Tenofovir disoproxil fumarate (TDF) is a nucleoside reverse-transcriptase inhibitor (NRTI) approved for the treatment of HIV infection. TDF-based regimens are widely used and represent recommended antiretroviral options for a large number of patients worldwide.1,2
TDF undergoes renal clearance by a combination of glomerular filtration and active tubular secretion.3 A decline of renal function and a proximal tubular dysfunction, including the development of a Fanconi syndrome, have been reported in patients initiating therapy with a TDF-based regimen, which are expected to develop over time.4–8 While the exact mechanism of this toxicity remains unclear, a subcellular decrease in mitochondrial DNA abundance has been recently described in mice as being involved in TDF-induced renal tubular toxicity.9 Several associated risk factors have been proposed, which include genetic predisposition (ABCC2 gene)10 and the combination with some ritonavir-boosted protease inhibitor (PI/r), expected to increase TDF plasma concentration and the risk of TDF-induced renal toxicity.4,11,12 Recently, Rodríguez-Nóvoa et al13 showed a dose-dependent effect of TDF plasma concentrations on kidney tubular dysfunction, which suggested that therapeutic drug monitoring of TDF might be useful to prevent renal toxicity and optimize TDF tolerance. On the other hand, several studies reported a greater effect of TDF on the decline in renal function.7,14
The objective of this study was to explore the correlation between TDF plasma trough concentration (Ctrough-TDF) and glomerular function in a population of patients on a combination of antiretroviral therapy (cART) containing TDF and to isolate independent variables predicting TDF-induced renal dysfunction.
This study was performed in an outdoor HIV clinical unit, where approximately 1000 patients [40% hepatitis C virus (HCV) coinfected] are followed. Data of HIV-infected patients receiving a TDF-containing cART regimen between December 2006 and January 2010 were retrospectively collected from the Nadis electronic medical record database.15
Patients with ≥1 available measurement of Ctrough-TDF performed at steady state and at least 17 hours after drug intake were included in the analysis. When several measurements were available, the highest value was retained for the analysis. In France, because of possible TDF-induced renal impairment, TDF monitoring is usually done on the occasion of renal function assessment (performed every month during the first year after TDF introduction). Information about the time of last drug intake was reported at the time of sample collection. Patients taking drugs expected to induce nephrotoxicity (eg, amphotericin B and nonsteroidal anti-inflammatory drugs) were excluded.
Main Outcome Measures
Ctrough-TDF was quantified by a validated high-performance liquid chromatography method (limit of quantification: 40 ng/mL, according to the ongoing Food and Drug Administration guidelines for bioanalytical method validation) with an expected interval ranging between 40 and 90 ng/mL according to the lower and upper limits mentioned in the summary of TDF characteristics.16
The main parameter monitored to assess renal function was glomerular function rate (GFR) assessed by creatinine clearance calculated with the Cockroft and Gault (CG)17 formula, and also by using the Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equations.18 Renal impairment was defined as GFR <60 mL/min. Renal filtration rate was assessed at the time of TDF initiation (BL-GFR), at the time of Ctrough-TDF determination (T1-GFR), and for patients still on the same cART, 12 months later (T2-GFR).
Patients were distributed into 3 subgroups according to Ctrough-TDF. Low-level (LL), normal-level (NL), and high-level (HL) groups were defined by Ctrough-TDF <40, between 40 and 90, and >90 ng/mL, respectively.
Other variables included in the analysis were age, gender, weight, body mass index, corporal surface, comorbidities (hypertension, diabetes, and hepatitis C), Centers for Disease Control and Prevention stage, time of exposure to TDF, association of a PI/r or atazanavir (boosted or not boosted by ritonavir), plasma HIV viral load, CD4 counts, and phosphoremia measured in the fasting state (normal values = 0.78–1.53 mmol/L). All those variables were collected at the time of Ctrough-TDF determination.
