Cardiovascular and cerebrovascular events among patients receiving omalizumab: Pooled analysis of patient-level data from 25 randomized, double-blind, placebo-controlled clinical trials.

Omalizumab is an anti-IgE antibody used in the treatment of moderate to severe persistent asthma. EXCELS (An Epidemiologic Study of Xolair [omalizumab]: Evaluating Clinical Effectiveness and Long-term Safety in Patients With Moderate-to-Severe Asthma; NCT00252135) was a prospective observational cohort study conducted as a postmarketing commitment to the US Food and Drug Administration to better understand the long-term safety of omalizumab in clinical practice. Results from EXCELS suggest that omalizumab is not associated with increased risk of malignancy.1Long A. Rahmaoui A. Rothman K.J. Guinan E. Eisner M. Bradley M.S. et al.Incidence of malignancy in patients with moderate-to-severe asthma treated with or without omalizumab.J Allergy Clin Immunol. 2014; 134: 560-567.e4Abstract Full Text Full Text PDF PubMed Scopus (164) Google Scholar In a subsequent analysis, however, we reported an imbalance in cardiovascular and cerebrovascular events.2Iribarren C. Rahmaoui A. Long A.A. Szefler S.J. Bradley M.S. Carrigan G. et al.Cardiovascular and cerebrovascular events among patients receiving omalizumab: results from EXCELS, a prospective cohort study of moderate-to-severe asthma.J Allergy Clin Immunol. 2016 Sep 14; ([Epub ahead of print]. http://dx.doi.org/10.1016/j.jaci.2016.07.038)PubMed Google Scholar The observational design of EXCELS precluded randomization and blinding to treatment, and thus differences in asthma severity between cohorts at baseline may have contributed to this imbalance, but some increase in risk could not be excluded. Here, we report results from a separate analysis of cardiovascular events using pooled data from randomized controlled trials (RCTs) of omalizumab. Twenty-five core trials and 2 extension studies met the inclusion criteria for this pooled analysis (see Table E1 in this article's Online Repository at www.jacionline.org); data included patients recruited from 26 countries. Common reasons for exclusion included open-label or single-arm (uncontrolled) design, limited drug exposure (eg, single dose), and pediatric-only studies. All potential events of interest were identified from the Adverse Event Report Form or Discontinuation Case Report Form using a prespecified search strategy (see Fig E1 in this article's Online Repository at www.jacionline.org). External adjudication was then performed by 5 cardiologists and 3 neurologists. Additional methodologic information and statistical analysis can be found in this article's Online Repository (see text and Table E2 in this article's Online Repository at www.jacionline.org). The pooled population of RCTs included 3342 omalizumab-treated patients and 2895 placebo-treated patients, of whom 72% (n = 2409) and 80% (n = 2320), respectively, were treated for allergic asthma. Total observation time was 1857 person-years (PYs) for omalizumab and 1681 PYs for placebo; median duration of treatment was 154 days (omalizumab) and 161 days (placebo). Baseline demographic and clinical characteristics were generally comparable between the treatment groups (see Table E3 in this article's Online Repository at www.jacionline.org). Using the broad search strategy, 351 potential events of interest occurring in 249 patients were identified for external adjudication (Fig E1). The prespecified primary analysis included all deaths and serious arterial thrombotic events (ATEs) (Table I). The results are shown in Fig 1. The Mantel-Haenszel rate difference per 1000 PYs, stratified by clinical trial, for ATEs overall (rate difference, 0.30; 95% CI, –1.6 to 2.2) was consistent with the primary analysis (rate difference, 0.30; 95% CI, −4.2 to 4.6). The largest difference observed was for arrhythmia, which was higher in the placebo versus omalizumab groups. Results were similar in the main subgroup treated for allergic asthma (data not shown), where the rate difference was 0.48 events per 1000 PYs (95% CI, −4.7 to 5.6) and the rate ratio (RR) was 