Risk Prediction Model for Basal Cell Carcinoma in Cardiac Allograft Recipients

N. Nair, D. Du, Z. Hu, E. Gongora

Research output: Contribution to journalArticlepeer-review


PURPOSE: Basal cell carcinoma (BCC) is the second most common skin cancer in post-transplant patients. Long-term immunosuppression predisposes them to higher risk. This study was undertaken to identify risk factors using the UNOS database. METHODS: The study cohort of 1133 patients derived from the UNOS database (2000-2015) was divided equally into testing and validation groups. Descriptive statistics were used to characterize the study cohort. Cox's proportional hazards regression was used to investigate the effect of different risk factors on the development of BCC post-heart transplantation. Univariate and multivariate analysis was used to test the correlation between BCC and potential risk factors. Hazard ratio (OR) and the 95% confidence interval (CI) were calculated for each risk factor. Statistical significance was set at p< 0.05. Forward sequential feature selection was done to select significant risk factors for multivariate modeling. The multivariate model was used to predict the probability to develop skin BCC at 5, 8 and 10 years post-transplant. Analyses were performed using MATLAB software from The MathWorks, Inc. Receiver Operating Curves (ROCs) were generated to derive Area Under the Curve (AUC) to assess the quality of the prediction model. RESULTS: Multivariate analysis showed that the presence of lower PRA at transplantation had an OR=0.96 (p<0.0001). Black, Hispanic and Other races as compared to whites had an OR of 0.01, 0.3, 0.18 (p=0.0000, p=0.0000, p=0.0002 respectively). Age >60 years had an OR of 2.2 (p=0.0000). Males had an OR of 1.8 (p=0.000). The ROCs showed AUC at 5, 8 and 10 years post-transplant of 0.73, 0.74, and 0.74 respectively in the testing set and 0.79, 0.76, 0.75 respectively in the validation set (figure1). CONCLUSION: Higher PRAs at transplant, male sex, white race, age >60 years are major risk factors for BCC. A risk prediction model has been generated for the first time for BCC with a c-statistic of >0.7 in both testing and validation sets making it a good tool for risk stratification.

Fingerprint Dive into the research topics of 'Risk Prediction Model for Basal Cell Carcinoma in Cardiac Allograft Recipients'. Together they form a unique fingerprint.

Cite this