Patients and Methods

The study was conducted from June 2002 through April 2006 (ClinicalTrials.gov reference NCT00412646).

Patients. Eligible patients included HIV-1–infected adults with a VL ⩾1000 copies/mL who were previously treated with NNRTIs, nucleoside reverse-transcriptase inhibitors (NRTIs), and protease inhibitors (PIs; each for ⩾3 months) and who were currently receiving a stable treatment regimen (for ⩾8 weeks). After a stage 1 protocol amendment, patients were eligible if they had had treatment interrupted for ⩾8 weeks and continued to not receive antiretroviral therapy until baseline. Patients were screened at baseline to determine the susceptibility of their infecting virus to antiretroviral agents; however, infection with NNRTI-resistant virus or prior receipt of a virologically failing regimen was not required for enrollment.

Exclusion criteria included an active AIDS-defining illness or clinically significant disease, history of a severe allergic or dermatological reaction leading to discontinuation of therapy with any NNRTI or abacavir, and use of an investigational medication (except tenofovir) within 30 days before ETR administration. The protocol was approved by independent ethics committees, and the trial was performed in accordance with Good Clinical Practice and the Declaration of Helsinki. All patients gave written consent.

Study design. The original primary objective was to evaluate antiretroviral activity. After new safety findings were reported in 2002 (Tibotec BVBA, unpublished data), the trial design and primary objective were revised to evaluate the safety and tolerability of ETR (experimental formulation TF035) BID plus an OBR over 48 weeks.

In stage 1 of the 2-stage dose-escalation study, patients were randomized centrally (1:1:1) to receive placebo, 400 mg BID of ETR, or 800 mg BID of ETR. Different patients were enrolled in stage 2 and were randomized (1:2:4) to receive placebo, 800 mg BID of ETR, or 1200 mg BID of ETR. After 48 weeks, patients who derived benefit from treatment could continue with ETR treatment (plus OBR) for an additional 96 weeks (2 optional 48-week extension periods).

Initially, permitted options for the investigator-selected OBR comprised combinations of NRTIs and/or PIs (ritonavir-boosted saquinavir or lopinavir), of which ⩾2 had to be scored as active on the basis of vircoTYPE HIV-1 (VirtualPhenotype) results at screening. Abacavir use was allowed only if therapy was continued from the screening period and there was no evidence of hypersensitivity for ⩾8 weeks. After a stage 1 protocol amendment, newly recruited patients could receive enfuvirtide (ENF). OBR dose adjustments and within-class substitutions were allowed for tolerability reasons.

After a 12-week double-blind treatment phase in both stages, a formal interim analysis was performed with use of all available data and was reviewed by the Data and Safety Monitoring Board. Investigators and patients remained blind to study treatment for the remaining 36 weeks.

Randomization in each stage was stratified by plasma VL (<10,000 copies/mL and ⩾10,000 copies/mL) at the time of screening and region (Canada, northern and central Europe, and southern Europe), with use of an adaptive minimization technique. Treatment adherence was assessed by pill counts.

Patient assessments. Patients were seen at weeks 1 and 2, then every 2 weeks until week 8, and then every 4 weeks from week 8 through week 24, with subsequent assessments every 8 weeks. Safety assessments included monitoring of the incidence of AEs and HIV-related events, measurement of vital signs, electrocardiograms, and clinical laboratory tests. Specific neuropsychiatric events that are commonly associated with efavirenz [5] were categorized as neuropsychiatric “events of interest.”

Plasma preparation tubes (PPTs) were used to collect samples for VL assessment. In addition, beginning in April 2004, EDTA tubes were used at specific times; during the course of the trial, data became available that revealed that, depending on the handling and storage of the samples, PPTs do not always provide an accurate assessment of plasma VL [19]. VL was determined using the Roche Amplicor HIV-1 monitor test, version 1.5. Viral phenotype and genotype were determined by Virco BVBA with use of the Antivirogram and vircoTYPE HIV-1 (previously VirtualPhenotype), respectively. The pharmacokinetics of ETR (trough plasma concentrations and 12-h exposures [12-h area under the curve]) were estimated by population-pharmacokinetic modeling with use of individual plasma concentrations measured in sparse samples obtained at predefined times during the trial.

Statistical analysis. Sample size was determined on the basis of predicted number of patients required to show a dose-response relationship for a change in log10 plasma VL between baseline and week 24. With the assumption of a change in plasma VL of 1.2 log10 copies/mL in the placebo group and an ETR dose-related response, 40 patients per arm were considered to be sufficient (with 5% significance and 2-sided 80%–95% power, depending on the dose-response relationship). However, to assess the dose-response relationship between the ETR group and the placebo group, an additional 20 patients per group were required in stage 1. Thus, a total of 180 patients was necessary.

