Development and Validation of a Risk Score for the Prediction of Cardiovascular Disease in Living Donor Kidney Transplant Recipients

Background. Cardiovascular disease (CVD) is a major cause of death in kidney transplant (KT) recipients. To improve their long-term survival, it is clinically important to estimate the risk of CVD after living donor KT via adequate pre-transplant CVD screening. Methods. A derivation cohort containing 331 KT recipients underwent living donor KT at Kyushu University Hospital from January 2006 to December 2012. A prediction model was retrospectively developed and risk scores were investigated via a Cox proportional hazards regression model. The discrimination and calibration capacities of the prediction model were estimated via the c-statistic and the Hosmer-Lemeshow goodness of fit test. External validation was estimated via the same statistical methods by applying the model to a validation cohort of 300 KT recipients who underwent living donor KT at Tokyo Women's Medical University Hospital. Results. In the derivation cohort, 28 patients (8.5%) had CVD events during the observation period. Recipient age, CVD history, diabetic nephropathy, dialysis vintage, serum albumin and proteinuria at 12months after KT were significant predictors of CVD. A prediction model consisting of integer risk scores demonstrated good discrimination (c-statistic 0.88) and goodness of fit (Hosmer-Lemeshow test P=0.18). In a validation cohort, the model demonstrated moderate discrimination (c-statistic 0.77) and goodness of fit (Hosmer-Lemeshow test P=0.15), suggesting external validity. Conclusions. The above-described simple model for predicting CVD after living donor KT was accurate and useful in clinical situations.