|Home | About | Journals | Submit | Contact Us | Français|
Tonus was a pilot study in which previously untreated human immunodeficiency virus type 1 (HIV-1)-infected patients received the combination of abacavir, lamivudine, and tenofovir once a day. There was a high rate of early virological failure, and the M184V and K65R mutations were frequently detected at week 12 (W12). The objective of this study was to examine the selection dynamics of the K65R and M184V/I mutations. Bulk sequencing of the reverse transcriptase (RT) gene was performed on plasma HIV-1 RNA at baseline, W4, and W12 for 21 patients with detectable viral loads. The RT genes from baseline, W4, and W12 plasma samples from five patients who developed both M184V and K65R but with different mutational patterns were also cloned and screened for the K65R mutation by selective real-time PCR. At baseline, bulk sequencing and clonal analysis showed only wild-type RT sequences. At W4, M184V/I was detected in 12/19 patients and K65K/R in 2 patients by bulk sequencing. At W12, M184V/I was found in 18/20 patient, together with the K65R in 13 patients. At W4, clonal analysis revealed the K65R mutation in 0.6 to 48% of clones in the five patients studied. At W12, the K65R mutation was found in 30 to 100% of clones. K65R and M184V/I seemed to arise in separate clones, followed by an enrichment of viruses containing both mutations. The clinical relevance of this independent evolution is unclear. M184V/I was selected more frequently than K65R at W4. However, K65R was also detected early using a clone-sensitive genotyping method. All three nucleoside analogs are known to select the K65R and/or M184V/I mutation. This convergent genetic pathway to resistance, associated with lower antiretroviral potency, may explain the high selection rate of these mutations in this trial.
The treatment aim for human immunodeficiency virus (HIV) infection is to durably suppress viral replication, preventing morbidity and mortality. Despite the availability of several antiretroviral classes and of several drugs within each class, durable viral suppression remains a challenge, owing at least in part to the selection of resistant viruses. Long-term treatment with protease inhibitor-based highly active antiretroviral therapy is often poorly tolerated, with numerous drug-drug interactions and the risk of lipodystrophy and insulin resistance. In 2001, first-line regimens containing three nucleoside reverse transcriptase (RT) inhibitors emerged as an attractive alternative to regimens containing nonnucleoside reverse transcriptase inhibitors or protease inhibitors. However, several trials of triple-nucleoside analog combinations were terminated prematurely because of high virologic failure rates. Zidovudine (AZT)-lamivudine (3TC)-abacavir (ABC) proved less effective than AZT-3TC-efavirenz (EFV) and AZT-3TC-ABC-EFV in a randomized trial (9). Early virologic failure was also observed with first-line dideoxyinoisine (ddI)-stavudine (d4T)-ABC (20). Tenofovir DF (TDF)-containing regimens including ddI-3TC or ABC-3TC also failed to durably suppress viral load (4, 5, 11).
The K65R mutation, resulting in an amino acid shift from lysine to arginine, is selected in vitro by tenofovir, abacavir, didanosine, and zalcitabine. This mutation confers low-level phenotypic resistance in vitro to tenofovir, abacavir, didanosine, and lamivudine (8, 24, 27). Large genotype databases show that K65R remains infrequent, although its incidence rose from 0.8% in 1998 to 3.8% in 2003, probably owing to the increasing use of tenofovir and also abacavir (17, 18, 20). The M184V mutation is associated with abacavir, lamivudine, and emtricitabine therapy and results in high-level resistance to lamivudine and emtricitabine and low-level resistance to abacavir. The prevalence of the M184V mutation is notably higher than that of K65R and is observed in approximately 40% of HIV samples in large databases.
The Tonus trial was an open-label pilot study in which 38 antiretroviral-naive patients with CD4 cell counts of below 350/mm3 received a once-a-day regimen of abacavir, lamivudine, and tenofovir, three nucleoside inhibitors. The trial was terminated prematurely because of a high rate of virological failure at week 12 (12/36; 33%) and frequent selection of the M184V and K65R mutations, as observed in other similar studies (4, 5). The objective of this study was to document the selection dynamics of K65R and associated RT gene mutations in Tonus study patients with treatment failure.
