Chemotherapy in Advanced Ovarian Cancer: an Overview of Randomised Clinical Trials BMJ: First Published As 10.1136/Bmj.303.6807.884 on 12 October 1991
Total Page:16
File Type:pdf, Size:1020Kb
Chemotherapy in advanced ovarian cancer: an overview of randomised clinical trials BMJ: first published as 10.1136/bmj.303.6807.884 on 12 October 1991. Downloaded from Advanced Ovarian Cancer Trialists Group Abstract cytotoxic drugs, usually doxorubicin and cyclophos- Objectives-To consider the role of platinum and phamide with or without hexamethylmelamine.45 the relative merits of single agent and combination When, however, doubt was cast on the effectiveness chemotherapy in the treatment of advanced ovarian of doxorubicin by both randomised phase III trials6-8 cancer. and phase II studies in patients not responding to Design-Formal quantitative overview using cisplatin9 several major centres adopted cisplatin plus updated individual patient data from all available cyclophosphamide as standard. Furthermore, when randomised trials (published and unpublished). other trials failed to find significant survival differences Subjects-8139 patients (6408 deaths) included in between single agent cisplatin and cisplatin in combi- 45 different trials. nation with other drugs'0 some centres reverted to Results-No firm conclusions could be reached. using single agent platinum as routine first line treat- Nevertheless, the results suggest that in terms of ment. Recently a large number oftrials have compared survival immediate platinum based treatment was cisplatin with its less nephrotoxic and neurotoxic better than non-platinum regimens (overall relative analogue carboplatin, and although follow up times risk 0-93; 95% confidence interval 0-83 to 1-05); were often short, many institutions have now adopted platinum in combination was better than single agent carboplatin as standard. platinum when used in the same dose (overall Currently it is unclear what constitutes optimal relative risk 0-85; 0*72 to 1-00); and cisplatin and chemotherapy for advanced disease and treatment carboplatin were equally effective (overall relative strategies vary both nationally and internationally. risk 1-05; 0-94 to 1-18). What is clear is that to date no individual clinical trial Conclusions-In the past, randomised clinical has been large enough to detect survival differences of trials of chemotherapy in advanced ovarian cancer the magnitude that could reasonably be expected with have been much too small to detect the degree of available treatment.'2 Consequently the inconclusive benefit which this overview suggests is realistic for results of over 50 such trials reported could be con- currently available chemotherapeutic regimens. sistent with moderate treatment benefits.'3 The British Hence a new trial comparing cisplatin, doxorubicin, Medical Research Council Gynaecological Cancer and cyclophosphamide (CAP) with carboplatin has Working Party realised the need to synthesise the been launched and plans to accrue 2000 patients. information from these trials to evaluate currently used chemotherapeutic regimens. Given the problems asso- ciated with a qualitative review of published work,'3 http://www.bmj.com/ Introduction this group initiated an overview which used formal Ovarian carcinoma is the seventh most common cluantitative methods to combine the results from all cancer ofwomen in the world. Some 140 000 new cases Available randomised trials examining the role of are diagnosed every year and the disease is responsible platinum and of combination chemotherapy in the for the greatest number of deaths from gynaecological treatment ofadvanced ovarian cancer. At the outset the malignancy in Europe and North America.' Despite MRC overview secretariat contacted the investigators over 50 randomised clinical trials having examined responsible for each trial, inviting their collaboration. the relative efficacy of different chemotherapeutic In so doing it established the Advanced Ovarian on 24 September 2021 by guest. Protected copyright. regimens in advanced disease (FIGO (International Cancer Trialists Group, under whose auspices the Federation of Gynaecology and Obstetrics) stages III overview was conducted. and IV), individually these trials have been too small to show clear benefit of one type of chemotherapy over another. Nevertheless, many ofthese trials have had an Methods and data important influence on clinical practice, and conse- The relative merits of single agent and combination quently the type and intensity of chemotherapy used chemotherapy and the role of platinum in disease routinely for patients with advanced disease have management were sought, and five comparisons fluctuated greatly. between different forms of chemotherapy were identi- Ovarian cancer was one of the first solid malignant fied as being of interest. These were: (I) single