GW433908/ritonavir once daily in antiretroviral therapy-naive HIV-infected patients

absence of protease resistance at 48 weeks

Abstract

Objectives: 

To investigate the emergence of resistance to GW433908 (908), a protease inhibitor (PI) with demonstrated antiviral efficacy, safety and tolerability, when administered once daily (q.d.) with low dose ritonavir (908/r).

Design: 

A 48-week Phase III open-label study (SOLO, APV30002) in which antiretroviral therapy-naive patients (n = 649) were treated with 908/r, (1400 mg/200 mg, q.d.) or nelfinavir [1250 mg, twice daily (b.i.d.)] with two nucleoside reverse transcriptase inhibitors (NRTI), abacavir (300 mg, b.i.d.) and lamivudine (150 mg, b.i.d.).

Methods: 

Viral genotype and phenotype were analysed at baseline and on treatment up to 48 weeks and beyond.

Results: 

Emergence of genotypic resistance was significantly different between the 908/r q.d. and the nelfinavir b.i.d. treatment arms for both PIs (0 versus 50%; P < 0.001) and the NRTI (13% versus 69%; P < 0.001) received. In the nelfinavir arm the key protease mutations D30N and/or L90M were frequently observed. The absence of protease resistance mutations and reduced incidence of NRTI resistance mutations in the 908/r q.d. arm was confirmed by phenotyping, which showed a lack of PI cross-resistance.

Conclusions: 

The absence of resistance to 908 or cross-resistance to other PIs, and reduced NRTI resistance, following a 908/r q.d. regimen supports the use of this boosted PI early in therapy.

Introduction

GW433908 is a potent and durable HIV protease inhibitor (PI) with a favourable safety profile as demonstrated by two Phase III clinical trials in antiretroviral-naive patients [^1,2]. Co-administration of 908 with low dose ritonavir boosts exposure of amprenavir (APV) which facilitates flexible dosing, including once daily (q.d.) dosing, and requires no food or fluid restrictions [^3,4].

Resistance studies with APV, the active moiety of 908, have identified four distinct genetic pathways leading to reduced susceptibility to APV [^5–7]. The pathways are characterized by the protease substitutions I50V, I54L/M, V32I+I47V, or, less commonly, I84V. Other secondary mutations include L10F, L33F or M46I/L in protease and L449F or P453L in the P1/P6 Gag cleavage site [⁸]. Little or no cross-resistance to other PIs was detected in virus from APV failures [⁷].

Efficacy data from the Phase III SOLO study comparing 908/r q.d. and nelfinavir (NFV) twice daily (b.i.d.) when taken with abacavir (ABC)/lamivudine (3TC) b.i.d. has been reported [²]. Briefly, 69% and 55% in the 908/r q.d. arm and 68% and 53% in the NFV b.i.d. arm had plasma HIV-1 RNA < 400 copies/ml, and < 50 copies/ml respectively at week 48 (intent-to-treat rebound/discontinuation = failure). More of the NFV b.i.d. arm (57/327, 17%) experienced virological failure than in the 908/r q.d. arm (22/322, 7%). Resistance analysis of the virological failures in SOLO comparing 908/r q.d. with NFV b.i.d. is presented.

Methods

Study design

SOLO was a randomized, parallel, open-label, multicentre, study of 908/r 1400 mg/200 mg q.d. (n = 322) versus NFV 1250 mg b.i.d. (n = 327), in combination with ABC b.i.d. and 3TC b.i.d. Patients could remain in the study until the last patient enrolled completed the week 48 study visit. Baseline resistance was analysed by genotyping at day 1 in a random subset (169) of patients. Virologic failure was defined as an HIV-1 RNA level (vRNA) of ≥ 1000 copies/ml at two consecutive visits from week 12 or beyond, after having previously achieved vRNA of < 400 copies/ml; or having on-going viral replication defined as failing to achieve vRNA of < 400 copies/ml at week 12. The first of two consecutive samples with vRNA of ≥ 1000 copies/ml at week 12 or beyond was analysed together with the corresponding baseline sample. Evolution of resistance during therapy was analysed using the last timepoint on randomized therapy where the vRNA was > 1000 copies/ml.

