Evaluating Glomerular Filtration Rate Variability: A Clinical Medicine Approach to Early Kidney Dysfunction Detection

Abstract

Background:
Early detection of kidney dysfunction is essential for improving clinical outcomes, yet conventional single-measure estimated glomerular filtration rate (eGFR) assessments may fail to identify early physiological instability. This study evaluates the performance of major variability indices using a computational modelling framework designed to simulate realistic renal function trajectories.

Methods:
A synthetic cohort of 150 simulated subjects (75 stable renal function, 75 early dysfunction) was generated using calibrated epidemiological parameters and validated renal physiology distributions. Each subject provided eight longitudinal eGFR measurements over a two-year period. Variability metrics, including standard deviation (SD), coefficient of variation (CV), variability independent of the mean, and visit-to-visit variability, were calculated. Logistic regression assessed associations with early dysfunction, and receiver operating characteristic (ROC) analysis evaluated diagnostic performance.

Results:
Early dysfunction trajectories demonstrated significantly greater variability than stable trajectories. SD values ranged from 5.9-9.6 mL/min/1.73 m² in the dysfunction group versus 3.8-4.6 mL/min/1.73 m² in stable subjects. CV also showed clear separation (0.075 vs. 0.045). Both SD and CV significantly predicted early dysfunction (p < 0.001). CV exhibited the highest diagnostic accuracy (AUC = 0.93), outperforming SD (AUC = 0.86) and slope-based decline metrics (AUC = 0.72).

Conclusions:
GFR variability metrics, particularly CV, demonstrate strong discriminatory ability for identifying early renal dysfunction. These findings support integrating variability-based assessments into early CKD detection frameworks and highlight the need for future clinical validation.