University of Washington Seattle, WA, United States
Disclosure information not submitted.
Xavier Sendaydiego1, Laura Gold2, Jean liew3, K Wysham4, Maureen Dubreuil5, James Andrews1, Pankti Reid6, David Liew7, Radjiv Goulabchand8, Abha Singh9, Grant Hughes1, Mathilde Pioro1, Jeffrey Sparks10, Jeffrey Jarvik2, Siddharth Singh9 and Namrata Singh11, 1University of Washington, Seattle, WA, 2Department of Radiology and University of Washington Clinical Learning, Evidence, and Research (CLEAR) Center for Musculoskeletal Disorders, Seattle, WA, 3Boston University, Boston, MA, 4VA Puget Sound/University of Washington, Seattle, WA, 5Department of Rheumatology, Boston University School of Medicine, Milton, MA, 6University of Chicago Medical Center, Chicago, IL, 7Austin Health, Heidelberg, Australia, 8St. Eloi Hospital, Department of Internal Medicine and Multi-Organic Diseases, Montpellier, France, 9University of California San Diego, San Diego, CA, 10Division of Rheumatology, Inflammation, and Immunity, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, 11University of Washington, Bellevue, WA
Background/Purpose: In the ORAL Surveillance trial, cancer risk was higher among patients with rheumatoid arthritis (RA) on tofacitinib, a Janus kinase inhibitor (JAKi), compared to tumor necrosis factor inhibitors (TNFi), in a population enriched for cardiovascular disease risk [1]. However, less is known regarding the comparative safety of non-TNFi biologics relative to TNFi. We assessed the comparative safety of individual non-TNFi and JAKi relative to TNFi for the risk of incident cancer in patients with RA.
Methods: We performed a cohort study using Merative MarketScan databases (2012-2021) of patients with RA identified using ≥1 ICD9/10 codes, age 18-64 years, who initiated treatment with TNFi, non-TNFi (rituximab, abatacept, tocilizumab and sarilumab), or JAKi (tofacitinib, baricitinib) on or after November 2012. Patients with past cancer diagnoses were excluded. We used Cox proportional hazard models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for developing incident cancer (excluding non-melanoma skin cancer) within 2 years of treatment initiation in patients on non-TNFi or JAKi relative to TNFi, adjusting for potential confounders, including demographics, geographic region, year of initiating a biologic, Charlson Comorbidity Index, frailty status measured using claims-based frailty index (2), healthcare utilization within 12 months prior to starting treatment, and proxies for RA severity. Cancer diagnoses were identified using validated administrative algorithms [3]. Patients were allowed to switch from one drug class to the other, allowing a single patient to contribute person-time over different drug classes. Patients were censored if they did fill of a prescription of any of these drugs for >90 days ( >180 days for rituximab), or end of study period (12/31/2021).
Results: We included 37,026 patients involving 78% female patients with a mean age of 47.6 ± 10.3 years, of whom 72% initiated TNFi, 10% JAKi, 8% abatacept, 5% rituximab, 3% tocilizumab, and 2% sarilumab (Table 1). The mean follow-up time was 360 days for TNFi, 250 days for non-TNFi, and 280 days for JAKi. There were 379 incident cancers observed during follow-up (Table 2). In multivariable models, exposure to rituximab or abatacept had a significantly higher risk of incident cancer (HR 2.2, 95% CI 1.5, 3.3; HR 1.7, 95% CI 1.3-2.4, respectively), compared with exposure to TNFi (Figure 1). While the hazard ratio for incident cancer was higher with exposure to JAKi compared with TNFi, this difference was not statistically significant (HR 1.3; 95% CI 0.9-1.9).
Conclusion: While we observed a lower hazard ratio for incident cancer with exposure to TNFi compared to non-TNFi and possibly JAKi in this generally younger and predominantly female population, potential for residual confounding by indication and the small number of outcomes per drug class limit interpretation of these results. Larger studies with longer follow-up are needed for better comparison of cancer risk between these drug classes.
References: 1. Ytterberg SR, et al: N Engl J Med 2022 2. Kim DH, et al: J Gerontol A Biol Sci Med Sci 2018 3. Setoguchi S, et al: Cancer causes & control : CCC 2007
Table 1. Baseline characteristics of the cohort stratified by drug initiation.
Table 2. Number of cancer outcomes (95% confidence intervals) within 2 years of initiating biologic drug per 10,000 person-years at risk, stratified by drug category that the patient initiated.
Figure 1. Kaplan-Meier curves showing the adjusted hazard ratio* for incident cancer per drug exposures.
X. Sendaydiego: None; L. Gold: None; J. liew: None; K. Wysham: None; M. Dubreuil: Amgen, 2, Pfizer, 5, UCB Pharma, 2; J. Andrews: None; P. Reid: None; D. Liew: None; R. Goulabchand: Novartis, 2; A. Singh: AbbVie/Abbott, 5, Novartis, 5, Pfizer, 5; G. Hughes: Janssen, 3; M. Pioro: None; J. Sparks: AbbVie, 2, Amgen, 2, Boehringer Ingelheim, 2, Bristol-Myers Squibb, 2, 5, Gilead, 2, Inova Diagnostics, 2, Janssen, 2, Optum, 2, Pfizer, 2, ReCor, 2; J. Jarvik: GE Healthcare, 12, Travel reimbursement for Faculty Board of Review for GE-Association of University Radiologists Radiology Research Academic Fellowship (GERRAF), Springer Publishing, 9, Wolters Kluwer/UpToDate, 9; S. Singh: None; N. Singh: None.
