Systematic Review of Observational Studies and Randomised Controlled Trials PRS6 of Omalizumab in Severe Persistent Allergic Asthma and Meta-analysis Feasibility Assessment

Bergrath E1, Ong SH2, Bousquet J3, *Baldwin M2, Manga V2, Rao S2, Cope S4 1Mapi, Inc., Boston, MA, USA; 2Novartis Pharma AG, Basel, Switzerland; 3University of Montpellier, Montpellier, France; 4Mapi, Inc., Toronto, ON, Canada

zz Substantial variation was observed in the rate of clinically significant exacerbations across Introduction RCTs in baseline risk (SoC: 0.40 - 2.86) and treatment effect (Rate ratio [RR]: Event rate zz Asthma is one of the most common long-term inflammatory disorder of the airways in experimental arm/Event rate in the control arm: 0.39 - 0.75), and across observational characterised by symptoms including breathlessness, chest tightness, wheezing, sputum studies in baseline risk (before omalizumab: 3.48 - 6.00) and treatment effect (RR: 0.12 - 0.46) production, airflow obstruction, hyper-responsiveness of airways, and cough.1 (Figure 3). zz According to The Global Initiative for Asthma (GINA), approximately 300 million people suffer zz The rates were based on study durations ranging from 6.4 to 12 months across the RCTs and from asthma worldwide.2 from 4 to 24 months across the observational studies. zz Out of the several types of asthma recognised clinically, allergic asthma, which relates to Figure 3. Rate ratio of clinically significant exacerbations an immunoglobulin E (IgE) response to environmental allergens, is the most common form of asthma. It accounts for approximately two thirds of asthmatics and nearly half of the 24 RCT (OMA vs. PLBO) patients suffering from severe asthma.3-5 13 14 RCT (OMA vs. SoC) zz OBS (after vs. before OMA) In Europe, omalizumab is used as an add-on to existing treatment to improve the control of 20 severe persistent asthma in patients with a positive skin test or in-vitro reactivity to a perennial Legend aeroallergen and symptoms that are inadequately controlled with high doses of inhaled 1. Vignola 2004 (SOLAR) 6 corticosteroids (ICS) plus a long-acting inhaled β-2 agonist (LABA). 16 2. Humbert 2005 (INNOVATE) zz Although randomised controlled trials (RCTs) have established the efficacy of omalizumab in ) 3. Bousquet 2010 (EXALT) 4. Hanania 2011 comparison to standard of care (SoC), there is uncertainty regarding real world effectiveness 12 5. Ayres 2004 (IA-04) of omalizumab, particularly with respect to the long-term effects. 9–11 6. Vennera 2012 512 7. Molimard 2008 zz 4 Therefore, an assessment of the available observational evidence is required in order to 8. Subramaniam 2013

understand real world effectiveness of omalizumab, as well as to compare the results to Time (months 8 9. Vennera 2012 evidence from RCTs. 10. Braunstahl 2011/2013 (eXpeRience) 3 1 2 11. Cazzola 2010 8 12. Barnes 2013 (APEX (AIC) 4 7 Objective 6 13. Braunstahl 2011/2013 (eXpeRience) 14. Vennera 2012 zz To assess and compare the relative efficacy and real world effectiveness of omalizumab in combination with SoC versus SoC alone among adults with uncontrolled severe persistent 0 allergic asthma based on RCTs and non-randomised observational trials. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 Rate ratio of clinically significant exacerbations Methods OBS: Observational study; OMA: Omalizumab; PLBO: Placebo; RCT: Randomised Controlled Trial; SoC: Standard of Care zz A systematic literature search was performed as an update to the original review that was conducted in October 2011 by Norman et al.7 zz In contrast to the rates for clinically significant exacerbations, the variation in severe zz Search strategy: MEDLINE® and EMBASE® databases and Cochrane library were searched for exacerbation rates across RCTs was limited with respect to baseline risk (SoC: 0.42 - 0.48) studies published between January 2011 and January 2014. and the treatment effect (RR: 0.50 - 0.56). However, considerable differences were identified for this outcome across the observational studies in terms of baseline risk (before omalizumab: zz Selection criteria: Table 1 The predefined selection criteria presented in were used to identify 2.20 - 4.50) and treatment effect (RR: 0.05 - 0.39) (Figure 4). relevant studies. zz The study durations for which these rates were reported ranging from 6.4 to 7.4 months Table 1. Study selection criteria across the RCTs and 3.7 to 48 months across the observational studies.