The primary analysis used the intention-to-treat population (all patients who received at least 1 dose of study medication). For the safety analyses, data from both stages were combined for each dose. The changes in log10 VL and CD4 cell count from baseline were analyzed with use of analysis of covariance models.

Analysis of dose trend for virologic response parameters at week 48 and comparison of active dose groups with placebo groups were performed with use of the Cochran-Armitage trend test and Fisher's exact test, respectively; adjustments for baseline log10 plasma VL were analyzed in a logistic-regression model.

Results

Patient disposition and baseline characteristics. Fifty investigators from 11 countries enrolled patients. The last patient was randomized to a treatment arm in July 2004. Of 408 patients screened, 240 patients were randomized and received treatment (166 in stage 1 and 74 in stage 2) (figure 1). Of these 240 patients, 182 (75.8%) had a week 24 visit, and 166 (69.2%) had a week 48 visit. Demographic parameters and baseline antiretroviral mutations were well balanced across groups (table 1).

Treatment interruption at screening and ENF use during the treatment period were more frequent during stage 2 than during stage 1 (24.3% vs. 11.4% and 39.2% vs. 14.5%, respectively). During both stages, the frequency of ritonavir-boosted lopinavir use was lower among the ETR groups, compared with the placebo groups. During stage 2, the frequency of de novo ENF use was lower among the placebo group (18.2%) than it was among the ETR groups (30.0% among the patients who received 800-mg doses and 44.2% among the patients who received 1200-mg doses). The primary efficacy analysis was adjusted for differences in ritonavir-boosted lopinavir and ENF use, because this may have affected the relative potency of the OBR. During both stages, there was a median 1.6-fold change (overall range, <0.1-fold to 377.8-fold) in the 50% effective concentration for ETR from baseline.

Safety and tolerability. The safety results include all information available at the time of analysis. A substantial number of patients were treated for >48 weeks (table 2).

No deaths occurred during the treatment period. No statistically significant differences in overall incidence and severity of AEs between the ETR group and the placebo group were observed (table 2). Overall, the most common AEs were diarrhea, abdominal pain, headache, nausea, and rash (any type) (table 2). The incidence of diarrhea was lower among ETR-treated patients than it was among patients who received placebo (28.7% vs. 42.4%; P<.05). The most common grade 3 or 4 AEs in the combined ETR group were hypertriglyceridemia (which occurred in 3.4% of patients in the ETR group and 3.0% of the patients in the placebo group) and clinical pancreatitis (which occurred in 2.3% of patients in the ETR group and 1.5% of patients in the placebo group).

There was a trend toward a higher incidence of certain AEs (including skin and subcutaneous tissue disorders, psychiatric disorders, musculoskeletal and connective tissue disorders, and renal and urinary disorders) among the patients who received 1200 mg BID of ETR, compared with the patients who received the other dosages of ETR. However, this trend did not reach statistical significance, and there was no clear relationship between dose and any individual AE.

The overall incidence of neuropsychiatric events of interest was comparable across groups (46.6% in the combined ETR group and 45.5% in the placebo group; P=.886). Most of these events had a severity grade of 1 or 2. The most common nervous system disorders (i.e., those that occurred in >2 patients) were headache (21.3% in the ETR group vs. 19.7% in the placebo group), dizziness (10.9% vs. 6.1%), and memory impairment and somnolence (each 1.1% vs. 1.5%). Although no relationship was apparent between ETR dose and any individual neuropsychiatric AE, there was a trend for an increasing incidence of psychiatric disorders with increasing ETR dose. The most common psychiatric disorders (i.e., those that occurred in >2 patients) were insomnia (8.0% in the ETR group vs. 4.5% in the placebo group), depression (4.6% vs. 6.1%), anxiety (2.3% vs. 3.0%), sleep disorder (1.7% vs. 3.0%), stress symptoms (1.7% vs. 0.0%), abnormal dreams and aggravated depression (each 1.1% vs. 1.5%), and loss of libido (1.1% vs. 0.0%). The incidences of nervous system disorders (1.7% in the ETR group vs. 0.0% in the placebo group) and psychiatric disorders (1.1% vs. 0.0%) were low; however, such events led to treatment discontinuation.

No clinically relevant difference in the incidence of hepatic AEs (or related laboratory abnormalities) was noted between the ETR group and the placebo group (3.4% vs. 6.1%; P=.469); none of these events led to trial discontinuation. No grade 3 or 4 hepatobiliary disorders were reported.

