Minor changes in calculated creatinine clearance and anion-gap are associated with tenofovir disoproxil fumarate-containing highly active antiretroviral therapy

Authors

  • A Winston,

    Corresponding author
    1. National Centre in HIV Epidemiology and Clinical Research, University of New South Wales, Sydney, Australia, and
    2. HIV, Immunology and Infectious Diseases Clinical Services Unit, St Vincent's Hospital, Sydney, Australia
    Search for more papers by this author
  • J Amin,

    1. National Centre in HIV Epidemiology and Clinical Research, University of New South Wales, Sydney, Australia, and
    Search for more papers by this author
  • PWG Mallon,

    1. National Centre in HIV Epidemiology and Clinical Research, University of New South Wales, Sydney, Australia, and
    2. HIV, Immunology and Infectious Diseases Clinical Services Unit, St Vincent's Hospital, Sydney, Australia
    Search for more papers by this author
  • D Marriott,

    1. HIV, Immunology and Infectious Diseases Clinical Services Unit, St Vincent's Hospital, Sydney, Australia
    Search for more papers by this author
  • A Carr,

    1. HIV, Immunology and Infectious Diseases Clinical Services Unit, St Vincent's Hospital, Sydney, Australia
    Search for more papers by this author
  • DA Cooper,

    1. National Centre in HIV Epidemiology and Clinical Research, University of New South Wales, Sydney, Australia, and
    2. HIV, Immunology and Infectious Diseases Clinical Services Unit, St Vincent's Hospital, Sydney, Australia
    Search for more papers by this author
  • S Emery

    1. National Centre in HIV Epidemiology and Clinical Research, University of New South Wales, Sydney, Australia, and
    Search for more papers by this author

*Correspondence and present address: Dr Alan Winston, 2nd Floor St Stephens Centre, Chelsea and Westminster Hospital, 369 Fulham Road, London SW10 9NH, UK. Tel:+44 208 846 6508; Fax: +44 208 746 5628e-mail: alan_winston71@yahoo.co.uk

Abstract

Background

Tenofovir disoproxil fumarate (Tenofovir DF, TDF), the first nucleotide reverse transcriptase inhibitor approved for the treatment of HIV disease, has been associated with renal dysfunction in isolated cases. The aim of this study was to assess changes in renal parameters in individuals receiving TDF- and non-TDF-containing highly active antiretroviral therapy (HAART).

Methods

All individuals on HAART attending our clinic were included in the analysis. Time-weighted changes in serum creatinine, calculated creatinine clearance (CCrCl) and anion-gap were assessed for individuals on TDF- and non-TDF HAART.

Results

Of 948 individuals on HAART, 290 (31%) and 618 (65%) were on TDF- and non-TDF HAART, with 40 (4%) having ceased TDF HAART. Baseline values for serum creatinine, CCrCl and anion-gap were similar for those on TDF- and non-TDF HAART. In a multivariate analysis, statistically significant differences were observed in time-weighted change from baseline in anion-gap and CCrCl between individuals on TDF- and non-TDF HAART [mean difference in change between groups: anion-gap 0.78 mmol/L (standard error, 0.19) and CCrCl−6.80 (standard error 2.2); P=0.005 and P=0.032, respectively] after adjusting for baseline anion-gap and CCrCl, respectively. Two cases of TDF-associated renal failure were observed.

Conclusion

Overt renal failure with TDF HAART is rare. However, subtle but statistically significant changes in anion-gap and CCrCl were observed which were associated with TDF HAART. These parameters may be of use in monitoring individuals on HAART.

Introduction

Tenofovir disoproxil fumarate (tenofovir DF, TDF) is the first nucleotide analogue reverse transcriptase inhibitor approved for the treatment of HIV infection [1]. The potency of TDF has been demonstrated in treatment-experienced [2,3] and in treatment-naive patients [4].

