Impact of HIV genotyping and drug levels on the response to salvage therapy with saquinavir/ritonavir

A total of 139 HIV-infected adults failing a protease inhibitor (PI) regimen were recruited in a multicentre, prospective, open trial. All began saquinavir soft gel 1000 mg and ritonavir 100 mg twice a day plus two nucleoside analogues. Ten discontinued treatment because of adverse events. Plasma HIV-RNA levels fell > 1 log at 24 weeks in 44.5% of patients in an intent-to-treat analysis. The number of PI resistance mutations and saquinavir plasma levels independently predicted the virological response. The efficacy and safety of a rescue intervention based on saquinavir 1000 mg boosted by ritonavir 100 mg, both twice a day, plus two reverse transcriptase (RT) inhibitors, was examined in an open, multicentre, prospective trial conducted in Spain on patients failing a protease inhibitor (PI) triple combination. As a secondary goal, the predictive value of baseline genotyping as well as of plasma levels of saquinavir on the therapeutic response were assessed. Patients were enrolled in 2000 at 20 clinical centres. HIV-infected adults with plasma HIV-RNA levels above 1000 copies/ml, while on two nucleoside analogues and at least one PI (excluding saquinavir) for longer than 6 months, were eligible for the study. In all instances, saquinavir/ritonavir was prescribed in combination with two RT inhibitors. Patients were assessed at inclusion (baseline) and every 3 months. The main endpoints for efficacy were plasma HIV-RNA and CD4 cell measurements. A virological response was considered to be significant when plasma HIV-RNA reductions were greater than 1 log or fell to below 50 HIV-RNA copies per ml. The presence of drug resistance mutations was examined in plasma HIV RNA using the ABI 3100 DNA sequencer (Applied Biosystems, Foster City, CA, USA). A set of eight codon substitutions at the protease (L10I/R/V, G48V, I54V/L, A71V/T, V77I, V82A, I84V, and L90M) were considered as saquinavir resistance mutations, according to the most recent guidelines [1,2]. Blood samples for measuring saquinavir concentrations were collected 12 weeks after beginning treatment (to ensure steady state) and just before the morning dose (to estimate trough concentrations or Cmin). Trough concentrations of saquinavir less than 0.1 μg/ml were considered to be suboptimal [3–5]. A total of 139 patients were recruited into the study. The main patient characteristics are summarized in Table 1. The median time on previous antiretroviral therapies was 52 months; the mean time of exposure to PI was 28 months. Up to 86% of subjects had been exposed to two or more PI, indinavir and nelfinavir being the most frequently used. Finally, 60% of patients had been exposed to nevirapine or efavirenz.Table 1: Baseline characteristics of the study population.Saquinavir/ritonavir was mostly prescribed with two nucleoside analogues, but 12 patients (8%) received one nucleoside analogue plus one non-nucleoside analogue. Overall, abacavir was the drug most frequently prescribed with saquinavir/ritonavir. There were no other combinations with saquinavir/ritonavir, such as those including an extra PI or three or more RT inhibitors. The mean plasma HIV-RNA copy number before beginning saquinavir/ritonavir was 4.9 logs. The mean CD4 cell count was 368 cells/mm3. Forty-one individuals did not complete 24 weeks on therapy. The reasons were: one died of HIV dementia, nine had voluntary withdrawal, eight complained of gastrointestinal symptoms, two had liver toxicity, and 21 were lost to follow-up. Overall, 44.5% subjects (62/139) showed a significant virological response at 24 weeks in an intent-to-treat analysis. Considering only patients on treatment, 68% (67/98) showed a virological response at 24 weeks. Up to 50% obtained plasma HIV-RNA values below 50 copies/ml. The mean CD4 lymphocyte increase was +127 cells/μl at 24 weeks. The median number of baseline resistance mutations was four at the RT gene (range 0–13), and three at the protease gene (range 0–10). More than four PI resistance mutations were recorded in 40.4% of patients. The most prevalent primary PI mutation at baseline was 82A/T/F (23%). Saquinavir plasma concentrations were measured in 73 patients with sufficient volume of plasma available 12 weeks after being on treatment. The median saquinavir plasma concentration was 1.43 μg/ml (range 0.2–6.3), and did not differ significantly when comparing virological responders