Pediatric Nephrology

, Volume 28, Issue 7, pp 1011–1023

Update on tenofovir toxicity in the kidney

Authors

    • UCL Centre for Nephrology, Royal Free Hospital
Review

DOI: 10.1007/s00467-012-2269-7

Cite this article as:
Hall, A.M. Pediatr Nephrol (2013) 28: 1011. doi:10.1007/s00467-012-2269-7
  • 1.4k Views

Abstract

Tenofovir (TFV) is a widely used and effective treatment for HIV infection. Numerous studies have shown that TFV exposure is associated with small but significant declines in estimated glomerular filtration rate (eGFR). However, TFV toxicity is targeted mainly at the proximal tubule (PT), and in severe cases can cause the renal Fanconi syndrome or acute kidney injury. Severe toxicity occurs in a minority of patients, but milder PT dysfunction is more common; the long-term significance of this on kidney and bone health is uncertain. Recent work suggests that changes in eGFR on TFV therapy might be explained by inhibition of PT creatinine secretion rather than actual alterations in glomerular function. Risk factors for nephrotoxicity include pre-existing kidney disease, increased age, and low body mass. Mitochondria in the PT are the targets of TFV toxicity, but the exact mechanisms remain unclear. Substantial improvement of renal function occurs in many patients with TFV toxicity upon stopping therapy, but function does not always return to baseline. In recent years, TFV usage has been extended to new clinical spheres, including pediatrics, resource-poor settings and treatment of Hepatitis B infection; theoretical reasons exist as to why some of these patients might be at higher or lower risk of TFV toxicity. Finally, strategies have been proposed to prevent TFV toxicity or enhance recovery.

Keywords

Hepatitis BHIVProximal tubuleRenal Fanconi syndromeTenofovir

Abbreviations

AKI

Acute kidney injury

ART

Anti-retroviral therapy

CKD

Chronic kidney disease

CrCl

Creatinine clearance

eGFR

Estimated glomerular filtration rate

FGF 23

Fibroblast growth factor 23

FS

Renal Fanconi syndrome

HBV

Hepatitis B virus

MDRD

Modification of Diet in Renal Disease formula

MRP

Multidrug resistance-associated protein

MtDNA

Mitochondrial DNA

NaPi-IIa

Sodium-phosphate co-transporter subtype IIa

NRTI

Nucleoside reverse transcriptase inhibitor

OAT

Organic anion transporter

PI

Protease inhibitor

PPAR

Peroxisome proliferator-activated receptor

PT

Proximal tubule

TFV

Tenofovir

WHO

World Health Organization

Introduction

Tenofovir (TFV) disoproxil fumarate is an orally bio-available pro-drug of TFV, a nucleotide analogue of adenosine monophosphate, which is structurally similar to adefovir and cidofovir, both of which are established nephrotoxins. TFV is now widely used as an effective first-line therapy for both HIV and Hepatitis B virus (HBV) infection; it has a good overall safety profile, with less metabolic side-effects that are associated with nucleoside reverse transcriptase inhibitors (NRTIs). TFV has a relatively long half-life, allowing once daily dosing and making compliance easier for patients; an important issue in diseases that require life-long therapy. Initial trials performed in HIV-infected patients did not show evidence of significant nephrotoxicity, but numerous cohort studies and case reports have since described the onset of renal dysfunction in association with TFV exposure. The renal proximal tubule (PT) is the main target of TFV toxicity; although the pathogenesis is incompletely elucidated mitochondria appear to be a major target, as they are with other established tubular toxins such as gentamicin, cisplatin, and ifosfamide [1].

Several detailed reviews of TFV-associated nephrotoxicity have been published previously [28]. Although much has been learned in the last 10 years, important questions still remain: for instance, do changes in estimated glomerular filtration rate (eGFR) observed in TFV-exposed patients reflect alterations in glomerular function or impaired tubular secretion of creatinine? Could chronic urinary phosphate wasting, due to mild sub-clinical PT toxicity, lead to a decline in bone mineral density in the long term? Can findings from studies performed on adult patients in the developed world be translated to pediatric practice and/or resource-poor settings? Will nephrotoxicity occur in HBV-infected patients to the same extent as in the HIV-infected population? If so, can it be prevented or reversed, and does an acute toxic insult lead to chronic kidney disease (CKD) even after drug withdrawal?

In this article, a concise summary will be provided of what is known about TFV toxicity in the kidney, along with some discussion of more recent developments in the field relevant to these questions. Drug toxicity is an important cause of kidney disease across many specialties, and valuable lessons for the future could be learned from past experiences with TFV.

