icon star paper   HIV Articles  
Back grey_arrow_rt.gif
 
 
Atazanavir (Reyataz) and the I50L Mutation
 
 
  "Identification of I50L as the Signature Atazanavir (ATV)Resistance Mutation in Treatment-Naive HIV-1Infected Patients Receiving ATV-Containing Regimens"
 
The Journal of Infectious Diseases 2004;189:1802-1810
 
Richard Colonno,1 Ronald Rose,1 Colin McLaren,1 Alexandra Thiry,1 Neil Parkin,2 and Jacques Friborg1
 
1Bristol-Myers Squibb Pharmaceutical Research Institute, Wallingford, Connecticut; 2ViroLogic, South San Francisco, California
 
ABSTRACT
 
Atazanavir (ATV) is a once-daily human immunodeficiency virus (HIV) protease inhibitor (PI) shown to be effective and well tolerated.
 
ATV has a distinct resistance profile relative to other PIs, with susceptibility maintained against 86% of isolates resistant to 12 PIs.
 
Clinical isolates obtained from PI-naive patients designated as experiencing virologic failure while receiving ATV-containing regimens contained a unique isoleucine-to-leucine substitution at amino acid residue 50 (I50L) of the HIV-1 protease.
 
The I50L substitution, observed in all isolates exhibiting phenotypic resistance to ATV, emerged in a variety of different backgrounds and was most frequently accompanied by A71V, K45R, and/or G73S.
 
Viruses containing an I50L substitution were growth impaired, displayed ATV-specific resistance, and had increased susceptibilities (0.4 of reference strain) to other PIs.
 
Comparison of viruses bearing I50L with those bearing I50V revealed specific resistance to ATV and amprenavir, respectively, with no evidence of cross-resistance. The unique I50L substitution is the signature mutation for resistance to ATV.
 
The clinical significance of the observations described here have yet to be determined. It appears that future PI-treatment options would be preserved with the emergence of the I50L substitution, but the clinical relevance of the observed increases in susceptibility to other PIs, and whether continued treatment with ATV might be required to maintain selective pressure for the I50L substitution will need to be defined in future clinical studies.
 
INTRODUCTION
 
HIV protease (PR) inhibitors (PIs) are potent and effective antiretrovirals. However, their extensive use has led to the emergence of HIV-1 variants exhibiting cross-resistance to multiple PIs. The correlation between genotypic changes within the PR gene and phenotypic resistance remains poorly understood, and secondary substitutions appear to play a major role in expression of a resistance phenotype.
 
Atazanavir (ATV; Reyataz; BMS-232632) is a once-daily HIV-1 PI. ATV has a 50% effective concentration (EC50) of 35 nmol/L against a variety of HIV-1 isolates in different cell types and is a highly selective and effective inhibitor of the HIV-1 PR in vitro (Ki, <1 nmol/L) [7]. In vitro passage of HIV-1 in the presence of ATV results in the selection of resistant variants. Genotypic analysis of 3 different HIV-1 strains with in vitro selected resistance to ATV indicated that an N88S substitution in the viral PR appeared first during the selection process in 2 of the 3 strains. An I84V change appeared to be an important substitution in the third strain, along with an isoleucine (Ile)toleucine (Leu) change at amino acid residue 50 (I50L), and all 3 variants required multiple changes to achieve significant resistance (>10-fold decrease in susceptibility) levels. Substitutions were also observed at the PR cleavage sites, after drug selection. The evolution to resistance was somewhat distinct for each of the 3 strains used, suggesting that multiple pathways to resistance are possible and confirming the importance of viral genetic background in development to resistance.
 