1.2 (95% CI, 0.25-5.9).Table IPrimary and secondary analyses: Rates of AEs, ATEs, and other CV SAEsPrimary analysisAll ATEs: Deaths and SAEsOmalizumab (n = 3342)Placebo (n = 2895)Events, n54PYs, n18561681ATE rate per 1000 PYs (95% CI)2.7 (0.88-6.3)2.4 (0.65-6.1)Event rate per 1000 PYs (95% CI) CV death0.00 (0.00-2.0)1.8 (0.37-5.2) MI1.1 (0.13-3.9)0.00 (0.00-2.2) Stroke0.54 (0.01-3.0)0.00 (0.00-2.2) TIA0.54 (0.01-3.0)0.00 (0.00-2.2) Unstable angina0.54 (0.01-3.0)0.60 (0.02-3.3)Primary analysisOther CV SAEsOmalizumab (n = 3342)Placebo (n = 2895)Events, n311PYs, range1852-18571674-1681Event rate per 1000 PYs (95% CI) Arrhythmia0.54 (0.01-3.0)4.8 (2.1-9.4) Heart failure0.54 (0.01-3.0)0.0 (0.0-2.2) Pulmonary HTN0.00 (0.00-1.2)0.59 (0.02-3.3) PE/DVT0.54 (0.01-3.0)1.2 (0.14-4.3)Secondary analysisAll AEsOmalizumab (n = 3342)Placebo (n = 2895)ATEs, n64PYs, n18551680Rate per 1000 PYs (95% CI)3.2 (1.2-7.0)2.4 (0.65-6.1)Stroke, n20PYs, n18561681Rate per 1000 PYs (95% CI)1.08 (0.13-3.9)0.00 (0.00-2.2)Arrhythmia, n1324PYs, n18521674Rate per 1000 PYs (95% CI)7.0 (3.7-12)14 (9.2-21)PE/DVT, n33PYs, n18551681Rate per 1000 PYs (95% CI)1.6 (0.33-4.7)1.8 (0.37-5.2)AE, Adverse event; CV, cardiovascular; DVT, deep vein thrombosis; HTN, hypertension; MI, myocardial infarction; PE, pulmonary embolism; SAE, serious adverse event; TIA, transient ischemic attack. Open table in a new tab AE, Adverse event; CV, cardiovascular; DVT, deep vein thrombosis; HTN, hypertension; MI, myocardial infarction; PE, pulmonary embolism; SAE, serious adverse event; TIA, transient ischemic attack. The secondary analysis included all deaths and all ATEs, serious and nonserious (Table I). The rate difference was 0.85 events per 1000 PYs (95% CI, –3.7 to 5.3), with an RR of 1.4 (95% CI, 0.32-6.6). Results were similar in the main subgroup treated for allergic asthma (data not shown), where the rate difference was 1.1 events per 1000 PYs (95% CI, −4.2 to 6.4) and the RR was 1.4 (95% CI, 0.33-6.8). Further subgroup analyses were not generally informative owing to the small number of overall ATEs (see Table E4 in this article's Online Repository at www.jacionline.org). Taken together, this analysis of pooled safety data from more than 6200 participants in 25 RCTs (>3500 PYs of observation) revealed few cardiovascular events in either the omalizumab group (n = 5) or the placebo group (n = 4). Rates for ATE, as well as for individual events, were similar between the 2 groups; however, the CIs were wide, reflecting statistical uncertainty of these estimates. The secondary analysis, which included nonserious events, showed higher rates in both the omalizumab and placebo groups, but the rate differences were also small and the CIs were wide and included the null hypothesis. Findings from this analysis should be interpreted in the context of its own limitations. First, owing to the duration of clinical trials, the results are relevant only as they relate to shorter-term exposure to omalizumab. Second, patients participating in clinical trials often are different than those treated in actual practice. Third, although randomization is intended to balance unmeasured confounders in any given individual trial, it does not guarantee balanced groups when the data are pooled. To address this concern, a sensitivity analysis stratified by clinical trial also was performed and yielded similar results. Finally, although every attempt was made to keep adjudicators blinded to treatment status, the blinded case narratives were written by a third party (ie, neither sponsor nor adjudicator), who was unblinded to treatment allocation. In conclusion, in this analysis of pooled RCT data, the rates of observed ATEs were similar between the omalizumab and placebo groups. However, the low number of events and wide CIs limit the ability to exclude small differences in risk. We acknowledge former Novartis employees Janice Canvin, Martin Blogg, John Powell, and Guy Peachey for their invaluable contributions to the study design, data collection and analysis, and interpretation of the results. Download .docx (.05 MB) Help with docx files Online Repository text