| |
Virological response to a triple nucleoside/nucleotide analogue regimen (AZT/3TC/tenofovir) over 48 weeks in HIV-1-infected adults in Africa
|
| |
| |
AIDS: Volume 20(10) 26 June 2006 p 1391-1399
DART Virology Group and Trial Team
Abstract
Objectives: To evaluate virologic response up to 48 weeks, and emergence of HIV-1 resistance mutations at 24 weeks, in therapy-naive adults initiating zidovudine/lamivudine/tenofovir DF.
Design: A cohort within the DART trial.
Methods: Plasma HIV-1 RNA was assayed in 300 adults with baseline CD4 cell count < 200 cells/μl from sites in Uganda and Zimbabwe using the Roche Amplicor assay v1.5. Samples with HIV-1 RNA > 1000 copies/ml at 24 weeks were sequenced in the pol region.
Results: Median baseline CD4 cell count was 101 cells/μl and HIV-1 RNA 279 910 copies/ml (mean, 5.4 log10). At 48 weeks, 61% (165/272) had HIV-1 RNA < 50 and 72% (196/272) < 400 copies/ml, compared with 59% (167/281) and 79% (221/281) at 24 weeks. At 24 and 48 weeks, 15 and 24% respectively had HIV-1 RNA > 1000 copies/ml (6 and 17% > 10 000 copies/ml), and mean CD4 cell count increases were 103 and 127 cells/μl, respectively. Higher baseline CD4 cell count was the most important predictor of virological suppression at 48 weeks, with little effect of baseline viral load. Eighteen of 20 genotypes from week 24 samples with HIV-1 RNA > 1000 copies/ml showed key resistance mutations in reverse transcriptase. Fourteen had M184V [10 with one to four additional nucleoside analogue mutations (NAMs)]; one had three NAMs only; and the remaining three had K65R. One participant with M184V had major non-nucleoside reverse transcriptase inhibitor-associated mutations, despite no disclosed treatment with this class.
Conclusion: Zidovudine/lamivudine/tenofovir has good virological efficacy in advanced HIV disease. In this population, who were infected with HIV-1 subtypes A, C or D, M184V with or without NAMs was the most common route to resistance, whereas K65R was identified less often.
Introduction
The introduction of antiretroviral therapy (ART) in resource-limited countries is now progressing. The most commonly utilized first-line regimen comprises two nucleoside reverse transcriptase inhibitors (NRTIs) and a non-nucleoside reverse transcriptase inhibitor (NNRTI). This is in accordance with WHO recommendations [1], which are based on efficacy, simplicity (adherence, supply chain management) and preservation of protease inhibitors (PIs) for second-line. Nevertheless, limitations to this standard first-line strategy are increasingly encountered. For example, interactions between NNRTIs and anti-tuberculosis treatment, toxicities of some NNRTIs in those with higher CD4 cell counts (particularly women) or co-infected with hepatitis C, and contraindications of other NNRTIs in pregnancy can all lead to the replacement of the NNRTI with a nucleoside/nucleotide analogue (WHO simplification strategy). This still preserves PIs for second-line. Such limitations have also led to triple nucleoside/nucleotide regimens being considered as first-line therapy.
It is therefore important to assess the efficacy of novel triple nucleoside/nucleotide regimens within this setting. DART is a large multi-centre randomized trial assessing strategies for antiretroviral therapy in Africa, with the majority of patients receiving combivir/tenofovir first-line. This particular combination has the advantages of avoiding interactions with anti-tuberculosis treatment, having a low pill burden and a good tolerability/toxicity profile, and being an NNRTI and PI-sparing triple combination. Routine viral load, as measured by HIV-1 RNA, and resistance testing are not undertaken within this trial, in line with WHO recommendations that such intensive laboratory support is not feasible within resource-limited environments. Within the DART study, plasma specimens are stored, but not tested. Treatment switch is guided by immunological (CD4 cell) and clinical parameters.