(This work was presented in part at the XIII International HIV Drug Resistance Workshop: Basic Principles and Clinical Implications, 8 to 12 June 2004, Tenerife, Canary Islands, Spain [abstr. 155].)
The primary end point of the Tonus trial was virologic suppression, defined as a plasma HIV type 1 (HIV-1) RNA level of below 400 copies at week 48 (16). As the study was terminated prematurely, a new definition of virologic failure was defined as applying to patients who never reached a HIV-1 RNA level of below 400 copies/ml or had a rebound of 0.7 log10 HIV-1 RNA or greater after achieving suppression of viral replication to <400 copies/ml (16). In the virologic substudy of the Tonus trial presented here, we focused on 21 patients who had a detectable viral load of above 50 copies/ml at least once while receiving the allocated treatment between inclusion in the trial and week 12 (due to either rebound or incomplete HIV-1 RNA suppression).
Blood samples were obtained at baseline, week 4 (W4), and W12. Plasma and cell pellets were stored at −80°C. Plasma HIV-1 RNA was quantified using the ultrasensitive Amplicor HIV-1 Monitor, version 1.5 (Roche Diagnostics), with a detection limit of below 50 HIV-1 RNA copies/ml of plasma. All plasma samples with detectable viral loads during treatment were screened for the K65R resistance mutation by bulk sequencing at baseline, W4, and W12. On the basis of these standard genotyping results, five patients with different mutational pathways were selected for clonal analysis of the RT gene in plasma samples collected at baseline, W4, and W12. In order to screen large numbers of clones, we developed a K65R-selective real-time PCR method. All the clones generated were tested with the K65R-selective real-time PCR. The RT genes from a subset of representative clones expressing the K65R mutation or not expressing K65R were sequenced (codons 35 to 240).
Viral RNA was extracted from plasma (0.5 to 1 ml depending on viral load) with the automated nucleic acid extractor Magnapure and the Total NA Large Volume kit (Roche Diagnostics GmbH, Manheim, Germany) and stored at −80°C. Ten microliters of RNA was used for RT-PCR (Titan One-Tube RT-PCR kit; Boehringer, Manheim, Germany) according to the manufacturer's instructions, with primers MJ3/MJ4 and A35/NE1, as described previously (19). RT-PCR products were stored at −20°C.
The sequencing conditions used here have been described in detail elsewhere (19). Sequencing was performed on a ABI Prism 3100 GA (Applied Biosystems, Foster City, CA). Nucleotide sequences were aligned using Sequence Navigator software (Applied Biosystems). RT gene mutations listed by the IAS-USA expert panel (www.iasusa.org) were recorded. Polymorphisms at codon 68, described by Roge et al. (20) as emerging together with the K65R mutation, were also recorded.
To obtain clones, as we have previously described (10), serial 10-fold dilutions (10−2 to 10−6) of RT-PCR products were prepared, and 5 μl of each dilution was amplified using primers A35/NE1 (19) in 5 μl of 10× buffer (25 mM MgCl2), 0.4 μl of deoxynucleoside triphosphates (100 μM each), 2.5 μl each of 10 μM primers, 2 U of AmpliTaq Gold DNA polymerase, and 45 μl of PCR-grade water. The cycling parameters were 95°C for 10 min; 15 cycles at 95°C for 30 s, 55°C for 30 s, and 72°C for 1 min each; and a final step at 72°C for 15 min. After electrophoresis on agarose gels, ethidium bromide-stained products were examined, and the most dilute initial sample that yielded a visible band under UV light was used for cloning. The PCR products were cloned into pCR4-TOPO (Invitrogen, Carlsbad, CA) and used to transfect Escherichia coli. Colonies were grown on imMedia Amp agar plates (Invitrogen). To determine the proportion of clones containing K65R and wild-type sequences, individual colonies were transferred to 200 μl of Luria-Bertani medium as we have previously described (10) Following overnight incubation at 37°C, cultures were resuspended and diluted 1:50 with water, and then 10-μl aliquots were tested by real-time PCR as described below.
