Skip to main page content

• HOME
• Current Issue
• Archive
• Alerts
• About ASM
• Contact us
• Tech Support
• Journals.ASM.org

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Baxter, J. D. Articles by Mayers, D. L. Search for Related Content PubMed PubMed Citation Articles by Baxter, J. D. Articles by Mayers, D. L.

 Previous Article  |  Next Article 

Journal of Virology, November 2006, p. 10794-10801, Vol. 80, No. 21
0022-538X/06/$08.00+0     doi:10.1128/JVI.00712-06
Copyright © 2006, American Society for Microbiology. All Rights Reserved.

Genotypic Changes in Human Immunodeficiency Virus Type 1 Protease Associated with Reduced Susceptibility and Virologic Response to the Protease Inhibitor Tipranavir

John D. Baxter,^1* Jonathan M. Schapiro,² Charles A. B. Boucher,³ Veronika M. Kohlbrenner,⁴ David B. Hall,⁴ Joseph R. Scherer,⁴ and Douglas L. Mayers⁴

Cooper University Hospital/UMDNJ-Robert Wood Johnson Medical School, Camden, New Jersey,¹ Stanford University, Stanford, California,² Eijkman-Winkler Center, University Medical Center Utrecht, Utrecht, The Netherlands,³ Boehringer-Ingelheim Pharmaceuticals, Inc., Ridgefield, Connecticut⁴

Received 7 April 2006/ Accepted 14 August 2006

Top Abstract Introduction Materials and Methods Results Discussion References

 
Tipranavir is a novel, nonpeptidic protease inhibitor of human immunodeficiency virus type 1 (HIV-1) with activity against clinical HIV-1 isolates from treatment-experienced patients. HIV-1 genotypic and phenotypic data from phase II and III clinical trials of tipranavir with protease inhibitor-experienced patients were analyzed to determine the association of protease mutations with reduced susceptibility and virologic response to tipranavir. Specific protease mutations were identified based on stepwise multiple-regression analyses of phase II study data sets. Validation included analyses of phase III study data sets to determine if the same mutations would be selected and to assess how these mutations contribute to multiple-regression models of tipranavir-related phenotype and of virologic response. A tipranavir mutation score was developed from these analyses, which consisted of a unique string of 16 protease positions and 21 mutations (10V, 13V, 20M/R/V, 33F, 35G, 36I, 43T, 46L, 47V, 54A/M/V, 58E, 69K, 74P, 82L/T, 83D, and 84V). HIV-1 isolates displaying an increasing number of these tipranavir resistance-associated mutations had a reduced phenotypic susceptibility and virologic response to tipranavir. Regression models for predicting virologic response in phase III trials revealed that each point in the tipranavir score was associated with a 0.16-log₁₀ copies/ml-lower virologic response to tipranavir at week 24 of treatment. A lower number of points in the tipranavir score and a greater number of active drugs in the background regimen were predictive of virologic success. These analyses demonstrate that the tipranavir mutation score is a potentially valuable tool for predicting the virologic response to tipranavir in protease inhibitor-experienced patients.

Top Abstract Introduction Materials and Methods Results Discussion References

 
Current treatment of human immunodeficiency virus (HIV) infection involves the concomitant administration of at least three antiretroviral medications among the classes of nonnucleoside reverse transcriptase inhibitors (NNRTIs), nucleoside reverse transcriptase inhibitors (NRTIs), protease inhibitors (PIs), and to a lesser extent entry inhibitors (1a). Antiretroviral drugs have limited efficacy when used as monotherapy, in part due to the rapid emergence of drug resistance mutations that often have a significant impact on the efficacy of other members of that same class of inhibitors (13). This effect is most apparent within the class of NNRTIs, for which a single mutation in viral reverse transcriptase can confer cross-resistance to all other members of the current NNRTI class. Various degrees of cross-resistance among the PIs have been observed. This has resulted in limited or no options for the use of another PI following multiple PI failure. There is clearly a need for novel PIs with activity against PI-resistant HIV type 1 (HIV-1) and distinct patterns of PI cross-resistance. Tipranavir demonstrates unique characteristics that offer potential therapeutic advantages to PI-experienced patients.

Tipranavir is a novel, nonpeptidic inhibitor of the HIV-1 protease manufactured by Boehringer-Ingelheim Pharmaceuticals, Inc. Tipranavir's antiviral activity was evaluated with several cell culture systems and against clinical HIV-1 strains, and it was found to be a potent inhibitor of wild-type HIV-1 replication, with 50% inhibitory concentrations (IC₅₀s) ranging from 0.03 to 0.07 µM and 90% inhibitory concentrations ranging from 0.07 to 0.18 µM (1, 12, 17, 19). In vitro passage studies showed that the development of resistance to tipranavir is slow, requiring up to 9 months in culture (5). The presence of six mutations, I13V, V32I, L33F, K45I, V82L, and I84V, was required to confer a >10-fold decrease in tipranavir susceptibility. The long duration of time combined with the high number of genotypic mutations required to evolve resistance to tipranavir in vitro suggest a high genetic barrier for tipranavir.

An in vitro study using 105 clinical HIV-1 isolates from patients with prior exposure to other PIs demonstrated that tipranavir retains activity against more than 90% of isolates resistant to other PIs (11). A phase II clinical trial of ritonavir-boosted tipranavir in highly treatment-experienced patients with documented PI resistance demonstrated the potent antiviral activity of ritonavir-boosted tipranavir (10). Subsequently, phase III clinical trials demonstrated significantly greater rates of virologic and immunologic responses for ritonavir-boosted tipranavir compared to other ritonavir-boosted PIs (2, 7). These data indicate that the resistance profile of tipranavir is distinct from that of other PIs in that mutations that lead to resistance to other PIs do not have the same impact on tipranavir susceptibility.

In this analysis, we describe the relationship between genotypic changes in HIV-1 protease and tipranavir phenotypic susceptibility in HIV-1 isolates from highly treatment-experienced patients. We also examine the relationship between specific protease mutations and the virologic response to tipranavir in phase II and III clinical trials. The objective of this analysis is to identify a set of HIV-1 protease mutations that can be used to predict either susceptibility or response to ritonavir-boosted tipranavir treatment.

