RISK OF NON-MELANOMA SKIN CANCER WITH TNFi AND OTHER bDMARDs IN PATIENTS WITH PSORIATIC ARTHRITIS: A SWEDISH AND DANISH COHORT STUDY

BACKGROUND: Non-melanoma skin cancer (NMSC) comprises one third of all malignancies diagnosed worldwide each year. NMSC primarily consists of two subtypes: basal cell carcinoma (BCC) and squamous cell carcinoma (SCC). Current literature suggests that psoriasis is associated with an increased risk of NMSC, but studies on the risk of NMSC in patients with psoriatic arthritis (PsA) remain sparse.[1] Under particular scrutiny is the impact of treatment with biologic disease-modifying anti-rheumatic drugs (bDMARDs)—including tumour necrosis factor inhibitors (TNFi)—on NMSC risk, which in patients with PsA is largely unknown. OBJECTIVES: To explore the risk of NMSC overall, BCC, and SCC in patients with PsA treated with TNFi or other bDMARDs compared with bDMARD-naïve PsA patients in a real-world clinical setting. METHODS: We performed an observational cohort study using nationwide health care and clinical rheumatology registers (SRQ, DANBIO) in Sweden and Denmark. With a study period from 1 January 2010 to 31 December 2021, using SRQ and DANBIO we identified all patients with PsA above 18 years of age and without any prior cancer diagnosis who were: 1) bDMARD naïve, 2) a TNFi initiator, or 3) an initiator of other bDMARDs. Follow-up started on the date of the first registered visit in SRQ/DANBIO, first ever start of TNFi, and first ever start of any other bDMARD, respectively. Each patient was followed for NMSC overall as well as for BCC and SCC outcomes separately. We employed an ‘ever-treated’ approach, although follow-up in the bDMARD naïve group was stopped upon start of TNFi or start of any other bDMARD. This allowed for each unique patient to contribute person years (PY) of follow-up and outcomes to several treatment groups. Country-specific hazard ratios (HR) with 95% confidence intervals (95%CI) for NMSC overall, BCC, and SCC were calculated using Cox regression with attained age as time scale, hereby comparing both TNFi and other bDMARDs with bDMARD naïve PsA patients as the reference group. All Cox models were adjusted for calendar period, sex, and PsA disease duration. We used a fixed-effects meta-analysis to estimate pooled HRs and 95%CIs. Results: For Sweden and Denmark combined, we identified 17 354 bDMARD-naïve, 8458 TNFi-treated, and 2012 other bDMARD-treated patients with PsA, contributing 87 161, 44 317, and 7598 PY of follow-up, respectively. A total of 651 NMSC overall, 505 BCC, and 87 SCC occurred during follow-up. In the pooled analyses (see Figure 1), we found a HR of 1.25 (95%CI 1.04 to 1.50) for NMSC in TNFi treated compared with bDMARD naïve patients with PsA. The HR for NMSC in those treated with other bDMARDs was 1.19 (95%CI 0.82 to 1.72). Specifically looking at BCC, the TNFi group showed a HR of 1.23 (95%CI 0.99 to 1.52) compared with bDMARD naïve, while the HR of BCC with other bDMARDs was 0.98 (95%CI 0.62 to 1.57). The HR for SCC was numerically increased in the TNFi group with a HR of 1.56 (95%CI 0.91 to 2.65) compared with bDMARD naïve, and albeit based on few events in the SCC group (n=8) we also observed an increased HR for SCC in PsA patients treated with other bDMARDs: HR 2.95 (95%CI 1.26 to 6.91). CONCLUSION: In this Swedish-Danish cohort study on patients with PsA, treatment with TNFi was associated with a small increased risk of NMSC and with non-significant numeric risk increases for both BCC and SCC compared with bDMARD naïve. Treatment with other bDMARDs was not associated with an increased risk of BCC, whereas we found a threefold increase in risk for SCC compared with bDMARD naive. Although based on few events and the fact that we cannot rule out the presence of channeling bias and surveillance bias, this signal of increased SCC risk in PsA patients treated with other bDMARDs calls for further investigations.