|Home | About | Journals | Submit | Contact Us | Français|
We evaluated the human immunodeficiency virus type 1 (HIV-1) integrase coding region of the pol gene for the presence of natural polymorphisms in patients during early infection (AHI) and with triple-class drug-resistant HIV-1 (MDR). We analyzed selected recombinant viruses containing patient-derived HIV-1 integrase for susceptibility to a panel of strand transfer integrase inhibitors (InSTI). A pretreatment sequence analysis of the integrase coding region was performed for 112 patients identified during acute or early infection and 15 patients with triple-class resistance. A phenotypic analysis was done on 10 recombinant viruses derived from nine patients against a panel of six diverse InSTI. Few of the polymorphisms associated with in vitro InSTI resistance were identified in the samples from newly infected individuals or those patients with MDR HIV-1. We identified polymorphisms V72I, L74I, T97A, V151I, M154I/L, E157Q, V165I, V201I, I203M, T206S, and S230N. V72I was the most common, seen in 63 (56.3%) of the AHI samples. E157Q was the only naturally occurring mutation thought to contribute to resistance to elvitegravir, raltegravir, and L-870,810. None of the patient-derived viruses demonstrated any significant decrease in susceptibility to the drugs tested. In summary, the integrase coding region contains as much natural variation as that seen in protease, but mutations associated with high-level resistance to existing InSTI are rarely, if ever, present in integrase naïve patients, especially those being used clinically. Most of the highly prevalent polymorphisms have little effect on InSTI susceptibility in the absence of specific primary mutations. Baseline testing for integrase susceptibility in InSTI-naïve patients is not currently warranted.
The continued development of novel antiretroviral agents for the treatment of human immunodeficiency virus type 1 (HIV-1) has most recently culminated in the introduction of a new therapeutic class, the strand transfer integrase inhibitors (InSTI); these compounds have shown dramatic results in clinical trials. Raltegravir (RAL; MK-0518) has demonstrated excellent efficacy in both treatment-experienced and naïve patients (2, 6, 31) and is the first InSTI to be approved for use in treatment-experienced patients in the United States, the European Union, and other developed countries. Elvitegravir (EVG; GS-9137) has also shown potent antiviral activity both in vitro and in vivo (7, 37), and phase III clinical trials are under way.
InSTI act by targeting the integrase protein. Integrase has the following two catalytic functions: (i) it removes a dinucleotide from each 3′ end of viral DNA (the 3′ processing reaction), and (ii) in the host nucleus, it mediates the transfer of the proviral DNA strand and covalently links the 3′ ends into the host DNA (the strand transfer reaction) (11). These steps create the provirus, a state that, at once, achieves transcriptional competency for the retrovirus and provides for the stable maintenance and integrity of the viral genome throughout the life span of the infected cell and for all subsequent daughter cells. All current compounds in clinical development preferentially target the strand transfer reaction of integrase, despite representing distinct chemical classes (7, 12, 15, 19, 30).