Tenofovir toxicity in HIV-infected patients

Effect on glomerular function

An initial randomized controlled trial of TFV versus an NRTI (stavudine) showed that the former was well tolerated, and revealed no evidence of significant nephrotoxicity, as measured by serum creatinine and estimated creatinine clearance (CrCl, calculated by Cockcroft-Gault method) [9]. However, this study used patients with good baseline kidney function and without significant co-morbidities; furthermore, evaluation of renal toxicity was not the primary aim. In a subsequent pooled analysis from this and another more recent trial (TFV versus the NRTI zidovudine), a modest but significant decline in eGFR was observed in the TFV-exposed patients [10]; a finding that was also reported in several cohort studies, in which it might be argued that the participants are more representative of ‘real life’ patients [1116]. Overall, a meta-analysis that included data from 17 studies concluded that TFV exposure is associated with a mean difference in estimated CrCl of −3.9 ml/min over the course of treatment [3]. However, this meta-analysis also found a high degree of statistical heterogeneity in the published data, due to variability in parameters such as follow-up time, previous anti-retroviral therapy (ART) exposure, and concomitant usage of protease inhibitors (PIs) [3]. Since then, two further large cohort studies, one European (n = 7,378) [17] and one North American (n = 10,841) [18], have reported that TFV exposure is associated with a small but significant increase in the relative risk of developing CKD (eGFR < 60 ml/min).

Although TFV only appears to have a modest effect on average eGFR values, it is possible that sizeable changes occurring in a minority of individuals could be masked within a large study population. In a cohort study of more than 10,000 patients starting TFV with baseline creatinine <1.5 mg/dl (<130 μmol/l), increases in serum creatinine level >2 mg/dl (>177 μmol/l) were observed in 0.6 % of patients [19]; this suggests that the incidence of large alterations in glomerular function is relatively low, however, the absolute number of individuals affected may not be inconsiderable given how many patients world-wide are now taking TFV as long-term therapy. Studies using serum creatinine level and eGFR should be interpreted with caution in HIV-infected patients, as the former may be affected by alterations in muscle mass, which can change over time on ART. Furthermore, commonly used formulae for calculating eGFR, such as MDRD (Modification of Diet in Renal Disease), are inaccurate at higher values and are not well validated in HIV-infected populations; a recent large cohort study concluded that the CKD-EPI (Chronic Kidney Disease Epidemiological Collaboration) formula may be superior [20].

Effect on tubular function

Serum creatinine and eGFR are predominantly measurements of glomerular function, however, the main target of TFV nephrotoxicity is the PT; in severe cases leading to a breakdown of solute transport in this nephron segment (renal Fanconi syndrome—FS) or acute kidney injury (AKI). Animal studies have revealed that TFV can cause PT damage in mice [21], rats [22, 23], and non-human primates [24], while numerous case reports and case series have described FS or AKI in HIV-infected patients taking TFV [2532]. FS is characterized by urine wasting of solutes normally exclusively reabsorbed in the PT (i.e., phosphate, bicarbonate, glucose, low molecular weight proteins, and amino acids), of which phosphate loss is the most clinically important, as osteomalacia, bone pain, and fractures can result. The exact incidence of TFV-induced FS is unknown, and attempts at accurate estimates are hampered by under-reporting and a lack of clear diagnostic criteria, but based on the available data it is probably <1 % [19, 30]. Renal biopsy specimens from patients with TFV toxicity typically show acute tubular damage, with misshapen and swollen mitochondria in the PT on electron microscopy [30, 32].

While cases of FS and AKI are relatively infrequent in patients taking TFV, these represent the most severe end of the scale of PT toxicity; both cross-sectional [3336] and longitudinal [3739] studies have demonstrated that mild or sub-clinical PT dysfunction is much more common. The reported prevalence varies among studies, partly because of a lack of standardized definitions, but may be greater than 20 % [35]. It is currently unknown whether mild PT toxicity will lead to progressive CKD over time in these patients, but one credible concern is that chronic phosphate wasting might cause a decrease in bone mineral density (see below—“Effect on bone density”). An important conclusion from some of these studies is that PT dysfunction can occur in the absence of a significant abnormality in eGFR [35, 40]; future studies of drug toxicity in the kidney should therefore include measures of both glomerular and tubular function in order to be considered comprehensive.

Effect on tubular secretion of creatinine

Serum creatinine is widely used to calculate CrCl/eGFR, however, in addition to glomerular filtration, about 10-40 % of creatinine clearance occurs by secretion across the PT epithelium. One striking feature in several studies of TFV and kidney function is that most of the apparent decline in CrCl/eGFR seems to occur within the first 2–3 months of commencing therapy, with very little further change over time [12, 16, 41]. Therefore, given the typical magnitude and temporal pattern of CrCl/eGFR changes reported in patients taking TFV, it is plausible that they might be due to impaired PT creatinine secretion, rather than alterations in actual GFR; a phenomenon well recognized with other drugs such as cimetidine and trimethoprim. To explore this hypothesis, a recent small study of 19 HIV-infected patients, either remaining on zidovudine therapy or switching to TFV, looked in detail at changes over time in actual GFR (using isotopic measurements), calculated CrCl, and urine excretion of tubular protein (a sensitive measure of PT function) [42]. In the patients switching to TFV, mean CrCl was significantly decreased, while urine excretion of tubular protein was significantly increased after 48 weeks; however, there was no corresponding change in actual GFR, and no changes in any of the three parameters were observed in the zidovudine group. The authors therefore concluded that the decrease in CrCl in the TFV patients was more likely to be due to impaired PT creatinine secretion rather than a change in glomerular function; larger studies are now required to confirm or refute these findings. Organic anion transporters (OATs) have a role in tubular creatinine transport [43], and they are also the main uptake pathway for TFV into PT cells [44]; competitive inhibition of OAT-mediated transport thus provides a potential mechanism via which TFV could inhibit PT creatinine secretion.

