The 3-year renal safety of a tenofovir disoproxil fumarate vs. a thymidine analogue-containing regimen in antiretroviral-naive patients
AIDS:Volume 22(16)18 October 2008p 2155-2163
Gallant, Joel Ea; Winston, Jonathan Ab; DeJesus, Edwinc; Pozniak, Anton Ld; Chen, Shan-Shane; Cheng, Andrew Ke; Enejosa, Jeffrey Ve
aJohns Hopkins University School of Medicine, Baltimore, Maryland, USA
bMount Sinai School of Medicine, New York, New York, USA
cOrlando Immunology Center, Orlando, Florida, USA
dChelsea and Westminster Hospital, London, UK
eGilead Sciences, Foster City, California, USA.
Background: Cases of renal dysfunction in patients receiving tenofovir disoproxil fumarate (TDF) have been reported. We analyzed the renal safety of TDF compared with thymidine analogue-containing (control) regimens through 144 weeks from two clinical trials in antiretroviral-naive HIV-infected patients.
Methods: We evaluated the changes in renal parameters in 1111 patients (TDF, n = 556; control, n = 555) who were enrolled in two randomized, controlled trials (Studies 903 and 934) comparing TDF vs. either stavudine or zidovudine in combination with efavirenz and either lamivudine or emtricitabine. The studies included patients with serum creatinine less than 1.5 mg/dl, serum phosphorus at least 2.2 mg/dl and estimated glomerular filtration rate by Cockcroft-Gault at least 60 and at least 50 ml/min at screening.
Results: Baseline characteristics were similar between groups.
No patient discontinued due to renal abnormalities in the TDF arm.
Through 144 weeks, the proportion of patients who experienced confirmed abnormalities in serum creatinine (>1.5 mg/dl) or serum phosphorus (<2.0 mg/dl) was less than 1% in both groups; a similar proportion of patients experienced urine proteinuria at least 100 mg/dl (TDF, 5%; control, 6%).
The median change from baseline to week 144 in glomerular filtration rate was -2 and 3 ml/min by Cockcroft-Gault, and -2 and -1 ml/min per 1.73 m2 by modification of diet in renal disease in the TDF and control groups (P < 0.05), respectively.
Conclusion: In two randomized, controlled trials, small differences in glomerular filtration rate over time were noted but no clinically relevant renal disease or adverse events were demonstrated in antiretroviral-naive patients treated with TDF through 144 weeks. Additional studies on renal health and renal safety in HIV are important goals for future clinical trials.
Tenofovir disoproxil fumarate (TDF) is a nucleotide reverse transcriptase inhibitor that has demonstrated efficacy as part of highly active antiretroviral therapy (HAART) in both treatment-naive [1,2] and treatment-experienced patients [3,4]. TDF is excreted by means of glomerular filtration and active renal tubular secretion . TDF is structurally similar to the nucleotide analogues cidofovir and adefovir that at high doses have been associated with nephrotoxicity . Renal abnormalities were observed in toxicology studies with animals given high doses of tenofovir  and multiple spontaneous cases of renal dysfunction associated with its use have been reported [7-10]. In randomized clinical trials, TDF has not been associated with increased nephrotoxicity compared with other regimens [1,2,11,12], although small decreases in estimated glomerular filtration rate (GFR) have been reported in cohort studies of TDF-treated HIV-infected patients [13-17]. The cumulative data support the renal safety of TDF, and it is recommended in combination with either emtricitabine or lamivudine as part of a preferred antiretroviral regimen for treatment-naive patients .
Single case report or small series of reports of renal disease with TDF include cases of acute renal failure, nephrogenic diabetes insipidus and proximal renal tubular dysfunction [7-10]. In some instances, renal dysfunction occurred in patients with underlying renal impairment, preexisting systemic conditions, or in patients taking nephrotoxic agents. In a case-control study from the Chelsea and Westminster cohort , the probability of developing serum creatinine elevation was found to be similar in TDF-exposed patients compared with those taking other antiretroviral regimens. In the Johns Hopkins observational clinical cohort study, patients who received TDF had greater increase in serum creatinine and decrease in GFR than a group of patients receiving alternative nucleoside reverse transcriptase inhibitors (NRTIs). Although statistically significant, this difference is difficult to interpret as the changes were small, not progressive, and were not associated with a higher rate of treatment discontinuation [14,15]. An updated analysis from this cohort suggested that long-term use of TDF was associated with modest declines in GFR that occur in the first 6 months in treatment-experienced patients and were nonprogressive. The difference between the TDF and other NRTI groups was accounted for by treatment-experienced patients, but was not observed in treatment-naive patients .