Cidofovir and adefovir, compounds structurally related to TDF, have been associated with an increased risk of acute renal insufficiency. In particular, a dose- and time-dependent development of a Fanconi-like syndrome has been reported [5,6]. In pivotal studies assessing the safety and efficacy of TDF, renal safety profiles were found to be similar in individuals receiving and in those not receiving TDF-containing highly active antiretroviral therapy (HAART) [2–4]. In addition, a recent large cohort has described no significant evidence of renal dysfunction with the use of TDF in clinical practice [7].

However, there are mounting numbers of anecdotal reports describing Fanconi's syndrome and renal insufficiency induced by TDF therapy [8–18]. Predisposing factors for the development of acute renal failure with the use of TDF remain obscure.

Several groups reporting no significant evidence of renal dysfunction with the use of TDF have relied on serum creatinine as a marker of renal function [7]. A recent report utilizing glomerular filtration rate (GFR) calculated on the basis of a 24-h urine collection has observed a lower GFR associated with TDF treatment [19].

The aim of this study was to investigate changes in anion-gap and calculated creatinine clearance (CCrCl) in a large cohort of HIV-1-infected individuals receiving TDF- and non-TDF-containing HAART.

Methods

Study design

All individuals currently on HAART attending the Department of Immunology, HIV and Infectious Diseases, St. Vincent's Hospital, Sydney, Australia were included in the analysis. Patients currently treated with TDF-containing HAART (TDF-HAART) were compared with patients on antiretroviral therapy never treated with TDF (non-TDF-HAART). HAART was defined as therapy consisting of at least three antiretroviral agents, in accordance with recent guidelines [20]. For the purpose of the analysis, baseline was considered the time current HAART was commenced with no subsequent changes to any antiretroviral agent. Routine data on individuals who attend our department are collected at each clinic visit in our clinic database. From this database the following parameters were extracted: current HAART, concomitant medication, hepatitis B virus (HBV) infection status (HBsAg) and hepatitis C virus (HCV) infection status (HCV Ab). In addition, we collected the following data for each clinic visit from baseline until 1 January 2005: weight (kg), CD4 lymphocyte count (cells/μL), plasma HIV RNA (copies/mL), serum urea and electrolytes.

Individuals who had ceased a TDF-containing HAART regimen were also included in the analysis (ceased-TDF-HAART), where baseline was taken as the date on which the TDF-containing HAART regimen was commenced, and data were collected until this regimen was ceased. These individuals were included in the analysis only in this group irrespective of further antiretroviral therapy.

CCrCl was calculated using the Cockcroft and Gault equation [21]:

image

where IBW is ideal body weight, F= 1.04 for female patients and F= 1.23 for male patients. Ideal body weight (IBW) kg−Y+ (0.906 to each centimetre greater than 152.4cm), where Y= 45.4 (female patients) and Y= 50 (male patients).

Anion-gap was calculated using a standard formula where:

anion-gap (mmol/L) = [serum sodium (mmol/L) + serum potassium (mmol/L)]−[serum chloride (mmol/L) + serum bicarbonate (mmol/L)].

We also performed a clinical record search of any individual in whom renal failure during TDF therapy was documented.

Statistical analysis

Predictors of time-weighted change in serum creatinine, CCrCl and anion-gap from baseline to last available values were determined using linear regression modelling. The potential predictors assessed in univariate analysis were current TDF use, age (years), gender, weight (kg), time on current HAART (years), baseline CD4 lymphocyte count (cells/μL), baseline plasma HIV RNA (copies/mL), baseline serum creatinine (μmol/L), baseline CCrCl (mL/min), baseline anion-gap (mmol/L), HBV and HCV infection status, concomitant antiretroviral therapy and other concomitant medications. For the purpose of the analysis, concomitant antiretroviral therapies were grouped in the following categories: thymidine analogues (zidovudine or stavudine), didanosine, non-nucleoside reverse transcriptase inhibitors (NNRTIs), protease inhibitors (PIs) and enfuvirtide. Other concomitant medications were grouped according to drug class. Drug class groups were analyzed individually where 5% or more of individuals were currently receiving an agent from one class.