Previous studies of first-line triple nucleoside or nucleoside/nucleotide combinations have raised concerns regarding antiviral potency and emergence of drug resistance. Lower virological suppression rates have been demonstrated with zidovudine/lamivudine/abacavir (as combivir/abacavir as well as trizivir) in randomized comparisons with NNRTI-containing regimens [2], although a large observational cohort study did not demonstrate any differences in virological suppression between individuals treated with zidovudine/lamivudine/abacavir or zidovudine/lamivudine/NNRTI over 24 months [3]. With regard to nucleotide/nucleoside-containing regimens, there have also been concerns about the rate of virological failure and emergence of resistance observed in small observational series, predominantly with once daily dosing [4-6], although a small study of zidovudine/lamivudine/tenofovir demonstrated good suppression [7]. Further, co-administration of zidovudine with tenofovir appears to limit the emergence of K65R, probably because this mutation confers hyper-susceptibility to zidovudine, with the virus therefore taking alternative routes to resistance [8].
In the light of the paucity of virological data on zidovudine/lamivudine/tenofovir in previously untreated individuals, particularly in the context of resource-limited settings, we undertook a retrospective virological substudy on the first 300 patients receiving combivir/tenofovir within the DART trial ( www.ctu.mrc.ac.uk/dart ). Routine virological monitoring was not undertaken. In this study we focused on assessing virological response up to 48 weeks in a population with advanced HIV disease (of whom 25% had current or prior tuberculosis diagnosis), and resistance patterns emerging at 24 weeks.
Methods
Trial design and participants
DART is a randomized, controlled trial evaluating two ART management strategies in symptomatic HIV-infected adults (aged 18 years or older) with CD4 cell count < 200 cells/μl initiating ART in Uganda (two clinical sites) and Zimbabwe (one clinical site). Trial entry requires that participants had not previously received ART other than to prevent mother-to-child HIV transmission. The primary randomization compares clinical monitoring only (CMO) with laboratory plus clinical monitoring (LCM). All participants have a full blood count and tests of liver and kidney function, lymphocyte subsets are measured, blood specimens are stored and they see a doctor routinely at weeks 4 and 12, and then every 12 weeks. However, the laboratory results are not returned to clinicians caring for CMO participants unless there is a grade 4 toxicity or the treating clinician has requested them for a clinical reason. Lymphocyte subsets are never returned for CMO participants, and HIV-1 RNA measurements are only performed retrospectively, and so that are unavailable to guide therapy in any patient. A second randomization compares structured treatment interruptions (STIs: 12 weeks on, 12 weeks off ART) with continuous ART in those achieving CD4 cell count ≥ 300 cells/μl at 48 or 72 weeks on continuous therapy. Individual informed consent was obtained from every participant, and the trial received ethics approval in Uganda, Zimbabwe and the UK. The trial began in January 2003 with a planned total duration of 5 years. All participants initiated triple combination antiretroviral therapy with twice-daily combivir (co-formulated zidovudine and lamivudine) plus either tenofovir (74%, once-daily), nevirapine (17%, twice-daily) or abacavir (9%, twice-daily) in line with WHO guidelines at the time the trial started [1].
Laboratory measurements
Three hundred patients taking zidovudine/lamivudine/tenofovir were included in this virology substudy, 100 from each of the Entebbe, Harare and Kampala sites. At each site, 50 consecutively enrolled patients with baseline CD4 cell count 100-199 cells/μl and 50 with CD4 cell count 0-99 cells/μl were included after the first 100 patients had entered the trial overall and, in each individual site, after 20 patients had entered the trial.
Stored plasma samples taken at baseline, 4, 12, 24, 36 and 48 weeks were assayed for HIV-1 RNA at the Joint Clinical Research Centre, Kampala using the Roche Amplicor v1.5 assay (Roche Molecular Systems, Almeda, California, USA) for baseline samples (lower limit of detection 400 copies/ml), and the Roche ultrasensitive assay for other samples (50 copies/ml). Samples with HIV-1 RNA > 1000 copies/ml at 24 weeks were genotyped using reverse transcriptase-polymerase chain reaction and subsequent sequencing of a contiguous region of the pol gene, encompassing the whole of the protease gene and codon 1-230 of reverse transcriptase with appropriate sets of primers with a Beckman capillary sequencer in the MRC/UVRI Uganda Research Unit on AIDS, Entebbe, the Regional Virus Laboratory, Glasgow, and University College London, UK [9]. Key mutations were identified by reference to the most recent IAS-USA algorithm [10]. Subtype was derived from the pol sequences obtained at week 24 using the STAR algorithm [11].
