Abstract
The estimated glomerular filtration rate (eGFR) at 1 year post-transplantation is a well-established predictor of long-term graft survival; however, its predictive accuracy needs improvement. We retrospectively analyzed data from 51 kidney transplant recipients at Kanazawa Medical University Hospital (January 2001-February 2015). Cox regression was used to identify risk factors for death-censored graft loss and create a nomogram to predict graft survival at 15 years post-transplantation. The predictive factors ultimately included in the nomogram included eGFR and serum glucose-regulated protein 78 (GRP78) at 1 year post-transplantation. In terms of discrimination, assessed by area under the receiver operating characteristic curve (AUC-ROC), no significant difference was noted between the eGFR model (AUC 0.84 [0.67-1.00]) and nomogram (AUC 0.92 [0.82-1.00]) (p = 0.38). However, calibration, evaluated by the calibration plot, indicated superiority of the nomogram over the eGFR model, confirmed in the internal validation cohort using the Bootstrap method. Regarding clinical value evaluated by decision curve analysis, the nomogram showed a greater net benefit than the eGFR model, especially at wider diagnostic thresholds (particularly important lower thresholds). Our findings suggest the added predictive value of serum GRP78 at 1 year post-transplantation for long-term graft survival prediction.