treat.d3=c(1,3,1,2,2,3), treat.d4=c(1,3,1,2,2,3), treat.nv3=c(1,3,1,2,2,3,2,4,5,6,6,7), treat.nv4=c(1,3,1,2,2,3,4,5), treat.se3=c(1,3,1,2,2,3,4,5), treat.se4=c(1,3,1,2,2,3,4,5), treat.f3=c(1,3,1,2,2,3,4,5), treat.f4=c(1,3,1,2,2,3,4,5))

407 APPENDIX 11: Details of the economic model

Table 31: Survival data used in the sensitivity analysis

Name Overall trial Platinum sensitive Platinum Source population resistant/refractory Sensitivity Overall Topotecan ‘v’ paclitaxel 0.914 1.01 (0.663,1.541) 0.738 (0.498,1.093) 039 analysis survival (0.681,1.226) including 30- hazards Topotecan ‘v’ PLDH 1.216 (1,1.478) 1.432 1.069 (0.823,1.387) 30-49 57 overall (1.066,1.923) survival data PLDH ‘v’ paclitaxel 0.93 (0.70,1.23) 1.05 (0.66, 1.67) 0.87 (0.61, 1.24) 30-57 Progression Topotecan ‘v’ paclitaxel 0.811 0.823 0.749 (0.501,1.121) 039 free (0.603,1.092) (0.538,1.261) survival Topotecan ‘v’ PLDH 1.118 (0.928, 1.287 0.992 (0.77,1.279) 30-49 hazards 1.347) (0.977,1.694) PLDH ‘v’ paclitaxel - - -

Sensitivity Overall Topotecan ‘v’ paclitaxel - 1.01 (0.663,1.541) 0.738 (0.498,1.093) analysis using survival Topotecan ‘v’ PLDH - 1.432 1.069 (0.823,1.387) sub-group hazards (1.066,1.923) treatment PLDH ‘v’ paclitaxel - 1.05 (0.66, 1.67) 0.87 (0.61, 1.24) effects Progression Topotecan ‘v’ paclitaxel - 0.823 0.749 (0.501,1.121) free (0.538,1.261) survival Topotecan ‘v’ PLDH - 1.287 0.992 (0.77,1.279) hazards (0.977,1.694) PLDH ‘v’ paclitaxel - - -

N.B other sensitivity analysis on Analysis 1 did not involve changing any of the survival estimates

408 Table 34: Resource use associated with adverse events (taken from Schering submission(ref))

Adverse event Resource use item Cost per Distribution item (£) Stomatitis/pharyngitis Outpatient visit 90 Triangular (0, 1.5, grade 3 2) Paracetamol 1 Fixed mouthwash Sulcralfate 0.37 Triangular (7,8.5, 10) Oramorph 4.62 Fixed Stomatitis/pharyngitis Inpatient stay 293 Uniform (5,7) grade 4 Fluconzole iv 204.96 Fixed Saline 42 Fixed Paracetamol 1 Fixed mouthwash Sulcrafate 0.37 Triangular (7,8.5, 10) Oramorph 4.62 Fixed

PPE grade 3 Outpatient visit 90 Fixed Paracetamol 3.09 Fixed Pyridoxine 0.29 Fixed PPE grade 4 Inpatient stay 293 Triangular (5,7,14) Flucoxacillin iv 57.12 Fixed Saline 42 Fixed Paracetamol 3.09 Fixed Pyridoxine 0.29 Fixed

Neutropenia grade 3 Outpatient visit 90 Uniform (0,1) Ciprofloxacin 1.50 Uniform (5,7) Neutropenia grade 4 Inpatient stay 293 Uniform (2,5) Ciprofloxacin 1.50 Uniform (5,7) G-CSF 77.03 Uniform (3,7)

Thrombocytopenia grade - - - 3 Thrombocytopenia grade Day case admission 293 Fixed 4 Platelet transfusion 705 Fixed Type and cross 18 Fixed

Anaemia grade 3 Day case admission 293 Fixed Blood transfusion 78.80 Fixed Type and cross 18 Fixed Anaemia grade 4 Inpatient stay 293 Uniform (3,6) Blood transfusion 78.80 Fixed Type and cross 18 Fixed

Diarrhoea grade 3 Inpatient stay 1465 Fixed

409 Buscopan 1.39 Fixed Ciprofloxacin 1.50 Uniform (5,7) Codine 0.33 Fixed Loperamide 0.08 Uniform (2,3) Diarrhoea grade 4 Inpatient stay 1465 Fixed Buscopan 1.39 Fixed Ciprofloxacin 1.50 Uniform (5,7) Codine 0.33 Fixed Loperamide 0.08 Uniform (2,3)

Nausea/vomiting grade 3 Outpatient visit 90 Fixed Dexamethasone 0.51 Uniform (5,7) Granisetron 383.95 Fixed Nausea/vomiting grade 4 Inpatient stay 1465 Fixed Saline 42 Fixed Dexamethasone iv 6.60 Fixed Granisetron iv 360 Fixed Cyclizine 8.55 Fixed

Sepsis grade 3 Inpatient stay 293 Uniform (3,6) Gentamicin 61.25 Fixed Tazocin 368.48 Fixed Sepsis grade 4 Inpatient stay 293 Triangular (3.5,5,7) ICU 558 Triangular (3.5,5,7) Gentamicin 61.25 Fixed Tazocin 368.48 Fixed Saline 42 Fixed Fluconzole iv 204.96 Fixed

Fever grade 3 Inpatient stay 293 Uniform (3,6) Gentamicin 61.25 Fixed Tazocin 368.48 Fixed Fever grade 4 Inpatient stay 293 Triangular (3.5,5,7) ICU 558 Triangular (3.5,5,7) Gentamicin 61.25 Fixed Tazocin 368.48 Fixed Saline 42 Fixed Fluconzole iv 204.96 Fixed

410

411