Abstract
BACKGROUND: Severe tubular atrophy/interstitial fibrosis are critical pathological features associated with poor prognosis in IgA nephropathy (IgAN). The early identification of patients at high risk for severe tubular damage could guide clinical management and improve outcomes. OBJECTIVE: This study aimed to construct and validate a predictive model for assessing the risk of severe tubular atrophy and interstitial fibrosis in patients diagnosed with IgAN. METHODS: A total of 3276 patients from the Hankou branch of Tongji Hospital were retrospectively enrolled for model development. A predictive model for severe tubular atrophy/interstitial fibrosis was constructed based on independent predictors identified through univariate analysis, least absolute shrinkage and selection operator regression, and stepwise logistic regression. Furthermore, the model underwent internal and external validation using an independent dataset (n=1062), and performance evaluation using six machine learning algorithms: random forest, generalized linear model, decision tree, gradient boosting decision tree, extreme gradient boosting, and support vector machine. RESULTS: In this study, 8 variables were identified as independent predictors and used to construct a predictive model for severe tubular atrophy/interstitial fibrosis: Logit (P)=0.011×age (years)+0.324×hypertension history-0.302×education+.111×coefficient of variation of red cell distribution width-0.152×direct bilirubin (μmol/L)+0.003×uric acid (μmol/L)-0.021×estimated glomerular filtration rate (ml/min/1.73m²)+1.151×ln(24 h urine microalbumin) (mg/24h). The AUC for the predictive model was 0.860 (95% CI 0.847-0.873). The AUCs (95% CI) of the six machine learning algorithms ranged from 0.793 (0.765-0.822) to 0.880 (0.859-0.902) in internal validation and from 0.785 (0.756-0.814) to 0.862 (0.839-0.885) in external validation. CONCLUSIONS: We developed a concise and clinically useful model for predicting severe tubular atrophy/interstitial fibrosis in IgA nephropathy. It offers a non-invasive tool for risk assessment when biopsy is not feasible, aiding personalized treatment decisions.