As multiple InSTI have been assessed for the development of viral resistance in vitro and in vivo, it appears that there are several mutational pathways for the virus to achieve InSTI resistance, even between compounds of related chemical structure (18, 19, 24, 43). Clinical trials have demonstrated the following two primary mutational routes conferring high-level resistance to RAL: N155H in combination with L74M, E92Q, or G163R and Q148H/R/K with E138K or G140S/A (18, 28). More recently, the Y143R/C mutation has also been reported as conferring substantial resistance to RAL (17). In the case of EVG, patients experiencing treatment failure also developed mutations at E92Q, E138K, Q148H/R/K, and N155H. EVG patients also developed compound-specific mutations at T66I/A/K and S147G (24, 33). Furthermore, other drugs such as the diketo acid inhibitors and the naphthyridine carboxamide inhibitor L-870,810 appear to select for viruses that proceed to resistance using other mutational pathways (13, 19), even though considerable overlap exists between the resistance profiles of all drugs examined. For example, L74M and E92Q have been shown to confer substantial resistance to the diketo acid inhibitors S-1360 (12) and L-708,906 and the chemically diverse compound L-870,810 (22) as well as RAL and EVG. Most of the amino acid substitutions are located within the catalytic core domain (CCD) of integrase (amino acids [aa] 50 to 212) and cluster near a three-dimensional interface thought to exist between the integrase protein, two divalent metal cations (i.e., Mg2+), and target DNA. The mutations associated with drug resistance are believed to decrease drug susceptibility by disrupting the binding of the drug to this region adjacent to the catalytic core (18). Included in this set are substitution mutants H51Y, V72I, F121Y, T125K, G140S, Q146K/R, S147G, V151I, S153A/Y, M154I, G163R, V165I, V201I, and S230N (6, 18, 20-22, 31, 44). Some resistance-conferring mutations may be associated with a decrease in viral fitness; for instance, the N155H was associated with a 75% reduction in viral infectivity (9, 21). However, as seen for the reverse transcriptase (RT) and protease coding regions of the pol gene, compensatory mutations can develop which effectively eliminate impairment in fitness while maintaining high levels of drug resistance (37).
With the introduction of integrase inhibitors as viable clinical candidates, we sought to examine the integrase coding region from treatment-naïve patients identified during acute or early infection over the past 10 years in efforts to determine the prevalence of mutations associated with in vitro and clinical resistance to InSTI. We believed that such an analysis was valuable in assessing the need for pretreatment resistance testing when considering the use of InSTI. These samples were selected for the relative homogeneity of the viral population and for potential information about transmissibility. We compared these results to those derived from 15 treatment-experienced, multiclass-resistant patients who, at the time of the study, were naïve to InSTI therapy. We also sought to determine whether alterations in pol outside of integrase affected the prevalence of polymorphisms within the integrase coding region. Finally, to determine the phenotypic effect of common polymorphisms or amino acid substitutions associated with resistance to selected InSTI compounds in vitro, we compared the drug susceptibility profiles of a panel of otherwise isogenic recombinant viruses derived from select patient-derived integrase coding regions against a diverse panel of InSTI.
We analyzed the integrase coding region in 112 treatment-naïve patients identified during acute and early HIV-1 infection (AHI), defined as being infected for less than 1 year, between 1996, which 2007 (36). Subjects had been referred to the clinical service of the Aaron Diamond AIDS Research Center for evaluation since 1996, which had stored peripheral blood mononuclear cells (PBMCs) or plasma samples from their initial or pretreatment visits. Fifteen treatment-experienced patients with triple-class resistant virus (MDR) who were naïve to InSTI therapy were also similarly studied. All patients had provided written informed consent upon study enrollment. All guidelines from the Rockefeller University Institutional Review Board on human experimentation were followed.
The integrase coding region (867 bp) encodes a 288-aa protein, the carboxyl-terminal portion of the Gag-Pol polyprotein, and corresponds to nucleotides 4230 to 5096 of HXB2. Pretreatment proviral DNA from frozen PBMCs or viral RNA from frozen plasma was isolated using the Qiagen QIAamp DNA blood minikit or the QIAamp viral RNA extraction kit (Qiagen, Valencia, CA) as per protocol. Five microliters of RNA elute was reverse transcribed to cDNA using SuperScript III reverse transcriptase (Invitrogen, Carlsbad, CA). Extracted DNA and viral cDNA were then amplified by nested PCR using Phusion Hot Start High-Fidelity DNA polymerase and reagents from Finnzymes (Finnzymes, Woburn, MA). The outer primers were IN-1 (5′-CATGGGTACCAGCACACAAAGG-3′, corresponding to positions 4150 to 4171) and IN-2 (5′-CCCAAATGCCAGTCTCTTTCTCCTG-3′, corresponding to positions 5285 to 5261) (12); the inner primers were Sp5.2 (5′-AGGRATTGGAGGAAATGAACA-3′, corresponding to positions 4169 to 4189) and SpR8.2 (5′-GGGATGTGTACTTCTGAACTTA-3′, corresponding to positions 5213 to 5192). All positions are matched to HIV-1HXB2 (GenBank accession number K03455). For both rounds of PCR, the initial denaturation was done at 98°C for 30 s. Subsequently, 40 cycles of nested PCR were performed with the following conditions: 8 s of denaturation at 98°C and 30 s of annealing at 62°C for the first round and 30 s of annealing at 55°C for the second round with a 30-s extension at 72°C for both, with final extension performed for 5 min at 72°C.