Effect on bone density

Bone disease (rickets in children and osteomalacia in adults) is the main clinical sequela of FS and can occur in patients with severe TFV toxicity, due to reduced reabsorption of phosphate by the PT, leading to urinary phosphate wasting and de-mineralization of bone [30]. Activation of 25-hydroxyvitamin D to 1,25-dihydroxyvitamin D by 1α-hydroxylase takes place in mitochondria in the PT, and is likely to be impaired in cases of FS, further compounding the effects of TFV on bone health. Fractures are the most serious complication of bone disease; in a review of 164 patients with TFV-associated FS fractures were reported in 2 % [45].

In a prospective cohort study, the incidence of grade 2 hypophosphatemia (2.0–2.4 mg/dl [0.64–0.77 mmol/l]) was approximately double in TFV-exposed HIV-infected patients compared to non-exposed; however, this difference was not statistically significant, perhaps due to the relatively small number (n = 255) of patients in the cohort [46]. In a large cross-sectional study of HIV-infected patients, TFV exposure was not associated with hypophosphatemia, however, urinary excretion of phosphate was higher in patients taking TFV-containing regimes [35]. This raises the possibility that in patients with mild PT toxicity serum phosphate level might be maintained by chronic de-mineralization of bone; an important issue in HIV-infected populations where the prevalence of osteoporosis is high [47], and potentially also a concern for patients subsequently treated with bisphosphonates. In support of this hypothesis, a study of 90 patients taking TFV found evidence of impaired PT phosphate transport in 29/90 (32 %), but of these only 8/29 (28 %) had hypophosphatemia [48]. TFV usage has been strongly correlated with an increase in serum alkaline phosphatase, a marker of metabolic bone activity [49], and in a randomized controlled trial of TFV versus stavudine (an NRTI), a greater loss of bone mineral density was observed in the TFV-exposed group [9].

Mechanistically, urinary phosphate wasting in patients taking TFV could occur either due to a direct inhibitory effect of the drug on PT phosphate reabsorption, or via alterations in levels of hormones that regulate phosphate homeostasis. Vitamin D deficiency and/or secondary hypo-parathyroidism may contribute to phosphaturia in a minority of TFV-exposed patients [48]. Fibroblast growth factor 23 (FGF 23) is a phosphaturic hormone produced by osteocytes that has an important role in bone metabolism in kidney disease, and increased FGF 23 production can lead to phosphaturia and osteomalacia [50]. It has therefore been postulated that FGF 23 could play a role in the pathogenesis of urinary phosphate wasting in TFV toxicity, however, in the cases reported so far, FGF 23 levels were low or normal in individuals with TFV-induced FS [5153], implying that this is probably not the case.

Pathogenesis of tenofovir toxicity

Pharmacokinetic factors

Studies in both animals [54] and humans [55] have demonstrated that TFV-induced nephrotoxicity is dependent on the plasma drug concentration, which in turn is determined by the dose ingested, the volume of distribution (body mass), and the rate of excretion (via the kidneys); all three of these pharmacokinetic factors can therefore affect the likelihood of toxicity occurring.

It is well established that TFV toxicity is dose-dependent; administration of high, but not low, doses to macaques causes FS [24], and in clinical practice patients with TFV-induced nephrotoxicity respond to a dose adjustment. Furthermore, in a recent large cohort study of over 10,000 HIV-infected patients, after multivariable adjustment the risk of developing proteinuria and/or CKD increased with each cumulative year of exposure to TFV [18]. The volume of distribution of TFV is dependent on body mass, and low body weight has been identified as a risk factor for TFV nephrotoxicity in multivariate analyses of both cross-sectional [56] and cohort studies [19, 57].

TFV is renally excreted via a combination of glomerular filtration and tubular secretion; excretion of TFV will therefore be impaired in individuals with pre-existing CKD, and a dose adjustment should be made in those with eGFR < 50 ml/min [58]. Renal function declines with age, which may explain why increased age is also a risk factor for TFV toxicity [19, 38]. In patients established on therapy for some time and with stable renal function, TFV toxicity can subsequently occur if renal function rapidly deteriorates for another unrelated reason (e.g., sepsis, hypotension, or toxicity from another drug) before a dose adjustment can be made.