The apparent discrepancy between the results seen from prospective randomized clinical trials [1,2,11,12], in which no renal disease was detected, and from cohort studies [13-17], in which small declines in renal function have been reported, raises the possibility that renal dysfunction may occur in certain high-risk patients, particularly those with underlying renal impairment or coexisting systemic disease. We therefore analyzed pooled data from Studies 903 and 934 to evaluate the effect on renal function of long-term treatment with TDF compared with a thymidine analogue (stavudine or zidovudine) in combination with efavirenz and either lamivudine or emtricitabine. This combined analysis is specifically focused on renal safety and represents data from over 1000 patients, thereby providing greater power to detect renal events as well as small changes in renal function.
Study 903 was a Phase III, 144-week, multicenter, randomized, double-blind, active-controlled trial designed to evaluate the efficacy and safety of TDF compared with stavudine, in combination with lamivudine and efavirenz in antiretroviral-naive patients. Study 934 is an ongoing Phase III, multicenter, randomized, open-label trial designed to evaluate the efficacy and safety of TDF and emtricitabine compared with fixed-dose zidovudine/lamivudine, in combination with efavirenz in antiretroviral-naive patients. Complete details on the study design and eligibility criteria for both studies have been reported previously [1,2]. Patients were included if they had a baseline serum creatinine less than 1.5 mg/dl, serum phosphorus at least 2.2 mg/dl and calculated creatinine clearance using Cockcroft-Gault equation  at least 60 ml/min (for Study 903) and at least 50 ml/min (for Study 934).
The analysis reported here focuses on combined data through 144 weeks from patients who were enrolled in Studies 903 and 934. Patients randomized to the TDF-containing regimens formed the TDF group, whereas patients randomized to stavudine in Study 903 or zidovudine/lamivudine in Study 934 comprised the thymidine-analogue (control) group. Changes from baseline through week 144 in serum creatinine, serum phosphorus and urine protein were obtained. The proportions of patients developing a maximum graded toxicity in serum creatinine, serum phosphorus and proteinuria through 144 weeks were also obtained. Using the patient weights collected at each study visit, we calculated and summarized the median change in estimated GFR from baseline among patients in both the TDF and control arms by both the Cockcroft-Gault equation  and the modification of diet in renal disease (MDRD) formula [20,21]. An analysis of individual data using a repeated measures data analysis was conducted to examine the trend in GFR in each individual over time using proc mixed procedure (SAS Institute Inc., Cary, North Carolina, USA) adjusting for treatment, time on treatment, age, sex, race, hypertension, diabetes mellitus, baseline proteinuria and BMI. Individual plots of GFR by Cockcroft-Gault and MDRD for patients whose baseline GFR was in the lowest quartile (25th percentile) of the overall population were created. In addition, descriptive analyses of changes in estimated GFR from subgroups of black patients, patients who had mild renal impairment at baseline, and patients who were taking concomitant antihypertensive or antidiabetic medications during the 144-week study period were also performed. Mild renal impairment was defined as an estimated GFR of 50-80 ml/min at baseline using the Cockcroft-Gault equation for creatinine clearance. The concomitant antihypertensive and antidiabetic medications received during the study were angiotensin-converting enzyme inhibitors, angiotensin II receptor antagonists, beta blockers, diuretics, calcium-channel blockers, alpha adrenergic agonists, other antihypertensives, sulfonylureas, metformin, insulin and thiazolidinediones.
Baseline demographic and HIV disease characteristics were summarized using medians and interquartile ranges (IQRs). Wilcoxon Rank Sum test was used for comparison of the changes from baseline to week 144 in serum creatinine, phosphorus and estimated GFR between treatment groups. Repeated measures analysis was used for change from baseline in estimated GFR in each treatment group. All statistical analyses were performed using SAS version 8.2 statistical software (SAS Institute Inc., Cary, North Carolina, USA).
TDF has become one of the most frequently used antiretroviral agents for the treatment of HIV. In several clinical trials, the incidence of nephrotoxicity with TDF was not different from the non-TDF control [1,2,11,12]. Likewise, several cohort and case-controlled studies have shown low absolute rates of renal dysfunction or no differences in significant renal adverse events when TDF was compared with other antiretroviral regimens [13,22,23]. In the Hopkins cohort, more decreases in GFR were seen among patients on a TDF-containing regimen compared with alternative NRTIs. However, the decrease in GFR was modest, did not result in discontinuation of TDF, and the difference was accounted for entirely by treatment-experienced patients [14-16]. A similar effect of TDF on GFR was reported in an analysis of the HOPS cohort . Although it is reassuring that the overall risk for renal toxicity remains extremely low in published reports irrespective of study design, there are differences in renal outcomes when data from randomized controlled trials are compared with cohort studies. For this report, we specifically designed our analysis to improve the power and precision of assessing renal safety in two similarly designed randomized controlled trials that were originally powered and performed to test for efficacy.