In a previous paper (10), we have shown that when mixtures containing 20% viral RNA with mutated sequences and 80% viral RNA with wild-type sequences were evaluated for possible recombination, 10 of 10 clones identified as carrying the mutation L90M in the protease gene also carried all the other mutations and polymorphisms initially present in the mutated sequence, indicating no recombination during the amplification.
The assay used allele-specific oligonucleotides for the detection of wild-type and K65R mutant sequences. The K65R drug resistance mutation consists of an AAA-to-AGA transversion at codon 65 of the RT gene. The primers used for real-time PCR were designed on the basis of published sequences in the pol region of HIV-1HXB2, and the 3′ end of the forward primers was located on codon 65. Real-time PCR was performed in parallel reactions with locked nucleic acid (PROLIGO Primers and Probes, France SAS) forward primers K65 (5′-CTCCAGTATTTGCCATAAAGA+A-3′) and 65R (5′-CTCCAGTATTTGCCATAAAGA+G-3′), permitting preferential amplification of wild-type and K65R-containing sequences, respectively. The reverse primer 65AS (5′-TGGGAAGTTCAATTAGGAATA-3′) and probe [5′-(6-FAM)CTTGAGTTCTCTTTATTAAGTTCTCTG(TAMRA)-3′] were used in both amplification systems. Ten microliters of diluted clonal culture (1:50) was used for real-time PCR. Reaction conditions were those recommended by the manufacturer of the TaqMan Universal PCR Master Mix (Applied Biosystems). The final concentration of each primer was 0.5 μM, and the final probe concentration was 0.2 μM. Real-time PCR was performed with a Perkin-Elmer 7700 sequence detection system (PE Applied Biosystems).
Phylogenetic analyses were performed to determine the HIV-1 subtypes of the viruses by estimating the relatedness of pol sequences and reference sequences of HIV-1 genetic subtypes and circulating recombinants obtained from the Los Alamos database (http://hiv-web.lanl.gov). Nucleotide sequences were aligned with the CLUSTAL W program version 1.7 (23). Phylogenetic reconstruction was performed using a Kimura two-parameter model and the neighbor-joining method with 1,000 bootstrapped data sets.
We studied 21 patients who had detectable viral loads (above 50 copies/ml) at least once while receiving the allocated treatment (Table (Table1),1), between inclusion in the trial and week 12. At baseline, the median viral load and CD4 cell count values were 106,000 copies/ml (range, 21,000 to 757,000) and 223 cells/mm3 (range, 61 to 348), respectively.
At baseline, 18 patients had wild-type RT gene sequences, while 3 patients had a polymorphism at codon 68 (S68G), which was associated in one case with T215E. At week 4, genotyping results were unavailable for two patients (no plasma specimen in one case and viral load below 100 copies/ml in one case). A mutation at codon 184 was found in 12 (63%) of the remaining 19 patients. The K65R mutation was found in two cases (10.5%), in association with M184I in one case.
At week 12, genotyping results were available for 20 patients. Eighteen patients (86%) had the M184V/I mutation. A mutation at codon 65 was found in 13 patients (65%), always in conjunction with M184V/I. Five patients had the M184V/I mutation alone. V118I was present in one of the patients with both the K65R and M184V mutations. Two of the three patients with a baseline polymorphism at codon 68 had acquired the K65R mutation at W12. The S68G mutation was observed in one additional patient (patient 9) who had a silent polymorphism at this position at baseline (AGC instead of AGT).
Phylogenetic analyses showed that 11 viruses were subtype B, 5 were CRF02_AG, 2 were A, 1 was D, and 1 was F2. The subtype of the remaining virus was not determined. All the viruses used in clonal experiments were subtype B viruses.
