HIV Articles  
Gemini: A Noninferiority Study of Saquinavir/Ritonavir Versus Lopinavir/Ritonavir as Initial HIV-1 Therapy in Adults
  JAIDS Journal of Acquired Immune Deficiency Syndromes:Volume 50(4)April 2009pp 367-374
Walmsley, Sharon MD*; Avihingsanon, Anchalee MD; Slim, Jihad MD; Ward, Douglas J MD; Ruxrungtham, Kiat MD; Brunetta, Jason MD_; Bredeek, U Fritz MD, PhD; Jayaweera, Dushyantha MD#; Guittari, Carol Jean MS, RPh**; Larson, Peter MD**; Schutz, Malte MD**; Raffi, Francois MD From the *University of Toronto, Toronto, Canada; HIV Netherlands Australia Thailand, Thai Red Cross AIDS Research Center, Chulalongkorn University, Bangkok, Thailand; Saint Michael's Medical Center, Newark, NJ; Dupont Circle Physicians Group, Washington, DC; _Maple Leaf Medical Clinic, Toronto, Canada; El Rio, Special Immunology Associates, Tucson, AZ; #University of Miami Miller School of Medicine, Miami, FL; **Roche, Nutley, NJ; and University Hospital, Nantes, France.
Introduction: Direct comparison of the efficacy and safety of different agents is needed to guide selection of optimal treatment regimens for therapy-naive HIV-1-infected patients.
Methods: Gemini was a 48-week, multicenter, open-label, noninferiority trial in treatment-naive HIV-1-infected adults randomized to either saquinavir/ritonavir (SQV/r) 1000 mg/100 mg twice a day or lopinavir/ritonavir (LPV/r) 400 mg/100 mg twice a day, each with emtricitabine/tenofovir 200 mg/300 mg every day.
Results: A similar proportion of participants in the SQV/r (n = 167) and LPV/r (n = 170) arms had HIV-1 RNA levels <50 copies per milliliter at week 48: 64.7% vs 63.5% and estimated difference in proportion for noninferiority: 1..14%, 96% confidence interval: -9.6 to11.9 (P < 0.012), confirming that SQV/r was noninferior to LPV/r treatment. There were no significant differences in week 48 CD4 counts between arms. The rate and severity of adverse events were similar in both groups. There were no significant differences in the median change from baseline between arms in plasma lipids except for triglyceride levels, which were significantly higher in the LPV/r at week 48.
Conclusions: In treatment-naive, HIV-1-infected patients, SQV/r treatment was noninferior in virologic suppression at 48 weeks to LPV/r treatment and offered a better triglyceride profile.
participants were allowed to begin or switch treatment to the Meltrex formulation of LPV/r as it became available in their respective countries. Participants switching formulation during the study made comparison of certain AEs more difficult.
Highly active antiretroviral therapy has resulted in striking decreases in HIV-1-related morbidity and mortality.1,2 In addition to antiviral potency and tolerability, many factors are considered in the choice of an initial therapeutic regimen for a patient with HIV-1. These factors include anticipated compliance with therapy, risk of baseline resistance, concurrent illness, concomitant use of other medications, tolerability, and concerns about side effects among others. Thus, it is important that clinicians and patients have multiple, well-characterized therapeutic options to enable individualized therapy.
Regimens based on combinations of 2 nucleoside reverse transcriptase inhibitors (NRTIs) and either a nonnucleoside reverse transcriptase inhibitor (NNRTI) or a ritonavir-boosted protease inhibitor (PI/r) are recommended for initial therapy for the treatment of antiretroviral (ARV)-naive patients.3 In practice, NNRTI-based regimens are used more frequently. Until recently, ARV treatment guidelines recommended boosted lopinavir [lopinavir-ritonavir (LPV/r)] as the only preferred protease inhibitor (PI) for initial treatment of HIV-1-infected patients.4 Numerous recent clinical studies have directly compared other boosted PI regimens with LPV/r.5-9 In the KLEAN trial, treatment with ritonavir-boosted fosamprenavir 700 mg/100 mg twice a day was shown to have similar antiviral efficacy to that of LPV/r twice a day in ARV-naive patients.8 In the ARTEMIS study, darunavir/ritonavir was generally well tolerated and proved noninferior to LPV/r in virologic response,7 whereas in the CASTLE study, a regimen based on atazanavir/ritonavir was noninferior to LPV/r.9 However, a limitation to the use of PI-containing ARV regimens is that they can be associated with metabolic abnormalities such as dyslipidemia and insulin resistance, which may increase the risk of cardiovascular disease.10-12 Metabolic changes have also been attributed to the NRTI backbones, especially those containing stavudine or azidothymidine, and to NNRTIs including efavirenz.13-15
Saquinavir is a well-established PI also in clinical use, but before the new formulation it was not practical for first-line HIV-1 therapy given the high pill burden. The development of Invirase, a 500-mg saquinavir mesylate film-coated tablet formulation, reduced pill burden to that of other PIs used in initial HIV-1 therapy and thus warranted the determination of its relative efficacy compared with other boosted PIs.