Seventy-seven of these 300 patients entered a structured treatment interruption (STI) pilot study at 28 weeks (based on their CD4 cell response at week 24) and interrupted all ART for 12 weeks. In order to assess longer-term virological response on continuous therapy to 48 weeks, these 77 patients were replaced with the next patients enrolled in the trial who did not undergo an STI before 48 weeks matched on centre, CD4 at baseline (± 25 cells/μl) and CD4 at week 24 (± 35 cells/μl).
Statistical analysis
Virological response up to 48 weeks is presented for the 300 patients who did not interrupt zidovudine/lamivudine/tenofovir as part of an STI (similar results obtained to 24 weeks in the 77 patients subsequently undergoing the pilot STI). Change in log10 HIV-1 RNA was estimated using normal interval regression, replacing undetectable values with the interval in which the true value could lie (i.e. 0-limit of detection) [12]. Three analyses were performed to assess the proportions achieving virological suppression: an intention-to-treat (ITT) analysis based on all available HIV-1 RNA results regardless of ART received (all missing results were at random); an ITT missing equals failure (M = F) analysis regardless of ART received but counting missing HIV-1 RNA results due to death, missed visits or no sample as 'failures'; and an 'on-treatment' (OT) analysis based on all available HIV-1 RNA results when patients had been taking zidovudine/lamivudine/tenofovir continuously without interruption. We used univariable and multivariable (backwards elimination, exit probability 0.05) logistic regression models to identify baseline factors associated with HIV-1 RNA suppression < 50 and < 400 copies/ml at 24 and 48 weeks, also considering ART interruptions, self-reported adherence, and changes in CD4 cell count as potential predictors.
Results
Baseline characteristics are similar to the DART trial overall (Table 1). Sixty-six percent were women and 66% had HIV-1 RNA > 150 000 copies/ml at ART initiation. Mean baseline HIV-1 RNA was 5.4 log10 copies/ml (SD, 0.8). As expected baseline HIV-1 RNA was 0.22 log10 copies/ml higher in patients initiating ART with WHO stage 3 or 4 disease [95% confidence interval (CI), 0.005 to 0.44; P = 0.05 adjusted for all other baseline factors in Table 1a] and tended to be higher in older patients (+0.12 log10 copies/ml higher for every 10 years older (-0.02 to +0.25), P = 0.09), although there was no association with baseline CD4 cell count (P = 0.19) or sex (P = 0.61) in this population (all baseline CD4 cell counts < 200 cells/μl). However, patients from Harare had 0.33 log10 copies/ml lower HIV-1 RNA (95% CI, 0.11 to 0.56; P = 0.003, adjusted for all baseline factors) compared with those from Uganda.
Overall, HIV-1 RNA results were obtained from 1689 samples of a possible 1800 (94%), with death (3%), missed visits (1%), or no stored sample (2%; for example, due to insufficient blood draw) accounting for missing results. Nineteen patients died before 48 weeks (6%), four and 11 of them before 4 and 24 weeks, respectively. Twelve of these 19 had baseline CD4 cell counts below 50 cells/μl and the last HIV-1 RNA was < 6000 copies/ml in 12 of the 15 who died after 4 weeks (six < 50 copies/ml). Only 17 (6%) patients had a change in ART: all substituted stavudine for zidovudine, predominantly for anaemia and/or neutropenia. No patient switched to second line therapy (based on clinical/immunological criteria of failure). 38 patients (13%) interrupted ART for a total of 4 days or longer (median, 12 days; range, 4-167), for toxicity (57% of interruptions), personal reasons (20%), inability to attend clinic (16%), or other illnesses/pregnancy (7%). In total 69 (23%) patients interrupted ART or substituted ART or died (or any combination of these). Thus 231 (77%) of the original 300 patients were known to be alive at 48 weeks having been prescribed zidovudine/lamivudine/tenofovir without interruption.