The susceptibility profile of ATV was subsequently determined by use of a panel of 950 clinical isolates exhibiting a wide array of PI-resistance profiles and genotypic patterns. In general, reductions in susceptibility to ATV required several amino acid changes and were modest in degree, and susceptibility was retained among isolates resistant to 1 or 2 other PIs. Although ATV displayed a distinct resistance pattern relative to the 6 PIs tested, there was a clear trend toward loss of susceptibility to ATV as isolates exhibited increasing levels of cross-resistance to multiple PIs, with the percentage of isolates resistant to 1, 2, 3, 4, or 5 PIs remaining susceptible to ATV determined to be 88%, 81%, 34%, 16% and 5%, respectively. Genotypic analysis of 943 PI-susceptible and -resistant clinical isolates identified a correlation between the presence of specific amino acid substitutions in the HIV PR (L10I/V/F, K20R/M/I, L24I, L33I/F/V, M36I/L/V, M46I/L, G48V, I54V/L, L63P, A71V/T/I, G73C/S/T/A, V82A/F/S/T, I84V, and L90M) and decreased susceptibility to ATV. Although no single substitution or combination of substitutions appears to be predictive of resistance to ATV, the presence of at least 4 of these substitutions correlated strongly with decreased susceptibility to ATV. The primary objective of the present study was to characterize the phenotypic profile and emerging genotypes of clinical isolates from PI-naive patients who were designated as having virologic failure and who developed resistance to ATV while receiving ATV-containing regimens.
 
Study population. Virus samples (obtained at baseline and during treatment, where available) were obtained from patients who experienced virologic failure while enrolled in the AI424-007/041 (ATV plus stavudine [d4T] and didanosine [ddI]), AI424-008/044 (ATV plus d4T and lamivudine [3TC]), and AI424-034 (ATV or efavirenz [EFV] in combination with zidovudine and 3TC) studies. Subjects who experienced a rebound in virus load (generally defined as a confirmed increase in HIV RNA to levels >400 copies/mL) and nonresponders (those who did not achieve HIV RNA levels <400 copies/mL by week 16 or 24 [study AI424-034]) were designated as having experienced virologic failure. They had paired plasma specimens (obtained at baseline and during treatment) evaluated for phenotypic and genotypic changes. In general, virologic-failure specimens were selected for further resistance evaluation only if they contained >1000 HIV RNA copies/mL and there was no indication that the specimen came from a patient at a time of transient nonadherence to study drug(s). During-treatment samples from patients experiencing a confirmed rebound in virus load were obtained at the first visit at which the rebound was apparent; if the HIV RNA level of that sample was <1000 copies/mL, the next available sample with HIV RNA >1000 copies/mL was used. For nonresponders, samples from week 16 or 24 were tested. Drug adherence was determined according to pill counts of returned study medications at each scheduled visit.
 
Drug-susceptibility assays. Clinical isolates were subjected to phenotypic (PhenoSense) and genotypic (GeneSeq) analysis by Virologic (South San Francisco, CA), as described elsewhere. Over the time period (several months) in which the samples reported here were tested, the ATV IC50 measured for the NL4-3 reference strain was 0.92.0 nmol/L. The success rate in obtaining phenotypic and genotypic results for patients experiencing virologic failure was 30%50%, largely because of low levels of HIV RNA (the assays used required levels of 500 copies/mL), insufficient sample volume, or unavailability of a sample.
 
Recombinant viruses (generated separately from those made for PhenoSense assays) were tested in an MT-2 cell susceptibility assay at Bristol-Myers Squibb. MT-2 cells were infected with recombinant viruses, at an MOI of 0.005 TCID50/cell, followed by incubation in the presence or absence of serially diluted PIs for 47 days. Virus yields were quantified by use of a reverse transcription assay, as described elsewhere. The results from at least 4 independent experiments were used to calculate the EC50.
 
RESULTS
 
Identification of I50L as the signature mutation for resistance to ATV. Previous in vitro studies indicated that ATV has an overall resistance profile distinct from that of other PIs. To better understand how viruses become resistant in ATV-treated patients, clinical isolates from treatment-naive patients experiencing virologic failure while enrolled in clinical studies AI424-007/041, AI424-008/044, and AI424-034 were analyzed. Overall, the detection of ATV-resistant variants appeared to be low (2% of treated patients) and was comparable to, if not somewhat lower than, that of the study control drugs nelfinavir (NFV) and EFV. Resistance to ATV was defined as a >2-fold decrease in susceptibility to ATV from baseline, a fold change (FC) of >2.5 relative to the NL4-3 reference strain, or evidence of the I50L mutation. Resistance levels for NFV and EFV were also defined as an FC of >2.5.
 