The incidence of rash (any type) among patients who received ETR was 19.5%, compared with 12.1% among patients who received placebo (P=.25); there was no relationship between incidence of rash and ETR dose. Rash that was considered to possibly be related to ETR generally occurred early (median time from treatment initiation to rash onset, 11 days; range, 8–770 days) (figure 2) and was of limited duration (median duration of rash, 10 days; range, 3–64 days). The most common types of rash were erythema and papular rash. One case of maculopapular rash with fever in the group of patients who received 400 mg BID of ETR and 1 case of toxic skin eruption in the placebo group were considered to be severity grade 3; both resulted in permanent discontinuation of study medication. An additional 5 ETR-treated patients (3 who received 800 mg BID and 2 who received 1200 mg BID) permanently discontinued treatment because of rash. No case of Stevens-Johnson syndrome, toxic necrolysis, or erythema multiforme was observed.

The most common serious AEs reported in the combined ETR group were pancreatitis (in 4 patients [2.3%] in the ETR group and in 1 patient [1.5%] in the placebo group) and pneumonia (in 2 patients [1.1%] in the ETR group and in no patients in the placebo group). Three serious pancreatitic AEs were considered to be possibly related to study medication (in 2 patients who received 400 mg BID of ETR and in 1 patient who received placebo). Associated risk factors for pancreatitis, including use of didanosine, history of pancreatitis, and/or biliary obstruction, were present in all patients.

No consistent, clinically relevant changes in laboratory parameters were observed; mean and median changes from baseline were generally small and comparable across groups. No important differences were noted with respect to the incidence of grade 3 or 4 laboratory abnormalities or laboratory test results among the ETR groups or between the combined ETR group and the placebo group (40.2% in the combined ETR group vs. 39.4% in the placebo group).

Virologic and immunologic efficacy. Because of the limitations of using PPTs for VL assessment, we reported only data for samples collected in EDTA tubes. In contrast to stage 2, not every patient in stage 1 had samples collected in EDTA tubes at weeks 24 (131 of 166 patients) and 48 (128 of 166 patients), because the trial was ongoing when confirmatory EDTA sampling was instituted.

During stage 1, no statistically significant difference was noted between the ETR group and the placebo group for the least square means estimates of the (imputed) change in log10 plasma VL between baseline and weeks 24 and 48 (figure 3A). However, in subgroup analyses of stage 1 data for patients who either had virus with >2 primary PI mutations or were not receiving an active PI in the OBR, an additional antiviral effect was noted in the group that received 800 mg BID of ETR, compared with the group that received 400 mg BID of ETR and with the placebo group.

During stage 2, statistical superiority for the least square means estimates of the (imputed) change in log10 plasma VL between baseline and week 24 was observed for 800 mg BID and 1200 mg BID of ETR versus placebo (P=.009 and P=.049, respectively) (figure 3B). There was a trend showing an additional benefit with either ETR dosage over placebo at week 48 (figure 3B).

During stage 1, mean CD4 cell count increased significantly between baseline and week 48 in the groups that received 400 mg BID and 800 mg BID of ETR and the placebo group (for each group at week 48 vs. baseline, P<.001) (figure 4A). No statistically significant differences in CD4 cell count were observed between groups during stage 1. In contrast, in stage 2, both ETR groups, but not the placebo group, showed a statistically significant increase in CD4 cell count between baseline and week 48 (800 mg BID of ETR baseline vs. week 48, P=.039; 1200 mg BID of ETR baseline vs. week 48, P=.02) (figure 4B); these increases were also statistically significantly different between the ETR groups and the placebo group (800 mg BID of ETR vs. placebo, P=.025; 1200 mg BID of ETR vs. placebo; P=.019).

Baseline NNRTI resistance (fold change in the 50% effective concentration of ETR and the number and/or type of NNRTI mutations [20, 21]) did not influence the virologic results in either stage. In the 400-mg BID and 1200-mg BID ETR groups, virologic response was lower when >2 baseline NNRTI resistance–associated mutations were present (from a list identified by a comprehensive literature review [21]) or if there was a >10-fold change in the 50% effective concentration of ETR (arbitrary threshold). However, these factors did not influence response in the 800-mg BID ETR group.

Pharmacokinetics. ETR pharmacokinetic parameters increased in a dose-proportional manner between the 400-mg BID ETR group and the 800-mg BID ETR group, but less than a dose-proportional increase was observed between the 800-mg BID ETR group and the 1200-mg BID ETR group. Sex, age, body weight, and race (white vs. black patients), and concomitant use of PIs and/or tenofovir did not have statistically significant effects on ETR exposure. No relationships between ETR pharmacokinetics and safety parameters were observed.