Samples were assessed for viral genotype, including coding regions, protease (PRO), reverse transcriptase (RT) and p7/p1 and p1/p6 Gag cleavage sites, and for drug-susceptibility phenotype, by ViroLogic Inc. (South San Francisco, California, USA) using the GENESEQ and the PHENOSENSE assays. Genotypic comparisons were based on the wild-type HIV-1 strain, HXB2. Phenotypes were reported as fold reduction in 50% inhibitory concentration relative to wild-type strain NL4-3. Resistance observed was evaluated in relation to vRNA profiles of virological failures which were categorized based on the slope of vRNA rebound. Rebound slopes were defined as rapid, if the rebound reading was > 10 000 RNA copies/ml, unless the rebound was < 20% of baseline viral load; or gradual if the rebound reading did not exceed 10 000 RNA copies/ml, unless it was > 80% of baseline viral load. A third category included those with ongoing replication.

Results

Resistance mutations in baseline virus

Virus from only two patients (2/196; ∼1%), both in the 908/r q.d. arm, had evidence at baseline of any PRO mutations [⁹]. In one patient, an I50i/v mixture was detected at baseline but not following subsequent failure at week 20 of 908/r q.d. therapy. An indinavir resistance mutation L24I [^9,10], and an RT revertant mutation T215D [^11,12] were evident at day 1 in another patient (patient 1, Table 2). Between week 1 and week 4, additional PRO mutations M46L, I54V and V82A and an RT mutation M41L were detected. This patient responded but failed by week 12 with only M184V added after failure. The incidence of baseline nucleoside reverse transcriptase inhibitor (NRTI) resistance mutations was 6% (12/191, data unavailable for five patients) with four in the 908/r q.d. arm, (T215D; T215S; M41L+V118i/v; T215L+L100i/l+K103k/n) and eight in the NFV b.i.d. arm (T215D; L210W; E44D; V118i/v (n = 2); K219k/e; M41L+T215S; L210W+T215D).

Table 2: Baseline and emergent reverse transcriptase (RT) inhibitor and protease (PRO) inhibitor (PI) resistance-associated substitutions and related phenotypic sensitivities during therapy with 908/r daily (q.d.).

Emergence of resistance mutations during therapy

Statistically significant differences (Fisher's Exact Test) were observed between the 908/r q.d. and NFV b.i.d. arms in both the incidence of selection of PRO (0 versus 50%; P < 0.001) and RT (13% versus 69%; P < 0.001) mutations (Table 1). Natural polymorphisms in the absence of any other PRO mutations were excluded from this analysis (NFV, n = 3; 908/r, n = 1). Correlation of resistance data with vRNA rebound slopes showed that for NFV failures RT and PRO mutations were most frequent in the gradual rebounds (19/27) and ongoing replication groups (16/18), but not in the rapid rebounds (2/9). For 908 the few RT mutations observed were also either in the gradual rebounds (3/11) or in the ongoing replication group (1/9), not in the rapid rebounds (0/12).

Table 1: Incidence of the emergence of mutations during therapy at the first failure time-point.

Profile of PI resistance mutations selected (week 12 onwards)

Selection of PRO or cleavage site mutations by 908/r q.d. was not observed in virus from any of the first failure samples analysed (n = 32, Table 2), or from 12 patients who continued to receive 908/r q.d. despite evidence of ongoing replication, sampled at the last time-point on therapy (median, 8 weeks on continued failing regimen; range, 4–40 weeks). Similarly there was no evidence of reduced susceptibility to APV, or cross-resistance to any other PIs (Table 2). Only patient 1 who had baseline resistance and archived mutations, had reduced susceptibility to APV and saquinavir and resistance to indinavir, lopinavir (LPV), NFV and ritonavir (RTV) (Table 2).