Top Abstract Introduction Materials and Methods Results Discussion References

 
Viral isolates. HIV-1 resistance data were analyzed from six clinical trials conducted in North America, Latin America, and Europe with HIV-positive adult patients to evaluate the efficacy and safety of tipranavir boosted with ritonavir. Three phase II dose-ranging trials (1182.2, 1182.4, and 1182.52), two phase III controlled trials (1182.12 and 1182.48), and a phase IIb/III pharmacokinetic and safety trial of dually boosted PI use (1182.51) were included. Genotyping and phenotypic resistance testing was conducted with stored baseline plasma samples of patients entered in trials 1182.2 and 1182.4. In all other trials, baseline genotyping was conducted during the screening phase. All baseline samples from these trials were obtained from patients while on a failing drug regimen. Resistance testing of on-treatment samples was conducted with stored samples selected on the basis of treatment failure. Trials 1182.2 and 1182.4 differed from the other trials in that only clinical virologic failure entry criteria were required, while subsequent trials required genotypic resistance to be documented prior to entry. All patients were antiretroviral treatment experienced. However, by design, patients in trial 1182.2 were NNRTI naive; patients in trial 1182.4 had exposure to only a single PI; and patients in all other trials were triple-class experienced, including at least two PI-based regimens.

The nucleotide sequence data for isolates in these analyses has been deposited in GenBank (accession numbers DQ875942 to DQ880940).

Laboratory assays. Plasma HIV-1 RNA genotyping was conducted using either the TruGene 6.0 and 7.0 [Bayer] or Virtual Phenotype assays (version 3.6; [VIRCO]). The TruGene assay produced a 1.3-kb sequence that included the entire protease gene and the first 250 amino acids of the reverse transcriptase gene, whereas the Virtual Phenotype assay sequenced the entire protease gene and the first 400 amino acids of the reverse transcriptase gene. Codons with mixtures of mutant and wild-type (WT) amino acids were considered mutant for these analyses.

VIRCO Laboratories (Mechelen, Belgium) was the designated reference laboratory for the HIV-1 phenotypic analyses using the Antivirogram recombinant virus assay. Results are reported as n-fold WT values, comparing the IC₅₀ for the clinical sample to that for the concurrently run WT viral control. Plasma HIV-1 RNA levels were measured using the Roche Amplicor HIV-1 Monitor assay, version 1.5 (Roche Diagnostic Systems, Inc., Ontario, Canada) or the Roche UltraSensitive method, version 1.5.

Statistical methods. (i) Introduction. The analysis plans consider the tipranavir phenotype (IC₅₀ n-fold change (FC) from WT value) and the change from the baseline in log₁₀ plasma HIV RNA levels to be dependent variables. A screening regression analysis of every deviation at each protease codon from subtype B (the assay reference), as a single independent variable, was the basis for selecting the mutations considered for the stepwise linear regression. Phase II study data sets were used to create a tipranavir mutation score. Validation included evaluating phase III study data sets to see whether the same mutations would be selected when repeating these analyses and checking to see how the tipranavir score contributes to multiple-regression models of phenotype and virologic response.

(ii) Data sets used for tipranavir score derivation. The five data sets used to derive the tipranavir mutation score were labeled and identified as models A through E. The relationship between protease genotype and phenotypic susceptibility to tipranavir was investigated with the first three data sets. The first data set, model A, addresses resistance selected by other protease inhibitors prior to treatment with tipranavir. It includes patients with both a drug resistance genotype and phenotype (n = 287). The second data set, model B, addresses resistance selected by exposure to tipranavir on a background of mutations selected by other protease inhibitors. This model includes all patients with at least one available on-treatment specimen that yielded both a drug resistance genotype and phenotype (n = 72), with the last available assay results used for analysis. The third data set, model C, addresses extensive protease exposure, including exposure to tipranavir. Therefore, it includes the last available specimen with both assay results for all patients in either model A or B or both (n = 293.)

The relationship between protease genotype and virologic response to tipranavir was investigated with two data sets. To examine the association between baseline genotype and the 2-week change from baseline in plasma HIV-1 RNA levels, the data set model D included all patients enrolled in the phase II dose selection trial (trial 1182.52) with a 2-week virologic assessment of tipranavir functional monotherapy (n = 204). The data set model E was designed to examine the association between the baseline genotype and the 24-week change from baseline in plasma HIV-1 RNA levels and thus included all tipranavir patients enrolled in phase II trials for whom both results were available (n = 298).

Furthermore, the relationship between protease genotype and virologic rebound was investigated with a data set that included all tipranavir patients enrolled in phase II trials who had both a baseline and at least one on-treatment protease genotype determination. The existence of an on-treatment genotype, requiring a specimen with >1,000 copies/ml HIV-1 RNA, was considered evidence of virologic rebound. Baseline and on-treatment genotype sequences were compared to identify emergent mutations.

(iii) Regression methods for tipranavir mutation score derivation. Initial screening of observed mutations was conducted using linear regression models, with the independent variable being an indicator for each mutation and the dependent variable being the log₁₀ tipranavir IC₅₀ degree of change from the WT or change from baseline in log₁₀ plasma HIV-1 RNA levels. The estimated regression coefficient, median IC₅₀ n-fold change (or change in HIV-1 RNA level) for those isolates with the mutation and those without it, and the nominal statistical significance of the difference between the slope coefficient and zero were calculated for all identified amino acid mutations.

The resulting list of mutations considered for forward stepwise multiple regression was then selected on the basis of at least five observations, a P value of <0.05 for the regression slope, and a positive difference between the median log₁₀ tipranavir IC₅₀ n-fold change from the WT value or change from baseline in log₁₀ plasma HIV-1 RNA level for those isolates with the mutation compared to those without the mutation. These criteria eliminate any rare mutations and mutations associated with increased susceptibility. Selection for or against a mutation by exposure to tipranavir was considered in refining the list. Although the L90M and 82A mutations did not meet our analysis criteria, these mutations were also considered due to their previously reported association with resistance to tipranavir and other PIs as well (4, 8, 11).

When multiple substitutions occurred at a given codon, whether to consider each separately or several as a group in the stepwise multiple-regression analysis was based on (i) whether there were differences in the simple linear regression models for the different amino acids and (ii) whether the P value for the combined amino acids was smaller than those for the individual amino acids (which would indicate that all were having consistent effects). When there were differences or the P value was larger, the mutations were considered separately.

