Basal cell carcinoma risk prediction in survivors of childhood cancer

Background Survivors of childhood cancer face excess risk of developing basal cell carcinoma. Age-specific basal cell carcinoma risk prediction models for survivors may support targeted screening recommendations. Methods We developed models predicting basal cell carcinoma risk by ages 40 and 50years featuring detailed cancer treatment predictors, utilizing statistical and machine-learning algorithms and data from 23 166 five-year survivors in the Childhood Cancer Survivor Study, a multi-institutional retrospective cohort study. Selected models were externally validated in 5314 survivors in the St Jude Lifetime Cohort. Model discrimination and precision were evaluated using the area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC) and benchmarked against the current Children’s Oncology Group Long-Term Follow-Up Guidelines (COG LTFU, v6.0) for skin cancer screening. Results By ages 40 and 50years, basal cell carcinoma cumulative incidence was 5% and 15% in the Childhood Cancer Survivor Study and 7% and 21% in the St Jude Lifetime Cohort, respectively. The XGBoost algorithm-based models with treatment dose-specific predictors performed best, showing good external discrimination (age 40 years: AUROC = 0.75; age 50 years: AUROC = 0.76) and precision (age 40 years: AUPRC = 0.20; age 50 years: AUPRC = 0.52), outperforming COG LTFU Guideline-directed risk stratification (age 40 years: AUROC = 0.65; age 50 years: AUROC = 0.62; age 40 years: AUPRC = 0.09; age 50 years: AUPRC = 0.26; P<.01). These novel models reclassified 37% of survivors with COG-recommended skin cancer screening as low risk by age 40 years and 29% of survivors without COG-recommended screening as moderate or high risk by age 50 years, suggesting these recommendations overestimate risk in younger survivors and miss relevant predictors (eg, attained age, chemotherapy). Conclusions In this study, we present validated basal cell carcinoma risk prediction models for childhood cancer survivors that outperform current practice guidelines. The associated online risk calculator can inform risk- and age-based screening recommendations.