HOW TO DECIDE WHETHER CREATININE OR CYSTATIN C IS MORE ACCURATE FOR KIDNEY FUNCTION ASSESSMENT IN OLDER ADULTS

Abstract BACKGROUND AND AIMS The gold standard for evaluation of kidney function is measured GFR [mGFR] which is challenging to perform in clinical practice. Serum creatinine and cystatin C are used to estimate GFR clinically, but neither is a perfect marker. Within any given individual, the estimated glomerular filtration rate (eGFR) can be widely different when using serum creatinine [eGFRCr] vs. cystatin C [eGFRCys]. We investigated which was closest to mGFR in elderly outpatients when results between these two eGFRs were discrepant. METHOD About 657 older ambulatory participants of the population-based Berlin Initiative Study (BIS) had GFR measured by iohexol plasma clearance (mGFR), as well as serum creatinine and cystatin C levels. eGFRCr and eGFRCys were calculated using the 2009 and 2012 CKD-EPI equations, respectively. The cohort was divided into two groups based on which of these eGFR estimates was lowest. Bias (defined as ‘eGFR-mGFR’), imprecision (standard deviation [interquartile range] of bias), as well as proportion within 30% (P30) were calculated. RESULTS Mean (±SD) age was 78 (±6) years, eGFRCys was 59 (±23), mean eGFRCr was 64 (±20) and mGFR was 55 (±20) mL/min/1.73 m2. Mean (±SD) body mass index was 28 (4) kg/m2 58% were men, and 26% had diabetes mellitus. Two-thirds of participants (448/657 = 68%) had eGFRCys < eGFRCr. Regardless of which group participants were in, the lower of the two eGFR estimates was the one less biased compared to mGFR. Among those in whom eGFRCr ≤ eGFRCys, 80% (=168/209) had eGFRCr closer to mGFR than eGFRCys; among those in whom eGFRCys < eGFRCr, 74% (=333/448) had eGFRCys closer to mGFR than eGFRCr (Figure). P30 was 86% for eGFRCr when it was the lower estimate, and was 90% for eGFRCys when it was the lower estimate (Table). CONCLUSION In a population of community-dwelling older adults, the lower value between eGFRCys and eGFRCr was closest to mGFR in 76% of participants.