Median values and interquartile ranges (IQRs) (first and third quartiles) were calculated for the description of quantitative variables. The χ2 test (or Fisher exact test, when χ2 test was not applicable) was used to compare proportions. The Kruskal–Wallis and Mann–Whitney nonparametric tests were used for group comparison of quantitative variables. A multivariate analysis to identify the variables associated with GFR changes was performed by using a linear regression model, adjusting to baseline GFR values, and testing for interactions. All significance levels were set at 0.05. Statistical analyses were performed using SPSS software, version 15 (SPSS Inc, Chicago, IL).
In this retrospective study, 168 patients receiving a TDF-based regimen were identified. Five patients were excluded because of chronic renal failure or concomitant administration of other nephrotoxic drugs. The remaining 163 patients were included in the analysis. Several plasma TDF measurements, performed in 51 patients, showed that Ctrough-TDF was relatively stable over the follow-up period, indicating a low intraindividual variability. Moreover, although only one value was retained, Ctrough-TDF was always >90 ng/mL in patients from the HL group. At the time of Ctrough-TDF determination, the median time since HIV diagnosis was around 17 years (IQR: 11.3–21.1), and 13% of the patients had progressed to AIDS. The median duration of ART was 8.7 years (IQR: 4.3–11.9), and the median period of TDF-based therapy was 21.1 months (IQR: 0.3–89.8). One-quarter of the population was coinfected with HCV, whereas diabetes and hypertension were reported in approximately 14% of the patients. The median value of CD4 counts was 382 cells/mm3 (IQR: 236–586) and 68% of the patients had an undetectable plasma viral load (<40 copies/mL). The cART regimens included non-NRTI (NNRTI) in 15% of the patients; PI/r in 64%, which consisted of atazanavir/r (45%), darunavir/r (13%), fosamprenavir/r (10%), lopinavir/r (8%), or saquinavir/r (2%); nonboosted atazanavir in 8%, whereas 13% of the patients were treated with other TDF containing-regimens.
Distribution of Ctrough–Tenofovir Disoproxil Fumarate
Median Ctrough-TDF in the whole population was 92 ng/mL (IQR: 61–135) and was not influenced by gender [men, n = 99, 100 ng/mL (IQR: 72–135); women, n = 45, 106 ng/mL (IQR: 74–164); P = 0.465]. Median Ctrough-TDF was higher in patients treated with PI/r but the difference did not reach statistical significance [NNRTI-combined regimen: 85 ng/mL (IQR: 73–126); PI/r-combined regimen: 103 ng/mL (IQR: 73–149); P = 0.461]. The proportions of patients in LL, NL and HL groups were 11.7% (n = 19), 36.8% (n = 60), and 51.5% (n = 84), respectively. Median Ctrough-TDF in NL and HL groups were 70 ng/mL (IQR: 55–78) and 130 ng/mL (IQR: 106–178), respectively. Ctrough-TDF values in LL group were <40 ng/mL, which is the limit of both the expected interval and quantification. As shown in Table 1, the 3 groups were comparable with respect to gender ratio, age, body mass index, and corporal surface. Regarding common renal impairment risk factors, the prevalence of diabetes, hypertension, or HCV coinfection was not statistically different between groups. There was no association between Ctrough-TDF and the time exposure to TDF. Although the proportion of TDF-NRTI + NNRTI regimens varied among groups, the proportions of patients receiving a PI/r based regimen were similar in the 3 groups (between 63% and 74%). Centers for Disease Control and Prevention stages were significantly different but had no correlation with the Ctrough-TDF. Median viral load and CD4 cell count and the nadir of CD4 cell count were comparable in the 3 groups.
Changes in Estimated Glomerular Function Rate
At the time of TDF introduction, BL-GFR values estimated by means of the CG formula were <60 mL/min [median: 55.5 mL/min (IQR: 51.5–58.3)] in 3.7% (n = 6) of the patients. Proportion of patients with BL-GFR between 60 and 90 mL/min (median: 78.1 mL/min [IQR: 72.2–83.5]) and >90 mL/min [median: 110.9 mL/min (IQR: 100.3–123.6)] were 29% (n = 47) and 58% (n = 95), respectively. To exclude an inaccurate estimation linked to the GFR formula used, we also evaluated the GFR using MDRD or CKD-EPI formula for each patient. We observed a high variability of GFR values, especially for the lowest range (Fig. 1). The median (IQR) BL-GFR values measured with MDRD equations were 68.7 (62.4–71.9), 87.7 (76.5–96.2), and 106.8 (96.7–121.5) mL/min in LL, NL, and HL groups, respectively. Corresponding values obtained with CKD-EPI were 70.2 (62.2–76.6), 88.7 (82.9–97.8), and 107.2 (101.7–113.6) mL/min.