TFV is taken up into PT cells via basolateral membrane organic anion transporters (OAT1 and OAT3) [44], and subsequently exits across the apical membrane via multidrug resistance-associated protein 4 (MRP4) [59] and possibly also via MRP2 [60] (Fig. 1). It has been suggested that polymorphisms in genes encoding these transporters might affect TFV handling in the PT, and could help to explain why toxicity occurs in some individuals but not others [60]. In support of this hypothesis, a polymorphism in ABCC4, which encodes MRP4, was associated with higher plasma TFV level [61], while polymorphisms in ABCC2, which encodes MRP2, have been associated with tubular dysfunction in patients taking TFV [56, 62]. More recently, MRP7, encoded by ABCC10, has also been associated with TFV transport in the PT and risk of tubular toxicity [63]. Although potentially informative, these studies have been criticized for small participant numbers and inconsistent definitions of tubular dysfunction; it therefore remains unclear to what extent genetic variability in PT transporter function determines the incidence and severity of TFV toxicity in any given population.
https://static-content.springer.com/image/art%3A10.1007%2Fs00467-012-2269-7/MediaObjects/467_2012_2269_Fig1_HTML.gif
Fig. 1

Tenofovir toxicity in the proximal tubule. Tenofovir (TFV) enters proximal tubule (PT) cells via basolateral membrane organic anion transporters (OAT), and is transported across the apical membrane by multidrug resistance-associated protein 4 (MRP4) and possibly MRP2 as well. Phosphate is reabsorbed from the glomerular filtrate across the apical PT membrane by the sodium-phosphate co-transporter subtype IIa (NaPi). TFV toxicity in the PT can lead to the presence of damaged and dysmorphic mitochondria with loss of matrix cristae, and in some cases giant mitochondria (GM); energetic failure also impairs phosphate transport leading to urinary wasting. Probenecid inhibits TFV uptake into PT cells by OATs, reducing the intracellular concentration of the drug. Rosiglitazone binds to peroxisome proliferator-activated receptor-γ (PPAR) and causes an increase in NaPi expression, ameliorating renal phosphate wasting in a rat model of TFV toxicity

PT transporters implicated in TFV handling are involved in the excretion of drugs often given concomitantly, such as ritonavir (MRP2) and didanosine (OAT), and in many reports of TFV toxicity patients have been taking one or both of these agents (e.g. see [30]). The plasma TFV drug concentration is also increased by other PIs, such as darunavir, atazanavir, and lopinavir. In a review of the FDA Adverse Reporting System from 2001 to 2006, 164 cases of TFV-induced FS were identified; of these 74 % were receiving a ritonavir-boosted PI and 43 % were taking didanosine [45]. After adjusting for renal function, TFV clearance via the kidney was found to be 17.5 % slower in patients taking a ritonavir-boosted PI [64], and the usage of this treatment combination has been associated with a greater decline in eGFR [16, 65]. The combination of TFV with didanosine is no longer recommended.

Mitochondrial toxicity

Both HIV infection and NRTIs (especially thymidine analogues) have been associated with mitochondrial toxicity [66]; in the case of the latter, toxicity is thought to be mediated via inhibition of DNA polymerase γ (polγ), leading to depletion of mitochondrial DNA (mtDNA) levels. However, TFV is a weak inhibitor of polγ, and experiments using human PT cell cultures reported that TFV caused significantly less depletion of mtDNA in comparison to other NRTIs [67], and was less cytotoxic than its analogue cidofovir [68]. Furthermore, in animal studies, TFV exposure did not lead to a reduction in mtDNA content in the kidneys of rats or rhesus monkeys [69], however, wide variations in mtDNA content were reported among animals within the same treatment group.

In contrast to the original work described above suggesting that TFV is not toxic to mitochondria, the appearance of dysmorphic and swollen mitochondria in the PT on electron microscopy is a characteristic feature of TFV toxicity in humans [30, 32]. Furthermore, depletion of mtDNA in human kidney has been described in patients taking TFV in combination with the NRTI didanosine [70], suggesting that NRTI exposure might potentiate TFV-induced mitochondrial toxicity. In a recent comparison study of patients taking TFV or abacavir, exposure to the former drug was associated with a significant increase in urinary excretion of the mitochondrial respiratory chain enzyme cytochrome c [71]. Mitochondrial toxicity in the PT has now also been described in both mice [21] and rats [23] exposed to TFV, but the mechanism(s) remain to be elucidated. Of note, the PT is intrinsically vulnerable to aerobic insults as it lacks the capacity to perform anaerobic glycolysis.

Tenofovir usage in children, resource-poor settings, and HBV infection

Most of the data acquired to date on the effect of TFV on kidney function has come from studies performed in HIV-infected adults in the developed world. However, TFV is now increasingly being used to treat children and also patients in resource-poor settings; furthermore, it is now a first-line therapy for HBV infection. Theoretical reasons exist as to why these patient groups might be at higher or lower risk of TFV toxicity, so caution should be exercised in extrapolating data from previous studies.