The discriminatory capacity of the assay was evaluated with wild-type (n = 180) and K65R mutant (n = 55) sequenced clones which were amplified with appropriate and inappropriate oligonucleotides. Amplification of a K65R mutant sequence with the wild-type K65 primer was shown by a positive fluorescent signal that appeared ~15 cycles after the signal obtained with the 65R mutant primer. Amplification of a wild-type sequence with the 65R mutant primer was shown by a positive fluorescent signal that appeared ~7 cycles after the signal obtained with the wild-type K65 primer. These differences in threshold fluorescence values reflect lesser amplification efficiency when a mismatch is present at the 3′ end of the forward primer (14). Clones with and without the K65R mutation were easily distinguished by comparing the number of cycles required to reach threshold fluorescence in reactions performed in parallel with the K65 and 65R forward primers. We studied between 126 to180 clones per sample. Among the patients with the K65R mutation in plasma at week 4 or 12, the following five patients with different mutational pathways were selected for clonal analysis of the RT gene: the two patients (patients 19 and 20) with mixed viral populations at codon 65 (K65K/R) at week 4 and 3 of the 11 patients (patients 10, 12, and 16) with only M184V/I at week 4. Screening by K65R-selective real-time PCR was performed on clones generated from plasma samples at baseline, week 4, and week 12. The results are shown in Table Table2.2. None of the clones from baseline specimens bore the K65R substitution. At week 4, K65R was found in a variable number of clones in all five patients studied. This mutation was detected in 82/171 (48%) and 21/163 (13%) clones from patients 19 and 20, respectively, in whom bulk sequencing had shown mixed K65K/R populations at week 4. The K65R mutation was detected in only 2/135 (1.5%), 11/173 (6%), and 1/180 (0.6%) clones from patients 10, 12, and 16, in whom bulk sequencing had failed to detect the K65R mutation at W4. At week 12, in keeping with bulk sequencing, the K65R mutation was found in almost all the clones, except in patient 10 (30% of clones).
Results of clonal sequencing at weeks 4 and 12 are compared with bulk genotyping results for each of the five selected patients in Fig. Fig.1.1. In every case, sequencing confirmed the results of selective K65R screening by real-time PCR. At week 4, clonal sequencing raised the possibility that the K65R and M184V/I mutations were present on separate genomes. In fact, K65R and M184V/I were mutually exclusive in all mutant clones from patient 20 and in 5 of 11 mutant clones from patient 12. At week 12, the vast majority of clones were dual mutants (K65R and M184V/I). Interestingly, the M184A (ATG-to-GCG) substitution was detected in two clones from patient 10 at week 12.
Polymorphisms at codon 68 and the T215E mutation were detected in patient 19 at weeks 4 and 12. Additionally, patient 12 acquired the S68G substitution (in one clone) at week 4, together with a new polymorphism at codon 68 (S68N, in four clones). The Q151M mutation was never detected in a minority population. Of note, one clone obtained from patient 19 at week 4 bore the Q151R substitution. This mutation was not observed in any clones at week 12. The V118I mutation was detected in 45 of 76 clones from patient 12 at week 12. Several other mutations, such as A62V, T69S, and Y115F, were detected in some clones from patient 10. In addition, the thymidine analogue mutation D67N was detected in one clone from patient 20, in association with K65R.
The selection dynamics of the K65R mutation in vivo are poorly documented. In this study, we examined the kinetics of emergence of the K65R mutation in patients enrolled in the Tonus clinical trial. The Tonus trial investigated the safety and efficacy of once-daily treatment with tenofovir, abacavir, and lamivudine in treatment-naive patients. The clinical results of this trial showed a high rate of virologic failure with the development of resistance, as described elsewhere (16).