The Gemini study was designed to assess the noninferiority of virologic response at <50 HIV-1 RNA viral copies per milliliter after treatment with saquinavir/ritonavir (SQV/r) twice a day vs the preferred PI at the time, LPV/r twice a day, [both SQR/r and LPV/r were administered with concomitant emtricitabine/tenofovir (FTC/TDF) once daily] in treatment-naive patients infected with HIV-1. Saquinavir was administered twice daily using the new formulation, offering a similar pill burden of 6 tablets per day for both treatment arms of the study.
Disposition of Participants

Of 337 participants, 167 were randomized to receive SQV/r (plus FTC/TDF) and 170 were randomized to receive LPV/r (plus FTC/TDF) (Fig. 1). Six participants had no postbaseline safety assessment: 4 (2.4%) in the SQV/r arm and 2 (1.2%) in the LPV/r arm, and were excluded from the safety analysis. Overall, 39 (23..4%) and 35 (20.6%) of participants discontinued treatment from the SQV/r and LPV/r arms, respectively. Eight participants (4.8%) in the SQV/r arm and 13 participants (7.6%) in the LPV/r arm withdrew for safety reasons, and 31 (18.6%) and 22 (12.9%) withdrew in the SQV/r and LPV/r arm for nonsafety reasons. Of these, 11 (6.6%) and 5 (2.9%) participants withdrew due to VF in the SQV/r and LPV/r arms, respectively. Of the 170 participants randomized to the LPV/r arm, 36 participants (21.2%) took the new LPV/r formulation (Meltrex) exclusively, 78 participants (45.9%) did not receive Meltrex, and 56 participants (32.9%) changed LPV/r formulation at some point during the study.
Demographic and Baseline Disease Characteristics
Demographic and baseline disease characteristics for the ITT population (N = 337) were similar between treatment groups (Table 1). Overall, the median age was 38 years (range: 18-66 years), and most participants were male (79.2%). The majority of participants were white (46.9%), with 32.6% black and 19.6% Asian or Pacific islanders. The overall median baseline CD4 cell count was 142 (range: 1-583) cells per cubic millimeter. (Note that the upper limit of the range exceeds the entry criteria limit of 350 cells/mm3 because these patients also had high viral loads and thus were included as they met the need-to-treat priority criterion recommended in the treatment guidelines available at the time of the trial design and conduct.)4 The baseline plasma HIV-1 RNA level was >100,000 copies per milliliter for 65.6% of participants, with a mean ± SD HIV-1 RNA level of 5.18 ± 0.58 log10 copies per milliliter, and 26.9% of participants had prior AIDS-defining events.
Efficacy Assessments
In the ITT analysis, a similar proportion of participants in each of the SQV/r (n = 167) and LPV/r (n = 170) groups had confirmed plasma HIV-1 RNA levels <50 copies per milliliter at week 48, 64.7% vs 63.5%, respectively; estimated difference in proportion for noninferiority was 1.14%, 96% CI: -9.6 to 11.9; P < 0.012. The time course of the proportion of patients with viral loads <50 copies per milliliter is shown in Figure 2A. The results in the PP population at week 48 were consistent with those of the ITT population: in the SQV/r group (n = 148), 65.5% of the participants achieved HIV-1 RNA <50 copies per milliliter compared with 62.1% in the LPV/r group (n = 145; estimated difference in proportion for noninferiority: 3.47%, 96% CI: -8.1 to 15.0; P < 0.006). Because the lower limit of the 2-sided 96% CIs (equivalence to 1-sided 98% CI) for the difference in response between treatment arms in both populations was not ≦-12%, this study met the set criterion of demonstrating noninferiority of SQV/r treatment vs LPV/r treatment.