non- tumours to be treated by chemotherapy, and the single platinum agent versus non-platinum combination; (II) over 30 combina- Advanced Ovarian Cancer alkylating agents that were first used years ago single non-platinum agent versus platinum Trialists Group were considered optimal treatment until the mid- tion; (III) addition of platinum to a regimen; (IV) Members of the trialists 1970s. The past 15 years, however, have seen many single agent platinum versus platinum combination; group and the organisations changes in disease management. In 1978 a small (V) cisplatin versus carboplatin. and groups that collaborated randomised trial in advanced disease found that Hexa- Trials were eligible for inclusion in the overview in the overview are listed at CAF, a combination of cytotoxic drugs (hexamethyl- if they examined first line treatment for advanced the end of this report. melamine, cyclophosphamide, methotrexate, and epithelial ovarian carcinoma and made one or more of fluorouracil), achieved higher response rates than the comparisons listed above. Each had also to be Correspondence to: Dr L A the single alkylating agent melphalan and suggested unconfounded and believed to have been randomised Stewart, MRC Cancer Trials the of a in in a manner that ofthe next Office, 1 Brooklands possibility corresponding improvement precluded prior knowledge Avenue, Cambridge survival.2 At about the same time phase II studies treatment assignment. To avoid publication bias'4 it CB2 2BB. suggested that cisplatin was the most promising new was essential that both published and unpublished drug then available.' These results rapidly led to the studies were included, and various methods were BMJ_ 1991;303:884-93 standard use of cisplatin in combination with other employed to identify relevant trials. A bibliographic 884 BMJ VOLUME 303 12 OCTOBER 1991 101 individual trials. Virtually all of this information was recovered, so that 99 5% of all patients randomised in the studies included in the overview were available for analysis. After collection, processing, and checking of 0 8- the data individual trialists were sent a printout oftheir data set as stored in the overview database together BMJ: first published as 10.1136/bmj.303.6807.884 on 12 October 1991. Downloaded from with calculated survival curves so that they could check for errors. 06- All analyses were carried out on an intention to treat basis, patients being analysed according to their allocated treatment irrespective of whether they 0I) actually received that treatment. For each comparison 0*4- survival analyses were stratified across all trials to generate log rank statistics. The individual observed and expected numbers of deaths as calculated in these actuarial survival analyses were pooled to provide an L=Combination overall relative risk for the relevant comparison. This 0 2- Single agent value gives the relative risk of death associated with two types of treatment. For example, when comparing treatment A with treatment B a relative risk of 0-8 represents a 20% decrease in the risk of death when 0 1 2 3 4 5 6 i 6 lbX1 using treatment A whereas a value of 1-2 represents a Years 20% increase in the relative risk of death when using treatment A. A relative risk of unity represents no Single non-platinum Numbers at risk difference in the risk of death associated with the two agent 1379 649 315 185 152 126 104 90 76 66 50 treatments. Appendix C explains the graphical dis- Non-platinum 1767 893 427 258 182 138 l116 94 82 70 61 plays used in the figures to present the relative risks for combination the various trials. FIG 1-Survival curves for comparison I (single non-platinum agent v nonI-platinum combination) All p values are two sided, and the x2 values were calculated on one degree of freedom unless otherwise specified. review by means of MEDLINE, CancerLit, and published texts was carried out. This was supple- mented by examining the trial registers produced by Results the National Cancer Institute (PDQ ClinProt) and the Table I summarises the numbers of patients and United Kingdom Co-ordinating Committee on Cancer deaths in comparisons I-V. Research.'I The proceedings of relevant clinical meet- ings were also consulted. In addition, questionnaires were sent to the principal authors of published trials TABLE I -Numbers oftrials, deaths, and patients per comparison and to international members of the Royal College of No of No of Obstetricians and Gynaecologists asking them to No No of available unavailable of http://www.bmj.com/ supplement a provisional list of trials. Pharmaceutical Comparison trials trials deaths patients companies concerned in ovarian cancer treatment were 16 6 2817 3146 also approached in this way. II 11 2 1136 1329 were III 8 0 1134 1408 Fifty three eligible randomised trials identi- IV 6 0 712 925 fied, two other potentially eligible