Children

TFV has recently received FDA approval for usage in patients aged 12–18 years weighing more than 35 kg; the safety in patients under 12 years has not been established [58]. It has been used as salvage therapy in cases of ART resistance or in patients with unacceptable side-effects from other drugs. Given that identified risk factors for TFV-associated nephrotoxicity include low body mass and increased age, it is theoretically possible that toxicity could be either more or less frequently observed in children than adults. Accurate dosing is an important issue; historically, this was a challenge as TFV was previously only available as a once-daily adult dose of 300 mg, but a children’s formulation has now been made available by the manufacturer.

Several studies have examined the effect of TFV on kidney function in HIV-infected children (summarized in Table 1) [7279]. The outcomes of these studies were variable, with some reporting evidence of renal impairment (including decrease in eGFR, hypophosphatemia, and PT dysfunction) and others not. In general, the numbers of patients within each study were small, and there have been no randomized controlled trials comparing the effects of TFV with other forms of ART. Caution should therefore be exercised in interpreting results and larger studies are needed, ideally incorporating measures of both glomerular and tubular function. In addition to the studies listed in Table 1, isolated case reports have been published of severe PT toxicity in children taking TFV in combination with didanosine [80, 81] or ritonavir [82].
Table 1

Studies of tenofovir (TFV) and kidney function in HIV-infected children

Study reference

Number of patients

Design

Age range (years)

Proportion on TFV (%)

Exposure to TFV

Outcome

Hazra R et al., 2005

18

Phase I trial

8–16

100

48 weeks

Median decrease in eGFR of −27 ml/min per 1.73 m2. Maximal increase in serum creatinine 0.3 mg/dl (26.4 μmol/l). No cases of overt PT dysfunction.

Vigano A et al., 2007

27

Prospective cohort

5–18

100

96 weeks

No significant change in mean serum creatinine, urine protein:creatinine ratio or tubular reabsorption of phosphate

Papaleo A et al., 2007

92

Cross-sectional

<10–<20

48

15 months

TFV exposure significantly associated with increased urinary β2-microglobulin excretion.

Riordan A et al., 2009

1,253

Retrospective cohort

2–21

13

n/a

Adverse event reporting: 6 cases of renal toxicity (including 3 of PT dysfunction). Overall rate of renal serious adverse effects in those on TFV was 2.2 per 100 child years.

Judd A et al., 2010

456

Nested case–control

2–18

29

2.1 years

TFV exposure significantly associated with hypophosphatemia on multivariate analysis (OR 4.81, p = 0.01)

Soler-Palacin P et al., 2011

40

Prospective cohort

8–17

100

77 months

No change in CrCl. Significant decrease in serum phosphate and potassium, and also in tubular phosphate reabsorption. Negative correlation between serum phosphate and time on TFV

Vigano A et al., 2011

26

Prospective cohort

5–17

100

60 months

Moderate reduction in eGFR in 1 patient. No significant change in tubular phosphate reabsorption or urinary excretion of α1-microglobulin

Pontrelli G et al., 2012

49

Prospective cohort

36

63

2 years

Decline in mean eGFR and serum phosphate after 2 years in whole cohort, but no difference between TFV-exposed and non-exposed patients

CrCl creatinine clearance, eGFR estimated glomerular filtration rate, n/a data not available, OR odds ratio, PT proximal tubule

Children have a higher rate of bone turnover than adults, and might therefore be more prone to bone toxicity associated with ART. However, as with kidney function, studies on the effects of TFV exposure on bone health in HIV-infected children have generally involved small numbers of participants and have reported conflicting results. In a cohort study of 15 children commenced on TFV, increases were observed in urinary excretion of both phosphate and calcium; these changes were associated with a significant decrease in median bone mineral density, and 2/15 patients discontinued therapy due to bone loss [83]. Decreases in bone mineral density were also reported in a case series of five children receiving TFV, and in the smallest child the magnitude of the decrease was 27 % [84]. However, in contrast, two other small cohort studies (n = 16 [85] and n = 21 [86]) of children exposed to TFV reported no significant adverse effect on bone mineral density. Vitamin D deficiency is common in children with HIV infection, and this may be an additional factor in the development of bone disease; supplementation may be required in some cases.

Resource-poor settings

The largest burden of HIV infection in the world is in countries with limited resources (e.g., in Sub-Saharan Africa); the potential risks of TFV usage in these environments include the high prevalence of pre-existing CKD, malnutrition and low body weight, and a lack of facilities to perform appropriate monitoring for nephrotoxicity [87]. The prevalence of pre-existing stage 3 CKD (eGFR <60 ml/min) in HIV-infected patients commencing ART has been estimated to be around 25 % in studies performed in Tanzania [88], Uganda [89], and Zambia [90]. Renal function has been reported to improve after commencing ART (not including TFV) [89], but pre-existing CKD is a risk factor for mortality in the first 90 days of treatment [90].