Multivariate analysis was performed on parameters with P-values less than 0.10 in univariate models using a step-wise forward method. All P-values are two-tailed and values less than 0.05 regarded as significant. Data were analysed using stata statistical software (v 8.0; STATA Corp., College Station, TX, USA).

Results

Baseline characteristics

At this institution, 1346 HIV-positive individuals are currently attending for care, with 948 individuals identified as currently on HAART (Fig. 1). A total of 290 (31%) and 618 (65%) individuals were on TDF-HAART and non-TDF-HAART regimens, with 40 (4%) in the ceased-TDF-HAART category.

Figure 1.

 Flow diagram for individuals currently on highly active antiretroviral therapy (HAART). TDF, tenofovir.

Baseline characteristics are shown in Table 1. The three groups as described above were similar for age, gender, weight, baseline CD4 lymphocyte count and plasma HIV RNA at baseline. However, the non-TDF-HAART group had been on their current antiviral regimen for a mean of 2.7 years [95% confidence interval (CI) 2.5–2.8] compared with 1.7 years (95% CI 1.5–1.8) and 1.0 years (95% CI 0.04–2.6) in the TDF-HAART and ceased-TDF-HAART groups, respectively. Ninety-five individuals (10%) were recorded as taking at least one non-antiretroviral concomitant medication. No medication or group of medications was being taken by more than 5% of the cohort.

Table 1.   Baseline characteristics and concomitant medication in individuals on tenofovir (TDF)- and non-TDF-containing highly active antiretroviral therapy (HAART) regimens and individuals who ceased a TDF-containing-HAART regimen
ParameterTDF HAARTnon-TDF HAARTCeased TDF
  • *

    CCrCl, calculated creatinine clearance; CI, confidence interval; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.

Baseline characteristics (n=948)
Total number of patients [n (%)]290 (31)618 (65)40 (4)
Age (years) [mean (range)]46 (23–82)45 (21–75)45 (29–70)
Gender (male) [n (%)]279 (97)585 (95)44 (100)
Weight (kg) [mean (range)]74 (45–162)75 (42–121)75 (48–103)
Time on current HAART (years) [mean (95% CI)]1.7 (1.5–1.8)2.7 (2.5–2.8)1.0 (0.04–2.6)
Baseline CD4 count (cells/μL) [mean (95% CI)]381 (345–416)461 (437–485)367 (204–529)
Current CD4 count (cells/μL) [mean (95% CI)]460 (425–496)523 (499–547)402 (241–564)
Baseline HIV RNA number detectable [n (%)]203 (70)325 (52)17 (68)
Current HIV RNA number detectable [n (%)]116 (40)246 (40)14 (53)
Baseline serum creatinine (μmol/L) [mean (95% CI)]80 (73–89)80 (75–84)72 (61–83)
Current serum creatinine (μmol/L) [mean (95% CI)]83 (78–88)83 (77–88)107 (61–153)
Baseline anion-gap (mmol/L) [mean (range)]14.0 (6.9–26)14.3 (5.1–26)14.8 (9.5–23)
Current anion-gap (mmol/L) [mean (range)]14.3 (5.4–23)13.9 (7.1–24)13 (7.6–21.1)
Baseline CCrCl (mL/min) [mean (95% CI)]*115 (110–120)110 (107–114)120 (100–141)
Current CCrCl (mL/min) [mean (95% CI)]*110 (101–119)110 (105–115)104 (82–126)
Hepatitis B sAg positive [n (%)]32 (11)15 (2)1 (2)
Hepatitis C Ab positive [n (%)]26 (9)37 (5)4 (10)
Concomitant medication [n (%)]
Thymidine analogues69 (24)258 (41)3 (8)
Didanosine32 (11)77 (12)2 (5)
NNRTIs55 (19)114 (18)8 (20)
PIS204 (70)310 (50)19 (40)
Enfuvirtide20 (7)6 (1)4 (10)
Non-antiretroviral concomitant medication27 (9)60 (9)8 (20)

