Impact of race-independent equations on estimating glomerular filtration rate for the assessment of kidney dysfunction in liver disease

Abstract

Abstract Background Altered hemodynamics in liver disease often results in overestimation of glomerular filtration rate (GFR) by creatinine-based GFR estimating (eGFR) equations. Recently, we have validated a novel eGFR equation based on serum myo-inositol, valine, and creatinine quantified by nuclear magnetic resonance spectroscopy in combination with cystatin C, age and sex (GFRNMR). We hypothesized that GFRNMR could improve chronic kidney disease (CKD) classification in the setting of liver disease. Results We conducted a retrospective multicenter study in 205 patients with chronic liver disease (CLD), comparing the performance of GFRNMR to that of validated CKD-EPI eGFR equations, including eGFRcr (based on creatinine) and eGFRcr-cys (based on both creatinine and cystatin C), using measured GFR as reference standard. GFRNMR outperformed all other equations with a low overall median bias (-1 vs. -6 to 4 ml/min/1.73 m2 for the other equations; p < 0.05) and the lowest difference in bias between reduced and preserved liver function (-3 vs. -16 to -8 ml/min/1.73 m2 for other equations). Concordant classification by CKD stage was highest for GFRNMR (59% vs. 48% to 53%) and less biased in estimating CKD severity compared to the other equations. GFRNMR P30 accuracy (83%) was higher than that of eGFRcr (75%; p = 0.019) and comparable to that of eGFRcr-cys (86%; p = 0.578). Conclusions Addition of myo-inositol and valine to creatinine and cystatin C in GFRNMR further improved GFR estimation in CLD patients and accurately stratified liver disease patients into CKD stages.

Keywords

• eGFR
• GFRNMR
• Creatinine
• Cystatin C
• Chronic liver disease