Prediction Model for Monoclonal Gammopathy of Renal Significance Risk in Nonmalignant Monoclonal Disorders.

BACKGROUND

The aim is to identify independent risk factors for monoclonal gammopathy of renal significance (MGRS) and develop a predictive model for optimizing renal biopsy decision-making in suspected patients, thereby improving biopsy positivity rates.

METHODS

We retrospectively enrolled 291 patients from The First Affiliated Hospital of Wenzhou Medical University (January 2015 - December 2022). Clinical and laboratory indicators were collected. Independent predictors of MGRS were screened using Least Absolute Shrinkage and Selection Operator (LASSO) regression and multivariate logistic regression, followed by the construction of a nomogram model. Patients were randomly divided into training (n = 204, 70%) and validation (n = 87, 30%) cohorts. Model performance was evaluated in the independent validation set via ROC analysis, calibration curves, decision curve analysis (DCA), and Hosmer-Lemeshow (HL) test.

RESULTS

Among 291 patients, 132 (45.4%) were MGRS-positive. Five independent predictors were identified: abnormal free light chain (FLC) ratio, advanced age, abnormal white blood cell count, hypoglobulinemia, and 24-hour urinary protein > 1.5 g. The model exhibited excellent discrimination, with an AUC of 0.823 (95% CI: 0.767 - 0.879) in the training set and 0.912 (95% CI: 0.854 - 0.971) in the validation set. Calibration parameters approxi-mated ideal values in both sets (training: intercept = -0.00, slope = 1.00; validation: intercept = -0.05, slope = 1.07). HL test confirmed optimal goodness-of-fit (training: χ² = 7.395, p = 0.495; validation: χ² = 3.631, p = 0.889). DCA demonstrated significant net clinical benefit across threshold probabilities.

CONCLUSIONS

This validated MGRS predictive model (AUC > 0.9 in independent validation) shows high accuracy and clinical utility for noninvasive screening of high-risk patients and individualized renal biopsy decisions.