The evolution of renal parameters between baseline and the time of Ctrough-TDF determination based on CG formula is presented in Table 2. A significant decrease in GFR between baseline and time of Ctrough-TDF determination was observed only in the HL group, which appeared significant (P < 0.001) whatever the formula used to assess renal filtration (see Table, Supplemental Digital Content,http://links.lww.com/QAI/A377). When assessed according to gender, the delta BL-T1-GFR was more pronounced in women with a significant difference between HL and NL groups [−11.2 (IQR: −18.5 to −3.1) vs. +0.78 (IQR: −6.1 to +15.1) mL/min; P = 0.009].
Median levels of phosphoremia at the time of Ctrough-TDF were within the normal range in the 3 groups, but a significant difference was observed between NL and HL groups (Table 2). Overall, 10 patients had phosphoremia levels less than the threshold of 0.78 mmol/L, 3 in the LL group (0.51, 0.68, and 0.77 mmol/L), 1 in the NL group (0.65 mmol/L), and 6 in the HL group (0.33, 0.57, 0.72, 0.73, 0.74, and 0.74 mmol/L).
A multivariate analysis of factors associated with a decrease in GFR was performed at the time of Ctrough-TDF determination (Table 3). The final regression model regarding the relation between Ctrough-TDF and a decrease in GFR was stratified by gender. Whatever the GFR formula used, an older age and a higher level of GFR at the time of TDF initiation were significantly associated with a decrease in GFR in both sexes. However, a Ctrough-TDF >90 ng/mL was significantly associated with a decrease in GFR in women but not in men, independently of the time of exposure to TDF. The decrease in GFR was not associated with the presence of diabetes, hypertension, HCV-coinfection, or with the presence of boosted PI or atazanavir (boosted or not) in the antiretroviral regimen.
Longitudinal Assessment of GFR Changes
Data from 98 patients still on the same cART regimen after Ctrough-TDF determination (14, 33, and 51 patients in groups 1, 2, and 3, respectively) were analysed. Women accounted for around one-third of the population. Patient characteristics were similar in the 3 groups, and the median time of exposure to TDF was 38.4 months (IQR: 14.7–71.1). Similar results were observed after 12 months of follow-up (10–14 months) for both sexes, with a significantly lower T2-GFR (P < 0.001) and a higher decrease in GFR (P = 0.002) in patients with a Ctrough-TDF >90 ng/mL, whatever the formula used. Median delta BL-T2-GFR values (CG formula) in LL, NL, and HL groups were −1.69 (IQR: −8.47 to +10.9), −1.28 (IQR: −8.59 to +11.9), and −8.27 (IQR: −20.3 to −0.98) mL/min, respectively (P = 0.003). This significant difference was observed whatever the formula used. Median delta BL-T2-GFR in the HL group, estimated by MDRD and CKD-EPI formulae, were −11.8 mL/min (IQR: −25.5 to −1.45, P < 0.001 vs. other groups) and −7.40 mL/min (IQR: −16.9 to −2.15, P < 0.001 vs. other groups), respectively. Furthermore, the proportion of patients with a decrease in GFR > −10 mL/min was significantly higher in the HL group, whatever the formula used [17% in NL group (whatever the formula used) vs. 44% (CG), P = 0.017; 56% (MDRD), P < 0.001; 46% (CKD-EPI), P < 0.001 in the HL group].
