Risk Prediction Model for Skin Cancers in Cardiac Transplant Patients: A UNOS Database Analysis

Purpose Post-transplant patients are at increased risk for cancers of the skin. This study attempts to generate a risk prediction model using the UNOS database population. Methods Of the 24,735 adults (age 18 years and above) heart transplant recipients between 2000 and 2015 in the UNOS database 2,625 recipients developed skin cancer. Uni- and multi-variate Cox regression analysis were performed to assess the association of different risk factors with post-transplant skin cancer, and p-values, hazard ratios and their confidence intervals derived. . Variables in the univariate analysis were selected as inputs to the multivariate analysis. Stepwise forward selection was conducted to select the final multivariate model. The multivariate model was used to predict the probability of developing skin cancer in 5, 8, and 10 years after heart transplantation. The model accuracy was assessed using Receiver Operating Characteristics (ROC) curves and Area Under Curves (AUCs). All the analysis was performed using MATLAB software from the MathWorks, Inc. Results Multivariate analysis showed that whites had an OR of 35.5 as compared to blacks (p=0.001). Males had an OR of 2.79 (p=0.001). Malignancy at listing showed an OR of 1.73 (p=0.001) while malignancy at transplant showed an OR of 2.42 (p=0.001). OKT3 had an OR of 1.54 (p=0.001) as compared to other induction agents The ROC curves generated using these risk factors showed AUC at 5, 8 and 10 years post-transplant of 0.78, 0.78, 0.77 respectively in the training set and 0.77, 0.76, 0.75 respectively in the validation set (figure 1). Conclusion Male sex, white race, malignancy at the time of listing or at transplantation and OKT3 induction are major risk factors for skin cancers post transplantation. The risk prediction model generated for the first time for skin cancers in the post-cardiac transplant patients has a c-statistic of 0.75.