Performance of creatinine and cystatin C GFR estimating equations in an HIV-positive population on antiretrovirals

Abstract

Objective: To evaluate the performance of Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine, cystatin C, and creatinine-cystatin C estimating equations in HIV-positive patients.

Methods: We evaluated the performance of the Modification of Diet in Renal Disease (MDRD) Study and CKD-EPI creatinine 2009, CKD-EPI cystatin C 2012, and CKD-EPI creatinine-cystatin C 2012 glomerular filtration rate (GFR) estimating equations compared with GFR measured using plasma clearance of iohexol in 200 HIV-positive patients on stable antiretroviral therapy. Creatinine and cystatin C assays were standardized to certified reference materials.

Results: Of the 200 participants, median (IQR) CD4 count was 536 (421) and 61% had an undetectable HIV viral load. Mean (SD) measured GFR (mGFR) was 87 (26) mL/min per 1.73 m. All CKD-EPI equations performed better than the MDRD Study equation. All 3 CKD-EPI equations had similar bias and precision. The cystatin C equation was not more accurate than the creatinine equation. The creatinine-cystatin C equation was significantly more accurate than the cystatin C equation, and there was a trend toward greater accuracy than the creatinine equation. Accuracy was equal or better in most subgroups with the combined equation compared to either alone.

Conclusions: The CKD-EPI cystatin C equation does not seem to be more accurate than the CKD-EPI creatinine equation in patients who are HIV-positive, supporting the use of the CKD-EPI creatinine equation for routine clinical care for use in North American populations with HIV. The use of both filtration markers together as a confirmatory test for decreased estimated GFR based on creatinine in individuals who are HIV-positive requires further study.

Figure 1. Performance of estimating equations by clinical subgroups

Panel a: Bias:Median difference between measured and estimated GFR; Panel b: Accuracy: Percentage of estimates greater than 30% of measured GFR (1-P₃₀). eGFRcr, estimated GFR by the CKD-EPI equation; eGFRcys, estimated GFR by the CKD-EPI cystatin C equation; eGFRcr-cys, estimated GFR by the CKD-EPI creatinine-cystatin C equation. The number of participants in each subgroup are age ≤50 (132), > 50 (68); female (55), male (145); Blacks (104), non Blacks (96); body mass index (kg/m²) <22 (35), 22–30 (129), >30 (36); diabetes yes (16), no (184).

Figure 2. Performance of estimating equations by tenofovir, CD4 and viral load subgroups

Panel a: Bias: Median difference between measured and estimated GFR; Panel b: Accuracy: Percentage of estimates greater than 30% of measured GFR (1-P₃₀) The number of participants in each subgroup are tenofovir status yes (125), no (75); CD4 count (cells/ul) <350 (56), ≥ 350 (144); Viral load (copies/mL) <1000 or undetected (168), ≥ 1000 (14); serum albumin (mg/dl) <3.6 (43), ≥ 3.6 (157); serum C- reactive protein (mg/L) <4.8 (150), ≥ 4.8 (50).