Background: Reliable estimates of glomerular filtration rate (GFR) are important in the clinical management of HIV-positive patients. Data on the performance of widely used estimating equations (eGFR) relative to exogenously measured GFR are sparse in this population.
Methods: We evaluated cross-sectional and longitudinal accuracy and bias of eGFR, based on creatinine and cystatin C, relative to disappearance of infused iohexol from plasma (iGFR) in a cohort of participants followed annually for up to 7 years.
Results: A total of 222 HIV-positive and 139 HIV-negative participants contributed 1240 visits with valid iGFR and eGFR measures. Estimated GFR based on both creatinine and cystatin C performed the best. Estimated GFR based on creatinine alone overestimated iGFR by 9 mL·min·1.73 m on average and was significantly less accurate in HIV-positive than HIV-negative individuals. The performance of equations based on either creatinine alone or cystatin C alone were significantly affected by participant factors (eg, non-suppressed HIV RNA, nadir CD4 count, hepatitis C virus coinfection). The average iGFR slope was -4% per year in HIV-positive participants. In both HIV-positive and HIV-negative participants, eGFR slope measures were generally unbiased but inaccurate, with only 60%-74% of observations falling within ±5% points of iGFR slope.
Conclusions: Both creatinine and cystatin C have limitations as GFR indices in HIV-positive individuals. Estimated GFR based on both creatinine and cystatin C performed best in our study and may be preferred in HIV-positive persons with kidney disease or comorbidities that place them at high risk for kidney disease.