Phenotypic analysis of virus samples from the 208 patients designated as having virologic failure while receiving ATV-containing regimens initially identified 14 pairs of isolates whose during-treatment virus exhibited a >2-fold decrease in susceptibility to ATV relative to baseline. Genotypic analysis of these isolates showed that all isolates obtained during treatment contained an I50L substitution in the absence of other recognized primary mutations within the viral PR. Nine isolates also showed the emergence of an A71V substitution. The I50L substitution has not previously been reported in isolates from patients, whereas A71V is a common secondary PI mutation and naturally occurring polymorphism. An additional 4 isolates obtained during treatment from patients designated as having virologic failure also exhibited an ATV-specific resistance phenotype and had the I50L substitution; however, baseline isolates were unavailable. Three additional isolates with an I50I/L mixture at amino acid residue 50 and susceptible phenotypes, as well as 3 other during-treatment I50L-containing isolates for which no phenotypic data were available, were reported from patients not designated as having virologic failure. Thus, there are 23 isolates with either genotypic and/or phenotypic data implicating an I50L substitution as a primary substitution in resistance to ATV.
 
Amino acid substitutions in the P7-P1 and P1-P6 processing sites of the gag precursor have been detected in vitro and in vivo during treatment with HIV-1 PIs. To determine whether similar mutations at the P7-P1 and P1-P6 cleavage sites developed during treatment with ATV, we examined the gag sequence that spans the P7-P1-P6 junctions of the 7 available ATV-resistant viruses from study AI424-007. The substitution A431V at the P2 position of the P7-P1 cleavage site associated with resistance to indinavir (IDV) and ritonavir (RTV) emerged in 3 viruses during treatment, whereas changes were not observed at these cleavage sites in the remaining 4 viruses. These results suggest that cleavage-site changes are not required for the expression of the resistance to ATV phenotype.
 
The appearance of I50L was noted as early as week 16 of treatment, although the mean and median time receiving treatment to emergence of this substitution were 62 and 54 weeks, respectively, and the mean time to virologic failure in this group of patients (n = 10) was 50 weeks. ATV phenotypic resistance levels were 1.529, relative to the NL4-3 reference strain, with a median FC of 9.6, with increased susceptibility to the other PIs tested consistently noted. Resistance to 3TC related to the emergence of an M184I/V substitution was observed in all 11 I50L isolates from patients designated as having virologic failure whose treatment regimen contained 3TC.
 
The emergence of I50L occurred in a variety of genetic backgrounds, with baseline isolates averaging nearly 7 PR substitutions, including L10F/I/V, K20R, M36I/L, and L63P. The I50L substitution was frequently accompanied by changes at residues A71V (52%), K45R/Q/N (30%), and G73S (26%) and, to a lesser degree, with the substitutions M46I/L, I64V, and V82A (13% for each). The strong relationship between the emergence of I50L and the development of resistance to ATV was further supported by 2 isolates in which the progression of I50 to I50I/L to I50L correlated directly with increasing levels of resistance to ATV (data not shown). Where viral subtype information was available, I50L-containing isolates were observed among both subtype B (n = 16) and subtype C (n = 6) viruses.
 
Effect of I50L on susceptibility to ATV and other PIs. The emergence of I50L correlated with ATV-specific resistance, whereas susceptibility to other PIs was either unchanged or increased. The median decrease in susceptibility to ATV from baseline was 9.6-fold (n = 15), whereas the median susceptibility from baseline increased by 1.5-fold (n = 15) for amprenavir (APV), 2.0-fold (n = 9) for IDV, 3.2-fold (n = 4) for lopinavir (LPV), 1.6-fold (n = 15) for NFV, 4.1-fold (n = 15) for RTV, and 2.0-fold (n = 15) for saquinavir (SQV). APV and NFV displayed more-modest increases in susceptibility, compared with the other PIs, although most APV FC values were <1 at baseline. Susceptibility (FC) levels, relative to the reference strain, reached 0.4-fold in 29%100% of the I50L isolates for APV (43%), IDV (46%), LPV (100%), NFV (29%), RTV (81%), and SQV (67%). In contrast, the percentage of baseline isolates exhibiting an FC <0.4 ranged from 21% for APV to either 11% (for RTV and SQV) or 0% (for IDV, LPV, and NFV). Two isolates obtained during treatment, exhibiting ATV FC values <1, contained I50I/L mixtures. The 1 isolate displaying an FC of only 1.5 exhibited a baseline FC of 0.2 before developing the I50L substitution.
 