Sequencing of the PCR product was conducted by Genewiz with ABI Prism 3730xl DNA analyzers (Genewiz, South Plainfield, NJ) using primers Sp5.2, SpR8.2, Sp7 (5′-TTCCCTACAATCCCCAAAGTCAA-3′, positions 4651 to 4673) and SpR7 (5′-CCTTGACTTTGGGGATTGTA-3′, positions 4675 to 4656) to ensure complete coverage of both strands. The sequencing of the RT and protease was done using extracted HIV-1 RNA, which then underwent single-tube RT-PCR and nucleotide sequence analysis of protease (codons 4 to 99) and RT (codons 1 to 247) using the Trugene HIV-1 G9 genotyping kit (Bayer Diagnostics, Tarrytown, NY) and OpenGene DNA sequencing system as per protocol (36).
Nucleotide sequences were multiple aligned using Clustal W in MegAlign v. 7.2 (DNAStar, Madison, WI) and compared to the HIV-1 reference strain HXB2 sequence. Polymorphisms were defined as differences from HXB2. Phylogenetic trees were constructed with multiple reference strains from the Los Alamos database using the neighbor-joining method in MEGA 4.0 (38) to assess relatedness. Bootstrap analysis using 500 replicates was used to confirm tree topology. Evolutionary distances were determined using the maximum composite likelihood method. Viral subtype was determined using the REGA HIV-1 subtyping tool v2.0 (8).
Ten recombinant viruses were constructed from nine patient samples with common polymorphisms or mutations at sites associated with resistance (19, 20, 37). These included V72I, L74I, T112I, T125M/V, V151I, M154L/I, K156R, V201I, T206S, and S230N/G. To specifically analyze the effect of mutations at the highly variable site T125, two viral clones with polymorphisms T125M and T125V were used from patient B (27 and 28). Of note, selected polymorphisms in the integrase coding region were detected in individual patient clones (see Table Table2)2) but not by consensus sequencing (see Fig. Fig.2).2). They included K42T and N281S in subject F clone 15, R199K in subject A clone 6, N222D in subject R clone 6, and D278G in subject K clone 2. As positive controls, we constructed two recombinant viruses with resistance conferring amino acid substitutions associated with reduced susceptibility to InSTI, one containing T97A and the other N155H, both of which were created by site-directed mutagenesis using overlap PCR as previously described (42).
Six InSTI were selected for the evaluation of the effect of these mutations on inhibitory concentrations. The InSTI represented multiple chemical classes: EVG, a hydroxyquinolone (35); L-870,812 (21) and L-870,810 (19), which are both naphthyridine carboxamide inhibitors; a second generation naphthyridine carboxamide L-900,564 (Fig. (Fig.1),1), and the hydroxypyrimidinone carboxamide inhibitors L-900,525 (Fig. (Fig.1)1) and RAL (18).