Adults (≥12 years) with severe persistent allergic asthma who meet the following criteria zz For clinically significant exacerbations or severe exacerbations, the point estimates seem to • A positive skin test or in-vitro reactivity to a perennial aeroallergen suggest that heterogeneity was present within RCTs, observational studies, and across both study designs. Populations • Reduced lung function (FEV1 < 80%) • Frequent daytime symptoms or night-time awakenings Figure 4. Rate ratio of severe exacerbations • Multiple documented severe asthma exacerbations despite daily high dose ICS plus a LABA 48 RCT (OMA vs. PLBO) Interventions Omalizumab in addition to best standard therapy at step 4 or 5 according to GINA 11 RCT (OMA vs. SoC) Optimised standard therapy (step 4 or 5) OBS (after vs. before OMA) 40 • Daily high-dose ICS plus a LABA with the possible addition of leukotriene receptor antagonist, theophyllines, or slow releasing β-2 agonist tablets (Step 4) Comparators Legend • Daily high-dose ICS plus a LABA with the possible addition of leukotriene receptor 32 1. Humbert 2005 (INNOVATE) antagonist, theophyllines, or slow releasing β-2 agonist tablets plus frequent or continuous ) 2. Bousquet 2010 (EXALT) OCS (or plus methotrexate, cyclosporine, or gold) (Step 5) 3. Schumann 2012 (XCLUSIVE) 4. Tzortzaki 2012 Physician’s GETE, incidence of clinically significant exacerbations, incidence of severe 24 5. Korn 2009 Outcomes exacerbations which require unscheduled contact with healthcare professionals, emergency 10 6. Castello 2011 visits or hospitalisations, hospitalisations, lung function (FEV1 and PEF), AQLQ, and use of OCS Time (months 7. Braunstahl 2011/2013 (eXpeRience) 16 Observational studies (≥30 patients), RCTs, systematic reviews and meta-analyses of RCTs 8. Tzortzaki 2012 Study design and/or observational studies 9. Bruselle 2009 (PERSIST) 7 8–9 10. Braunstahl 2011/2013 (eXpeRience) AQLQ: Asthma Quality of Life Questionnaire; FEV : Forced Expiratory Volume in 1 second; GETE: Global Evaluation of 1 8 11. Tzortzaki 2012 Treatment Effectiveness; GINA: Global Initiative for Asthma; ICS: Inhaled Corticosteroids; LABA: Long-Acting β-2 Agonist; 2 OCS: Oral Corticosteroids; PEF: Peak Expiratory Flow; RCTs: Randomised Controlled Trials 5 6 1 4 3 zz Data extraction was performed by one researcher and checked independently by another 0 researcher for any new studies that were identified in the systematic review update using a 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 predefined data extraction form. For studies identified in the existing systematic review, the Rate ratio of severe exacerbations data were checked by one researcher against the original publication. OBS: Observational study; OMA: Omalizumab; PLBO: Placebo; RCT: Randomised Controlled Trial; SoC: Standard of Care zz A feasibility assessment was performed to evaluate whether it was possible to synthesise the RCTs alone, the observational studies alone, or both study designs combined. zz Limited variation was observed in the rate of hospitalisations for RCTs in the baseline zz The assessment included a comparison of the treatments evaluated in each study, differences risk (SoC: 0.12 - 0.17) and in treatment effect (RR: 0.12 [95%CI: 0.029 - 0.475] to 0.54 in outcome definitions, study characteristics, and patient characteristics that might affect [95%CI: 0.250 - 1.166]), whereas there was substantial variation across the observational the relative treatment effects, differences in the baseline risk across the trials, and observed studies in terms of baseline risk (before omalizumab: 0.32 - 4.45) and treatment effect treatment effects involving a visual inspection of the heterogeneity in the reported outcomes (RR: 0.09 - 0.71) (Figure 5). in terms of exacerbations and hospitalisations (Figure 1). zz The time points at which these rates were reported ranged from 3.7 to 7.4 months across the Figure 1. Summary of the feasibility assessment process for meta-analysis8 RCTs and from 4 to 24 months across the observational studies, involving a wider range in follow-up time compared to the RCTs. Are there differences within treatment comparisons in terms of: zz For the rate of clinically significant exacerbations and severe exacerbations, meta-analysis was not feasible because of the limitations of data reported in the included studies with Part A respect to different outcomes evaluated and lack of estimates required for deriving the The treatment uncertainty. The total number of events before and after (and/or total patient-years) in trials (doses/schedules) or Part B outcome definitions is required to estimate standard error of RRs. The distribution of that are expected each study or patient zz However, the RRs of hospitalisations, in both RCTs and observational studies can be synthesised to modify relative characteristics if additional data regarding the total number of events (and/or total patient-year) is available. treatment effect? that are expected Part C zz Although it is possible to combine the RRs from both RCTs and before and after observational to modify relative Baseline risk that is treatment effect? studies for these outcomes, the interpretation will differ substantially given the study design also associated with and therefore is only suggested as potential exploratory sensitivity analysis. the relative treatment effects? Part D Figure 5. Rate ratio of hospitalisations Observed treatment effect? RCT (OMA vs. PLBO) 24 OBS (after vs. before OMA) No Yes 8 RCT (OMA vs. SoC) Plan to perform meta-analysis Sufficient data to explain or adjust for 20 OBS (OMA vs. SoC) and list any assumptions differences using meta-regression? Legend Yes No 1. Bousquet 2010 (EXALT) 16 2. Bousquet 2010 (EXALT) ) Plan to perform meta-regression Sufficient data to explore 3. Humbert 2005 (INNOVATE) to explain or adjust for differences differences using subgroups? 4. Subramaniam 2013 12 5. Vennera 2012 Yes No 6 7 9 10-11 6. Vennera 2012 7. Cazzola 2010