and non-responders. Mean saquinavir plasma levels were approximately 30-fold the EC50 for wild-type viruses. Overall, 83% of patients harbouring less than four saquinavir resistance mutations at baseline reached a significant virological response, whereas it occurred in only 28.6% of those harbouring four or more saquinavir mutations (P = 0.009). Moreover, an association between the lack of virological response and specific PI resistance mutations was recognized only for L90M or V82A. However, the influence of these mutations on the rate of virological response was much lower than taking into account the total number of saquinavir resistance mutations. When the analysis was performed only considering those 73 patients for whom saquinavir plasma levels were available, a significant correlation between virological response and optimal levels of saquinavir (> 0.1 μg/ml) was noted. Fifty-one of the 59 patients (86.4%) with saquinavir plasma levels greater than 0.1 μg/ml reached a virological response. In contrast, it occurred in 50% of subjects (7/14) presenting with saquinavir plasma levels less than 0.1 μg/ml (P = 0.006). In the multivariate analysis, both HIV genotyping and saquinavir levels independently predicted the virological response to saquinavir/ritonavir at 24 weeks (see Table 2). More than 97% of patients harbouring less than four saquinavir resistance mutations together with saquinavir plasma levels greater than 0.1 μg/ml showed a favourable virological response. In contrast, only 12% of subjects with more than four saquinavir resistance mutations and suboptimal saquinavir levels showed a satisfactory virological response.Table 2: Predictive value of HIV genotyping and saquinavir drug levels on the virological response to a saquinavir/ritonavir salvage combination (multivariate analysis).Ritonavir-boosted PI therapy is currently widely used for treating patients who fail conventional highly active antiretroviral therapy regimens [6–10]. In contrast with what is seen with indinavir, amprenavir or lopinavir, the main effect of ritonavir on saquinavir pharmacokinetics is on the Cmax rather than on the elimination half-life [4]. In our trial, the performance of saquinavir/ritonavir in heavily pre-treated patients was quite satisfactory: 68% of patients on treatment showed a significant virological response, and half reached less than 50 HIV-RNA copies/ml. Moreover, the combination of saquinavir/ritonavir was relatively well tolerated. Only 8% of subjects discontinued therapy because of adverse events, mainly gastrointestinal symptoms. Patients who have failed multiple PI-containing regimens often accumulate a large number of resistance mutations at the protease gene. In this setting, cross-resistance to all current PI should be expected [1,2,11]. Several studies have shown that the presence of PI resistance mutations at baseline predicts the virological response to salvage therapy [12–17]. In our study, patients harbouring less than four saquinavir resistance mutations showed the best virological outcome. Similar studies have been conducted using other ritonavir-boosted PI combinations, and for example, for lopinavir/ritonavir (Kaletra), the breakpoint in the number of protease mutations over which the response is compromised seems to be five [18–20]. Interindividual variability in the exposure to antiretroviral drugs is widely recognized and could partly explain the different degrees of inhibition of viral replication provided by the same regimen in distinct patients [21,22]. In our trial, a significant relationship was clearly recognized between virological response and the attainment of adequate saquinavir plasma levels (Cmin > 0.1 μg/ml). All together, these data suggest that adding low doses of ritonavir to saquinavir is the best way to guarantee the achievement of optimal saquinavir levels in most instances. Only through this enhancement in saquinavir plasma concentrations may the inhibition of wild-type and low-level resistant viruses be presumed. In conclusion, our data support the integration of both therapeutic drug monitoring and resistance testing in clinical practice, particularly when managing antiretroviral-experienced patients who begin PI-containing regimens. These methods represent useful tools for optimizing therapy. Acknowledgements The authors would like to thank Drs Andrew Hill and Belén Garbayo for the critical review of the manuscript, and Susana Traseira for helpful assistance during the trial.