Estimation of renal function by CKD-EPI versus MDRD in a cohort of HIV-infected patients: a cross-sectional analysis

Chronic kidney diseases (CKD) have become one of the leading causes of mortality among patients with HIV infection, and are often associated with comorbidities that affect kidney function, such as diabetes mellitus and hypertension [1‐4]. The reported prevalence of renal diseases ranges from 3 to 24% [4‐8], depending on the definition of renal disease, study population, and access to health services. Furthermore, although the widespread use of antiretroviral therapy (ARV) has decreased the incidence of HIV-associated nephropathy, the overall prevalence of renal disease continues to increase among people living with HIV [9].

An accurate assessment of renal function in the HIV-infected population is essential, because HIV treatment with ARV involves complex drug regimens often accompanied by significant side effects and drug interactions. A significant number of these drugs undergo renal elimination and demand dose-adjustments according to kidney function [10].

Methods based on the clearance of exogenous markers, like inulin, ⁵¹Cr-EDTA and iohexol, are the gold standard for glomerular filtration rate (GFR) measurement, but are cumbersome and infrequently used in clinical settings [11]. Numerous equations used to estimate creatinine clearance or GFR have been used over the past few decades. Two of the most commonly used equations (Cockroft-Gault [CG] and Modification of Diet in Renal Disease [MDRD]) are derived from persons with GFR ≤90 mL/min/1.73 m², have not been validated among persons with normal kidney function, and tend to underestimate higher GFRs [12]. Meanwhile, the CKD-EPI (CKD Epidemiology Collaboration) equation has been evaluated in several populations and become the most accurate GFR estimating equation [13‐15]. Recently, the CKD-EPI equation has been recommended as the first-choice method to assess renal function by the 2016 European AIDS Clinical Society (EACS) Guidelines [16].

This study aimed to analyze eGFR in a large cohort of patients with HIV infection followed-up in a single center and to evaluate the renal function estimation obtained by the CKD-EPI versus the MDRD equation, also stratifying patients according to different degrees of chronic kidney disease.

This was an observational, cross-sectional, retrospective, single center study. The study project was reviewed and approved by the Institutional Review Board (CEIC Hospital Clinic i Provincial, Barcelona, Spain, IRB# 2014/1080). Eligible patients were all adult HIV-1-infected patients (>18 years old) with at least one routine visit during the calendar year of 2014. Patients were excluded if they had not had at least two measurements of serum creatinine during the year 2014 and/or if they had had a diagnosis of end stage renal disease, dialysis and/or transplantation before the diagnosis of HIV infection. Patients were considered as having hypertension, diabetes and/or dyslipidemia if they were receiving specific treatment for these conditions.

All available serum creatinine values were obtained from the hospital electronic patient record system and were converted into eGFR, using age, gender and the CKD-EPI and MDRD equations. We did not consider the African-American coefficient factor as applicable to black patients from Africa, Europe and the Antilles [17], because this information is not available in the electronic medical records, although the prevalence of black patients in our institution, regardless their HIV-status, is very low (<3%).

Impaired renal function (eGFR <90 mL/min/1.73 m²) was confirmed with a second measurement at least 3 months apart from the first “baseline” creatinine. We used clinical guidelines from the US National Kidney Foundation’s Kidney Disease Outcome Quality Initiative (K/DOQI) to grade renal impairment [18]. In this study, eGFR below 60 mL/min/1.73 m² was used as the cutoff point for chronic kidney disease (CKD).

Of the 4622 eligible patients with follow-up in 2014, 4503 were included for analysis. Excluded patients (six patients were diagnosed with end stage renal disease before HIV-infection, and the remaining 113 patients had no laboratory data in 2014) were less likely to be smokers, had a shorter duration of HIV infection (7.7 vs. 12.3 years), were less likely to be on antiretroviral treatment (91.5% vs. 97.3%), and had higher nadir CD4 counts and lower viral loads (data not shown).

In this large sample of HIV-infected patients, there was a good correlation between the CKD-EPI and MDRD equations. The CKD-EPI equation gave somewhat higher eGFR estimates than the MDRD formula (95.2 mL/min/1.73 m² and 90.4 mL/min/1.73 m² by CKD-EPI and MDRD creatinine equations, respectively; p < 0.001), resulting in smaller but non-significant numbers of patients with eGFR <90 mL/min/1.73 m² (27.4% vs. 34%). Although the clinical relevance of the difference found should be interpreted with caution, our data are consistent with observations in the general [19, 20] and HIV-infected populations [21], suggesting that MDRD may overestimate the severity of renal impairment. Studies in the general population comparing CKD-EPI and MDRD against isotopic GFR observed less bias, improved precision and greater accuracy with CKD-EPI [22], although others have found lesser mean bias with MDRD [23].

In this analysis, at least a quarter of patients were diagnosed with mild reduced GFR disease (eGFR 60–89 mL/min/1.73 m²). This is important in detecting subclinical renal disease and identifying those patients with a higher cardiovascular risk or at risk of developing declining renal function. In practical terms, detecting minor changes in eGFR allows the early adoption of lifestyle modification and medication therapies, as appropriate, to control hypertension, dyslipidemia and hyperglycemia. In addition, nephrotoxic medications should be avoided, and medication doses (particularly when prescribing ARV) must be adjusted for renal function. For example, Stribild® (combination of emtricitabine, tenofovir disoproxil fumarate, cobicistat and elvitegravir for a complete 1-pill, once-a-day HIV-1 treatment) prescription information contraindicates the use of this drug in patients with eGFR below 70 mL/min/1.73 m² [24]. The MDRD equation misdiagnosed patients with apparent ‘normal’ renal function (over 60 mL/min/1.73 m²).

Chronic kidney disease stages III-V (eGFR <60 mL/min/1.73 m2) were found in 3% of HIV-infected patients with both equations, a prevalence similar to the EuroSIDA study and the French [25, 26] cohorts. However, this prevalence was lower than that found in a recent American analysis, in which 7.5% of HIV-infected patients had CKD [27], and even lower than a Nigerian analysis [28] that showed 52.6% of patients with GFR <60 mL/min. Possible explanations are the higher proportions of blacks and hypertensive patients on American and Nigerian cohorts, the duration of untreated HIV infection, the length of exposure to nephrotoxic antiretrovirals, as well as lifestyle-related variables and socioeconomic factors reported to impact health care coverage [29].

This study has some limitations: first, its retrospective nature; second, the absence of a direct measure of glomerular filtration rate for comparison; third, the unavailability of urine protein/creatinine ratios; and, finally the absence of an outcome measure for those patients with a reduced eGFR, which fell outside the scope of this study.

In summary, the evaluation of renal function is essential for the clinical management of patients with HIV-infection. In line with observations in the general population, this study suggests that the MDRD equation may underestimate renal function in subjects with GFRs over 60 mL/min/1.73 m². The main contribution of this report was that the CKD-EPI equation identified 24% of patients with mild reduced eGFR, not identified using the MDRD equation. This finding has strong implications for clinical practice, as it provides a framework for identifying patients at risk of poorer outcomes (overt chronic kidney disease, cardiovascular disease and mortality) and allows better clinical management, especially in terms of the dose adjustment of antiretroviral drugs.

In conclusion, the CKD-EPI equation must be considered as the first-choice method to evaluate renal function in patients with HIV-infection.