Two large cohort studies have reported data on the effect of TFV exposure on kidney function in HIV-infected patients in Sub-Saharan Africa. In The DART (Development of Anti-Retroviral Therapy in Africa) study, which involved a cohort of 3,316 adults in Uganda and Zimbabwe, of whom just over two-thirds initiated ART with a TFV-containing regime, TFV exposure was associated with an increase in eGFR (1 ml/min) that was significantly lower than in patients taking abacavir (6 ml/min) or nevirapine (9 ml/min); TFV exposure was also associated with a higher incidence of CKD (defined as CrCl < 60 ml/min: 5.9 % versus 3.1 % and 2.1 %, respectively) after 4 years [91]. Eighteen deaths were reported with renal disease contribution, all in patients commenced on TFV. In another cohort study of 10,485 patients commencing ART in Zambia, 6,900 (66 %) were started on an ART regime including TFV; data on renal function was available from approximately half of the cohort, which revealed that TFV exposure was associated with a mean decrease of −14.7 ml/min in CrCl from baseline after 6 months; however, similar changes were also observed in patients taking an NRTI, implying that the effect was not specific to TFV [92]. In a smaller study of 933 patients commencing ART in Lesotho, 566 (57 %) were started on TFV, and after 12 months the median change in CrCl was +7 ml/min; however, CrCl decreased to <50 ml/min in 31/566 patients (5.5 %) [93]. Outside of Africa, a cohort study performed in Thailand of 130 patients taking TFV reported that median baseline eGFR decreased significantly by a small amount (−3 ml/min) after 3–6 months of follow-up, with 2/130 patients developing AKI [94].

Taken together, the results of these studies are not dissimilar to what has been reported from studies in the developed world; i.e., a wide variability in outcomes, but some association between TFV exposure and a modest effect on CrCl/eGFR, while severe changes in kidney function were infrequent. However, caution should be exercised when interpreting these studies, as some have been weakened by inconsistent monitoring of creatinine and inaccurate calculation of CrCl [93]. Furthermore, creatinine-based measures of kidney function might be confounded by changes in body weight on ART, and no studies to date have screened for tubular dysfunction, which, as discussed above, is a better marker of TFV toxicity in the kidney. Nevertheless, the World Health Organization (WHO) recommends TFV as a first-line therapy for HIV infection, on the grounds of simplicity of use (once-daily dosing) and treatment of HBV co-infection, and advises monitoring of CrCl every 6 months if possible [95]. Realistically, there are likely to be major logistical barriers to providing appropriate monitoring in remote and/or deprived communities; in these patients the potential risks of toxicity have to be balanced with the benefit of treating a virus that has a significant impact on life expectancy. In such scenarios, the WHO guidelines recommend that inability to perform CrCl should not a barrier to TFV use [95]. Given that TFV toxicity occurs more frequently in HIV-infected patients on concomitant PIs, combination with a non-nucleoside reverse transcriptase inhibitor may be a better strategy for first-line therapy.

Hepatitis B-infected patients

TFV is now widely used as a first-line therapy for HBV-infected patients, and in a randomized double-blind study versus adefovir, TFV was shown to have superior anti-viral efficacy with a similar overall safety profile [96]. In a longer-term follow-up of 542 TFV-exposed patients from this study, no significant deterioration in mean serum creatinine level was demonstrated after 144 weeks, with only 2/542 (0.4 %) patients experiencing a >0.5 mg/dl (>44 μmol/l) rise in creatinine, while 4/542 (0.7 %) developed a reduction in serum phosphate <2 mg/dl (<0.65 mmol/l) [97]. These data are encouraging, but must be viewed in the context that similar findings were also reported in initial studies performed in HIV-infected patients; furthermore, patients with pre-existing renal impairment were excluded from the trial, and no data was provided on proteinuria or PT function. A more recent European multi-center cohort study of 78 HBV-infected patients with a median TFV treatment duration of 76 weeks reported no significant change in the median serum creatinine concentration, with only one patient having a >1.5-fold increase from the baseline value; however, data for up to 12 months exposure was only available on 40 patients [98]. Further studies are now required, involving greater patient numbers, more detailed assessment of kidney function and bone health, and longer-term follow-up.

Although it remains to be seen whether the incidence and severity of TFV-associated PT toxicity in HBV-infected patients will be the same as in HIV-infected, there are credible theoretical reasons as to why there may be differences. For example, patient characteristics, such as gender, ethnicity, body mass and pre-existing renal impairment, might be significantly different between HBV-infected and HIV-infected populations, and could influence the baseline risk of toxicity within each. Furthermore, HIV is directly toxic to kidney tubules and could conceivably predispose individuals to developing TFV toxicity. Also, HIV-infected patients are more likely to be taking other drugs that can interact with TFV transport in the PT. One study has directly compared changes in kidney function over time in HBV-infected and HIV-infected patient cohorts, and found a greater rate of decline in median eGFR in the latter group, apparently supporting the hypothesis that they are at greater risk of TFV-associated nephrotoxicity [99]; however, the numbers of patients in the study cohorts were small (HIV: n = 120, HBV: n = 37), and larger studies will be required to confirm or refute this hypothesis.