Bulk sequencing at W4 for patients with detectable viral loads revealed the M184V/I mutation in 63% of patients and K65R in 10.5%. At W12, M184V/I and K65R were detected by standard genotyping in 90% and 65% of patients, respectively. Early virologic failures were also observed in two trials of the same first-line once-a-day TDF-ABC-3TC regimen, with frequent selection of the M184V/I mutation (45% and 98%, respectively) and the K65R mutation (36% and 53%, respectively) (4, 5). The K65R mutation was never found in isolation. In patients receiving a first-line TDF-ddI-3TC regimen, virologic failure at W16 was associated with selection of the M184V/I mutation in 20 of the 22 patients, of whom 10 also had the K65R mutation (10). The higher frequency of the K65R mutation in our study (65%) may be due to later genotyping than in other studies. In addition, we used a nested PCR-based genotyping method capable of detecting K65R mixed mutants with greater sensitivity. In fact, 6/13 patients with detectable K65R at W12 had mixtures of K65R with wild-type K65.
The current genotyping method cannot be used to study the early phase of the wild-type-to-K65R transition because viral populations representing fewer than 20% of total viruses cannot be reliably studied. Using clonal analysis of the RT gene, we detected K65R mutants earlier in patients in whom standard genotyping had failed to identify this mutation. Among the patients selected for clonal analysis, all five had K65R mutants at W4 by the clonal genotyping method, ranging from 0.6% to 48%. These findings demonstrate that the K65R mutation can be rapidly selected in previously untreated patients. The large number of K65R mutant clones present at W4 in one patient suggests that this mutation could probably be detected earlier by clonal analysis, in keeping with a recent study (21). Similarly, clonal analysis for the M184V mutation in earlier samples would also likely show M184V, earlier given the bulk sequence detection of M184V/I in the week 4 samples. However clonal analysis is a labor-intensive method and cannot be used routinely for resistance testing in HIV-1-infected patients. Moreover, it is not validated for clinical use, and there are no known clinical correlates to the mutant percentages that can be determined.
At week 4, K65R and M184V were usually detected in different clones prepared from a given sample. This might be due to the relative variability of intracellular tenofovir, 3TC, and abacavir concentrations and activity, leading to different levels of selective pressure in individual cells or tissue compartments. Additionally, the resensitizing effect of the M184V mutation on tenofovir sensitivity may provide some negative selective pressure for the addition of K65R to the M184V mutant (2, 27, 28). Eight weeks later, however, the two mutations were found in the same clones. Clonal analysis suggested that resistance arose from distinct viruses harboring K65R or M184V, followed by enrichment of viruses containing both mutations. Similar observations have been made from in vitro drug coselection experiments (M. Miller, unpublished observations). This could be due to genetic recombination in vivo or the stepwise addition of the K65R mutation to the M184V mutant. For tenofovir, the K65R mutation confers a weak selective advantage compared to the strong selective advantage of M184V for lamivudine. However, both mutations together provide a strong selective advantage against abacavir, with >6-fold resistance to abacavir (24). Thus, maximal resistance to all three drugs in this treatment regimen would select for both the M184V and K65R mutations under continued treatment pressure.
In order to minimize PCR-mediated recombination, we used optimal RT-PCR conditions for clonal analysis (3, 10, 12). Several findings indicate that artifactual recombination did not account for the observed genetic diversity. In particular, the K65R and M184V mutations in the week 4 sample of one patient were always mutually exclusive. The K65R and M184V mutations are present at the 5′ and 3′ ends of the amplified fragment, respectively, and these results would not be expected if recombination occurred frequently during PCR.
Several uncommon amino acid substitutions were detected by clonal analysis. An M184A substitution (ATG to GCG) requires a transversion which is not an intermediate of the M184V mutation (ATG to GTG). In vitro, M184A viruses are known to have a greatly reduced replicative capacity and seem to be only marginally resistant to 3TC; as a result, they have not so far been found in patients (13). Our findings suggest that this mutation can be selected in vivo but confers only a very weak selective advantage. However, as the nucleotide sequences of the two clones harboring the M184A substitution were identical, we cannot distinguish between emergence of an M184A variant or a PCR artifact.
The polymorphism at codon 68 is commonly present in approximately 3% of wild-type viruses. Roge et al. described the joint emergence of K65R and S68G in patients receiving a first-line ABC-ddI-d4T regimen, and they suggested that the S68G polymorphism could enhance the replicative capacity of K65R mutants (20). Garcia-Lerma et al. demonstrated that the introduction of the S68G mutation restored the replicative capacity of the Q151L intermediate and did not affect replicative capacity in a wild-type RT genetic background (7). In our study, patients harboring an S68G mutant at baseline developed K65R mutants during treatment, and two patients acquired the S68G and K65R mutations during the follow-up.
