Back
The Kidney Failure Risk Equation (KFRE) was designed as a clinical tool to forecast the likelihood of reaching end-stage kidney disease (ESKD) within 2 and 5 years in patients with chronic kidney disease (CKD) stages G3a-G5. Since its reliability and performance in cases of glomerulopathies have not been rigorously assessed and confirmed, we aimed to evaluate the 4-variable KFRE usefulness in subjects with biopsy-proven glomerulopathies.
In this retrospective, single-center cohort study, 1164 patients with a histological diagnosis of primary or secondary glomerulopathy between 1st January 2008 and 31st December 2017 and an estimated glomerular filtration rate (eGFR) below 60 ml/min/1.73/m2 at the moment of kidney biopsy (KB) were enrolled. Information on outcomes were collected until 31st December 2022, with a mean follow-up time of 67.5 (95%CI 66 to 70) months. Patients with other histological findings, inadequate biopsy samples, and missing data were excluded. The outcome of interest was the initiation of renal replacement therapy (RRT – dialysis or kidney transplantation). Data from the time of KB were used for the calculation of the 4-variable KFRE (age, gender, eGFR, and urinary albumin-to-creatinine ratio - uACR). Discrimination and calibration performance of the KFRE were assessed in the whole cohort and in subgroups defined based on distinct glomerulopathy patterns, using the area under the receiver operator characteristic curve (AUC) and calibration plots.
The majority of subjects were males (56.9%), with a median age of 50 (95% CI 48 to 51) years, median eGFR of 49 (95%CI 45 to 52) ml/min/ 1.73m2 and median uACR 1877 (95% CI 1706 to 2099) mg/g. The three most common histological patterns were IgA nephropathy (19.1%), membranous nephropathy (14.9%) and diabetic nephropathy (12.1%). During the follow-up, 380 subjects (32.6%) died, 233 (20%) of them in the first two years. RRT was started in 280 subjects (24.5%) in the first 2 years and in 386 subjects (33.2%) within 5 years. The median 2-year and 5-year Kidney Failure Risk Equation (KFRE) scores were 2.9 (95% CI, 2.6 to 3.1)% and 13.3 (95% CI, 12.8 to 13.8)%, respectively. The 4-variable KFRE provided good discrimination in predicting ESKD in the whole cohort with an AUC of 0.780 (95% CI, 0.752 to 0.809) at 2 years and 0.781 (95% CI 0.753 to 0.810) at 5 years (Figure 1A). The calibration plots also showed a good fit for the entire cohort and various glomerulopathy etiologies. However, for the 2-year model, the predicted risk for kidney failure was underestimated in subjects with moderate to severe calculated risk. In contrast, the calibration plot for the 5-year KFRE displayed a close alignment between the predicted and observed risks across the whole cohort (Figure 1B).
For diabetic nephropathy, amyloidosis, and membranous nephropathy, the AUC was not statistically significant, suggesting a lack of strong discrimination (Figure 2).
The 4 variable-KFRE appears to be a good predictive tool in adults with biopsy-proven glomerulopathies both for 2 and 5-years. However, its usefulness seems limited for diabetic nephropathy, amyloidosis and membranous nephropathy.