The integrase coding region was amplified by RT-PCR as described above from patient plasma samples using modified primers containing an XbaI restriction site at the 5′ end of int and a SacII restriction site near its 3′ end. Each of the integrase coding regions was recombined into an HXB2-derived proviral DNA clone, R7/3(X/S)Bsd (39). Notably, the integrase amino acid sequence of R7/3 differs from HXB2 at two loci, G123S and R127K. The env gene of R7/3(X/S)Bsd is inactivated by a frameshift mutation and the virus is capable of only a single round of infection. In the R7/3(X/S)Bsd plasmid DNA, the XbaI and SacII restriction sites that flank the integrase coding region are unique to the plasmid DNA and are silent with respect to the viral coding frames. This cloning strategy allows for unbiased comparisons to be made between the integrase variants against an otherwise isogenic background. The DNA sequence of the entire recombinant segment was determined for all proviral clones. To generate pseudotyped viral stocks, proviral DNAs were each cotransfected into HEK293T cells with pCI-VSV-G, a plasmid DNA expressing the vesicular stomatitis virus G envelope protein, with PEI (Polysciences, Inc., Warrington, PA). The cells were fed the following day, and virus was harvested 48 h posttransfection. The viral titer was determined by p24 enzyme-linked immunosorbent assay (Coulter, Miami, FL), and samples were stored at −80°C until use.
A single-cycle infectivity assay was employed to determine the relative 50% effective concentration (EC50) of different InSTI against our panel of recombinant HIV-1 variants. TZM-bl cells, an adherent human cell line expressing host receptors CD4, CXCR4, and CCR5 as well as long terminal repeat-driven luciferase and β-galactosidase, were seeded at 8,000 cells per well in 150 μl Dulbecco's modified Eagle's medium (DMEM)/10% fetal bovine serum in a 96-well plate. The supernatant was aspirated 24 h later, and 20 μg of Polybrene in DMEM solution was added to each well. The cells were incubated for 30 min at 37°C. Each InSTI was diluted into multiple concentrations. The Polybrene medium was removed, and 50 μl of serially diluted InSTI was added to each well, with 150 μl of medium, and incubated for 30 min at 37°C. Virus preparations (5 ng p24/50 μl of DMEM) were added to each well and incubated at 37°C. The cells were processed for β-galactosidase expression 48 h later as per Galacto-Star instructions (Applied Biosystems, Framingham, MA). The supernatant was aspirated, and the cells were washed once with phosphate-buffered saline. Lysis solution (60 μl) was added to each well, the plates were left for 30 min at room temperature, and 40 μl of the solution was transferred into a black plate. Diluted (1:50) reaction buffer (60 μl) was added and incubated for 45 min at room temperature. The light signal was then measured using an MLX microplate luminometer (Dynex Technologies, Inc., Chantilly, VA). The samples were run in triplicate, and the experiment was repeated for confirmation.
The analyses were conducted using the statistical packages Stata 10.0 (Statacorp, TX) and R (www.R-project.org). Changes over time in the distribution of specific polymorphisms were analyzed using the chi-square test for trend. Associations between clinical characteristics and polymorphisms at sites V72, T97, V151, M154, V201, and T206 were determined using Fischer's exact test and the Wilcoxon rank test was used for nonparametric continuous data. Associations between certain RT mutations and polymorphisms at sites V165, M154, V72, V151, and T206 were also tested using Fischer's exact test (4).
Drug susceptibility was determined as follows: for each combination of virus and drug, a dose-response curve was obtained, and the EC50 value was determined using Grafit software (Erithacus Software, Horley, United Kingdom). EC50 values were obtained for each virus-drug combination replicate. The comparisons in phenotypic susceptibility between recombinant viruses and reference virus R7/3 for each drug were performed using an analysis of variance model, with the log10 EC50 values as the response and the drug and virus combinations as predictors. Differences in the mean log10 EC50 values and 95% confidence intervals were obtained. These were back transformed and reported as ratios of n-fold change EC50 values. The P values were determined from the Wald test of analysis of variance, without adjustment for multiple testing. All P values are two-sided.
Nucleotide sequences obtained from the patients (ahi1 to ahi112 and mdr1 to mdr15) have been submitted to GenBank under accession numbers FJ786268 to FJ786400.