The selected mutations were then included in each of the five forward stepwise regression analyses of the data sets described above (models A to E). At each step, the mutation with the strongest relationship to the dependent variable after adjustment for previously selected mutations was identified on the basis of the P value for its regression coefficient. Selection continued until no remaining mutations reached a P value of <0.05 as the last term added to the model. The set of mutations contributing to the tipranavir mutation score was then derived by aligning the results of the five analyses and choosing those that were selected by at least two analyses or selected by one analysis and also selected in vivo by exposure to tipranavir.

(iv) Regression methods for tipranavir mutation score validation. Matched genotypic and phenotypic results (n = 523) as well as treatment response data (n = 569) were analyzed from three phase III trials of tipranavir boosted with ritonavir. Screening criteria included genotypic evidence of resistance to protease inhibitors.

An independent analysis of these data repeated the steps in the analysis of the phase II data. Univariable analyses of genotype versus phenotype and genotype versus response were the basis for selection of mutations included in stepwise multiple-regression analyses. These stepwise analyses identified mutations that made significant contributions to the prediction of the dependent variable, i.e., phenotype or viral response. Differences from the phase II analyses were identified and evaluated to determine whether to adjust the score.

The selected mutations from the phase II analyses were then used in multiple logistic regression analyses to evaluate how well the models predict the phase III data. The baseline tipranavir score, derived by counting the number of tipranavir-associated mutations observed (each considered as one point in the tipranavir mutation score), was included in regression models including enfuvirtide use and the number of active antiretrovirals in the optimized background regimen, which was based on the genotypic interpretation of susceptibility. This approach yielded evidence of the predictive value of the tipranavir mutation score.

Top Abstract Introduction Materials and Methods Results Discussion References

 
Screening analyses: phase II trials. Of 99 protease codons, the 5 sets of one-way regression analyses identified 45 amino acids at 29 codons that were statistically significantly associated with phenotypic resistance or reduced virologic response to tipranavir in at least one analysis.

The data sets for association of genotype and phenotype (models A, B, and C) identified 43 amino acids at 28 codons, of which 1, 10I, was not taken into the multiple-regression analyses because it was selected against by tipranavir exposure and was not identified by any genotype to response analysis. Mutations 20M, 20V, 35N, 37E, 39Q, 66F, 82A, 91S, 92K, and 92R were significant only when analyzing the genotype associated with exposure to other protease inhibitors (model A). Mutations 10V, 16A, 24M, 33I, 33V, 34D, 36L, 69K, 82L, 82T, 89M, and 89V were significant only when mutations associated with tipranavir exposure were included in the analysis (models B and C). Mutations 13V, 33F, 35D, 36I, 43T, 46L, 47V, 54V, 58E, 62V, 71V, and 84V were significant in all three analyses of phenotype.

The data sets for association of genotype and virologic response identified 17 amino acids at 16 codons, of which one, 13L, was not taken into the multiple-regression analyses because it was selected against by tipranavir exposure and was not identified by any of the genotype-to-phenotype analysis. Mutations 55Q, 71V, 84V, and 91S were significant for the 2-week response (model D); 13L, 16A, 36I, 46L, 69K, and 89V were significant for the 24-week response (model E); and 13V, 20M, 32I, 33F, 35G, 47V, and 54M were significant for both.

Three mutations were added to the list taken into multiple-regression analysis because they appeared to be selected for by tipranavir exposure: 66V, 74A, and 83D. The 90M mutation was taken into the multiple-regression analysis because of its prior association with phenotypic resistance to tipranavir (4, 11). The 82A mutation was also taken into the multiple regression because like 90M it is considered a major protease mutation for other PIs; however, it appeared that the 82A mutation was selected against with exposure to tipranavir. The list of mutations assessed in stepwise multiple regression was as follows: 10V, 13V, 16A, 20 M/R/V, 24M, 32I, 33F/I/V, 34D, 35D/N/G, 36I/L, 37E, 39Q, 43T, 46L, 47V, 54M/V, 55R/Q, 58E, 62V, 66F/V, 69K, 71V, 72E, 74A/P, 82A/F/L/T, 83D, 84V, 89M/V, 90M, 91S, 92K/R (47 mutations). The majority of these mutations identified by the screening analyses have not been reported as being natural polymorphisms in untreated patients (20).

Mutations at six positions were associated with significantly increased susceptibility to tipranavir compared with that of isolates that did not have the mutation (model C): 18L (median FC = 0.5; P = 0.035), 30N (median FC = 0.5; P < 0.001), 50V (median FC = 0.7; P = 0.021), 54L (median FC = 0.7; P = 0.018), 77I (median FC = 1.0; P = 0.011), and 88D (median FC = 0.6; P = 0.001). The most common, 77I, was observed in 102 (35%) isolates. The D30N mutation in 26 isolates (9%) and N88D in 26 isolates (9%) have been associated with nelfinavir resistance, and the I50V mutation in 19 isolates (6%) has been associated with amprenavir resistance (8). With the goal of developing a score based on mutations associated with tipranavir resistance, these six mutations were not considered further.

Multiple stepwise regression analyses: phase II trials. Forty-seven mutations, selected by screening analyses as described above, were taken into the multiple stepwise regression analysis models and reduced to 28 mutations identified in at least one of the five stepwise models (Table 1). Mutations identified in only one of the stepwise regression models or mutations which were not identified in emergence analyses were excluded. These were mutations 35N, 37E, 39Q, 55Q, 72E, 82F, and 91S, which are not included in determining the tipranavir mutation score. Thus, the 21 tipranavir-associated mutations (16 codons) identified from phase II trials consisted of the following: 10V, 13V, 20M/R/V, 33F, 35G, 36I, 43T, 46L, 47V, 54A/M/V, 58E, 69K, 74P, 82L/T, 83D, and 84V.

View this table: [in this window] [in a new window]

TABLE 1. Summary of stepwise multivariable regression analyses and mutations comprising the tipranavir mutation score: phase II trials

Table 1 demonstrates the mutations selected by each of the stepwise multiple-regression methods for the five analysis sets. With phenotype as the dependent variable, r² values achieved were from 0.49 to 0.82 (models A, B, and C). Effect estimates of the contribution of each mutation to reduced susceptibility ranged from a 0.10 to 1.28 log₁₀ IC₅₀ n-fold change in models that included baseline isolates (models A and C), while the range was from 0.37 to 1.73 in the model that included only on-treatment observations (model B).