Data from Study 934 comparing emtricitabine/TDF with zidovudine/lamivudine, both in combination with efavirenz, are available through 144 weeks of treatment. After 144 weeks, GFR decreased in both groups, although the decline in GFR was slightly greater in the emtricitabine/TDF group. The declines in GFR in the emtricitabine/TDF group were observed in the first 96 weeks, with minimal change from weeks 96 to 144 . Study 903, comparing TDF with stavudine, both in combination with efavirenz and lamivudine, showed small increases from baseline in GFR at 144 weeks in both groups, although the increase was slightly more in the stavudine group . By pooling data from these two studies, we can better track renal function in over 500 patients receiving TDF in these prospective clinical trials.
The combined analysis of Studies 903 and 934 presented here indicates that a low incidence of renal abnormalities is seen with long-term treatment with TDF. Through 144 weeks, no clinically relevant changes in serum creatinine and phosphorus were seen with both the TDF and control groups. The incidence of serum creatinine elevation or hypophosphatemia was less than 1% in both treatment groups. Likewise, a similar incidence of proteinuria (5-6%) was seen in both treatment groups. Small but statistically significant decreases in estimated GFR were observed in the TDF group through 144 weeks. A significant increase in GFR was seen in the control group using both formulae. The findings seen here in the TDF group are similar to the results observed from cohort studies in that declines in GFR seen with TDF were nonprogressive and did not result in renal adverse events or discontinuation of TDF treatment [13-17,22,23]. It is important to note that the small increases in GFR seen in the control group in this analysis is different from the progressive decreases in GFR over time reported in patients treated with non-TDF-containing regimens [14-17,22-24]. These small differences in GFR (either increases or decreases) are difficult to interpret, especially in the light of the relative inaccuracy of creatinine-based estimates when GFR is more than 60 ml/min .
Among patients whose baseline GFR was in the lowest quartile of the population, individual plots of GFR indicate that their renal function remained stable through 144 weeks in both the TDF and control groups. Concerns have been raised that nephrotoxicity may be more problematic in black patients and in patients with mild renal impairment or risk factors for renal disease such as hypertension or diabetes mellitus. Similar to the overall population, small decreases from baseline in estimated GFR were observed through 144 weeks among patients from these subpopulations who received TDF. Among black patients who enrolled in the two studies, no significant difference in the change from baseline in estimated GFR was observed between treatment groups through 144 weeks. However, among patients with mild renal impairment at baseline or taking medications for hypertension or diabetes mellitus, a statistically significant difference was seen in the changes in estimated GFR using the MDRD formula but not the Cockcroft-Gault equation. Again, these differences underscore the limitations of creatinine-based estimates in calculating GFR values in the normal range. In addition, patients with hypertension were mostly receiving angiotensin-converting enzyme inhibitors or angiotensin receptor II antagonists, which may further affect their renal function. In all three subpopulations, no patient discontinued TDF due to renal abnormalities.
The use of the Cockcroft-Gault and MDRD equations to estimate GFR has inherent limitations, as both formulae rely on serum creatinine, and neither has been validated in HIV-infected patients. In particular, the MDRD formula has only been validated in patients with chronic kidney disease [20,21]. In a study  evaluating GFR in HIV-infected patients, both formulae appeared to underestimate renal function when compared with creatinine clearance measured using 24-h urine collection. However, another recent study  comparing various GFR estimations with 125Iodine-iothalamate clearance in HIV-infected patients receiving HAART demonstrated that the Cockcroft-Gault equation and 24-h urine-based GFR measurement were reasonably accurate when compared with GFR measured by 125Iodine-iothalamate clearance. Despite their shortcomings, both formulae are used commonly in clinical practice to predict GFR primarily because measurement of creatinine clearance using 24-h urine collection is cumbersome, time consuming, and unreliable in an outpatient setting.
