Systematic literature review and meta-analysis informing the EULAR points to consider on the initiation of targeted therapies in patients with inflammatory arthritis and a history of cancer.

BACKGROUND: Targeted therapies have been associated with potential risk of malignancy, which is a common concern in daily rheumatology practice in patients with inflammatory arthritis (IA) and a history of cancer. OBJECTIVES: To perform a systematic literature review to inform a Task Force formulating EULAR points to consider on the initiation of targeted therapies in patients with IA and a history of cancer. METHODS: Specific research questions were defined within the Task Force before formulating the exact research queries with a librarian. We included studies reporting a relative risk measure of patients with a history of cancer initiating a targeted therapy or a conventional synthetic disease-modifying antirheumatic drug (csDMARD), regardless of the time since diagnosis of cancer. All relevant studies included in PubMed or Embase up to 15 July 2022 were included. Two reviewers independently performed standardised article selection, data extraction, synthesis and risk of bias assessment. RESULTS: 14 published articles and one ACR abstract fulfilled the inclusion criteria. All studies were high-quality observational studies, representing a median follow-up from treatment initiation of 4.52 years among 4428 patients and 15 062 patient-years of follow-up for new or recurrent cancer. All patients had a history of cancer, most frequently solid cancer, most frequently receiving treatment for rheumatoid arthritis and most frequently treated with tumour necrosis factor-alpha inhibitors. Across these studies, the overall HR of new incident cancer or cancer recurrence was 0.90 (95% CI 0.74 to 1.10) for patients receiving a targeted therapy versus a csDMARD. CONCLUSION: Overall, the targeted therapies and clinical contexts covered by the included studies were not associated with an increased risk of new incident cancer or cancer recurrence as compared with csDMARDs.