The AHI participants (n = 112) were mostly male (97.3%), white (91.5%), and U.S. born (87.5%) (Table (Table1).1). Their mean age at the time of the study was 35.5 years. The mean CD4+ T-cell count at the time the samples were drawn was 423 cells/μl (range, 32 to 1,000), and the mean log viral load was 5.4 log10 copies/ml (range, 3.8 to 7.6). The mean estimated time since infection was 60 days. The transmission risk for all subjects was male sexual contact. Proviral DNA was derived from 57 stored PBMCs and viral RNA from 55 plasma samples. The mean viral load (P = 0.19) and CD4+ T-cell counts (P = 0.56) did not differ by nucleic acid sample source. The 15 treatment-experienced subjects with MDR HIV-1 were older (mean age, 47.7 years; P < 0.001) and had lower median CD4+ T-cell counts (65 cells/μl; P < 0.001) and mean log10 viral loads (4.5 log10 copies/ml; P = 0.02).
Most viruses were subtype B (n = 109). Other isolates were subtype CRF02_AG (n = 2) and subtype A1 (n = 1). All MDR viruses were subtype B. Polymorphisms were identified at 121 of 288 (42.0%) amino acid sites (Fig. (Fig.2).2). Most amino acid changes occurred as a result of single nucleotide substitutions. There were no polymorphisms identified in the H12-H16-C40-C43 Zn+-binding motif or in the catalytic D64-D116-E152 triad (41), both of which have been identified as indispensable for integrase function (10). Furthermore, there was no variation at most residues known to interact with LEDGF. For example, we found no variation in A128, A129, W131, W132, Q168, E170, T174, and M178, residues all shown to interact directly with LEDGF (5). However, there were a substantial number of viruses (n = 89) with a D10E amino acid substitution. Though polymorphic, this substitution conserves an acidic amino acid residue which, as revealed by recent structural determination, forms a well-defined salt bridge with the arginine residue at position 405 in LEDGF (16).
There were polymorphisms identified at two sites within the dimeric interface that are essential for strand transfer activity: F181L and F185V (1). The highest degree of variability was identified in the C-terminal domain of the integrase coding region (aa 212 to 288): 38 out of 77 (49.3%) residues were polymorphic versus 61 out of 162 (37.7%) in the CCD (aa 50 to 211) and 21 out of 49 (42.9%) residues in the N-terminal domain (aa 1 to 49) (P = 0.23). However, within the CCD, aa 111 to 126 displayed the greatest variability. Almost no variation was seen from aa 174 to 180, or aa 235 to 250.
Within the subtype B samples, interpatient nucleic acid distances ranged from 0.1% in samples from known sexual partners to 8.5%. MDR samples did not cocluster (Fig. (Fig.33).
Few of the polymorphisms associated with in vitro integrase inhibitor resistance were identified in the samples from newly infected individuals or in MDR patients. We identified polymorphisms V72I, L74I, T97A, V151I, M154I/L, E157Q, V165I, V201I, I203M, T206S, and S230N. V72I (56.3%; n = 63) was the most common mutation among the AHI samples known to be associated with InSTI resistance. Although not statistically significant, the proportion of mutations at resistance conferring sites was greater in the AHI group (26.9%) than in the MDR group (19.5%) (P = 0.15). There were no significant associations seen between age, year of sample, viral load, or time since infection and the presence of V72I, T97A, V151I, M154L, V201I, or T206S.
There was strong evidence of an association between M154L and mutations at RT T215(Y/F/S/D) (P < 0.001), as previously noted by Ceccherini-Silberstein et al. (4). E157Q was the only naturally occurring mutation identified which has been shown to confer resistance to EVG (37), RAL (28), and L-870,810 (37); it was seen in five patients. V151I, which is involved in resistance pathways for L-870,810 (19) and RAL (18), was seen in one patient with AHI and two treatment-experienced patients. V75I, which confers fourfold resistance to GSK-364735 when present with T112S and Q146P (43), was observed in one AHI patient. L74I, T97A, and I203M, which were seen to possibly contribute to resistance to RAL in BENCHMRK-1/2 (6), were also seen in a small number of patients. S230N, which was selected for in serial passage experiments in the presence of L-870,810 (22), was fairly common, present in 19% of treatment-naïve patients and 7% of MDR patients. Contrary to a study of 243 patients from the Los Alamos database, we did not identify any variants with N155H, E138K, Q148H, or S153Y/A (27).