Prevention/treatment strategies

Effective and appropriate monitoring

At a population level, the best current preventative strategy for TFV toxicity is probably to ensure appropriate dosing, according to baseline kidney function and body weight, and to monitor patients carefully with tests including markers of PT function. While some risk factors have been identified, renal toxicity can occur in the absence of these [30]. In a recent retrospective cohort study of over 10,000 HIV-infected patients in the US, TFV exposure was associated with a significantly increased risk of developing proteinuria, rapid decline in renal function or CKD (eGFR < 60 ml/min); these associations were found to be similar regardless of baseline characteristics and co-morbidities [18]. Therefore, based on the available evidence, appropriate renal monitoring is required for all patients exposed to TFV, and cannot simply be targeted at particular high-risk sub-groups.

Methods used in previous studies to measure the effect of TFV exposure on kidney function are summarized in Table 2. Urinary excretion of tubular protein is the most sensitive measure of PT dysfunction, and proteins such as retinol-binding protein [34, 40] and β2-microglobulin [33, 37, 74] have been used in studies of TFV toxicity; potential drawbacks to this approach include limited availability of these tests, increased cost burden, and uncertainty surrounding the clinical significance and prognostic value of isolated tubular proteinuria. Urine protein:creatinine ratio, fractional excretion of phosphate, and dipstick testing for glycosuria are less sensitive tests for PT dysfunction, but are all relatively cheap and widely available. It has recently been suggested that the specificity of the urine protein:creatinine ratio for detecting tubular dysfunction might be enhanced by simultaneous measurement of urine albumin:creatinine ratio (a marker of glomerular disease) [100]. Identification of urinary phosphate wasting may be important with regards to long-term bone health. Phosphate homeostasis and bone mineral density are also affected by vitamin D status, which should be monitored, and supplements given if necessary. The HIV Medicine Association of the Infectious Diseases Society of America has recommended that patients receiving TFV should have eGFR and serum phosphate measured at least every 6 months, and should also be tested for proteinuria and glycosuria [101]; however, it may be prudent to monitor patients more frequently (e.g., every 3 months) in the first year of therapy to detect any rapid changes in kidney function [7].
Table 2

Measures of kidney function used in studies of tenofovir toxicity

Measure

Comments

Serum creatinine

Dependent on muscle mass. Affected by tubular secretion. Insensitive to mild/moderate changes in PT function

eGFR (MDRD formula)

Dependent on serum creatinine. Inaccurate at higher values. Not well validated in HIV-infected patients. Does not account for changes in body weight

Serum phosphate

Urine losses of phosphate may be compensated for by bone demineralization

FE phosphate or TmP/GFR

Calculated from paired serum/urine phosphate and creatinine concentrations. Urine phosphate leak may be important for long-term bone health

Glycosuria

Quick and cheap test. Not sensitive for mild/moderate PT dysfunction. Can be positive in diabetic patients

Urine PCR

Raised in moderate/severe cases of toxicity, less sensitive for mild PT dysfunction. Specificity for tubular injury might be improved by simultaneous measurement of urine ACR

Urine ACR

Predominantly a marker of glomerular dysfunction. Not sensitive for detecting mild/moderate PT dysfunction

Tubular protein excretion (e.g., RBP)

Most sensitive marker of PT dysfunction. Long-term clinical significance of isolated tubular proteinuria unclear

ACR albumin:creatinine ratio, eGFR estimated glomerular filtration rate, FE fractional excretion, MDRD modification of diet in renal disease formula, PCR protein:creatinine ratio, PT proximal tubule, RBP retinol binding protein, TmP/GFR maximal tubular reabsorption of phosphate

Inhibition of tenofovir uptake in the proximal tubule

In patients who would benefit from commencing TFV therapy (e.g., because of viral resistance to other ART), but who have a history of previous TFV toxicity and/or are predicted to be at high risk of toxicity for other reasons, one suggested preventative strategy is to inhibit TFV uptake in the PT by co-administering an OAT inhibitor, such as probenecid (Fig. 1). Experimentally, OAT1 knock-out mice are protected from TFV toxicity [102], and probenecid has been used previously to prevent tubular toxicity in patients receiving intravenous cidofovir for CMV infection [103]. In a case series of three patients with combined HIV/HBV infection, all of whom had a previous history of reversible severe PT toxicity on TFV, re-introduction of TFV in combination with probenecid lead to virological control in all three without recurrence of nephrotoxicity, suggesting that this may be a promising strategy [104]. However, this obviously needs to be confirmed in larger studies, and although probenecid is generally well tolerated, it has been associated with side-effects, including increased frequency of gout and development of urate kidney stones, and can also affect the excretion of other widely used drugs [103, 105]. An alternative, and simpler, approach is to lower the dose of the TFV administered, monitor serum drug levels, and screen carefully for any signs of renal toxicity.