At week 48, the proportion of participants in the SQV/r group who achieved HIV-1 RNA <400 copies per milliliter was similar to that in the LPV/r group: 72.5% vs 74.7%, respectively (estimated difference: -2.30%, 95% CI: -11.7 to 7.2). The log10 mean reduction from baseline in plasma HIV-1 RNA levels at week 48 was also similar in the 2 treatment arms: -3.39 log10 copies per milliliter in the SQV/r arm vs -3.36 in the LPV/r arm (estimated LSM difference: -0.026, 96% CI: -0.184 to 0.132; P = 0.621; Fig. 2B). The median changes from baseline in CD4 cell counts at week 48 were also similar between the 2 groups: 178 cells per cubic millimeter and 204 cells per cubic millimeter for the SQV/r and LPV/r arms, respectively; P = 0.33 (Fig. 2C).
TLOVR was analyzed based on response criteria of HIV-1 RNA <50 copies per milliliter and HIV-1 RNA <400 copies per milliliter. For participants who achieved a response of HIV-1 RNA <50 copies per milliliter during the study, 16 (9.6%) in the SQV/r arm and 14 (8.2%) in the LPV/r arm lost virologic response. For participants who achieved a response of HIV-1 RNA <400 copies per milliliter, 23 (13.8%) in the SQV/r arm and 23 (13.5%) in the LPV/r arm lost virologic response as defined by TLOVR. There were no statistically significant differences between the 2 treatment arms in the TLOVR analysis using either HIV-1 RNA <50 copies per milliliter (P = 0.439) or HIV-1 RNA <400 copies per milliliter (P = 0.590). The PP analysis of loss of virologic response was similar to that of the ITT analysis.
Virologic Failures
The number of participants who experienced VF did not differ significantly between the treatment arms [SQV/r arm 11/167 (7%) and LPV/r arm 5/170 (3%); P = 0.131] (Table 2). Of these, 4 of 11 in the SQV/r arm and 3 of 5 in the LPV/r arm were documented as poorly adherent, either based on a 4-day participant recall of missed doses of >15% or by specific investigator comment. The majority of participants experiencing VF had new reverse transcriptase mutations (all M184V/I). At VF, only 1 participant in the SQV/r arm and none in the LPV/r was found to have new major PI mutations per International AIDS Society-USA Panel guidelines16 (Table 2).
Adverse Events
There were few discontinuations due to adverse events (AEs): 5 participants (3%) in the SQV/r arm and 12 participants (7%) in the LPV/r arm. Most AEs occurred with similar and low frequencies in both arms with the following exceptions: bronchitis, 9 (6%) vs 2 (1%) (SQV/r vs LPV/r); diarrhea, 11 (7%) vs 24 (14%); headache, 4 (2%) vs 7 (4%); hypertriglyceridemia, 0 vs 5 (3%); nausea, 10 (6%) vs 15 (9%); and upper respiratory tract infections, 4 (2%) vs 8 (5%). The most frequently reported AEs of any grade that were considered related to study drug were gastrointestinal disorders 27 (17%) vs 45 (27%), SQV/r vs LPV/r, respectively. Because of the use of the new LPV/r formulation (Meltrex) by some participants in the LPV/r arm, it was not possible to make direct comparisons or determine the contribution of the LPV/r formulation to the incidence or the severity of gastrointestinal disorders between treatment groups. There were 4 deaths in the LPV/r arm and 3 in the SQV/r arm. In the LPV/r arm, only 1 of the deaths, due to hepatic failure, was considered possibly related to study drug. One death in the SQV/r arm (a crime victim) was considered by the investigator to be remotely related to study drug (Fig. 1).
