Article Commentary: Correcting eGFR for the Effects of ART on Tubular Creatinine Secretion: Does One Size Fit All?

HIV is a risk factor for chronic kidney disease (CKD), which is an important comorbidity in an ageing population [1]. Some antiretroviral therapy (ART), including tenofovir disoproxil fumarate (TDF), atazanavir and lopinavir, have been associated with an increased incidence of CKD [2]. As discontinuation of these agents is associated with recovery of kidney function [3–5], regular monitoring of kidney function to detect progressive decline in estimated glomerular filtration rate (eGFR) is advocated in guidelines, allowing for timely switches to more ‘kidney friendly’ regimens. ART not associated with renal injury (for example, dolutegravir [DTG], raltegravir [RAL] and rilpivirine [RPV]) or with an improved renal safety profile (for example, tenofovir alafenamide) is particularly attractive for older people with HIV and others at risk of CKD.

However, these options, along with many other antiretrovirals in current use (including the protease inhibitor [PI] boosting agents ritonavir and cobicistat) inhibit creatinine secretion at the proximal tubule [6] leading to variable, benign increases in serum creatinine. This occurs typically within 2–4 weeks of first exposure and is stable thereafter and not accompanied by (worsening) proteinuria, haematuria or glycosuria and, crucially, does not imply nephrotoxicity. However, these increases may translate into substantial reductions in eGFR, which clinicians use to monitor kidney function, and in creatinine clearance, which many pharmacists use to dose renally cleared medications, and potential misclassification of people as having CKD. Altogether, this may lead to anxiety among clinicians and patients, ART changes, additional and perhaps unnecessary monitoring or onward referral to nephrologists, and/or inappropriate dose reduction of antiretroviral or other medications.

On average, these reductions are less than 10 ml/min/1.73 m² for RPV, doravirine, RAL and ritonavir-boosted PI, 10–15 ml/min/1.73 m² for bictegravir and cobicistat-boosted PI, and more than 15 ml/min/1.73 m² for DTG [6,7]. A subset of recipients of these agents may experience substantially greater reductions in eGFR despite other aspects of kidney function (for example, glomerular filtration, tubular reabsorption and urinary concentration) being unaffected. An easily applicable correction factor, particularly for the worst offending agents, would thus be welcome.

In the current issue of Antiviral Therapy, Brunet et al. [8] examined one such approach in the Observational Pharmaco-Epidemiology Research and Analysis (OPERA) cohort: the addition of the median decrease reported in the literature to the calculated eGFR (ml/min/1.73 m²) for individuals on DTG (+17), RAL (+9), RPV (+9) or cobicistat-boosted regimens (+13). Study participants were young (median age 32 years), ART-naive with a median CD4⁺ T-cell count of 369 cells/mm³ and a median HIV RNA of 4.7 log copies/ml, and preserved kidney function (median eGFR 113 ml/min/1.73 m²). The study confirmed fewer cases of incident CKD (eGFR <60 ml/min/1.73 m²) with the correction factor applied, particularly for DTG and cobicistat-boosted regimens.

The authors compared the risk of developing CKD with these agents to efavirenz, an antiretroviral that does not affect tubular creatinine secretion. Using Cox proportional hazards models adjusted for age, gender, ethnicity, diabetes mellitus, hypertension, HIV RNA, hepatitis C coinfection, TDF and co-trimoxazole co-administration, without the correction factor, DTG, RAL, RPV and cobicistat were each associated with an increased risk of CKD. However, when the correction factor was applied, all four regimens were associated with significantly reduced risk of CKD as compared with efavirenz.

These results are interesting but also somewhat surprising as efavirenz has not been associated with nephrotoxicity, either directly or via an increase in tenofovir exposures when co-administered with TDF [6], perhaps suggesting that the applied correction factors may in fact have resulted in over-corrected eGFR estimates. Several factors may have contributed to this apparent over-correction. Baseline eGFR is a strong predictor of incident CKD, with those who have sub-normal eGFR prior to ART initiation most likely to progress to CKD [2]. The effect of inhibition of tubular creatinine secretion in the first instance is on serum creatinine concentrations; a rise in serum creatinine in individuals with normal renal function will lead to a greater reduction in eGFR as compared with individuals with impaired renal function. Hence, applying the same correction factor irrespective of baseline eGFR may result in overcorrection in those most at risk of incident CKD.

Furthermore, there is considerable inter-individual variability in the magnitude of serum creatinine increase with these agents. For example, in the ADVANCE trial, serum creatinine in the TDF/emtricitabine/DTG arm increased on average by 13 (from 65 to 78) μmol/l, which translates for the average male study participant in a reduction in eGFR of 30 (from 159 to 129) ml/min/1.73 m². However, those in the upper quartile experienced a creatinine increase of at least 53 μmol/l, which translates into a reduction in eGFR of at least 79 ml/min/1.73 m² [9]. By contrast, an increase in serum creatinine of 13 and 53 μmol/l in a 52-year-old White male with a baseline creatinine of 99 will result in a much smaller reduction in eGFR (10 and 28 ml/min/1.73 m², respectively).

However, there are few data on the true magnitude of creatinine inhibition in those with existing CKD, particularly in those with eGFR <30 ml/min/1.73 m² or renal transplants for whom small decreases in eGFR may be clinically significant. For example, the magnitude of creatinine increase with trimethoprim, which also reversibly inhibits creatinine secretion, was found to be greater in those with already elevated serum creatinine [10]. Therefore, while the application of a single, standard eGFR correction for a drug like DTG is clearly pragmatic and useful for epidemiological research, it is unlikely to be helpful to assess kidney function in people with HIV in routine clinical practice.

How might clinicians manage reductions in eGFR following the introduction of antiretrovirals that inhibit tubular creatinine secretion? In many individuals, the changes are small and the eGFR remains within the normal range (>90 ml/min/1.73 m²). These, and even large reductions in eGFR that are temporarily associated with the introduction of these agents per se should not be a cause for concern, although it may be important to confirm that kidney function remains stable 4–6 weeks after this initial decline in some individuals. A normal urine protein/creatinine ratio, absence of haematuria and normoglycaemic glycosuria, and unaltered serum urea levels would provide additional reassurance that no glomerular or tubular injury has occurred [11]. By contrast, in those whose eGFR continues to worsen at a rate of more than 3–5 ml/min/1.73 m²/year, a full medical review is warranted and tenofovir-nephrotoxicity should be excluded in those on TDF-containing ART regimens [12]. Other commonly used medications which reversibly inhibit creatinine secretion, such as trimethoprim and ranitidine, are best avoided in CKD to avoid confusion that may arise as a result of worsening eGFR.

In individuals on drugs that inhibit tubular creatinine secretion whose eGFR reaches a threshold for clinical decision making such as dose reduction of renally eliminated medications, obtaining a further measure of renal function such as cystatin C should be considered. Cystatin C is exclusively cleared by glomerular filtration and thus unaffected by such drugs; cystatin C measurements can also be converted into eGFR using the CKD-EPI equation [13]. A similar creatinine and cystatin-C-based eGFR provides confidence that the estimates are likely to accurately refect GFR, while a substantially higher cystatin-C-based eGFR estimate suggests that the creatinine-based eGFR is likely to be an underestimate [14–16]. However, as cystatin C levels are affected by immune activation and inflammation [17], these comparisons should not be performed in those with inter-current illness of uncontrolled HIV replication.