Reversal of tenofovir toxicity

The peroxisome proliferator-activated receptor-γ (PPAR-γ) is a member of the nuclear receptor superfamily of ligand-activated transcription factors, and is expressed in the PT [106]. The PPAR-γ agonist rosiglitazone induces increased expression of the sodium-phosphate co-transporter subtype IIa (NaPi-IIa) and the sodium/hydrogen exchanger 3 (NHE3) and in the kidney, both of which play key roles in solute transport in the PT. In a study of rats exposed to TFV, animals developed evidence of PT dysfunction with urinary phosphate wasting and renal tubular acidosis, and these abnormalities were mirrored by decreased expression of NaPi-IIa and NHE3 in membrane fractions prepared from renal cortices [22]. Treatment with rosiglitazone increased expression levels of NaPi-IIa and NHE3 and dramatically reversed the electrolyte and acid–base disturbances caused by TFV, suggesting that it could be a promising therapeutic strategy. However, PPAR-γ activators have pleiotropic affects, and rosiglitazone has been associated with fluid retention and increased risk of heart failure and stroke in humans [107]. It is therefore unlikely that rosiglitazone will provide a practical therapy in cases of TFV toxicity, but, as proof of concept, it appears that TFV toxicity can be pharmacologically reversed, at least in a rodent model.

Development of less nephrotoxic tenofovir analogues

A logical answer to the problem of TFV toxicity in the kidney is to develop an analogue of the drug with similar anti-viral efficacy, but that is not taken up to the same extent by PT cells. One such agent, GS-9148, has been developed by Gilead (the manufacturers of TFV) and is currently being evaluated as a treatment for HIV infection; experiments performed in vitro have demonstrated much lower efficiency of transport by OATs, while in vivo testing has shown limited accumulation in dog kidneys [108]. Another agent in development, CMX157 (Chimerix Inc), is a lipid conjugate of TFV, which has been designed to enhance uptake into target cells in order to reduce the effective concentration required in vivo (to which PT cells are exposed). This drug has been reported to have greater anti-viral activity than TFV in several different cell systems and an excellent in vitro cytotoxicity profile [109]; however, it is worth recalling that in vitro testing with TFV did not accurately predict subsequently tubular toxicity in humans.

Recovery after tenofovir toxicity

TFV-induced FS is reversible in animal models [24], and several case series of patients have described substantial improvement in kidney function upon cessation of the drug, even in patients requiring renal replacement therapy [30, 32, 110, 111]. This is perhaps not surprising, given the well-recognized ability of renal tubular cells to recover and regenerate after other severe insults such as ischemia. However, not all patients recover renal function after TFV withdrawal, and eGFR may not return to its baseline value [30, 32, 111, 112], leaving some patients with a degree of CKD, which is associated with increased cardiovascular mortality in the HIV-infected patients [113]. Recovery time post-TFV toxicity is highly variable and can take over 6 months in some patients [30].

Conclusions

Most currently used forms of ART have associated organ specific side-effects; this fact needs to be considered when objectively weighing the clinical impact of TFV-induced nephrotoxicity against the established benefits of the drug. Severe PT toxicity (FS and AKI) occurs in a minority of patients exposed to TFV, but the prevalence of milder PT dysfunction is much higher; the long-term consequences of this on kidney and bone health are unknown. The exact mechanism(s) of TFV toxicity remain unclear and warrant further study; risk factors have been identified, but toxicity can occur in the absence of these and all patients should be appropriately screened [7]. Data from studies of TFV toxicity in children are conflicting, and generally weakened by small numbers of participants. TFV is increasingly being used in resource-poor settings; the prevalence of background CKD is high in these populations, and effective monitoring of renal function will represent a huge logistical challenge. TFV is also now a first-line therapy for HBV infection and initial studies have shown no clear evidence of a deleterious effect on glomerular function (as they did not in equivalent HIV studies); it remains to be seen whether tubular toxicity will be a significant problem.

The PT is a major excretory route for xenobiotics; looking to the future, as new drugs enter the market over time it is likely that some will be associated with tubular toxicity, in which case important lessons could be learned from past experiences with TFV. Firstly, in vitro renal cell-based toxicity assays do not necessarily predict subsequent damage in vivo. Secondly, serum creatinine and eGFR are imperfect measures of kidney function and are not sensitive markers of tubular injury; future toxicity screens should incorporate tubular function tests (with consistent definitions of what constitutes tubular dysfunction). Thirdly, the incidence and severity of nephrotoxicity reported from clinical trials, which typically involve low-risk patients, may be lower than subsequently demonstrated in observational cohort studies, which are arguably more representative of ‘real-life’ patient populations. Lastly, risk factors identified for TFV toxicity (e.g., increased age, low body mass, and pre-existing CKD) are major pharmacokinetic determinants for all renally excreted compounds, and are thus likely to be applicable to other tubular toxic drugs. Successful implementation of these learning points will require closer collaboration between industry, pharmacists, nephrologists, and physicians from other specialties (e.g., HIV medicine).

Acknowledgments

Dr. Hall has received research funding from Kidney Research UK, St. Peter’s Trust for Kidney, Bladder and Prostate Research, The Academy of Medical Sciences, The British HIV Association, and The Wellcome Trust.

Disclosures

Dr. Hall previously has received a lectureship fee from Gilead Sciences, which markets tenofovir under the trade name Viread.

Copyright information

© IPNA 2012