Abstract
BACKGROUND: Identifying modifiable risk factors is essential for the prevention of skin cancer; however, establishing causality can be challenging in conventional epidemiological studies. This study aimed to determine the causal associations of potentially modifiable risk factors with skin cancer using Mendelian randomization (MR). METHODS: Genetic instruments for 53 risk factors, including socioeconomic status, dietary and lifestyle factors, anthropometric measures, medication use, and comorbidities, were identified from previous genome-wide association studies. Two-sample MR analyses were performed using summary statistics for three major types of skin cancer: melanoma, basal cell carcinoma (BCC), and squamous cell carcinoma (SCC). Findings were verified using multiple MR methods under different assumptions and replication datasets. RESULTS: Genetic liability to sunburn occasions, actinic keratosis, and prior skin cancers was associated with a higher risk of all three types of skin cancer, whereas genetic liability to vitiligo was associated with a lower risk. For specific skin cancer types, genetically predicted higher nevus counts and occupational class were associated with an increased risk of melanoma. Genetic liability to rheumatoid arthritis, type 2 diabetes, and increased physical activity were associated with a lower risk of BCC. Genetically predicated body mass index showed a negative association with BCC, and a positive association with SCC. CONCLUSIONS: Our study reaffirmed several previously established risk factors and identified novel potential risk factors for skin cancer. Further work is needed to unravel the biological pathways in different skin cancer types and translate our findings to inform public health policies.