Virologic response at 4 weeks was excellent (Fig. 1) with only 14 (5%) patients having less than a 1.5 drop in log10 HIV-1 RNA, of whom 10 had either at least a 1 log10 drop or achieved HIV-1 RNA < 1000 copies/ml. At 48 weeks, 61% (95% CI, 55-67%) had HIV-1 RNA < 50 copies/ml and 72% (95% CI, 66-77%) < 400 copies/ml (ITT M = F, 55 and 65%, respectively), in comparison with 59% (95% CI, 53-65%) and 79% (95% CI, 73-83%) at 24 weeks (ITT M = F, 56 and 74%) (Table 2). Correspondingly, 15 and 6% had HIV-1 RNA > 1000 or > 10 000 copies/ml, respectively, at 24 weeks and rising to 24 and 17%, respectively, at 48 weeks (Fig. 1b). Twenty-one (7%) patients never achieved suppression < 400 copies/ml, and 243 (81%) achieved and maintained < 400 copies/ml without confirmed rebound > 400 copies/ml. In the 36 (12%) patients who achieved < 400 copies/ml then had confirmed rebound > 400 copies/ml, the average rate of increase of log10 HIV-1 RNA after first rebound was estimated as only 0.25 per 12 weeks (95% CI, 0.11 to 0.40 increase; P = 0.001).
The most important independent predictors of achieving HIV-1 RNA < 50 copies/ml at week 24 were a smaller cumulative amount of time spent off ART up to week 24 and higher baseline CD4 cell count (Fig. 2), with similar results for < 400 copies/ml. However, by 48 weeks HIV-1 RNA < 50 copies/ml was more likely in patients with higher baseline CD4 cell count [adjusted odds ratio (AOR) = 1.33 per 50 cells higher (95% CI, 1.07-1.66); P = 0.01], who were female [AOR = 2.87 (95% CI, 1.66-4.98); P < 0.001], and who were older [AOR = 1.79 per 10 years older (95% CI, 1.20-2.67); P = 0.005]. Although there was a non-significant trend towards lower suppression rates in those with higher baseline HIV-1 RNA for the same baseline CD4 cell count [AOR = 0.74 per 1 log10 lower (95% CI, 0.52-1.06); P = 0.10] there was no evidence of lower rates of suppression in patients initiating zidovudine/lamivudine/tenofovir with HIV-1 RNA above thresholds defined in studies in industrialized countries, such as over 100 000 copies/ml [AOR = 0.99 (95% CI, 0.54-1.79); P = 0.96) or over 150 000 copies/ml (AOR = 0.81 (95% CI, 0.47-1.40); P = 0.45). There was, however a tendency to lower suppression rates in those with the highest viral loads at baseline > 700 000 copies/ml (AOR = 0.68 (95% CI, 0.36-1.29), P = 0.24) (similarly to Fig. 2).
Of the total of 377 patients with any viral load levels determined (including 77 in the initial phase of this virological evaluation who subsequently underwent a planned treatment interruption at week 28, see Methods), 53 (14%) patients had HIV-1 RNA > 1000 copies/ml at week 24. Of the 31 with HIV-1 RNA < 400 copies/ml at week 12, between 12 and 24 weeks 18 stopped ART for > 1 week (n = 2), had incomplete adherence (15, missing any pills according to self-report), adverse reactions (n = 3), or malaria (n = 6). Genotypes were successfully derived from 20 of the 53 patients with HIV-1 RNA > 1000 copies/ml at week 24, which comprised subtypes A(6), D (5), D/A(1) and C (8, all Zimbabwean patients). Two of the twenty had no mutations, but both had interrupted ART (one for the preceding 6 weeks because of pregnancy; the other from week 9 to 20 because of renal impairment). Eighteen had at least one major NRTI resistance mutation (Table 3): four had M184V alone and one had three nucleoside analogue-associated mutations (NAMS) alone; 10 had M184V and additional NAMs (mean, 2.4; range, 1-4); and three had K65R (one with T215Y, one with Y115F, and one K65R alone). Only the patient with K65R alone had substituted stavudine for zidovudine at week 14. The NAMs observed were M41L (8), D67N (6), K70R (5), T215F (1), T215N (3), T215Y (6). Only three (17%) of the 18 patients had ever achieved HIV-1 RNA < 50 copies/ml before 24 weeks [nine (50%) ever < 400 copies/ml], compared with 77% (96%) in the other 341 patients with week 24 viral load levels, and median baseline HIV-1 RNA was 504 150 copies/ml and 308 120 copies/ml, respectively.