The selected mutations include 17 amino acids at 14 codons that were significant in the most extensive univariable analysis of phenotypic resistance (model C). At all codons, 10V, 13V, 20R, 33F, 36I, 43T, 46L, 47V, 54A/M/V, 58E, 69K, 74P, 82L/T, and 84V had P values of <0.01 in the univariable analysis. The mutation 83D was one of three included in the analysis because of evidence that it was selected for by exposure to tipranavir. The mutation 35G was identified in univariable analysis models of virologic response.

Screening analyses: phase III trials. As with the phase II trials, each mutation observed at baseline was evaluated by comparing those viruses with the mutation present to those that were without the mutation. The baseline phenotype available for patients in trials 1182.12, 1182.48, and 1182.51 was examined to see which mutations were associated with reduced susceptibility to tipranavir. The baseline genotype for patients in phase III trials (1182.12 and 1182.48) was examined to see which mutations were associated with reduced 24-week HIV RNA response to tipranavir.

The baseline genotype and baseline phenotype analysis of phase III data identified 10 mutations that had not been identified as significant in the univariable phase II analyses: 11I, 19P, 22V, 60E, 70E, 72L, 82C, 83D, 85V, and 90M. The 83D mutation had been included in the phase II multivariable analysis on the basis of selection with exposure to tipranavir and was previously selected for inclusion in the tipranavir mutation score. The 90M mutation was included in the phase II multivariable analysis because of early observations that it was associated with tipranavir resistance but was not selected for inclusion in the tipranavir mutation score.

Thus, the string of mutations evaluated in multiple stepwise regression analyses of phase III included the following (the asterisk indicates added new mutations based on phase III univariable analyses): 10V, 11I*, 13V, 16A, 19P*, 20M, 20R, 20V, 22V*, 24M, 32I, 33I, 33F, 33V, 34D, 35D, 35N, 35G, 36I, 36L, 37E, 39Q, 43T, 46L, 47V, 54A, 54 M, 54V, 55R, 55Q, 58E, 60E*, 62V, 66F, 66V, 69K, 70E*, 71V, 72E, 72L*, 74A, 74P, 82A, 82C*, 82L, 82F, 82T, 83D, 84V, 85V*, 89M, 89V, 90M, 91S, 92K, and 92R. This new string is called the "modified string" in analyses described below.

Multiple stepwise regression analyses: phase III trials. The multivariable stepwise regression was performed once with the string of mutations identified by the phase II analyses. It was then repeated using the modified string as described above: the same string of mutations plus the mutations that were significant for the first time in the phase III univariable analyses. The purpose was to determine how much the resultant selected mutation set was dependent on the initial selection step.

The result for the genotype-to-phenotype multiple stepwise regression was a set of 16 mutations at 13 codons (10V, 35D/G, 43T, 46L, 47V, 54M/V, 58E, 60E, 74P, 82F/T, 83D, 84V, and 90M). Of these, 60E was the only mutation to be selected from among the eight that were added in the modified string. Only two other mutations were not previously seen in multiple stepwise regression for tipranavir: 35D and 90M. The 82F mutation, which had been significant in phase II phenotypic analyses but had not been included in the tipranavir mutation score, was again significant and remained a relatively rare mutation. Thus, four mutations that had not been included in the tipranavir mutation score were selected in these phase III models (35D, 60E, 82F, and 90M). These mutations were unsupported by anything other than genotypic-to-phenotypic relationships; neither a genotype-to-response relationship nor emergence with tipranavir exposure was noted. They were therefore not promoted to inclusion in the tipranavir mutation score because of this lack of corroboration. The remaining 12 mutations had been selected in phase II analyses and included in the tipranavir mutation score.

Applying both strings to the 24-week HIV RNA response (last observation carried forward [LOCF]) selected the same set of five mutations at four codons: 47V, 54M, 54V, 69K, and 82A. All except 82A are part of the tipranavir mutation score derived in the analyses of phase II trials. 82A was a significant predictor of good response, as evidenced by a negative parameter estimate, and therefore was not included in the tipranavir mutation score.

Nothing emerging in the phase III analyses met the standards set for inclusion in the tipranavir-associated mutations. Mutations found to be associated with the baseline phenotype were not associated with the response to tipranavir in any analyses and were not emergent in the phase II analyses of viral genotypes from patients exposed to tipranavir. Therefore, no modification of tipranavir mutation score determination was required on the basis of this analysis of the phase III baseline phenotype and 24-week response results.

Thus, the validated tipranavir score mutations include 16 protease positions, 10V, 13V, 20M/R/V, 33F, 35G, 36I, 43T, 46L, 47V, 54A/M/V, 58E, 69K, 74P, 82L/T, 83D, and 84V (Fig. 1). These mutations are distributed throughout the entire protease as follows: L33F, V82L/T, N83D, and I84V in the active site, E35G and M36I in the ear flap, K43T, M46L, I47V, I54A/M/V, and Q58E in the flap, L10V, I13V, and K20M/R/V in the cheek turn, and H69K and T74P in the cheek sheet (16, 22). The accumulation of these mutations is required to predict decreased phenotypic susceptibility or diminished antiviral responses to tipranavir.

View larger version (36K): [in this window] [in a new window]

FIG. 1. Backbone tracing of HIV-1 protease with tipranavir bound into the active site of the enzyme. The positions of the mutations that are part of the tipranavir score are highlighted in red or yellow on both monomers (although for clarity the labels appear only once per monomer).

Tipranavir mutation score versus phenotype: phase II and III trials. To further understand the relationship between the tipranavir mutation score and tipranavir phenotypic susceptibility, phenotypic data were combined for all phase II and III clinical trials. Table 2 shows the median IC₅₀ n-fold change from the WT value for baseline HIV-1 isolates according to the number of tipranavir score mutations. Tipranavir mutation scores of 0 or 1 were associated with a median n-fold change in tipranavir susceptibility of <1. Increasing tipranavir mutation score numbers were associated with a higher median degree of change from the WT value for tipranavir. A tipranavir mutation score of 4 was associated with a median n-fold change in tipranavir susceptibility of 2, while a score of 8 was associated with an n-fold change of 14.7. Although the number of isolates with tipranavir mutation scores of greater than 7 was limited, there was clearly a marked reduction in tipranavir susceptibility. Overall these pooled data demonstrate that with an increasing tipranavir mutation score there is a reduction in tipranavir susceptibility.