Several studies point to the association of K65R and Q151M mutations in genotypic databases (1, 20), suggesting that K65R may be a favorable genetic background for the development of the Q151 M multidrug resistance mutation. Two base changes (CAG to ATG) are required for the acquisition of the Q151M substitution, and Q151L (CAG to CTG) and Q151K (CAG to AAG) are possible intermediates. In our study, we never found the Q151 M mutation or possible intermediates at week 12, even as a mixed population in patients harboring K65R and S68G mutants. A recent study suggests that the association of K65R and Q151M is not linked to the use of TDF, especially in antiretroviral-naive patients, but is rather associated with long-term abacavir, ddI, and ddC therapy (29). This study also showed that development of the Q151M mutation occurred either first or at the same time as K65R. This pattern of resistance development is in contrast with the clinical trial data for TDF in both treatment-experienced and treatment-naive patients, which have shown only the K65R mutation to be associated with TDF treatment (6, 16). Nevertheless, since these mutations are compatible and can result in significant nucleoside reverse transcriptase inhibitor resistance, continued resistance monitoring of patients failing therapy and with a K65R mutation is warranted.
The K65R and M184V mutations increase overall RT fidelity (22, 25, 26). This, together with the convergent pathway to resistance of the associated drugs, may explain the small number of resistance mutations observed in the studied viral population. In addition, the K65R and M184V mutations have each been associated with decreased replication capacity in vitro (2). Replication capacity analyses performed for patients who developed the K65R and M184V mutations in the Tonus study have confirmed the effect of these mutations on decreasing viral replication capacity (15). The contribution of reduced replication capacity to the clinical course and plasma viral loads in these patients is unknown; however, the patients did appear to have a lower plasma HIV RNA viral load, despite acquisition of these resistance mutations, compared to their baseline (Fig. (Fig.1).1). The lower viral loads might also be due to some residual antiviral activity of TDF when the K65R and M184V mutations are present, since M184V increases susceptibility of the virus to TDF. Phenotypic data available for seven patients show that TDF susceptibility values were below the cutoff for 5/7 patients, indicating the potential for continued TDF drug activity (15).
In conclusion, we found that the M184V/I mutation was selected more frequently at week 4 than the K65R mutation in patients starting antiretroviral therapy with a triple-nucleoside combination containing TDF, ABC, and 3TC. However, the K65R mutation was also detected early by means of a sensitive clonal genotyping method. The clinical relevance of the independent acquisition of these two mutations by distinct viruses is unclear. TDF, ABC, and 3TC share a number of resistance mutations: TDF and ABC select K65R, while 3TC and ABC select M184V. This convergent genetic pathway to resistance, in conjunction with lower antiretroviral potency, may explain the high rate of selection of these two mutations in our patients. Follow-up clinical data from the Tonus study indicate that effective second-line regimens can be constructed for patients who have failed with both the M184V and K65R mutations, with 14/14 patients achieving plasma HIV-1 RNA levels of below 50 copies/ml on their subsequent regimens (AZT-ddI-lopinavir-boosted ritonavir [LPVr], n = 4; ABC-3TC-TDF, n = 3; ABC-ddI-LPVr, n = 2; AZT-3TC-ABC-LPVr, n = 1; AZT-EFV-LPVr, n = 1; AZT-3TC-EFV, n = 1; AZT-ddI-EFV, n = 1; LPVr-saquinavir, n = 1) through 48 weeks (15). However, the long-term implications of M184V plus K65R selection for nucleoside reverse transcriptase inhibitor treatment options remain to be determined.
We thank Sentob Saragosti (INSERM U552, Paris, France) and Vincent Calvez (Laboratoire de Virologie, Pitié-Salpétrière Hospital, Paris, France) for helpful discussion.