ATV-specific resistance maps exclusively to the I50L substitution. Direct evidence for the role of the I50L substitution in resistance to ATV was obtained by use of a series of clonal NL4-3 recombinant viruses containing only the PR gene from several resistant clinical isolates originating in study AI424-007. The ATV genotypes and phenotypes in MT-2 cells of this panel of recombinant viruses are shown in table 2. These results demonstrate that the ATV-resistant phenotype is carried within the coding sequence of the viral PR and does not require either cleavage-site changes or modifications in other viral proteins. To further demonstrate the specific effect of the I50L substitution on resistance to ATV, the I50L and A71V substitutions alone and in combination were inserted into the wt PR gene of 3 recombinant virus strains. Decreases in susceptibility to ATV of 2.15.4-fold were observed in all 3 viral backbones containing only the I50L substitution, and 5.710-fold with the combination of I50L and A71V. In contrast, the presence of the A71V substitution alone resulted in changes in susceptibility to ATV of 0.63.7-fold. These results provide the most-convincing evidence that the I50L substitution alone is critical to expression of the observed ATV-resistance phenotype and that the degree of resistance is enhanced by secondary substitutions, such as A71V. Overall, the results strongly suggest that the I50L substitution, sometimes combined with an A71V or other substitutions, is the signature mutation for resistance to ATV.
 
These results demonstrate that the ATV-resistant phenotype is carried within the coding sequence of the viral PR and does not require either cleavage-site changes or modifications in other viral proteins. To further demonstrate the specific effect of the I50L substitution on resistance to ATV, the I50L and A71V substitutions alone and in combination were inserted into the wt PR gene of 3 recombinant virus strains. Decreases in susceptibility to ATV of 2.15.4-fold were observed in all 3 viral backbones containing only the I50L substitution, and 5.710-fold with the combination of I50L and A71V. In contrast, the presence of the A71V substitution alone resulted in changes in susceptibility to ATV of 0.63.7-fold. These results provide the most-convincing evidence that the I50L substitution alone is critical to expression of the observed ATV-resistance phenotype and that the degree of resistance is enhanced by secondary substitutions, such as A71V. Overall, the results strongly suggest that the I50L substitution, sometimes combined with an A71V or other substitutions, is the signature mutation for resistance to ATV.
 
Phenotypic profiling of these recombinant viruses was performed to determine whether the I50L-induced resistance to ATV was also responsible for the enhanced susceptibility to other PIs. Figure 2 displays the change in susceptibility to all PIs, compared with the NL4-3 reference strain for each of NL4-3 recombinant viruses listed in table 2. Results closely resemble what was previously observed with clinical isolates (figure 1) ATV-specific resistance and a strong tendency toward increased susceptibility to the other PIs. Recombinant clones containing the baseline PR genes for these resistant isolates showed little change from the NL4-3 reference strain (data not shown). Moreover, enhanced susceptibility to APV, IDV, LPV, NFV, RTV, and SQV was observed in all 3 virus strains when the I50L or I50L/A71V substitutions were introduced by site-directed mutagenesis. In many cases, the increased susceptibilities of these recombinant viruses were 10-fold of the reference viruses.
 
Stability of viruses containing the I50L substitution. Genotypic and phenotypic progression after the emergence of an I50L substitution was monitored in a small series of isolates derived from 4 patients who continued to receive ATV-containing regimens. Results show that continued treatment with ATV did not result in rapid progression to broad cross-resistance or accumulation of additional amino acid changes, despite continued treatment with ATV for 1252 weeks. Apart from patient 100, who continued receiving treatment with ATV for an additional year, the observed genotypic changes appeared to be relatively benign and consistent with the susceptibility phenotypes that were available. The newly emerging K45R, V82A, and I93L substitutions observed in patient 100 were potentially of greater concern, because the V82A substitution is usually associated with cross-resistance to multiple PIs. However, all 4 patients showed stable virus loads (13506659 HIV RNA copies/mL). Clearly, additional data will need to be obtained before any definitive conclusions can be reached.
 