The incidence of renal AEs was low in both treatment arms, with 7 participants (4%) and 2 (1%) participants reporting events in the SQV/r and LPV/r arms, respectively. Two of the events were categorized as serious: 1 case of nephrolithiasis and 1 participant with hypokalemia, atrial tachycardia, and tuberculous pleuritis (both in the SQV/r group). No participant discontinued because of renal-related AEs. There were 2 participants in the LPV/r group (0.6% overall incidence), with elevated plasma creatinine levels >2 mg/dL that were attributed to FTC/TDF in whom the investigators decreased the dose of TDF and the creatinine levels normalized.
Plasma Lipid Levels
According to the National Cholesterol Education Program guidelines and the AIDS Clinical Trials Group guidelines, optimal or desirable ranges for fasting plasma lipid values are <200 mg/dL for TC, <100 mg/dL for LDL cholesterol, 60 mg/dL for HDL cholesterol, and <150 mg/dL for TG levels.12,17 At baseline, the proportion of participants with TC values meeting or exceeding guideline cutoffs was low and similar in the 2 study arms (Fig. 3A). The proportion of participants with elevated TC and LDL cholesterol values within each treatment arm increased at week 24 and again at week 48 in both treatment arms. In contrast, the proportion of participants with elevated TG values remained relatively stable in the SQV/r arm through week 48 but increased in the LPV/r arm at week 24 and remained increased at week 48. Lipid-lowering therapy was initiated in a small proportion of participants: 6 of the 167 participants in the SQV/r arm (all given statins: 2 before 24 weeks and 4 after 24 weeks) and 7 of the 170 participants in the LPV/r arm (5 given statins and 2 given fibrates; all after 24 weeks)._
FIGURE 3. Summary of participant plasma lipid values at baseline, week 24, and week 48 (ITT population). A, Percentage of participants with plasma lipid values exceeding guideline levels that consequently may warrant clinical intervention. B, Change from baseline in median fasting plasma lipid values at weeks 24 and 48. The respective baseline lipid values are listed at the bottom of this figure.



With the exception of HDL cholesterol levels, the laboratory values for fasting plasma lipid after SQV/r or LPV/r treatment were within the desirable range as described in the National Cholesterol Education Program and AIDS Clinical Trials Group guidelines. As Figure 3 shows, the median changes from baseline in TC, LDL cholesterol, and HDL cholesterol values were not significantly different between treatment groups; however, elevation of TG was significantly higher at both week 24 and week 48 in the LPV/r arm compared with the SQV/r arm. The median change (95% CI) from baseline of the TC:HDL ratio for SQV/r treatment arm was significantly greater than that for LPV/r at week 24: -0.40, (-0.6 to -0.1) vs -0.07 (-0.2 to 0.1; P = 0.024), respectively, but not at week 48: -0.27 (-0.5 to 0.2) vs -0.13 (-0.2 to 0..1; P = 0.473), respectively.
The results of Gemini indicate that SQV/r was noninferior (proportion of participants with viral loads <50 copies/mL at 48 weeks) to treatment with LPV/r when both were used in conjunction with fixed-dose FTC/TDF in ARV-naive HIV-1-infected patients. This result supports the consideration of SQV/r as a preferred option in first-line therapy for HIV-1-infected patients, as reflected in the European AIDS Clinical Society guidelines.18 Although the results of this study support the use of PIs as first-line treatment, NNRTI-based first-line regimens are often preferred due to their advantages in terms of ease of use, a low incidence of gastrointestinal toxicity, and a long half-life that permits tolerance of missed doses. Several trials have demonstrated that efavirenz in particular is an effective first-line treatment option. NNRTI-based regimens have been compared directly with PI-based regimens in clinical trials. In one study, virologic efficacy was comparable between ARV regimens based on efavirenz or atazanavir13 whereas others have found that efficacy was significantly greater with efavirenz compared with LPV/r.19,20 However, in the latter studies, patients with VF who received efavirenz were significantly more likely to develop a NNRTI-associated mutation or a mutation associated with 2 drug classes.19,20 Disadvantages of the currently available NNRTIs include the risk of primary infection with an NNRTI-resistant virus, a low genetic barrier for the development of resistance mutations and central nervous system side effects, and teratogenicity with efavirenz.3 Thus, this option is not appropriate for all patients.
Participants in Gemini, although ARV naive, had advanced HIV disease, evidenced by baseline viral loads, CD4 cell counts, and previous AIDS-defining events (Table 1), which are important considerations that may make it more difficult for participants to achieve undetectable HIV-1 RNA plasma levels at 48 weeks.