View this table: [in this window] [in a new window]

TABLE 2. Median n-fold change from WT value for tipranavir IC50 by tipranavir mutation score among baseline HIV-1 isolates from participants in phase II and III tipranavir trials

Comparison of phenotypic susceptibilities of baseline isolates to current protease inhibitors: phase II and III trials. Table 3 compares median HIV-1 phenotypic susceptibilities to tipranavir and the other currently available protease inhibitors from phase II and III clinical trials. Pooled baseline data from the phase II trials in which patients had exposure to one or more protease inhibitors show a median n-fold change of 1 for tipranavir, compared to 2.8 for amprenavir and 23.7 for lopinavir. In the large phase III trials, the median IC₅₀ n-fold change for tipranavir and the other protease inhibitors was higher than in earlier phase II trials, since patients in these studies had greater protease inhibitor exposure and more primary protease mutations. This is seen to an even greater extent in median n-fold change values for trial 1182.51, which included patients with the most highly protease inhibitor-resistant viruses.

View this table: [in this window] [in a new window]

TABLE 3. Baseline phenotypic susceptibilities of clinical HIV-1 isolates to tipranavir and existing protease inhibitors in phase II and III tipranavir trials

Tipranavir mutation score versus response: phase II and III trials. To examine the treatment response in relation to the tipranavir mutation score, HIV RNA reductions were examined for phase II and III trials. Table 4 shows the change in HIV RNA from the baseline through 2 and 24 weeks, combining data from the phase III 1182.12 and 1182.48 studies with data from patients receiving the same dose of boosted tipranavir in the 1182.51 and 1182.52 studies. Patients whose virus had 0 or 1 mutation achieved a median reduction in HIV RNA of 1.3 log₁₀ copies/ml at 2 weeks and 2.1 log₁₀ copies/ml at 24 weeks. Although 2-week responses were similar for those with two to seven tipranavir score-determining mutations, the virologic responses at 24 weeks were diminished. The dropoff in response at 24 weeks for these individuals has been attributed to the frequency of broad three-class drug-resistant viruses and a lack of active background therapy that could be administered with tipranavir (data not shown) (23). Although there were few patients whose viruses had baseline tipranavir mutation scores of 8, virologic responses were poor at 2 and 24 weeks for these patients.

View this table: [in this window] [in a new window]

TABLE 4. Treatment responses at 2 and 24 weeks by tipranavir mutation score in phase IIa and III trials

Regression models for predicting virologic response: phase III trials. The tipranavir mutation score at baseline was used along with variables for enfuvirtide use and genotypic susceptibility to optimized background drugs to predict a 1-log HIV RNA decline in patients receiving tipranavir in phase III studies. Logistic regression models considered the week 2 response and the week 24 response. These models showed that both enfuvirtide use and the tipranavir mutation score were consistently and independently associated with the likelihood of obtaining a 1-log₁₀-copies/ml HIV RNA reduction at weeks 2 and 24. At 24 weeks, patients who used enfuvirtide had a 3.6-fold greater chance of achieving a 1-log₁₀-copies/ml HIV RNA reduction than those who did not; in addition, for each drug used in the background that was active by genotype, patients had a 48% higher chance of achieving a virologic response. Finally, for each additional single mutation counted in the tipranavir mutation score, the chances of achieving a 1-log₁₀-copies/ml HIV RNA reduction at week 24 decreased by 21%.

When plasma HIV RNA reduction was analyzed as a continuous variable including the same explanatory variables as in the logistic regression models, again enfuvirtide use and the tipranavir mutation score were significantly and consistently associated with the magnitude of HIV RNA reduction (Table 5). The number of active drugs in the background regimen was also significant. Each point in the tipranavir mutation score was associated with 0.04-log₁₀-copies/ml-lower viral response at week 2 and a 0.16-log₁₀-copies/ml-lower viral response at week 24. In summary, evaluating the patients who participated in trials 1182.12 and 1182.48, mutations counted in the tipranavir mutation score were predictive of viral load responses to tipranavir-containing regimens.

View this table: [in this window] [in a new window]

TABLE 5. Multiple-regression models for virologic response in phase III tipranavir trialsa

Top Abstract Introduction Materials and Methods Results Discussion References

 
Tipranavir is a novel, nonpeptidic inhibitor of HIV-1 protease that offers potential therapeutic advantages for PI-experienced patients. Tipranavir has activity against HIV-1 isolates broadly resistant to other protease inhibitors. Phase III clinical trials have demonstrated that patients treated with tipranavir-ritonavir achieve a superior virologic and immunologic response compared to those patients treated with other ritonavir-boosted protease inhibitors (3). The results of these analyses suggest that the resistance profile of tipranavir is distinct from that of other protease inhibitors.

In this analysis of predominantly subtype B HIV-1 isolates, a unique string of 16 protease positions with 21 mutations (10V, 13V, 20M/R/V, 33F, 35G, 36I, 43T, 46L, 47V, 54A/M/V, 58E, 69K, 74P, 82L/T, 83D, and 84V) was identified as being associated with a reduced susceptibility and diminished virologic response to tipranavir. These tipranavir score mutations were identified based on univariable and multivariable analyses of phase II clinical trial data and were subsequently validated using phase III clinical trial data. Analyses included genotype-to-phenotype comparisons, virologic response to treatment with tipranavir, and the emergence of new mutations upon treatment with tipranavir. The viral isolates in these studies were obtained from patients with exposure to multiple other protease inhibitors, and most displayed markedly reduced susceptibilities to other commercially available PIs. Mutations counted in the tipranavir mutation score are not necessarily reflective of the de novo development of resistance for treatment-naive patients who receive tipranavir-ritonavir. An ongoing trial with this patient population will help define the protease resistance mutations that may emerge when tipranavir-ritonavir is used for initial therapy.