This retrospective study including HIV-infected patients followed-up in a French clinical unit showed that high-plasma Ctrough-TDF (>90 ng/mL) may be expected in half of the patients receiving a TDF-containing cART. Low Ctrough-TDF values in the low-level group were all <40 ng/mL, which is our limit of quantification. A possible bias due to the nonadherence to the treatment was unlikely, as indicated by plasma Ctrough values of concomitant PI or NNRTI found to be within the normal range, and also the high percentage of patients with undetectable HIV viral load (<40 copies/mL) in this group (close to 80%).
It is noteworthy that time exposure to TDF was not different in the 3 subgroups defined by TDF concentration. Furthermore, only 5% of the population analysed exhibited GFR <60 mL/min, whatever the method used, which limits misinterpretations due to a possible impact of previous renal failure on TDF concentrations. High Ctrough-TDF was found to be associated with renal dysfunction, which is consistent with data reported by Rodríguez-Nóvoa et al13 who showed a dose-dependent effect of TDF plasma concentrations on kidney tubular dysfunction. In this study, the authors proposed a threshold of 160 ng/mL at mid dose (C12) to discriminate a risk of kidney tubular dysfunction. Our results suggest that Ctrough-TDF >90 ng/mL could predict an impact on renal filtration as determined by creatinine clearance. This significant correlation between Ctrough-TDF and the decrease in GFR is in favour of a toxic concentration-dependent effect of TDF on glomerular filtration. In our study, the proportion of patients with a decrease in GFR > −10 mL/min was significantly higher in patients with Ctrough-TDF >90 ng/mL, whatever the formula used. To date, because of the absence of a reported correlation between drug plasma concentration and efficacy, therapeutic drug monitoring is not recommended for NRTI.1,2 Yet, adverse events resulting from the exposure to plasma NRTI cannot be excluded and monitoring TDF concentration remains questionable because of the impact of the drug on kidney function.19,20 Our data are consistent with pharmacogenetic13 and clinical20 studies supporting such a monitoring to prevent renal toxicity and optimize tolerance profile, especially in patients with a high risk of kidney dysfunction.
The multivariate analysis presented in our study identified higher baseline GFR and older age as independent variables associated with a higher decrease in GFR for men and women. The impact of age was not surprising, as it is known to be associated with chronic renal failure and decreased renal function.8,12,21–23 No association with hypertension, diabetes, and HCV coinfection, known to be associated with GFR decline,22 was noted in our study. However, one can postulate that the time since diabetes or hypertension diagnosis (several years) compared with the median duration of TDF exposure (18–26 months) in our study made the assessment of the impact of these comorbidities on TDF-associated renal impairment not feasible from a statistical viewpoint.
Our longitudinal assessment showed a negative impact of high Ctrough-TDF on renal filtration, whatever the sex of the patients. However, the multivariate analysis based on data measured early after TDF introduction (T1-GFR) identified Ctrough-TDF >90 ng/mL as an independent predictive factor of renal impairment in women but not in men. These data suggest an earlier and/or greater susceptibility of women regarding renal toxicity induced by high Ctrough-TDF. In a cohort of HIV patients receiving TDF or not, Goicoechea et al12 found no association between steady state TDF plasma Cmax and Ctrough and change in GFR over time (24 and 48 weeks), which was ascribed to a poor correlation between TDF plasma levels and renal cell toxicity. However, the cohort included >80% of men and, given the influence of female gender shown in our study, one cannot exclude that an impact of TDF concentrations on renal filtration in women may have been hidden. Furthermore, the authors reported that baseline GFR was lower in women than in men, which could further limit the interpretation and reinforce the hypothesis of a masked effect in women. In a recent study performed in 2 American military clinics, female gender was shown to be significantly associated with a reduction of GFR in HIV-infected patients using TDF, which supports a possible increased susceptibility of women regarding TDF-induced renal dysfunction.24
A recent meta-analysis of prospective studies suggests that the significant loss of renal function induced by TDF use may only have a modest clinical impact.25 On the other hand, long-term retrospective studies assessing HIV patients in “real life” showed worsening renal function and higher risks of proximal tubular dysfunction in HIV patients receiving TDF, which could significantly increase the probability of treatment discontinuation.8,14 Furthermore, the effect of TDF on renal impairment is expected to be enhanced in patients with preexisting renal impairment, which is observed in 4%–17% of HIV patients,22 and when TDF is used in association with a PI/r.3,11,12 In this study, Ctrough-TDF was not significantly increased in patients receiving a PI/r regimen as already reported.12 However, >70% of the patients had received a PI-based regimen, which could have biased the comparison with patients receiving an NNRTI.