No sample had mutations associated with PI resistance, but one of the 14 with M184V also had a major NNRTI-associated mutation (K103N), despite no disclosed treatment with this class (for example, as prophylaxis to prevent mother-to-child transmission). This viral mutant was identified within a 29-year-old woman with two children (one still alive) who did not have a recorded HIV diagnosis before screening for DART.
Discussion
Both PI and NNRTI-sparing drug regimens are highly relevant to resource-limited settings, where treatment-limiting co-morbidities are high and the available ART formulary is comparatively limited. Herein we have described the virological efficacy of zidovudine/lamivudine/tenofovir within the DART study. Seventy-nine percent and 72% achieved viral loads < 400 copies/ml at 24 and 48 weeks respectively (by ITT), and 59 and 61% < 50 copies/ml respectively. The population was advanced in their HIV disease. Indeed, 10% of DART participants were being treated for pulmonary or extra-pulmonary tuberculosis at ART initiation. Our results compare favourably with data derived from similarly immunosuppressed individuals in other cohorts. For instance, in the UK Collaborative HIV Cohort, 87 and 56% of patients initiating triple therapy (almost exclusively PI or NNRTI-based) in 2002 with CD4 cell counts < 200 cells/μl (median 90 cells/μl) achieved suppression < 400 and < 50 copies/ml, respectively, at 24 weeks (C Sabin, manuscript in preparation). Of more relevance to the DART population is an observational study of 449 South African miners who started efavirenz-containing triple therapy, with a median baseline CD4 cell count of 80 cells/μl. At 6 months, 333 (74%) and 268 (60%) individuals had viral load < 400 and < 50 copies/ml, respectively [13].
Although a large meta-analysis of observational studies [14] found lower rates of suppression when starting ART at lower CD4 cell counts, no consistent relationship between pre-ART CD4 cell count and short-term virological suppression has been reported. In contrast, slower virological suppression has generally been associated with higher pre-ART HIV-1 RNA [15,16]. In the DART population we found that baseline CD4 cell count was the strongest overall predictor of virological suppression at 24 and 48 weeks, and independent of baseline viral load. It is unclear why baseline CD4 cell count should be such a key determinant of virological success, but this phenomenon should be considered when evaluating the efficacy of antiviral strategies in resource-limited settings in order to avoid unrealistic expectations of virological response.
In contrast, the impact of baseline viral load on virological response may be influenced more by the particular drug combination used [17]. Based on virological dynamic concepts [18], this likely reflects the antiviral potency of the regimen. We found no evidence that baseline viral load significantly alters response to zidovudine/lamivudine/tenofovir within DART, suggesting good potency of this regimen in a population with high baseline viral load.
Recently, disappointing results have been reported in patients starting therapy with tenofovir plus two nucleoside analogues. Small pilot studies of once daily didanosine/lamivudine/tenofovir [6] and abacavir/lamivudine/tenofovir [5], as well as a randomized trial comparing abacavir/lamivudine/tenofovir with abacavir/lamivudine/efavirenz [3] have all shown a high rate of short-term virological failure with the tenofovir-containing regimens. Didanosine/tenofovir with efavirenz also appears poorly efficacious [19] and may increase lymphopenia [20]. By contrast with the drug combinations described above, tenofovir/lamivudine with efavirenz appears highly effective in a large number of patients for at least 3 years [21], illustrating that the actual drugs co-administered with tenofovir, rather than their class, have a major influence on the overall efficacy of the combination. We have previously demonstrated a relatively low rate of anaemia with zidovudine/tenofovir-based regimens [22] given the high proportion of women, low baseline haemoglobin and body mass index in our population, which confirm the safety of this approach.