Data from the TDF-expanded access programme found a low incidence of increased serum creatinine (<1%) through approximately 3700 person-years of exposure to TDF in treatment-experienced patients. In a multivariate analysis , the baseline factors associated with increased serum creatinine were elevated baseline serum creatinine, concomitant use of nephrotoxic agents, increasing age, lower weight and lower CD4 cell counts. Likewise, data from the DART trial , a prospective randomized clinical trial conducted in Africa, demonstrated that the incidence of severe renal impairment (GFR < 30 ml/min per 1.73 m2) through 96 weeks was low (1.5%) and similar between TDF-containing and non-TDF-containing regimens. The DART trial enrolled 3314 symptomatic antiretroviral-naive HIV-infected adults initiating therapy with fixed-dose zidovudine/lamivudine in combination with either TDF (74% of patients), nevirapine (17%) or abacavir (9%). Patients from the study were mostly women (65%) and had advanced HIV disease (median CD4 cell count 86 cells/_l at baseline). In the DART study , patients with baseline renal impairment experienced greater improvements in GFR than those without impairment, suggesting that renal benefits outweigh risks in this population of highly immunocompromised African patients. Similarly, data from Thai patients on HAART who enrolled in the Staccato study demonstrated no significant changes in GFR through a median of 21 weeks of treatment with TDF-containing regimens .
This analysis has several important limitations. First, the population of HIV-infected patients initiating antiretroviral therapy in these studies was predominantly men, young (median age of 36 years) and relatively healthy [1,2], whereas HIV-infected patients in the clinics are increasingly older and have other medical comorbidities that may affect renal function (i.e., hypertension, diabetes mellitus, and hyperlipidemia). Second, the inclusion criteria for these studies selected for patients with adequate renal function at baseline, with 1% of patients having a GFR of less than 60 ml/min prior to starting antiretroviral therapy. In contrast, data from recent studies [31,32] suggest that chronic kidney disease is a common complication of HIV infection, with 5-15% of patients having a GFR of less than 60 ml/min per 1.73 m2. Finally, the results from this analysis focus solely on patients initiating an efavirenz-based regimen. Although some cohort studies have shown greater declines in renal function with TDF-containing, protease inhibitor-based regimens , several prospective, randomized clinical trials evaluating the efficacy and safety of ritonavir-boosted protease inhibitors in combination with TDF have not demonstrated significant renal abnormalities or declines in renal function [33-35].
The results from this analysis demonstrate small decreases in GFR over time in antiretroviral-naive patients with adequate renal function at baseline who are treated with TDF. This is consistent with data reported from cohort studies. Although any decline in GFR could raise concerns, the estimating equations have demonstrated inaccuracies when GFR is in the normal range. The low incidence of serum creatinine elevation, hypophosphatemia or proteinuria seen in this analysis is also consistent with cohort data.
Although this analysis clearly demonstrated the low incidence of clinically significant renal abnormalities through 3 years of antiretroviral therapy with a TDF-containing regimen, the findings presented here underscore the importance of evaluating renal function prior to and during antiretroviral therapy, particularly in high-risk populations, including elderly patients, patients with medical comorbidities, and patients taking multiple concomitant medications. Consideration of therapeutic alternatives or appropriate dosing interval adjustment is recommended for patients on TDF who develop impaired kidney function. Moreover, as HIV-infected patients live longer, ageing becomes an important risk factor for renal impairment. It is therefore important to encourage long-term prospective studies to evaluate kidney function in HIV-infected patients, including patients taking TDF, and to conduct research on more accurate measures of GFR in patients with normal kidney function.
A total of 1111 patients enrolled in Studies 903 and 934, with 556 patients initiating TDF and 555 patients initiating a thymidine analogue (stavudine or zidovudine) in combination with efavirenz and either lamivudine or emtricitabine. Baseline demographic and HIV disease characteristics are shown in Table 1. There were no significant differences in age, sex and race at baseline between the two treatment arms. Likewise, the baseline plasma HIV-1 RNA levels and CD4 cell counts were similar between treatment arms.
A summary of changes from baseline in serum creatinine and estimated GFR is presented in Table 2. Baseline serum creatinine and estimated GFR were similar between the TDF and control groups. There was little to no change in median serum creatinine from baseline to week 144 in either treatment group. A summary of median GFR by study visit is presented in Fig. 1. The median change from baseline to week 144 in estimated GFR was -2 ml/min and -2 ml/min per 1.73 m2 in the TDF group by Cockcroft-Gault and MDRD, respectively. In the control group, the median change was +3 ml/min and -1 ml/min per 1.73 m2 by Cockcroft-Gault and MDRD, respectively. The difference in the change from baseline in GFR between treatment groups was statistically significant (P < 0.05). Individual plots of GFR for patients whose baseline GFR was in the lowest quartile (25th percentile) of the overall population are presented in Fig. 2a-d.