No evidence of cross-resistance between ATV and APV. Because an I50V substitution is a signature mutation sometimes associated with resistance to APV the cross-resistance relationship between ATV and APV was further examined. As noted above, none of the I50L-containing clinical isolates showed evidence of resistance to APV. A reciprocal evaluation was performed on a small panel of 8 clinical isolates that contained an I50V substitution and were resistant (4.214-fold) to APV and either LPV or RTV, since isolates displaying APV-specific resistance were not available. Seven of the 8 isolates resistant to APV had an ATV FC <1.0, and 3 isolates had an ATV FC 0.4. Thus, a specific change at amino acid residue 50 to either Leu or valine (Val), in the absence of other key primary and secondary substitutions, modulates resistance to ATV and APV, respectively.
 
Comparative studies of I50L- and D30N-containing viruses. The D30N substitution is a key signature mutation for NFV and is reported to have little effect on susceptibility to other PIs. An important distinction between the I50L and D30N substitutions is that the I50L change, to date, has been observed 100% of the time when resistance to ATV phenotype was evident in the treatment-naive patients. In contrast, the D30N resistance pathway accounts for resistance to NFV 50% of the time. A similar outcome was also observed in clinical studies AI424-007 and AI424-008, in which NFV was used as the PI control. Comparative studies were performed on the 18 phenotyped viruses containing an I50L substitution and on 75 clinical isolates containing a D30N substitution, without evidence of cross-resistance (FC, 2.5) to other PIs. Genotypic backgrounds for these 2 sets of viruses were similar, with a wide variety of additional substitutions present (data not shown). The only statistically significant changes (P < .001, Fisher's exact test) were at amino acid residues 13, 30, and 88 for NFV and at amino acid residues 50, 73, and 89 for ATV. The median FC value change from wt virus, for each of these isolates versus the 7 approved PIs, is presented in figure 3. Although the I50L isolates displayed selective resistance to ATV and increased susceptibility to all of the other 6 PIs, the D30N-containing isolates showed resistance to NFV and small median decreases in susceptibility to both ATV and IDV. The median FC of the D30N-containing viruses was never <0.4 for any of the other PIs, with only the level of susceptibility to APV reaching an FC as low as 0.6. In contrast, viruses containing the I50L substitution had median FC values 0.4 for IDV, LPV, RTV, and SQV. These results suggest that the emergence of either an I50L or D30N substitution in this group of viruses will likely preserve future PI treatment options, but that the I50L substitution has a greater potential to increase susceptibilities to other PIs. The clinical relevance of this finding is currently unknown.
 
I50L-containing viruses are growth impaired. Significant differences in HIV replication kinetics can be observed in parallel infections by measuring the amount of virus production over time via the detection of p24 antigen. To determine whether the I50L substitution has an effect on viral replication, growth curves for the NL4-3 and RF recombinant viruses described in table were compared with their respective wt viruses. In each case, insertion of the I50L substitution in either PR backbone resulted in a significant delay in viral replication, whereas addition of the A71V change with I50L restored some viability. Insertion of the A71V substitution alone, however, had no significant effect on the rates of viral growth. The single-cycle replicative capacities (RCs) of several I50L-containing clinical isolates were also determined and ranged from 0.3% to 42% of the reference strain, and, in the 3 cases in which paired isolates obtained at baseline and during treatment were available, the RC shifted from 92% to 4.2%, 150% to 12%, and 85% to 37% (authors' unpublished data). Taken together, these data suggest that the emergence of the I50L substitution impairs viral growth, most likely through a reduction in the enzymatic activity of the viral PR. The A71V change appears to be a compensatory substitution in this regard. The degree to which the impaired viral growth accounts for the observed increased susceptibility to other PIs when an I50L substitution is present is unknown.
 
DISCUSSION
 
ATV is a potent inhibitor of HIV-1 PR, with demonstrated effectiveness in clinical trials comparable to that of standard-of-care regimens, such as EFV or NFV. Its excellent oral bioavailability and pharmacokinetic profile enable once-daily dosing and a low pill burden, in the absence of added RTV. Previous studies involving a large panel of viruses resistant to other PIs demonstrated that ATV has a distinct resistance profile relative to the other available PIs. The present study was undertaken to identify the resistance markers and patterns of resistance in previously treatment-naive patients receiving ATV-containing regimens. Overall, virus variants resistant to ATV emerged infrequently in these patients. Despite the limited number of ATV-resistant isolates available, an obvious correlation became apparent between the development of ATV phenotypic resistance in patients designated as having virologic failure while receiving ATV-containing regimens and the emergence of a novel I50L substitution.
 