Baseline characteristics of the nine treatment-naïve patients whose primary viral isolates were selected for cloning and analysis are shown in Table Table2.2. The median plasma HIV-1 viral load was 4.9 log10 copies/ml (range, 4.2 to 6.6). A few of the patient-derived recombinants demonstrated significant changes in their susceptibilities to the six compounds (Table (Table3).3). The site-directed mutant with the known InSTI resistance-conferring amino acid substitution T97A had a sixfold decreased susceptibility to L-870,810 (P = 0.001) and a fourfold decrease in susceptibility to L-870,812 (P = 0.02). There were 2.9- and 1.9-fold reductions in susceptibility to EVG (P = 0.06) and RAL (P = 0.26), respectively, though neither of these affects achieved statistical significance. The other positive control, N155H, showed marked resistance to EVG (18-fold decrease in susceptibility; P < 0.001). There was a 13-fold decrease in susceptibility to L-870,810 (P < 0.001), a near 3-fold decrease in susceptibility to L-870,812 (P = 0.06), a slightly greater than 4-fold decrease in susceptibility to L-900,564 (P = 0.003), and a near 3-fold decrease to RAL (P = 0.03).
None of the patient-derived viruses studied demonstrated a significant decrease in susceptibility to the two compounds currently in clinical use (RAL or EVG). Furthermore, in total, none of the recombinant viruses tested revealed reduced susceptibilities to the panel of compounds tested. Changes in susceptibilities to the panel of InSTI were uniformly less than twofold (Table (Table33).
The addition of integrase inhibitors to the armamentarium of antiviral agents is a considerable advance in the management of patients infected with HIV-1. We undertook this study to determine whether inherent resistance to these new agents existed and, if so, whether there would be clinical implications. To answer these questions, we performed a sequence analysis of the integrase coding region in viruses from 112 acute and early infected individuals and 15 chronically infected patients with multidrug-resistant HIV-1. To understand whether amino acid substitutions in the integrase coding region were associated with changes in susceptibility, we constructed a panel of recombinants and tested susceptibility to a panel of InSTI in vitro.
Despite documenting substantial numbers of polymorphisms in the integrase coding regions similar to those seen in other analyses (1a, 29, 34), we did not discover amino acid substitutions associated with high-level resistance to the clinically relevant inhibitors RAL and EVG. Of note, there was an amount of amino acid variability similar to that seen in HIV-1 protease (47.5%) (26); however, at many positions, less than 2% of the patient isolates exhibited variation. Although none of the major mutations associated with primary resistance to InSTI in clinical use were identified (T66I, E92Q, Y143R/C, Q148K/R/H, and N155H), E157Q was observed in 5 out of 112 (4.5%) patients with AHI. This mutation has also been identified in a patient with RAL treatment failure (28) and has been selected by viral serial passage experiments in vitro in the presence of EVG (37). The treatment failure isolate was almost completely inactive in terms of 3′ processing and strand transfer when the purified mutant integrase protein was assayed for its catalytic function in vitro. In contrast, the mutant E157Q integrase derived from EVG-resistant virus obtained through serial passage was only slightly defective for strand transfer activity and partially restored activity to another strain when in combination with H51Y/E92Q/S147Q. It is therefore unclear whether this naturally occurring mutation impairs fitness or compensates for other mutations (42). As previously documented in studies of HIV-1 protease and the emergence of resistance to protease inhibitors, preexisting polymorphisms such as L63P also emerged during treatment and were deemed to be compensatory in the setting of primary resistance mutations. Integrase mutations V72I (19, 37, 40) and V201I (12) were highly prevalent, as was T206S (34). However, mutations associated with secondary resistance to clinically relevant drugs were less common: V151I was found in three patients; L74I, V75I, and T97A were found in one patient each; and I203M was found in eight patients (6). There were also multiple polymorphisms of unknown impact at sites associated with resistance (T125, G163, and S230). Although these mutations are believed to modulate the effects of the mutations associated with major drug resistance, further studies are required to better understand their contribution to viral fitness and mechanisms by which compensation may occur.