Limitations of this study include its open-label design, which may bias assessment of tolerability but would be unlikely to bias the virologic or immunologic end points. In addition, participants were allowed to begin or switch treatment to the Meltrex formulation of LPV/r as it became available in their respective countries. Participants switching formulation during the study made comparison of certain AEs more difficult. Another limitation is that twice-daily dosing of PIs was used, although once-daily regimens are currently indicated for some PIs; however, SQV/r has been listed as a recommended component for initial therapy in the 2008 IAS-USA guidelines despite the limitation of twice-daily dosing.21
The incidence and severity of clinical AEs were similar in the 2 treatment arms. Only 2 of 337 subjects (0.6%) had elevated plasma creatinine levels that were attributed to FTC/TDF; similar to the rates reported with TDF and lamivudine (2/296, 0.7%) or stavudine with lamivudine (2/301, 0.7%) when combined with efavirenz in Gilead study 903,14 and to those of TDF with emtricitabine combined with efavirenz (0/257, 0%) when compared with azidothymidine and lamivudine combined with efavirenz (2/254, 0.8%) in Gilead study 934.15
VF was uncommon in both treatment arms and not significantly different between arms.. In subjects who met criteria for VF, comparison of the results from HIV-1 genotypic testing of VF samples to their baseline samples showed that only 1 participant had developed new major PI-associated resistance mutations. These results confirm those from previous studies that have shown few emergent HIV-1 PI-resistant mutations among treatment-naive patients treated for 48 weeks with combination therapy that included a PI/r.7,8 Together the results from these studies and the current study indicate the low incidence in the emergence of PI-resistant mutations,22,23 and thus concerns over the emergence of such protease mutations should not limit the use of boosted PIs in combination therapy of treatment-naive, HIV-1-infected patients. Additionally, in the SQV/r and LPV/r treatment arms, 5 and 4 participants, respectively, showed emergence of NRTI-associated resistance mutations (all M184V/I). No patient developed a K65R HIV-1 mutation.
A growing concern with current HIV therapy is that treatment-related dyslipidemia typically associated with PI therapy may contribute to atherogenesis and thus increase the risk for coronary artery disease. This risk is increased further when other risk factors such as smoking and other HIV-associated metabolic abnormalities are present, such as insulin resistance10,24 and visceral adiposity.25 When designing an initial treatment regimen, the ideal PI would be one that is efficacious and minimally affects serum lipids. The results from this study show that both SQV/r and LPV/r treatment caused an increase from baseline values in fasting TC, LDL cholesterol, HDL cholesterol, and TG levels. However, participants in the LPV/r arm had significantly greater median increases from baseline TG levels at week 24 and week 48 compared with those in the SQV/r arm (P < 0.001 and P < 0.002, respectively). Elevated TG levels have been reported in LPV/r treatment arms in other clinical trials.26-28 High TG level is associated with metabolic syndrome and in some studies has been shown to independently increase the risk for coronary artery disease.29,30
In conclusion, the results from the Gemini study demonstrate that SQV/r is noninferior to treatment with LPV/r in efficacy, is similar in tolerability, and results in significantly lower increases in TG values. These data provide further information on therapeutic options for the physician who must weigh the risk factors associated with initial therapy for HIV-1 therapy patients. The results also add further support to the current treatment guidelines that boosted PIs are a viable component of highly active antiretroviral therapy in first-line therapy of HIV-1-infected patients.

HIV-1-infected participants were enrolled from multiple centers in the United States (including Puerto Rico), Canada, France, and Thailand, and were eligible if they had plasma HIV-1 RNA >10,000 copies per milliliter, CD4 cell counts of ≦350 cells per cubic millimeter, were male or nonpregnant, nonnursing females 18 years of age, and were naive to ARV therapy (<2 weeks total prior exposure). Participants were excluded if they had active hepatitis B infection (HBsAg+), evidence of an opportunistic infection, or an intercurrent illness.
This study was conducted in accordance with the Helsinki II declaration and the International Committee on Harmonization's Good Clinical Practice Guideline. Local institutional review boards or independent ethics committees approved the study protocols and all amendments. Written informed consent was obtained before any trial-specific procedure was performed.