Plan subgroup analyses to Sufficient data to explore differences Time (months 8 2 8. Vennera 2012 explore differences using sensitivity analyses? 9. Grimaldi-Bansouda 2013* 10. Grimaldi-Bansouda 2013 Yes No 4 3 11. Rottem 2012 4 12 12. Molimard 2008 Plan sensitivity analyses to Summarise why meta-analysis is not 5 explore differences feasible and identify required studies 1 *Adjusted rate ratio Adapted from Cope et al. 20148 0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 Rate ratio of hospitalisations Results OBS: Observational study; OMA: Omalizumab; PLBO: Placebo; RCT: Randomised Controlled Trial; SoC: Standard of Care zz Of the total 499 potentially relevant abstracts screened, 53 full-text articles were included in the systematic literature review. The 53 publications reported 38 separate studies, of which 12 were RCTs, 24 were observational studies, and 2 were systematic reviews (Figure 2). Conclusion

Figure 2. Study selection flow chart zz Real world evidence reinforces the efficacy of omalizumab in reducing clinically significant exacerbations or severe exacerbations rate in patients with severe allergic asthma derived ® ® Abstracts identified by MEDLINE and EMBASE (617) from RCTs, although differences in potential treatment effect modifiers were identified. Abstracts identified by Cochrane (30) There may be differences between the characteristics of the patients included in the observational studies compared to the RCTs in terms of severity. Duplicates excluded (148)

References Abstracts reviewed (499) 1. British Guideline on the Management of Asthma. British Thoracic Society. 2012. References excluded: (395) 2. GINA. Global Strategy for Asthma Management and Prevention. 2013. Duplicates (31) 3. Camarda LE et al. Am J Respir Crit Care Med 2011;184(2):152–3. Study design (99) 4. Yang W et al. Curr Med Res Opin 2010;26(9):2065–76. Patient population (177) 5. Knudsen TB et al. J Asthma 2009;46(1):91–4. Intervention (6) Observational study size (<30) (75) 6. European public assessment report (EPAR) summary for the public: Xolair, 2014. Available from URL: http://www.ema. Outcomes (7) europa.eu/docs/en_GB/document_library/EPAR_-_Summary_for_the_public/human/000606/WC500057293.pdf 7. Norman G et al. Health Technol Assess 2013;17(52). Full-text articles reviewed (104) 8. Cope S et al. BMC Medicine 2014;12:93.

Acknowledgements References excluded: (67) The authors thank Nishkarsh Likhar (Novartis) for assistance with poster content and Rajesh Nadella (Novartis) for designing 14 pre-2011 studies included Duplicates (5) the poster layout. from HTA (Norman et al, 2013), Study design (20) Patient population (19) 1 publication identified through Disclosure bibliographic searching Observational study size (<30) (10) Outcomes (9) This study was sponsored by Novartis Pharma AG. *Baldwin M was employed by Novartis at the time of development of Comparator (1) systematic literature review. Language (2) Unable to obtain (1)

53 Full-text articles included reporting 38 studies (12 RCTs, 24 observational Scan to download a reprint of this poster studies, and 2 systematic reviews) Downloading data may incur costs which can vary depending on your service provider and may be high if you are using your smartphones abroad. Please check your phone tariff or contact your service provider for more details. PHARMACEUTICALS

Poster presented at ISPOR 17th Annual European Congress, 8–12 November, 2014, Amsterdam, The Netherlands