The various formulae used to estimate GFR showed similar results in terms of changes between baseline and both the time of Ctrough-TDF measurement and 12 months later, despite a relative variability of baseline values, especially in the group of patients with the lowest rates of renal filtration. To date, none of the formulae available for the estimation of GFR is validated. Although a similarity of results has been previously reported in HIV-1–infected patients when using CG or MDRD formulae,22 differences in patients with GFR values between 60 and 90 mL/min have been reported.26 Studies to validate the most accurate equation for estimating GFR in HIV-infected patients for creatinine clearance calculation remain to be performed.
The main limitation of our study is that it was a retrospective and nonrandomized study, which provides fruitful information in a “real life” setting but was not designed to explore the causality of the relation observed. One the other hand, although the results of several measurements performed in the same patients support a good reproducibility of TDF levels, only one measurement was available for most of the patients over the analysis period. Furthermore, the population was characterized by a large variability in the duration of TDF use, a possible bias that seems, however, limited by the absence of association between Ctrough TDF and the duration of exposure.
In conclusion, this study shows that plasma Ctrough-TDF above a threshold of 90 ng/mL is associated with a decrease in glomerular function and that age and female gender are independent factors associated with an increased risk of renal impairment in HIV-patients receiving a TDF-containing regimen. The significant correlation between Ctrough-TDF and the decrease in GFR highlights a toxic concentration-dependent effect of TDF on glomerular filtration and supports previous studies reporting a tubular nephropathy in patients on a TDF-based regimen. The high prevalence of patients with Ctrough-TDF >90 ng/mL reported in this study emphasizes the interest of therapeutic drug monitoring for TDF, especially in women and older patients. Because TDF-based regimens are widely used in treatment-naive patients or in case of resistance or intolerance to older NRTIs, assessing plasma Ctrough-TDF may be useful to prevent the risk of renal dysfunction. Therefore, the impact of a dose reduction of TDF according to the residual concentration has to be evaluated. While a dose reduction is currently recommended only when GFR is <50 mL/min, the choice of alternative NRTI-based regimens over dose-adjusted TDF in this setting should be discussed to make sure that antiretroviral efficacy is maintained. Further prospective studies are required to investigate the interest of a dose-adjustment strategy in patients with a higher risk of developing renal impairment.
The authors wish to thank Hervé Bismut for editorial assistance. They are grateful to the patients who trusted them and accepted to belong to their cohort.
3. Ray AS, Cihlar T, Robinson KL, et al.. Mechanism of active renal tubular efflux of tenofovir. Antimicrob Agents Chemother. 2006;50:3297–3304.
4. Zimmermann AE, Pizzoferrato T, Bedford J, et al.. Tenofovir-associated acute and chronic kidney disease: a case of multiple drug interactions. Clin Infect Dis. 2006;42:283–290.
5. Karras A, Lafaurie M, Furco A, et al.. Tenofovir-related nephrotoxicity in human immunodeficiency virus-infected patients: three cases of renal failure, Fanconi syndrome, and nephrogenic diabetes insipidus. Clin Infect Dis. 2003;36:1070–1073.
6. Rodriguez-Nóvoa S, Alvarez E, Labarga P, et al.. Renal toxicity associated with tenofovir use. Expert Opin Drug Saf. 2010;9:545–559.
7. Horberg M, Tang B, Towner W, et al.. Impact of tenofovir on renal function in HIV-infected, antiretroviral-naive patients. J Acquir Immune Defic Syndr. 2010;53:62–69.