The I50L substitution is the signature resistance mutation for ATV. The appearance of the I50L substitution is frequently accompanied by A71V, a secondary substitution also observed among isolates resistant to other PIs. Apart from a potential relationship with the K45R and G73S substitutions, none of the other observed amino acid changes appeared to correlate with resistance to ATV. Supporting this conclusion is the observation that an ATV-resistant phenotype is expressed in all recombinant viral clones containing the I50L substitution in a variety of genetic backgrounds and independent of the A71V, K45R, or G73S substitutions. The finding that recombinant viruses carrying only the I50L substitution express ATV-specific resistance also demonstrates that the cleavage site changes observed in vitro and in isolates from patients are not required for expression of this resistance phenotype. However, the magnitude of resistance remains relatively low in the absence of other PR substitutions, and an I50L substitution has not been observed in clinical isolates in the absence of other substitutions.
 
Amino acid residue 50 is located in the flap region of the HIV PR and plays a key role in enzymatic function and PI binding. Seemingly small modifications, such as a change from the normal Ile to Leu or Val, can result in resistance to ATV or APV, and, in the case of Leu, appears to significantly enhance susceptibilities to the other PIs. The structural and molecular basis for this interesting observation are under study.
 
The finding of an exclusive pathway to resistance among virus isolates from previously treatment-naive patients who develop phenotypic resistance to ATV was somewhat surprising, since earlier in vitro selection studies predicted the possibility that multiple pathways exist and that an N88S substitution would be a primary resistance marker. Those studies did, however, identify an I50L resistance pathway and demonstrated that viruses containing the I50L substitution were particularly susceptible to other PIs. The increases in susceptibility are particularly noteworthy, because they resulted in a high percentage of isolates having susceptibilities to PI, as measured by FC values 0.4. Clearly, selection of the I50L substitution is the most efficient in vivo pathway to escape the inhibitory pressure of ATV. Despite the decreased viral fitness resulting from selection of an I50L change, initial results suggest that viruses containing an I50L substitution do not appear to rapidly accumulate additional amino acid changes or become cross-resistant to multiple PIs. Of interest, the I50L substitution has also been observed (at a reduced frequency) in treatment-experienced patients treated with ATV-containing regimens (authors' unpublished data).
 
The frequency and magnitude of the observed I50L substitutions are distinct from those observed for I50V and D30N, the signature resistance mutations for APV and NFV, respectively. Viruses containing I50V or I84V substitutions displayed the greatest reductions in susceptibility to APV. Four distinct genetic pathways to resistance to APV have been described among clinical isolates from patients treated with APV: I50V, I54L/M, I84V, and V32I plus I47V. Apart from the I84V pathway, the other 3 pathways to resistance to APV occur with nearly equal frequency, and the substitution observed most frequently in the presence of I50V was M46I/L. Although minimal cross-resistance to other PIs is observed, I50V contributes to reduced susceptibility to both RTV and LPV and is associated with increased susceptibility to SQV and IDV in vitro. In addition to I50V, selection of I54L/M appears to be somewhat unique to APV. With regard to the development of resistance to NFV, a D30N substitution emerges in 50% of clinical isolates from patients receiving NFV regimens for a median of 13 weeks, with a trend toward accumulation of additional amino acid changes at residues 35, 36, 46, 71, 77, and 88. The N88S substitution emerges in 20% of treated patients and can result in increased susceptibility to APV, although this effect can be neutralized by the presence of other substitutions, such as M46I/L, L63P, and V77I. Isolates containing the D30N substitution as the only primary PI mutation remained susceptible to APV, IDV, RTV, and SQV.
 
The clinical significance of the observations described here have yet to be determined. It appears that future PI-treatment options would be preserved with the emergence of the I50L substitution, but the clinical relevance of the observed increases in susceptibility to other PIs, and whether continued treatment with ATV might be required to maintain selective pressure for the I50L substitution will need to be defined in future clinical studies.
 
 
 
 
 
  icon paper stack View Older Articles   Back to Top   www.natap.org