Serum estradiol, reproductive timing, and melanoma prevalence among women: population-based evidence and single-cell transcriptomic insights

Evidence linking female reproductive factors and circulating estradiol with melanoma has been inconsistent. We conducted a cross-sectional analysis of US women in NHANES 2009-2023. Exposure-specific analytic samples were constructed for age at menarche, pregnancy history, natural menopause age, and serum estradiol. Associations with prevalent melanoma were examined using survey-weighted multivariable logistic regression. XGBoost with Shapley Additive exPlanations (SHAP) was used as a complementary prediction analysis. Two melanoma single-cell RNA-sequencing datasets (GSE269936 and GSE115978) were analyzed to examine the patient-level association between inferred ESR2 activity and E2F/G2M proliferation-related transcriptional activity. Low estradiol levels (0-20 pg/mL) were associated with higher odds of prevalent melanoma [adjusted odds ratio (OR) = 3.382; 95% confidence interval (CI) = 1.304-8.773; P = 0.013], and late menarche (>14 years) was also associated with higher odds (adjusted OR = 2.074; 95% CI = 1.118-3.846; P = 0.021). Pregnancy history was not significantly associated with melanoma prevalence. The association for premature menopause could not be estimated because no melanoma cases occurred in this category. Low estradiol was the highest-ranked encoded feature by global SHAP importance (mean |SHAP| = 0.157). Across 10 patients inferred ESR2 activity was inversely correlated with the composite proliferation score (pooled r = -0.250; 95% CI = -0.392 to -0.095; P = 0.006). Lower estradiol levels and later menarche were associated with higher odds of prevalent melanoma. The single-cell finding requires prospective and functional studies are needed to determine the biological and clinical relevance of these findings.