Laboratory parameters

Baseline and current serum creatinine, CCrCl and anion-gap were similar in the three groups (Table 1). The percentages of individuals with laboratory parameters outside the normal range according to our reference laboratory (serum creatinine>110 μmol/L, CCrCl<80 mL/min and anion-gap>18 mmol/L) were not significantly different amongst those in the TDF-HAART, non-TDF-HAART and ceased-TDF-HAART groups for the following laboratory parameters: baseline CCrCl (14, 10 and 19%, respectively; P=0.294), baseline anion-gap (10, 9.5 and 11%, respectively; P=0.952), current CCrCl (14, 16 and 27%, respectively; P=0.231) and current anion-gap (5.5, 9.5 and 7.5%, respectively; P=0.085). Individuals in the ceased-TDF-HAART group were more likely to have baseline and current serum creatinine values above the upper reference range compared with the other two groups (16% vs 4% in both other groups at baseline, P=0.002; 16% vs 5% and 6.5% for current values in the non-TDF- and TDF-HAART groups, P=0.023).

Factors associated with time-weighted change in CCrCl and anion-gap from baseline

Associations amongst time-weighted change from baseline in serum creatinine, CCrCl and anion-gap among the three groups are shown in Table 2. Individuals on TDF HAART compared with those on non-TDF HAART had a small but statistically significantly rise in mean time-weighted change in anion-gap from baseline (change of 0.53 mmol/L vs−0.24 mmol/L, respectively; P<0.001) and a statistically significant fall in mean time-weighted change in CCrCl from baseline (change of−5.6 mL/min versus 1.26 mL/min, respectively; P=0.002). No significant differences in time-weighted change in serum creatinine were observed, and no significant differences were found between the group who had ceased TDF HAART and the other groups.

Table 2.   Associations between time-weighted change from baseline to last follow up for anion-gap, calculated creatinine clearance and serum creatinine in the tenofovir (TDF)-highly active antiretroviral therapy (HAART), non-TDF-HAART and ceased-TDF-HAART groups
 Non-TDF-HAARTTDF-HAARTCeased-TDF-HAARTOverall P-value
  1. SE, standard error.

Mean time-weighted change
in anion-gap from baseline−0.240.530.24<0.001
(mmol/L) (SE)(0.11)(0.15)(0.52) 
P-valuereference< 0.0010.229 
Mean time-weighted change
in calculated creatinine clearance1.26−5.6−0.410.010
from baseline (mL/min) (SE)(1.0)(2.4)(6.7) 
P-valuereference0.0020.813 
Mean time-weighted change
in serum creatinine from baseline−1.70−2.15.60.740
(μmol/L) (SE)(2.5)(4.6)(2.7) 
P-valuereference0.9340.449 

As significant associations in time-weighted change from baseline for anion-gap and CCrCl were observed between those on TDF- and non-TDF HAART, other factors associated with these parameters were assessed. In a multivariate analysis, only TDF use and baseline anion-gap were significantly associated with a rise in time-weighted change in anion-gap from baseline (Table 3) and only TDF use was significantly associated with a fall in time-weighted change in CCrCl from baseline (Table 4). Time on current HAART regimen was not associated with these changes. It is interesting that lower baseline serum creatinine was associated with a small but significant rise in time-weighted change in CCrCl. This may be attributable to overall improvements in renal function associated with antiretroviral therapy [22].

Table 3.   Parameters associated with time-weighted change in anion-gap from baseline to last follow-up in individuals on TDF- and non-TDF- containing highly active antiretroviral therapy (HAART) regimens
ParameterUnivariate analysisMultivariate analysis*
Change AUCSEP-valueChange AUCSEP-value
  • P-values in bold are significant.