NFV-associated PRO mutations D30N and/or L90M were selected in 39% (21/54) of NFV-treated virological failures. Seventeen (31%) had D30N and three (6%) L90M and one (2%) combined D30N/L90M. In addition, seven patients developed the primary resistance mutation M46I/L, two in the absence of D30N or L90M.

Selection of NRTI resistance mutations

ABC/3TC RT mutations were detectable in virus from four out of 32 patients in the 908/r q.d. arm (Table 1). Only mutations at RT position 184 (V or I) were selected, and a mixture D67d/n was found in virus from patient 2 (Table 2). All shifts in ABC-susceptibility were less than fourfold in magnitude. In the NFV b.i.d. arm, significantly more patients (69% versus 13%; P < 0.001) developed the M184I, V, or X (where X = m/i/v) RT mutations and three patients developed ABC mutations (K65R, n = 2; L74V, n = 1; Table 1).

Discussion

Analysis of the virological failure samples from patients in SOLO have revealed significantly lower incidence of the emergence of resistance to both the PI and NRTI components of the regimen in the 908/r q.d. arm compared with the NFV b.i.d. arm.

A regimen containing 908 might be expected to select PRO mutations similar to those observed with the active moiety APV [⁷]. However, in the virological failure population of SOLO, there was no evidence of selection by the 908/r q.d. regimen of any PRO mutations, which was confirmed by further analysis of virus from 11 patients up to 64 weeks of therapy. In addition, vRNA of nine patients re-suppressed while on study therapy, which is consistent with the lack of genotypic changes and absence of reduced susceptibility to APV (Table 2). No PI cross-resistance was observed except in patient 1, where archival resistance mutations contributed to phenotypic resistance to the six PIs tested.

The absence of the selection of PRO mutations by 908/r in SOLO, is reminiscent of observations made from Study M98-863 [^13,14], in which antiretroviral-naive patients received d4T+3TC with either LPV/r b.i.d. or NFV b.i.d.. Isolates from 37 patients who received LPV/r for 48 weeks were analysed, and none developed resistance to LPV. Furthermore, of 76 NFV-treated patients analysed, 33% (25/76) demonstrated genotypic resistance (D30N or L90M) to NFV, compared with 39% (21/54) in SOLO.

Several factors may influence the emergence of resistance, including drug exposure, where absence of the development of resistance may be attributed either to very low or very high drug levels [^15,16]. Low drug exposure, as may result from poor adherence, could provide insufficient antiviral pressure for resistance selection and permit replication of unaltered virus. Higher drug exposure, as observed with the 908/r q.d. regimen [²], may require the acquisition of multiple mutations, thereby delaying resistance development by provision of a high genetic barrier [^13,14]. Evaluation of the viral load profiles in the NFV b.i.d. and 908/r q.d. arms revealed differences associated with the slopes of virological failure, and acquisition of mutations. This suggested that rapid rebounds which possibly result from stopping therapy are more frequently associated with few or no mutations, whereas gradual rebound or ongoing replication, possibly due to prolonged suboptimal drug exposure, are associated with acquisition of mutations. The high drug exposure of the 908/r regimen would be consistent with the greater genetic barrier producing a delay in the emergence of resistance to both PIs and NRTI. Questionnaire adherence data showed that the 908/r q.d. regimen had higher overall adherence than the NFV b.i.d. regimen, but that the 908/r q.d. virological failures were more frequently associated with poor adherence than the NFV b.i.d. virological failures, suggesting a lower barrier to resistance of the NFV arm [¹⁷].

In conclusion, the absence of emergent protease resistance to 908 or cross-resistance to other PI and reduced NRTI resistance during treatment with a 908/r q.d. regimen supports the use of this boosted PI early in the treatment continuum.

Acknowledgements

The authors thank the patients and investigators who participated in these studies, and ViroLogic Inc. for their contribution to the genotypic and phenotypic analyses. We wish to thank E. D. Blair for his technical writing assistance.