Discussion

Our results demonstrate that the tolerability of long-term ETR therapy was comparable to that of placebo in HIV-1–infected patients experienced with ⩾3 classes of antiretroviral agents. Among ETR-treated patients, the incidence of rash was similar to or less than that reported among patients who received efavirenz or nevirapine [5, 10]. ETR-associated rash was generally mild to moderate in severity, occurred early, was typically transient, and infrequently led to treatment discontinuation. No case of Stevens-Johnson syndrome, toxic epidermal necrolysis, or erythema multiforme was observed. Neuropsychiatric events or hepatotoxicity were not associated with ETR use in this study.

The incidence of diarrhea was comparable among ETR groups and was markedly lower in the ETR groups than in the placebo group. We speculated that the more frequent use of ritonavir-boosted lopinavir in the placebo group may have contributed to this finding. Because diarrhea is known to be associated with ritonavir-boosted lopinavir use and was the only event that was significantly different between groups, a post hoc analysis was performed to determine whether the incidence remained significantly different when accounting for ENF and ritonavir-boosted lopinavir use. Results revealed that 25 of 52 patients in the placebo group and 37 of 108 patients in the ETR group who received ritonavir-boosted lopinavir had diarrhea. Therefore, when ritonavir-boosted lopinavir use was accounted for, the incidence of diarrhea across groups was similar. The imbalance in the use of ritonavir-boosted lopinavir and ENF among groups was attributable to an amendment to the inclusion criteria during the recruitment of patients in stage 1. This amendment redefined the requirements for the OBR from an active PI and ⩾1 susceptible NRTI to ⩾2 susceptible antiretrovirals from any allowed class; the use of ENF was also allowed. This protocol amendment resulted in a heterogeneous population in stage 1 (between patients enrolled before and patients enrolled after the implementation of the protocol amendment) and imbalances among groups with respect to the underlying antiretroviral regimen. It was recognized that intergroup disparities in the OBR, particularly use of ritonavir-boosted lopinavir and ENF, had the potential to affect the efficacy results in the trial. Therefore, the primary efficacy analysis was adjusted to account for these factors.

This study offers additional evidence supporting the antiretroviral effect of ETR. However, the use of PPTs for sample collection limited our ability to fully explore this effect. During the trial, newly available data suggested that PPTs—depending on the handling and storage of samples—do not always provide an accurate assessment of plasma VL, particularly when HIV-1 RNA levels are <1000 copies/mL [22]. Consequently, during stage 1 recruitment, parallel VL assessments were conducted with plasma samples collected in EDTA tubes. During stage 1, no overall difference in antiviral efficacy was noted among groups. During stage 2, there was an added benefit in antiretroviral activity in both ETR groups, compared with the placebo group, when EDTA tubes were used. A statistically significant benefit in CD4 cell recovery was also observed during stage 2 with both ETR doses, compared with placebo.

Recent reports have demonstrated the strong antiretroviral efficacy of ETR among patients with advanced HIV-1 disease and multidrug-resistant virus over 24 weeks [16–18]. In this study, baseline NNRTI resistance did not influence VL results during either stage. Although responses were lower in patients who received 400 mg BID or 1200 mg BID of ETR, when >2 baseline NNRTI resistance–associated mutations were present, and if there was a >10-fold change in the 50% effective concentration of ETR, this finding should be interpreted with caution because of the small number of patients in each subgroup and the lack of impact on the response with 800 mg BID of ETR.

Antiretroviral efficacy was greater with 800 mg BID of ETR than with 400 mg BID of ETR or placebo among patients who did not receive an active PI in their OBR and among patients infected with virus with >2 primary PI mutations. Although no clear association between dose and ETR pharmacokinetics was identified, no additional pharmacokinetic or efficacy benefits were seen with 1200 mg BID of ETR, compared with 800 mg BID of ETR; a trend toward higher incidences of AEs in certain body systems was seen with 1200 mg BID of ETR. On the basis of data from this and additional phase IIb studies, a new formulation of ETR (200 mg BID; also known as F060) that provides exposure comparable to that of 800 mg BID of ETR was selected for further development [23]. This formulation was used in the phase III DUET trials [17, 18], which demonstrated superior virologic suppression over 24 weeks with 200 mg BID of ETR, compared with placebo, both with a robust background regimen. In addition, except for rash, ETR had a safety and tolerability profile comparable to that of placebo, confirming the findings from the current study.

In our study, long-term ETR therapy was generally safe and well tolerated in treatment-experienced HIV-1–infected patients. The addition of ETR to an antiretroviral regimen produced a safety and tolerability profile comparable to that of placebo. Despite confounders to the efficacy evaluation, further indication of antiretroviral activity in treatment-experienced patients was observed. Data from this study contributed to the selection of 800 mg BID of ETR for further clinical development (equivalent to 200 mg BID of ETR in the phase III formulation).