A number of the tipranavir score mutations correlated with positions identified in vitro to be associated with reduced susceptibility to tipranavir, namely positions 10, 13, 33, 36, 54, 82, and 84 (4). However, several positions were identified in these analyses (codons 13, 35, 43, 58, 69, 74, and 83) that have not been associated with resistance to other protease inhibitors (9, 18, 21) (http://hivdb.stanford.edu), demonstrating a novel resistance profile for tipranavir. Of particular clinical relevance is the fact that major protease mutations (30N, 48V, 50V/L, 82A/F, and 90M) associated with resistance to other protease inhibitors were not found to contribute to the tipranavir mutation score (9), and several (30N and 50V) were associated with increased susceptibility to tipranavir. Although associated with reduced susceptibility to tipranavir in testing of earlier clinical isolates (11), the 90M mutation was not included in the tipranavir mutation score due to the negative results of multiple multivariable analyses. The 90M mutation may serve as a marker for highly mutated viruses which are likely to contain other protease mutations associated with decreased tipranavir susceptibility.

As shown in Fig. 1, the residues that are part of the tipranavir mutation score are not concentrated solely around the active site but rather are distributed throughout most of the protease structure (encompassing the active site, the extended flap, and the cheek area) (16). Preliminary comparative analysis of the X-ray structure of tipranavir complexed with the wild type or the Q7K I13V V32I L33F K45I V82L I84V mutant (mutations counted in the tipranavir mutation score are underlined) suggests that the hydrogen bond network between tipranavir and the active site residues remains largely unaltered (data not shown). However, detailed thermodynamic measurements indicate that tipranavir behaves in a unique manner compared to other available protease inhibitors by gaining or sustaining only minimal loss in binding enthalpy when bound to mutant proteases (14, 15).

Increasing numbers of the tipranavir score mutations were associated with a reduced phenotypic susceptibility and a diminished virologic response to tipranavir. Although there were a limited number of isolates with a tipranavir mutation score of 8, clearly these viruses were associated with the highest n-fold change values, and a minimal response was seen in phase II and III trials. Therefore, these analyses would support a "cutoff" of 8 mutations of the tipranavir score group being predictive of full resistance and lack of efficacy with tipranavir. With 5 tipranavir score mutations, the overall median phenotypic FC was >3 and a diminished 24-week virologic responses were observed. Of note, as demonstrated with the multiple-regression models for virologic response in phase III trials, enfuvirtide use and the number of active background drugs also contributed to the virologic outcome.

The tipranavir mutation score is a potentially valuable tool for predicting phenotypic susceptibility and virologic response to tipranavir. However, for an individual patient the utility of the tipranavir score for correctly predicting the response may be somewhat limited given the range of phenotypic values possible with each score and the number of other active drugs available to administer with tipranavir. Furthermore, certain limitations should be acknowledged, such as confounding factors (e.g., use of efavirenz in NNRTI-naive patients in trial 1182.2, different doses of tipranavir and ritonavir in the three phase II trials, and possible undocumented nonadherence), which may have affected the predictive ability of the score. Thus, the potential for both false-positive and false-negative associations cannot be excluded. Validation of the tipranavir mutation score with the phase III clinical trial data supports the importance of this set of protease mutations; however, as more data become available in the future, further refinement of the score may be necessary, especially for patients infected with non-subtype-B HIV-1. The tipranavir mutation score was also highly predictive of virologic response with the multiple-regression model using phase III data. Of note, use of active drugs, such as enfuvirtide, along with tipranavir was associated with an improved virologic response. This emphasizes the importance of using tipranavir with other active agents with treatment-experienced patients to preserve drug activity and prevent the emergence of new resistance mutations.

In summary, the requirement that multiple specific protease mutations are needed to confer a reduced phenotypic susceptibility and a decreased virologic response to tipranavir suggests that there is a high genetic barrier to the development of resistance to tipranavir. The majority of HIV-1 isolates obtained from protease inhibitor-experienced patients in phase II and III clinical trials retained susceptibility to tipranavir, exhibiting an IC₅₀ n-fold change, relative to the WT value, of 3. Most patients failing therapy with other currently available protease inhibitors are likely to benefit from a treatment regimen containing tipranavir-ritonavir, particularly if used in combination with other active agents, as seen in the phase III trials. These trials also demonstrated that despite extensive prior protease inhibitor exposure, patients had favorable virologic responses to tipranavir-ritonavir, with attenuation of the response as the number of baseline tipranavir score mutations increased. The tipranavir mutation score derived from these analyses represents a unique set of mutations which are capable of predicting the virologic response to tipranavir-ritonavir in protease inhibitor treatment-experienced patients.

 
We thank Pierre Bonneau for helpful discussions and for providing the protease structure model and Peter Piliero for critically reviewing the manuscript. We are grateful to Paula McKenna (Virco, Ltd.) for providing phenotypes of the HIV-1 isolates.

 
* Corresponding author. Mailing address: Division of Infectious Diseases, Suite 513, Cooper Plaza, Camden, NJ 08103. Phone: (856) 757-7767. Fax: (856) 757-7803. E-mail: baxter{at}umdnj.edu.

Published ahead of print on 23 August 2006.

Top Abstract Introduction Materials and Methods Results Discussion References

 