We sought to understand whether a subset of the observed polymorphisms were associated with changes in drug susceptibility to a diverse panel of InSTI. All recombinant viruses were susceptible to the panel, particularly the clinically relevant inhibitors RAL and EVG. Notably the recombinant virus containing V151I, contrary to a recent study (37), did not demonstrate any decrease in susceptibility to any compound. The effect of V151I on susceptibility is inconsistent and may depend on the plasmid backbone used (15, 19, 25). It is also possible that other integrase polymorphisms within the isolate might have altered its response to the InSTI tested.
T97A, which has previously been identified in the resistance pathways of different InSTI (6, 14), was associated with an increase in EC50 values for three compounds from different classes: L-870,812, L-870,810, and EVG. This mutation, seen in only one of our patients, has been reported to be as prevalent as 2.3% (14) and therefore might have clinical implications for the use of InSTI.
Structural aspects of the integration complex remain poorly defined. Thus, any interpretation of the impact substitution mutations might have on integrase secondary, tertiary, or quaternary structure as it relates to strand transfer catalysis is worded with caution. We did not identify any mutations in areas deemed indispensable to integrase function (2, 41, 33); however, we have documented polymorphisms (K136N/Q/T and K156N/R) thought to be at the interface between integrase and the viral DNA ends (23, 32). Furthermore, amino acid positions of highest variability (Fig. (Fig.2),2), A124 and T125, have been previously defined as residues that alter InSTI susceptibility, exhibiting clade-specific differences in sequence and resistance to InSTI subsets (29). Indeed, patterns of covariation within the integrase coding region (34) and in other areas of the pol gene (4) should help us determine the effect of these mutations on integrase function and reveal compensatory mechanisms used by the virus to escape strand transfer inhibition.
Taken together, we believe that these data support the current contention that integrase inhibitor naïve subjects are likely to benefit fully from the incorporation of the current clinically relevant agents into a treatment regimen, without concern of inherent reduced susceptibility. It would appear that the integrase coding region contains as much natural variation as that of protease but that mutations associated with high-level resistance to existing InSTI are rarely if ever present in integrase-naïve patients. Most of the highly prevalent polymorphisms have little effect on InSTI susceptibility in the absence of specific primary mutations. However, the importance of these polymorphisms in determining barriers to resistance and genetic pathways needs to be determined by longitudinal genotypic evaluation of patients exposed to InSTI who either fail to respond or respond initially and subsequently fail. Recent reports have suggested the importance of T97A, K156N, G163R, or V165I (3). It must be emphasized that as the use of these potent agents expands, the potential for the transmission of InSTI-resistant viruses will increase and that it is therefore critical to include integrase genotypic and phenotypic analyses in studies that focus on the transmission of drug-resistant HIV-1 variants.
This work was supported by the National Institutes of Health grants AI041534 and AI047033; the Rockefeller University CCTA grant number UL1 RR024143 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH); and NIH Roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH.
We acknowledge Gilead Sciences for providing EVG and Damian McColl for critical review of the manuscript; John Wai at Merck Research Laboratories for the provision of compounds, chemical structures, and scientific guidance; the technical support of Wendy Chen for assistance in the preparation of figures and tables; the clinical staff at the Rockefeller University Hospital; and the study participants and their referring primary-care providers.
Published ahead of print on 3 August 2009.