Deep Learning Morphometric Analysis on Protocol Biopsies Predicts Future Graft Function.

INTRODUCTION: The predictive value of Banff classification in protocol transplant biopsies without specific lesions is limited. Morphometry provides precise data on microstructures, surpassing semiquantitative scores but is time-consuming. This study evaluates whether automated morphometric analysis with deep learning can predict glomerular filtration rate at 3 years using machine learning. METHODS: This retrospective study included kidney transplant recipients who underwent protocol biopsy without specific lesion. The models were trained and tested on the training/test cohort, with external validation on the application cohort. Eight deep learning algorithms extracted 23 morphometric parameters from whole-slide images (WSI). Ten machine learning models were tested for 3 years glomerular filtration rates prediction. RESULTS: A total of 367 patients were included. The means of the 3-year estimated glomerular filtration rates (eGFR) were 53 ± 23 and 53 ± 22 ml/min per 1.73 m2 in the training/test and application cohorts, respectively. In the training/test cohort, eGFR correlated negatively with interstitial fibrosis (r = -0.33; P < 0.001), tubular atrophy (r = -0.39; P < 0.001), and artery luminal stenosis (r = -0.29; P < 0.001), and positively with glomerular density (r = 0.16; P < 0.05) and glomerular epithelial (r = 0.33; P < 0.001), endothelial (r = 0.30; P < 0.001), and mesangial (r = 0.25; P = 0.002) cell densities. Kernel Ridge and Bayesian models achieved the best predictions (mean absolute error [MAE] = 11 ± 1 ml/min per 1.73 m2). External validation showed good association between predicted with Bayesian model and observed eGFR (MAE = 13 ± 11 ml/min per 1.73 m2, r = 0.68; P < 0.001). After correction with Bland-Altman bias, paired analysis showed no significant difference between predicted and observed eGFRs (P = 0.953). CONCLUSION: Integrating automated morphometric analyses into machine learning models may help predict glomerular filtration rates 3 years after protocol biopsies.