  • AUC, area under curve; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.

  • *

    Adjusted for baseline anion-gap and current TDF use.

Current TDF use0.780.19<0.0010.640.220.005
Age (years)0.010.010.128   
Gender−0.070.480.890   
Weight (kg)0.000.010.654   
Time on current HAART (years)0.140.060.0150.060.060.238
Baseline CD4 count, 100 cells/μL change0.040.040.278   
Baseline HIV RNA, detectable versus nondetectable−0.230.180.191   
Baseline serum creatinine, 100 μmol/L change0.070.120.592   
Baseline anion-gap mmol/L0.270.03<0.0010.180.03<0.001
Baseline calculated creatinine clearance−0.010.000.0460.000.000.932
Hepatitis B virus sAg positive−0.630.390.104   
Hepatitis C virus Ab positive−0.200.350.568   
Thymidine analogue0.480.220.0280.230.210.273
Didanosine0.430.290.137   
NNRTI−0.210.180.233   
PI0.020.140.874   
Enfuvirtide0.400.480.403   
Nonantiretroviral concomitant medication0.190.280.497   
Table 4.   Parameters associated with time-weighted change in creatinine clearance from baseline to last follow up in individuals on tenofovir (TDF)- and non-TDF-containing highly active antiretroviral therapy (HAART) regimens
ParameterUnivariate analysisMultivariate analysis*
Change AUCSEP-valueChange AUCSEP-value
  • P-values in bold are significant.

  • AUC, area under curve; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.

  • *

    Adjusted for baseline creatinine-clearance and current TDF use.

Current TDF use−6.802.20.003−4.902.200.032
Age (years)−0.080.110.466   
Gender−4.016.140.514   
Weight (kg)−0.090.090.334   
Time on current HAART (years)−0.630.700.368   
Baseline CD4 count, 100 cells/μL change−0.860.430.0450.580.420.171
Baseline HIV RNA, detectable versus nondetectable2.442.130.253   
Baseline serum creatinine, 100 μmol/L change−4.31.340.002−0.020.010.272
Baseline anion-gap mmol/L−0.320.330.325   
Baseline calculated creatinine clearance0.150.03<0.0010.140.03<0.001
Hepatitis B virus sAg positive7.354.500.103   
Hepatitis C virus Ab positive−3.233.920.411   
Thymidine analogues0.192.770.946   
Didanosine−2.213.630.542   
NNRTI−1.772.150.412   
PI1.641.780.357   
Enfuvirtide10.84.790.0256.144.830.204
Non-antiretroviral concomitant medication−4.813.000.104   

Cases of renal failure on TDF therapy

Three cases of acute renal failure in individuals on antiretroviral therapy containing TDF have been observed in our institution, representing 0.3% of all individuals on HAART. One case involved acute renal failure following a hospital admission with Streptococcus pneumoniae sepsis. Laboratory findings were consistent with acute tubular necrosis rather than related to TDF.

Renal failure and significant proteinuria were diagnosed in the other two cases and the likely aetiology was thought to be TDF. Both patients were white men. Patient A was aged 44 years and patient B was aged 42 years. Both patients were highly experienced with antiretroviral therapy, with a nadir CD4 lymphocyte count below 50 cells/μL. Prior to developing renal impairment, HAART comprised TDF-lamivudine-lopinavir-ritonavir in patient A and TDF-didanosine-lopinavir-ritonavir in patient B, and serum creatinine levels were within the normal laboratory limits (<110 μmol/L) in both individuals. After 14 and 17 months on treatment, respectively, CD4 lymphocyte counts were 630 and 300 cells/μL and plasma HIV RNA values were below 50 and 1000 copies/mL, respectively. In both cases, serum creatinine had risen to over 160 μmol/L, serum phosphate was low and significant proteinuria had developed (0.5 g proteinuria/day in patient A and 1.1 g/day in patient B). Concomitant medication comprised fluconazole 100 mg daily in patient A and sodium valproate 700 mg twice daily in patient B. TDF was ceased in both cases, with abacavir substituted in patient A. No other changes to medication were made. Three months later, serum creatinine had normalized and proteinuria was within the normal range (<0.15 g/day) in both individuals.