Primary and Secondary Objectives
The primary objective was to evaluate the efficacy of saquinavir (Invirase; Roche, Nutley, NJ) 1000 mg twice a day boosted with ritonavir 100 mg twice a day (Norvir; Abbott Laboratories, Chicago, IL) given with a fixed dose of FTC/TDF 200 mg/300 mg every day (Truvada; Gilead Sciences, Foster City, CA) vs fixed-dose LPV 400 mg twice a day boosted with ritonavir 100 mg twice a day (Kaletra capsules or Meltrex tablet formulation; Abbott Laboratories), also given with FTC/TDF 200 mg/300 mg every day, in treatment-naive adults infected with HIV-1. Primary efficacy was defined as the percentage of participants with HIV-1 RNA viral load of <50 copies per milliliter at week 48. Secondary efficacy assessments included the following: the proportion of participants with plasma HIV-1 RNA viral load of either <400 or <50 copies per milliliter at each study visit; the change from baseline in CD4 cell counts; and the time to loss of virologic response (TLOVR) through week 48, with virologic failure (VF) defined as any 2 consecutive measurements of HIV-1 RNA viral load >400 copies per milliliter at least 2 weeks apart at week 16 or thereafter. Safety assessments included the following: general safety and tolerability; the magnitude of change in fasting plasma levels of total cholesterol (TC), high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol (measured by direct assay), and triglyceride (TG); and the use of lipid-lowering agents.
Treatment allocation was based on a computer-generated centralized randomization schedule. Participants were allowed to begin or switch treatment to the new Meltrex formulation of LPV/r as it became available in their countries. Laboratory and clinical evaluations of participants were carried out at weeks 2, 4, 8, 12, 16, 20, 24, 36, 40, and 48. Immunologic differences were assessed by change in absolute CD4 cell count. HIV-1 RNA plasma concentrations were measured with the Roche Amplicor HIV-1 Monitor test (Roche Diagnostics, Indianapolis, IN). Samples for viral genotyping were taken at baseline and at all subsequent visits and stored. HIV-1 RNA genotypes were assessed in samples from those participants who met the criteria for VF during the trial and were compared with viral genotypes present in their baseline samples. Standard HIV genotyping was carried out by ICON Central Laboratories (Farmingdale, NY)..
Statistical Analysis
Primary efficacy outcomes were based on analyses of the intent-to-treat (ITT) population and the per-protocol (PP) population. The ITT population analysis included all participants exposed to study medication and with at least 1 postbaseline viral load evaluation. Participants with missing data or data obtained after discontinuation in the primary efficacy analysis were regarded as nonresponders. The PP population included all those participants in the ITT population except those who: (1) did not meet 1 or more of the entrance criteria; (2) took restricted medications during the treatment period; or (3) had other major protocol violations. Observed values were used for all secondary efficacy outcomes analysis that included the following: change from baseline in plasma HIV-1 RNA viral load (copies/mL); number and percentage of participants with an HIV-1 RNA viral load <400 copies per milliliter; and change from baseline in CD4 lymphocyte.
The primary efficacy analysis was defined as the proportion of participants who achieved HIV-1 RNA <50 copies per milliliter at 48 weeks (ITT population; missing data = failure). The trial was powered to assess noninferiority of treatment with SQV/r compared with treatment with LPV/r. The null hypothesis was tested based on a 1-sided 98% confidence interval (CI) for the difference in response between treatment arms of ≦-12%. CIs of 99.9% and 99.1% were used to determine noninferiority in 2 interim analyses. An analysis of covariance was used to model the treatment differences for each treatment group in the change from baseline in plasma viral load and CD4 cell count.
Secondary efficacy end points also included TLOVR analysis, which was determined as follows: in nonresponders or failures (participants who never achieved confirmed viral loads <50 copies/mL or <400 copies/mL), TLOVR was defined as the date of the first dose of study medication. In responders, TLOVR was defined as time from the first virologic response to the first of the following events: (1) VF; (2) suspected rebound in viral load; (3) death; (4) permanent study drug discontinuation; or (5) loss to follow-up.
Safety analysis was based on all randomized participants who received study medication and had at least 1 safety assessment.
  iconpaperstack view older Articles   Back to Top