8. Mocroft A, Kirk O, Reiss P, et al.; EuroSIDA Study Group. Estimated glomerular filtration rate, chronic kidney disease and antiretroviral drug use in HIV-positive patients. AIDS. 2010;24:1667–1678.
9. Kohler JJ, Hosseini SH. Subcellular renal proximal tubular mitochondrial toxicity with tenofovir treatment. Methods Mol Biol. 2011;755:267–277.
10. Rodríguez-Nóvoa S, Labarga P, Soriano V, et al.. Predictors of kidney tubular dysfunction in HIV-infected patients treated with tenofovir: a pharmacogenetic study. Clin Infect Dis. 2009;48:e108–e116.
11. Kearney BP, Mathias A, Mittan A, et al.. Pharmacokinetics and safety of tenofovir disoproxil fumarate on coadministration with lopinavir/ritonavir. J Acquir Immune Defic Syndr. 2006;43:278–283.
12. Goicoechea M, Liu S, Best B, et al.; California Collaborative Treatment Group 578 Team. Greater tenofovir-associated renal function decline with protease inhibitor-based versus nonnucleoside reverse-transcriptase inhibitor-based therapy. J Infect Dis. 2008;197:102–108.
13. Rodríguez-Nóvoa S, Labarga P, D'avolio A, et al.. Impairment in kidney tubular function in patients receiving tenofovir is associated with higher tenofovir plasma concentrations. AIDS. 2010;24:1064–1066.
14. Déti EK, Thiébaut R, Bonnet F, et al.; Groupe d'Epidémiologie Clinique du SIDA en Aquitaine. Prevalence and factors associated with renal impairment in HIV-infected patients, ANRS C03 Aquitaine Cohort, France. HIV Med. 2010;11:308–317.
15. Pugliese P, Cuzin L, Cabié A, et al.. A large French prospective cohort of HIV-infected patients: the Nadis Cohort. HIV Med. 2009;10:504–511.
17. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16:31–41.
18. Delanaye P, Cavalier E, Mariat C, et al.. MDRD or CKD-EPI study equations for estimating prevalence of stage 3 CKD in epidemiological studies: which difference? Is this difference relevant? BMC Nephrol. 2010;11:8.
19. Stöhr W, Reid A, Walker AS, et al.; DART Trial Team. Glomerular dysfunction and associated risk factors over 4-5 years following antiretroviral therapy initiation in Africa. Antivir Ther. 2011;16:1011–1020.
20. Menezes AM, Torelly J Jr, Real L, et al.. Prevalence and risk factors associated to chronic kidney disease in HIV-infected patients on HAART and undetectable viral load in Brazil. PLoS One. 2011;6:e26042.
21. Kirk O, Mocroft A, Reiss P, et al.; EuroSIDA Study Group. Chronic kidney disease and exposure to ART in a large cohort with long-term follow-up: the EuroSIDA Study [107LB]. Paper presented at: XVII Conference on Retrovirus and Opportunistic Infections; 2010; San Francisco, CA.
22. Mocroft A, Kirk O, Gatell J, et al.. Chronic renal failure among HIV-1-infected patients. AIDS. 2007;21:1119–1127.
23. Campbell LJ, Ibrahim F, Fisher M, et al.. Spectrum of chronic kidney disease in HIV-infected patients. HIV Med. 2009;10:329–336.
24. Crum-Cianflone N, Ganesan A, Teneza-Mora N, et al.. Prevalence and factors associated with renal dysfunction among HIV-infected patients. AIDS Patient Care STDS. 2010;24:353–360.
25. Cooper RD, Wiebe N, Smith N, et al.. Systematic review and meta-analysis: renal safety of tenofovir disoproxil fumarate in HIV-infected patients. Clin Infect Dis. 2010;51:496–505.
26. Guaraldi G, Zona S, Albertazzi V, et al.. Improved correlation of coronary atherosclerosis with estimated glomerular filtration rate assessed with chronic kidney disease epidemiology collaboration equation than with MDRD and Cockcroft-Gault in HIV-infected patients [A704]. Paper presented at: XVII Conference on Retroviruses and Opportunistic Infections; February 2010; San Francisco, CA.