1
1. Back, N. K. T., A. van Wijk, D. Remmerswaal, M. van Monfort, M. Nijhuis, R. Schuurman, and C. A. Boucher. 2000. In-vitro tipranavir susceptibility of HIV-1 isolates with reduced susceptibility to other protease inhibitors. AIDS 14:101-102.[CrossRef][Medline]
1
2. Bartlett, J. G., H. C. Lane, and the Panel on Antiretroviral Guidelines for Adults and Adolescents, Office of AIDS Research Advisory Council. 4 May 2006. Guidelines for the use of antiretroviral agents in HIV-1 infected adults and adolescents, p. 1-112. National Institutes of Health, Bethesda, Md. [Online.] http://aidsinfo.nih.gov/contentfiles/adultandadolescentGL.pdf.
2
3. Cahn, P., and the RESIST-2 study team. 2004. 24-week data from RESIST-2: phase 3 study of the efficacy and safety of background therapy plus tipranavir/ritonavir (TPV/r) or optimized ritonavir-boosted standard-of-care (SOC) comparator PI (CPI) in a large randomized multicenter trial in treatment-experienced HIV+ patients. Presented at the 7th International Congress on Drug Therapy in HIV Infection, Glasgow, Scotland, 14 to 18 November.
3
4. Cahn, P.I., and C. Hicks for the RESIST-2 and RESIST-1 study teams. 2005. RESIST-1 (R-1) and RESIST-2 (R-2) 48 week meta-analyses demonstrate superiority of protease inhibitor (PI) tipranavir+ritonavir (TPV/r) over an optimized comparator PI (CPI/r) regimen in antiretroviral (ARV) experienced patients. Presented at the 10th European AIDS Conference, Dublin, Ireland, 17 to 20 November.
4
5. Doyon, L., S. Tremblay, M. Cartier, and M. G. Cordingley. 2002. In vitro susceptibility of HIV-1 to tipranavir. Antivir. Ther. 7:S12.
5
6. Doyon, L., S. Tremblay, L. Bourgon, E. Wardrop, and M. G. Cordingley. 2005. Selection and characterization of HIV-1 showing reduced susceptibility to the non-peptidic protease inhibitor tipranavir. Antivir. Res. 68:27-35.[CrossRef][Medline]
6
7. Reference deleted.
7
8. Hicks, C., for the RESIST-1 study team. 2004. RESIST-1: randomized evaluation of strategic intervention in multi-drug resistant patients with tipranavir; phase 3 comparison of TPV/r and standard-of-care boosted-comparator PI (CPI/r) at 24 weeks. Presented at the 44th Interscience Conference on Antimicrobial Agents and Chemotherapy, Washington, D.C..
8
9. Johnson, V. A., F. Brun-Vezinet,, B. Clotet, B. Conway, R. T. D'Aquila, L. M. Demeter, D. R. Kuritzkes, D. Pillay, J. M. Schapiro, A. Telenti, and D. D. Richman. 2003. Drug resistance mutations in HIV-1. Topics HIV Med. 11:215-221.
9
10. Johnson, V. A., F. Brun-Vezinet, B. Clotet, B. Conway, D. R. Kuritzkes, D. Pillay, J. Schapiro, A. Telenti, and D. Richman. 2005. Update of the drug resistance mutations in HIV-1: 2005. Topics HIV Med. 13:51-57.
10
11. Kohlbrenner, V., J. Gathe, G. Pierone, K. Arasteh, P. Rubio, R. LaLonde, P. Piliero, S. McCallister, S. Garfinkel, R. Chaves, S. Spinosa, C. Dohnanyi, A. Quinson, and D. Mayers. 2003. Tipranavir/ritonavir (TPV/r) demonstrates potent efficacy in multiple protease-inhibitor (PI)-experienced patients at 24 weeks: BI 1182.52. Presented at the 9th European AIDS Conference, Warsaw, Poland, 26 to 29 October.
11
12. Larder, B. A., K. Hertogs, S. Bloor, C. van den Eyne, W. DeCian, W. Yenyun, W. W. Freimuth, and G. Tarpley. 2000. Tipranavir inhibits broadly protease inhibitor-resistant HIV-1 clinical samples. AIDS 14:1943-1948.[CrossRef][Medline]
12
13. Mayers, D., K. Curry, V. Kohlbrenner, and S. McCallister. 2004. Non-peptidic protease inhibitors (NPPIs): tipranavir. Adv. Antiviral Drug Res. 4:79-98.
13
14. Menendez-Arias, L. 2002. Targeting HIV: antiretroviral therapy and development of drug resistance. Trends Pharmacol. Sci. 23:381-388.[CrossRef][Medline]
14
15. Muzammil, S., L. W. Kang, A. A. Armstrong, A. Jakalian, P. R. Bonneau, V. Schmelmer, L. M. Amzel, and E. Freire. 2005. Binding energetics analysis shows a unique response of tipranavir to HIV-1 protease mutations associated with drug resistance. Presented at the 3rd International AIDS Society meeting, Rio de Janeiro, Brazil, 24 to 27 July.
15
16. Muzammil, S., L. W. Kang, A. A. Armstrong, A. Jakalian, P. R. Bonneau, V. Schmelmer, L. M. Amzel, and E. Freire. 2005. Unique response of tipranavir to multi-drug resistant HIV-1 proteases suggests new ways of combating drug resistance. Presented at the 14th International HIV Drug Resistance Workshop, Quebec City, Canada, 7 to 11 June.
16
17. Perryman, A. L., J. H. Lin, and J. A. McCammon. 2004. HIV-1 protease molecular dynamics of a wild-type and of the V82F/I84V mutant: possible contributions to drug resistance and a potential new target for drugs. Protein Sci. 13:1108-1123.[CrossRef][Medline]
17
18. Poppe, S. M., D. E. Slade, K. T. Chong, R. R. Hinshaw, P. J. Pagano, M. Markowitz, D. D. Ho, H. Mo, R. R. Gorman, T. J. Dueweke, S. Thaisrivongs, and W. G. Tarpley. 1997. Antiviral activity of the dihydropyrone PNU-140690, a new nonpeptidic human immunodeficiency virus protease inhibitor. Antimicrob. Agents Chemother. 41:1058-1063.[Abstract]
18
19. Puchhammer-Stoeckl, E., C. Steininger, E. Geringer, and F. X. Heinz. 2002. Comparison of virtual phenotype and HIV-SEQ program (Stanford) interpretation for predicting drug resistance of HIV strains. HIV Med. 3:200-206.[CrossRef][Medline]
19
20. Rusconi, S., S. la Seta Catamancio, P. Citterio, S. Kurtagic, M. Violin, C. Balotta, M. Moroni, M. Galli, and A. D'Arminio-Monforte. 2000. Susceptibility to PNU-140690 (tipranavir) of human immunodeficiency virus type 1 isolates derived from patients with multidrug resistance to other protease inhibitors. Antimicrob. Agents Chemother. 44:1328-1332.[Abstract/Free Full Text]
20
21. Shafer, R. W., O. Hsu, A. K. Patick, C. Craig, and V. Brendel. 1999. Identification of biased amino acid substitution patterns in human immunodeficiency virus type 1 isolates from patients treated with protease inhibitors. J. Virol. 73:6197-6202.[Abstract/Free Full Text]
21
22. Shafer, R. W., D. Stevenson, and B. Chan. 1999. Human immunodeficiency virus reverse transcriptase and protease sequence database. Nucleic Acids Res. 27:348-352.[Abstract/Free Full Text]
22
23. Thaisrivongs, S., H. I. Skulnick, S. R. Turner, J. W. Strohbach, R. A. Tommasi, P. D. Johnson, P. A. Aristoff, T. M. Judge, R. B. Gammill, J. K. Morris, K. R. Romines, R. A. Chrusciel, R. R. Hinshaw, K. T. Chong, W. G. Tarpley, S. M. Poppe, D. E. Slade, J. C. Lynn, M. M. Horng, P. K. Tomich, E. P. Seest, L. A. Dolak, W. J. Howe, G. M. Howard, F. J. Schwende, L. N. Toth, G. E. Padbury, G. J. Wilson, L. Shiou, G. L. Zipp, K. F. Wilkinson, B. D. Rush, M. J. Ruwart, K. A. Koeplinger, Z. Zhao, S. Cole, R. M. Zaya, T. J. Kakuk, M. N. Janakiraman, and K. D. Watenpaugh. 1996. Structure-based design of HIV protease inhibitors: sulfonamide-containing 5,6-dihydro-4-hydroxy-2-pyrones as non-peptidic inhibitors. J. Med. Chem. 39:4349-4353.[CrossRef][Medline]
23
24. Valdez, H., D. B. Hall, V. M. Kohlbrenner, C. A. Boucher, J. Schapiro, J. Baxter, J. Scherer, S. McCallister, and D. L. Mayers. 2005. Non-response to tipranavir is associated with pretreatment resistance characterized by tipranavir phenotype or genotypic tipranavir score. Antivir. Ther. 10:S29.

