Estimating the Glomerular Filtration Rate in the General Population Using Different Equations: Effects on Classification and Association

Abstract

Background/Aims: Several formulas for glomerular filtration rate (GFR) estimation, based on serum creatinine or cystatin C, have been proposed. We assessed the impact of some of these equations on estimated GFR (eGFR) and chronic kidney disease (CKD) prevalence, and on the association with cardiovascular risk factors, in a general population sample characterized by a young mean age. Methods: We studied 1,199 individuals from three Alpine villages enrolled into the MICROS study. eGFR was obtained with the 4- and 6-parameter MDRD study equations, the Virga equation, and with the three CKD-EPI formulas for creatinine, cystatin C, and the combination of creatinine and cystatin C. We assessed the concordance between quantitative eGFR levels, CKD prevalence, and in terms of association with total, LDL, and HDL cholesterol. Results: The highest and lowest eGFR levels corresponded to the cystatin C-based and MDRD-4 equations, respectively. CKD prevalence varied from 1.8% (Virga) to 5.8% (MDRD-4). The CKD-EPI based on creatinine showed the highest agreement with all other equations. Agreement between methods was higher at lower eGFR levels, older age, and in the presence of diabetes and hypertension. Creatinine-based estimates of eGFR were associated with total and low-density lipoprotein but not high-density lipoprotein cholesterol. The opposite was observed for the cystatin C-based GFR. Conclusion: GFR estimation is strongly affected by the chosen equation. Differences are more pronounced in healthy and younger individuals. To identify CKD risk factors, the choice of the equation is of secondary importance to the choice of the biomarker used in the formula. If eGFR is not calibrated to a gold standard GFR in the general population, reports about CKD prevalence should be considered with caution.

Estimation of Glomerular Filtration Rate Based on Serum Cystatin C versus Creatinine in a Uruguayan Population

“…The mean eGFR cys was 100.7 mL/min/1.73 m 2 and the eGFR epi was 95.1 mL/min/1.73 m 2 while the proportions of patients with eGFR below 60 mL/min/1.73 m 2 were 22.0% and 16.5%, respectively. Pattaro and colleagues also found similar results in a cross-sectional analysis of a population from three alpine villages [39]. Finally, in a recent meta-analysis Shlipak and colleagues noticed similar mean eGFR cys and eGFR epi estimates while the proportion of subjects under the threshold to define CKD was 13.7 and 9.7 mL/min/1.73 m 2 using cystatin and creatinine based equations, respectively [6].…”

Agreement between Chronic Kidney Disease Epidemiological Collaboration and Berlin Initiative Study equations for estimating glomerular filtration rate in older people: The Invecchiare in Chianti (Aging in Chianti Region) study

“…In the general population, CKD-EPI cre and Modification of Diet in Renal Disease equations showed an only fair agreement (κ = 0.50), similar to that obtained comparing CKD-EPI cre and CKD-EPI cre-cys (κ = 0.53). 25 Prevalence estimates of eGFR <60 in people aged 85 years or more participating in the Health and Anemia Study were 84.4% with BIS cre and 53.6 % with CKD-EPI cre , but a formal analysis of agreement between equations was not carried out in that study. 26 Finally, age, sex, weight and study setting were previously reported to represent important sources of discrepancy between equations.…”

Cystatin C– and Creatinine-Based Estimated Glomerular Filtration Rate, Vascular Disease, and Mortality in Persons With Diabetes in the U.S.

“…In the U.S. noninstitutionalized civilian population, kidney function estimated using cystatin C resulted in a reduced kidney function prevalence of 8.7% compared with an estimated 6.5% using creatinine (26). In a cross-sectional study of 1,360 inhabitants of the Alpine region in Europe, Pattaro et al (28) noted that the Lin concordance correlation coefficient of eGFR cr and eGFR cys was 0.56, with significant differences by age (0.57 in those ≥65 years old vs. 0.38 in those <65 years old) but not by diabetes status. Our results differ somewhat from those of this prior study; the presence of diabetes differentially affected kidney disease classification by eGFR cr and eGFR cys , at least in univariable analysis—an observation that may be attributable to our larger sample size and distinct American population.…”