In contrast to the poorly performing tenofovir/NRTI triple regimens described above, the results of our study suggest that a zidovudine/tenofovir combination provides a more robust basis for an effective PI and NNRTI-sparing antiretroviral regimen. The fact that efficacy was demonstrated even for those with high baseline viral loads (with only a trend to reduced suppression at HIV-1 RNA > 700 000 copies/ml) suggests that this triple combination is potent. Importantly, our data provide an evidence-base on which to develop triple nucleoside/nucleotide regimes as class-sparing simplification strategies.
Although different resistance patterns at time of virological failure do not necessarily provide an explanation for reasons for failure, we note that co-administration of zidovudine and tenofovir does appear to influence resistance patterns. In particular, it has been reported that virological failure on a tenofovir-containing regimen is not infrequently accompanied by the K65R mutation in reverse transcriptase. The prevalence of K65R reaches 25-60% in the small mainly observational studies of lamivudine/tenofovir and either didanosine or abacavir described above. By contrast, this mutation is very rare when tenofovir is prescribed with zidovudine [8], where regimen failure is more likely associated with the typical nucleoside analogue mutations (NAMs). The K65R mutation acts to reduce binding, as well as increase excision, of nucleoside analogue drugs to/from the reverse transcriptase enzyme. This explains the broad phenotypic cross resistance conferred against a range of nucleoside/nucleotide analogues. The one exception to this is zidovudine, for which the mutation appears to decrease excision following drug binding, thus making the enzyme hyper-susceptible to this drug [23]. This explains why the virus evolves along a different route to resistance under combined tenofovir/zidovudine selective pressure. Indeed, within the resistance testing undertaken within our study, K65R emerged in only three of 20 patients with viral load > 1000 copies/ml at 24 weeks in whom assays were undertaken. By contrast, M184V and one or more NAMS were the most common combination of mutations in 14 and 11 of 20 patients respectively. Certainly, the M184V mutation is expected at virological failure of any lamivudine-containing regimen. Perhaps less expected is the array of NAMS present at 24 weeks after initiating therapy in our study. One explanation may be that in those with virological failure this drug combination does encourage the more extensive acquisition of NAMs. However, only three patients with resistance test results at 24 weeks had previously suppressed viral load to < 50 copies/ml and baseline viral loads were high in DART (higher still in those with resistance test results). Therefore these patients would have been subject to a significant period of viraemia after initiating therapy, which could have encouraged more extensive resistance to emerge over time. Only a relatively low proportion of plasma samples with viral load > 1000 copies/ml at 24 weeks yielded viral sequences for resistance determination, probably because of sub-optimal sample storage and relatively low level viraemia. Indeed, the sequences obtained were derived from samples with higher viral loads, and the resistance patterns may not be representative of the population as a whole. However, the mutational patterns observed at 24 weeks do not preclude the construction of highly active second-line regimens of triple drug or two-drug two-class combinations [24]. Further, although numbers are small, the type of mutational patterns observed in HIV-1 subtypes A, C and D do not appear to differ substantially from those expected to arise in subtype B viruses [24]. Finally, we note the presence of the NNRTI mutation, K103N in one patient. The most likely explanation is that this represents prior and undisclosed NNRTI use. Although baseline genotypes have yet to be generated from DART patients described in this report, we note that no NNRTI resistance mutations were detected in a natural history (untreated) cohort study of around 150 patients recently undertaken in Entebbe, Uganda (Gale et al., manuscript submitted) or earlier studies [25], suggesting that pre-existing NNRTI resistance is rare in this population.
In conclusion, we demonstrate good antiviral efficacy of zidovudine/lamivudine/tenofovir over 48 weeks, in the context of a population in Africa with advanced HIV disease and a high prevalence of opportunistic infections infected with subtypes A, C and D. These data support the use of such a drug combination as first-line therapy within resource-limited settings.
|
|
| |
| |
|
|
|
