Lack of Minority K65R-Resistant Viral Populations Detected After Repeated Treatment Interruptions of Tenofovir/Zidovudine and Lamivudine in a Resource-Limited Setting

To the Editors: Intermittent antiretroviral therapy strategies have been demonstrated to be suboptimal to continuous therapy.1 Nevertheless, treatment interruption (due to toxicity, problems with drug supply, or as an agent to prevent mother-to-child transmission) remains a reality for many patients. It is therefore important to study its likely impact, one important aspect of which is the emergence of drug resistance. For instance, the long half-life of nonnucleoside reverse transcriptase inhibitors encourages the selection of pre-existing resistant mutants when stopping regimens containing this class of drug, a process associated with a higher incidence of nonnucleoside reverse transcriptase inhibitor resistance.2 Tenofovir (TDF) is another drug with an extended half life, and, is increasingly being used in developing and developed world settings. The critical mutation associated with TDF resistance is K65R, and we have assessed the risk of its emergence after repeated structured treatment interruptions (STIs) with a TDF-containing regimen in the Development of Antiretroviral Therapy in Africa (DART) trial. DART is a randomized study comparing monitoring strategies in Uganda and Zimbabwe in the absence of "real-time" virological testing. In a STI substudy, 813 patients with CD4 counts ≥300 cells per microliter at 48 or 72 weeks underwent randomization 4 weeks later to either repeated cycles of 12 weeks off/on treatment or continuous therapy.3 Most patients (547) received zidovudine/lamivudine (as combivir) plus TDF. However, due to adverse clinical events, the STI substudy was stopped such that few participants had undergone multiple STI cycles.3 Because the estimated intracellular half-life of Tenofovir diphosphate (approximately 95-150 hours)4,5 is so much more longer than that for zidovudine triphosphate (9 hours) or lamivudine triphosphate (17 hours),6 we hypothesize that treatment interruption will effectively lead to Tenofovir monotherapy, with selective pressure for emergence of K65R. We examined 18 participants in the STI arm who both underwent the maximum number of complete 12-week STI cycles (4) before the substudy was closed-by definition, all these participants were randomized at week 52-and initially received combivir/TDF and remained on this regimen throughout [allowing substitution of stavudine for zidovudine (n = 2). CD4 count and viral load (HIV-1 RNA)] were measured 8 weeks into each off/on cycle (at weeks: 48, 60, 72, 84, 96, 108, 120, 132, and 144). Population sequencing and minority species assay for the mutations K65R and M184V, which are associated with resistance to TDF and lamivudine, respectively, were undertaken wherever VL >1000 copies per milliliter. For population genotypic analyses, plasma-derived HIV-1 RNA was reverse-transcribed using one-step reverse transcriptase-polymerase chain reaction (Qiagen; West Sussex, United Kingdom). cDNA encoding the entire protease gene and codons 1-320 of reverse transcriptase (RT) were amplified by nested polymerase chain reaction and sequenced using an Applied Biosystems capillary sequencer. K65R and M184V minority sequencing was performed by pyrosequencing, using PyroGold SQA sample preparation kits and a PSQ96MA analyzer (Biotage, Uppsala, Sweden). cDNA generated by reverse transcriptase-polymerase chain reaction was polymerase chain reaction-amplified using subtype-specific, allele-specific primers. Pyrosequencing was found to reproducibly detect down to mixtures of 2% mutant: 98% wild-type, consistent with previously described assays in the literature.7 At 48 weeks, that is, 4 weeks before the first STI cycle, the median (interquartile range) CD4 count was 394 (321-443) cells per cubic millimeter and all patients had a viral load <400 HIV-1 RNA copies per milliliter (14 of 18 participants had <50 HIV-1 RNA copies/mL). Viral load and CD4 plots for individual participants followed the expected patterns of CD4 cell decline and viral rebound associated with an STI strategy (representative participant shown in Fig. 1A). The level of viral rebound and resuppression was similar across successive STIs. Mean HIV-1 RNA ranged from 4.5 to 4.7 log10 copies per milliliter off treatment and declined to 2.3-2.6 log 10 copies per milliliter when treatment was resumed.FIGURE 1: Viral load and CD4 count profiles (A) for a representative participant (B), the only participant who developed resistance (M184V).M184V was first detected during the third on-treatment cycle in 1 patient both by population sequencing and by pyrosequencing (34%), and persisted as a minority species during the fourth off-treatment cycle (13%) (Fig. 1B). On resumption of therapy (during the fourth on-treatment cycle), M184V was not detected by either method. M184V has been observed by others as the major mutation that appears during repeated STIs,8-11 and drug resistance selection is most likely to occur during STIs, when few mutations are required to confer resistance.12 M184V has been reported in 1 study to emerge as a minority at different times during an STI in 14 of 25 subjects.9 By contrast, the K65R mutation was never detected at any time point, either by population sequencing or minority species assay (upper 95% confidence limit for risk of resistance per cycle = 4.1%). This may be because persisting intracellular Tenofovir diphosphate levels are insufficient to select for resistance. In addition, because zidovudine (AZT) is known to counteract the selection of K65R,13 there may be an impact of very low levels of intracellular zidovudine triphosphate (AZT-TP). On the other hand, the fitness cost to K65R14 may lead to a rapid disappearance after selection, particularly during treatment interruption, leading to an underascertainment of resistance even with sensitive assays. The HIV-1 subtypes in this study were A (n = 7), AE (n = 4) and D (n = 7), however, research findings have documented preferential emergence of the K65R mutation in subtype C-infected patients.15 In vitro studies have supported this view, due to the presence of polymorphisms at positions 64, 65, and 66 in RT of subtype C viruses.16 Interestingly, highly active antiretroviral therapy-experienced patients infected with subtype A virus showed a lower propensity to develop K65R compared with subtype B and C viruses, despite similar exposure to antiretroviral agents that select for this mutation,17 which could partly explain lack of detectable K65R in the STI patients infected with subtype A HIV-1 isolates. Whatever the mechanism, the inability to detect K65R in our study, even as a minority species during an STI, may help allay fears about widespread emergence of this mutation on stopping TDF-containing therapy. In conclusion, our preliminary data suggest a low risk of K65R and M184V emergence after treatment interruption on a combivir/TDF regimen. We caution that this does not mean that treatment interruption strategies are beneficial; indeed, the STI randomization in DART was stopped early on the basis of increased clinical progression in the STI group. ACKNOWLEDGMENT We would like to thank all the patients and staff from all the centers participating in the DART trial. Adele L. McCormick, PhD* Ruth L. Goodall, PhD† Aengus Joyce, BSc* Nicaise Ndembi, PhD‡ Mike Chirara, PhD§ Pauline Katundu∥ Sarah Walker, PhD† David Yirrell, PhD¶ Charles F. Gilks, DPhil, FRCP# Deenan Pillay, MD, PhD* on behalf of the DART Virology Group and Trial Team *Division of Infection and Immunity, University College London, London, United Kingdom †Medical Research Council Clinical Trials Unit, London, United Kingdom ‡Medical Research Council/UVRI Uganda Research Unit on AIDS, Entebbe, Uganda §University of Zimbabwe, Clinical Research Centre, Harare, Zimbabwe ∥Joint Clinical Research Centre, Kampala, Uganda ¶Ninewells Hospital and Medical School, Dundee, United Kingdom #Imperial College London, London, United Kingdom