Development and Internal Validation of a Model to Predict Malignancy within a Year After Heart Transplantation in Adults

Abstract

<h3>Purpose</h3> Malignancy after heart transplantation (HT) is associated with poor outcomes. A model to predict the risk of malignancy within the first year after HT is urgently required. <h3>Methods</h3> We included adults (>18 years of age) who underwent HT included in the Scientific Registry of Transplant Recipients (SRTR) between January 2000 and May 2021. Candidate predictors were identified based on their association with malignancy in literature, included recipient age, gender, race, height, weight, kidney function, heart failure etiology, prior history of hypertension, diabetes, and smoking, induction therapy use, various histocompatibility results (CMV, EBV, DR mismatches), and transplant year. A multivariable regression model for predicting malignancy development within a first year after HT was developed and internally validated via 500 bootstrapped samples to estimate the optimism-corrected measures of model accuracy and performance. <h3>Results</h3> Among the 47,212 HT recipients with a median (25^th, 75^th percentile) age of 56 (46, 62) years, 25.6% were female, 75.8% were White, and 7.2% had prior malignancy. 2.3% developed malignancy within the first year. Recipient age at transplant, history of malignancy, recipient White race, and heart failure etiology were the strongest predictors of malignancy. The optimism-corrected model had modest discrimination (C-statistic: 0.699, 95%CI 0.688-0.714) and good calibration and performance (calibration slope: 0.967; Nagelkerke R² 0.065), particularly at lower predicted risk (Figure). <h3>Conclusion</h3> Despite selecting variables previously linked to malignancy, our model was modestly predictive of the development of malignancy in the first year after heart transplantation. These findings highlight the importance of future work to identify factors that may increase prediction of malignancy and may benefit from incorporation of time-to-event data.