Discussion

This cohort demonstrates that renal failure associated with TDF HAART is rare (0.3%), as described in other cohorts. However, subtle but statistically significant changes in CCrCl and anion-gap were observed, associated with TDF HAART, whereas no significant changes in serum creatinine were observed. Since performing this analysis, other groups have also reported subtle changes in CCrCl associated with TDF HAART [23,24].

Other nucleotide analogues, such as cidofovir and adefovir, may induce renal damage as a result of mitochondrial DNA depletion in renal tubular cells [25]. Cell culture studies have shown TDF to be only minimally toxic to renal tubular cells [26]. If mitochondrial DNA depletion is time-dependent then overt renal failure may not become apparent for some months after treatment is begun. The subtle changes we have observed in renal function in our cohort may be related to the low-grade toxicity of TDF on renal tubular mitochondrial DNA and may play a role in monitoring for the development of overt renal dysfunction over time.

Mild renal dysfunction as determined by GFR measurement from 24-h urine sampling has been shown to be associated with TDF HAART where changes in serum creatinine were not always apparent [19]. CCrCl as a measurement of GRF, although not as accurate as GRF measured by timed urine collection, is simple and can be calculated from routine laboratory parameters.

Other markers of proximal renal tubular function may be of importance in monitoring for the development of a Fanconi-like syndrome with TDF use. Anion-gap can be calculated from routine serum electrolytes and avoids the need for arterial blood or urine collections, and subtle changes are associated with TDF use in this cohort.

This study is limited by its retrospective design, and prospective controlled studies are needed to confirm these observations. Furthermore, our data are constrained by the information in our clinic database. We have not assessed data on serum phosphates, which may also be a sensitive marker of proximal tubular dysfunction. Only 10% of patients were recorded as receiving non-antiretroviral concomitant medication in our database, which may be an underestimate. Both HBV and HCV infections have been associated with renal disease and chronic renal failure [27,28]. Within our cohort, seroprevalence of HBV and HCV is low. In other areas with a higher prevalence of viral hepatitis, renal failure with TDF therapy in this group may be observed more frequently. Although changes in anion-gap and CCrCl were observed in this cohort, no changes outwith the normal laboratory range were observed prior to the development of renal failure in the two cases of suspected TDF-related renal insufficiency. No measurements of bone-mineral density are available for patients in this cohort, and hence any associations between changes in anion-gap and bone-mineral density cannot be assessed. Lastly, this cohort comprises subjects with a lack of renal impairment at baseline. This is similar to the previous studies assessing the safety of TDF [3,4] and provides little information regarding changes in CCrCl and anion-gap in patients with even modest renal impairment.

In our cohort, we have described two cases of clinically overt renal toxicity considered to be related to TDF HAART. Reasons for this conclusion included no evidence of renal failure in these subjects prior to commencing TDF HAART and resolution of renal failure on cessation of these regimens. It is interesting that one of these subjects developed renal toxicity when receiving a nucleoside backbone consisting of TDF-didanosine, which has previously been reported with this regimen [10]. Other causes of renal toxicity such as HIV-associated nephropathy cannot be excluded.

In conclusion, renal failure with the use of TDF HAART is rare but we have observed subtle changes in CCrCl and anion-gap in individuals on TDF therapy. Prospective work is required to assess the use of these parameters as predictors for the development of overt renal dysfunction with the use of TDF.

Acknowledgements

The National Centre in HIV Epidemiology and Clinical Research is funded by the Australian Government Department of Health and Ageing, and is affiliated with the Faculty of Medicine, The University of New South Wales.

We would like to thank Karl Hesse for extraction of data from the clinic database.

Ancillary