Using individual data on estimated GFR by repeated measures analysis of both Cockcroft-Gault and MDRD, similar changes in GFR (Table 2) were noted. After adjusting for treatment group, time on treatment, age, sex, race, BMI, hypertension, diabetes and baseline proteinuria, the mean change in estimated GFR over time using repeated measures data analysis was -2 ml/min by Cockcroft-Gault (P = 0.20 for change from baseline) and -6 ml/min per 1.73 m2 by MDRD (P < 0.05) by week 144 in the TDF group. In the control group, the change was +5 ml/min and +5 ml/min per 1.73 m2 by Cockcroft-Gault and MDRD, respectively (P < 0.05). The difference in the change in GFR over time between treatment groups was statistically significant (P < 0.05).
Black patients comprised 21% of the population enrolled in both studies; 130 received TDF and 104 received a thymidine analogue. Among the 234 black patients, the median baseline GFR by Cockcroft-Gault was 117 ml/min for both treatment groups (P = 0.30), whereas the median baseline GFR by MDRD was lower in the TDF group compared with the control group (116 vs. 124 ml/min per 1.73 m2, P = 0.02). A summary of median GFR by study visit in this subpopulation is presented in Fig. 3a. The median change from baseline to week 144 in estimated GFR by Cockcroft-Gault was -2 and 0 ml/min in the TDF and control groups, respectively (P = 0.81), and was -2 ml/min per 1.73 m2 in both treatment groups by MDRD (P = 0.09).
Among 136 patients taking concomitant antihypertensive or antidiabetic medications during the 144-week study duration, 63 initiated TDF and 73 initiated a thymidine analogue. Baseline median GFR by Cockcroft-Gault (118 ml/min in the TDF group and 115 ml/min in the control group, P = 0.75) and MDRD (104 ml/min per 1.73 m2 in the TDF group and 99 ml/min per 1.73 m2 in the control group, P = 0.50) were similar between treatment groups. A summary of median GFR by study visit in this subpopulation is presented in Fig. 3b. No significant difference in the median change from baseline to week 144 in GFR was seen using the Cockcroft-Gault equation (-4 ml/min in the TDF group and +5 ml/min in the control group, P = 0.20). However, a statistically significant difference was observed in the change from baseline in GFR by MDRD (-5 ml/min per 1.73 m2 in the TDF group and +8 ml/min per 1.73 m2 in the control group, P < 0.05).
A total of 57 patients were identified as having mild renal impairment at baseline, defined as an estimated GFR of 50-80 ml/min at baseline using the Cockcroft-Gault equation. Of these patients, 28 received TDF and 29 received a thymidine analogue. Baseline median GFR by Cockcroft-Gault (72 ml/min in the TDF group and 73 ml/min in the control group, P = 0.89) and MDRD (78 ml/min per 1.73 m2 in the TDF group and 74 ml/min per 1.73 m2 in the control group, P = 0.29) were similar between treatment groups. A summary of median GFR by study visit in this subpopulation is presented in Fig. 3c. Through week 144, no significant difference in the median change from baseline in GFR by Cockcroft-Gault was observed (+1 ml/min in the TDF group and +8 ml/min in the control group, P = 0.07), whereas a statistically significant difference was observed in the change from baseline in GFR by MDRD (-1 ml/min per 1.73 m2 in the TDF group and 12 ml/min per 1.73 m2 in the control group, P < 0.05).
A summary of change from baseline in serum phosphorus in the overall population is presented in Table 2. No statistically significant difference in the change from baseline to week 144 in serum phosphorus was observed between treatment groups (-0.1 mg/dl in the TDF group and 0 mg/dl in the control group, P = 0.08). The incidences of laboratory toxicities in serum creatinine, phosphorus and urine protein are shown in Table 2. A low incidence (<1%) of confirmed abnormalities in serum creatinine (>1.5 mg/dl) and phosphorus (<2.2 mg/dl) was seen in both treatment groups. No confirmed grade 3 or 4 abnormalities in serum phosphorus (<1.0 mg/dl) were reported through 144 weeks in either group. No confirmed grade 3 or 4 abnormalities in serum creatinine (>1.5 mg/dl) were reported through 144 weeks in the TDF group, whereas one patient in the control group reported a confirmed grade 3 elevation in serum creatinine (>3 mg/dl). A similar incidence of proteinuria (≥100 mg/dl) was observed in both treatment groups (TDF, 5%; control, 6%).
In the overall population, 28 and 37% of patients in the TDF and control groups, respectively, discontinued study regimen by week 144. No patient in the TDF group discontinued due to renal adverse events, whereas one patient in the control group discontinued due to acute renal failure.