---

Journal of Virology, November 2006, p. 10794-10801, Vol. 80, No. 21
0022-538X/06/$08.00+0     doi:10.1128/JVI.00712-06
Copyright © 2006, American Society for Microbiology. All Rights Reserved.

This article has been cited by other articles:

• Shahriar, R., Rhee, S.-Y., Liu, T. F., Fessel, W. J., Scarsella, A., Towner, W., Holmes, S. P., Zolopa, A. R., Shafer, R. W. (2009). Nonpolymorphic Human Immunodeficiency Virus Type 1 Protease and Reverse Transcriptase Treatment-Selected Mutations. Antimicrob. Agents Chemother. 53: 4869-4878 [Abstract] [Full Text]  
• Iqbal, H. S., Solomon, S. S., Madhavan, V., Solomon, S., Balakrishnan, P. (2009). Primary HIV-1 Drug Resistance and Polymorphic Patterns Among Injecting Drug Users (IDUs) in Chennai, Southern India. J Int Assoc Physicians AIDS Care (Chic Ill) 8: 323-327 [Abstract]  
• Bartolo, I., Rocha, C., Bartolomeu, J., Gama, A., Fonseca, M., Mendes, A., Cristina, F., Thamm, S., Epalanga, M., Silva, P. C., Taveira, N. (2009). Antiretroviral Drug Resistance Surveillance among Treatment-Naive Human Immunodeficiency Virus Type 1-Infected Individuals in Angola: Evidence for Low Level of Transmitted Drug Resistance. Antimicrob. Agents Chemother. 53: 3156-3158 [Abstract] [Full Text]  
• Tartaglia, A., Saracino, A., Monno, L., Tinelli, C., Angarano, G. (2009). Both a Protective and a Deleterious Role for the L76V Mutation. Antimicrob. Agents Chemother. 53: 1724-1725 [Full Text]  
• Marcelin, A.-G., Masquelier, B., Descamps, D., Izopet, J., Charpentier, C., Alloui, C., Bouvier-Alias, M., Signori-Schmuck, A., Montes, B., Chaix, M.-L., Amiel, C., Santos, G. D., Ruffault, A., Barin, F., Peytavin, G., Lavignon, M., Flandre, P., Calvez, V. (2008). Tipranavir-Ritonavir Genotypic Resistance Score in Protease Inhibitor-Experienced Patients. Antimicrob. Agents Chemother. 52: 3237-3243 [Abstract] [Full Text]  
• de Requena, D. G., Bonora, S., Calcagno, A., D'Avolio, A., Siccardi, M., Fontana, S., Milia, M. G., Sciandra, M., Garazzino, S., Di Garbo, A., Baietto, L., Trentini, L., Di Perri, G. (2008). Tipranavir (TPV) Genotypic Inhibitory Quotient Predicts Virological Response at 48 Weeks to TPV-Based Salvage Regimens. Antimicrob. Agents Chemother. 52: 1066-1071 [Abstract] [Full Text]  
• Lambert-Niclot, S., Flandre, P., Canestri, A., Peytavin, G., Blanc, C., Agher, R., Soulie, C., Wirden, M., Katlama, C., Calvez, V., Marcelin, A.-G. (2008). Factors Associated with the Selection of Mutations Conferring Resistance to Protease Inhibitors (PIs) in PI-Experienced Patients Displaying Treatment Failure on Darunavir. Antimicrob. Agents Chemother. 52: 491-496 [Abstract] [Full Text]  
• Poon, A. F. Y., Kosakovsky Pond, S. L., Richman, D. D., Frost, S. D. W. (2007). Mapping Protease Inhibitor Resistance to Human Immunodeficiency Virus Type 1 Sequence Polymorphisms within Patients. J. Virol. 81: 13598-13607 [Abstract] [Full Text]  
• Youle, M. (2007). Overview of boosted protease inhibitors in treatment-experienced HIV-infected patients. J Antimicrob Chemother 60: 1195-1205 [Abstract] [Full Text]  
• King, M. S., Rode, R., Cohen-Codar, I., Calvez, V., Marcelin, A.-G., Hanna, G. J., Kempf, D. J. (2007). Predictive Genotypic Algorithm for Virologic Response to Lopinavir-Ritonavir in Protease Inhibitor-Experienced Patients. Antimicrob. Agents Chemother. 51: 3067-3074 [Abstract] [Full Text]  

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Reprints and Permissions Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Baxter, J. D. Articles by Mayers, D. L. Search for Related Content PubMed PubMed Citation Articles by Baxter, J. D. Articles by Mayers, D. L.