Skin Cancer Risk Associated With Statin Treatment: A Target Trial Emulation Study

Background: There are still conflicting results about the association between statin treatment and skin cancer risk. Studies focusing on Caucasian population might not be applicable to the Taiwanese population because the incidence of skin cancer varies across ethnic groups. Additionally, many current studies compared the effect of statin use on the skin cancer risk between statin users and non-users, which was potentially susceptible to confounding effect.

Objectives: This study aimed to evaluate the association of continuous statin use with the risk of skin cancer among Taiwanese population.

Methods: We employed a target trial framework to emulate a pragmatic, open-label, clinical trial. We included new users of statins aged ≥ 20 years and older from Taiwan's National Health Insurance Research Database from 2013 to 2016. Patients were assigned into different groups based on their medication possession ratio (MPR) of statin use within the first two years after starting statins. We defined those MPR above 80% as continuous users and those MPR less than 80% as comparator. We followed up patients from the first day after the two-year window for the treatment assignment. The outcomes were non-melanoma skin cancer (NMSC) risk and melanoma skin cancer (MSC) risk. We applied the stabilized inverse probability of propensity score to balance the baseline covariates and confounders. We used the Cox proportional hazards model to generate hazard ratios (HR) and 95% confidence intervals (CI) to evaluate the risk of skin cancers between groups. We conducted sensitivity analyses that changed the threshold of continuous user of statins from 80 to 70 and 90%, to investigate the dose-response relationship.

Results: A total 53713 eligible patients were included, with 6264 were assigned as continuous user of statin as continuous users of statins, while the remaining 47449 patients served as the comparator group. Continuous use of statin was significantly associated with increased risk of NMSC compared with comparator group (HR=2.68, 95% CI=1.64-4.40), while no association was found with MSC risk (HR=1.05, 95% CI=0.16-7.00). We also observed a heightened risk of NMSC with an increased MPR for statin usage (MPR threshold 70 with HR= 2.04, 95% CI=1.24-3.35; MPR threshold 90 with HR=3.03, 95% CI=1.78-5.15), strengthening the causal inference between use of statin and NMSC.

Conclusions: The results indicated the use of statins was associated with an increased risk of NMSC but not MSC in the Taiwanese population when all measured confounders were controlled. A dose-response relationship strengthened the causality between the use of statins and the risk of NMSC.