ABSTRACT
Skin cancer risk is increased by exposure to ultraviolet radiation (UVR). Because UVR exposure accumulates over time and lighter skin is more susceptible to UVR, age and skin tone are risk factors for skin cancer. However, measurements of somatic mutations in healthy-appearing skin have not been used to calculate skin cancer risk. In this study, we developed a noninvasive test that quantifies somatic mutations in healthy-appearing sun-exposed skin and applied it to a 1038-subject cohort. Somatic mutations were combined with other known skin cancer risk factors to train a model to calculate risk. The final model (DNA-Skin Cancer Assessment of Risk) was trained to predict personal history of skin cancer from age, family history, skin tone, and mutation count. The addition of mutation count significantly improved model performance (OR = 1.3, 95% confidence interval = 1.14-1.48; P = 5.3 × 10-6) and made a more significant contribution than skin tone. Calculations of skin cancer risk matched the known United States population prevalence, indicating that DNA-Skin Cancer Assessment of Risk was well-calibrated. In conclusion, somatic mutations in healthy-appearing sun-exposed skin increase skin cancer risk, and mutations capture risk information that is not accounted for by other risk factors. Clinical utility is supported by the noninvasive nature of skin sample collection through adhesive patches.
ABSTRACT
Purpose: HLA-B∗15:02 is strongly associated with life-threatening severe skin hypersensitivity reactions in patients treated with carbamazepine (CBZ) and structurally related medications. FDA-approved labeling recommends HLA-B∗15:02 screening before CBZ therapy in patients of Asian ancestry. In this study, we aimed to (a) identify a direct method for screening HLA-B∗15:02, and (b) evaluate prevalence in a large cohort of United States patients. Methods: Candidate genetic markers were identified by mining public data. Association was tested in 28,897 individuals by comparing SNP results with high-resolution HLA typing. Retrospective analysis of de-identified SNP and ethnicity data from 130,460 individuals was performed to evaluate the ethnic distribution of HLA-B∗15:02 in the United States. Results: 28,897 United States individuals showed 100% concordance between HLA-B∗15:02 and the minor allele of rs144012689 (100% sensitivity/99.97% specificity). Retrospective analysis of 160 positive individuals (66 with physician-reported ethnicity) notably included 28 Asians (42%), 15 African Americans (22%), 11 Caucasians (17%), 2 Hispanics (3%), and 10 "Other" (15%). Conclusion: Screening United States patients for HLA-B∗15:02 without ethnicity-based preselection identifies more than twice the number of carriers at risk of CBZ-related adverse events than screening patients of Asian ancestry alone. Risk assessment based on ethnicity assumptions may not identify a large portion of at-risk patients in the ethnically diverse United States population.
