MaxCmin1 Study: SQV/rtv 1000/100 mg twice daily vs IDV/rtv
Randomized Trial to Evaluate Indinavir/Ritonavir versus Saquinavir/Ritonavir in Human Immunodeficiency Virus Type 1Infected Patients: The MaxCmin1 Trial
Ulrik Bak Dragsted,1 Jan Gerstoft,2 Court Pedersen,3 Barry Peters,5 Adriana Duran,8 Niels Obel,4 Antonella Castagna,12 Pedro Cahn,9 Nathan Clumeck,13 Johan N. Bruun,14 Jorge Benetucci,10 Andrew Hill,6 Isabel Cassetti,11 Pietro Vernazza,15 Mike Youle,7 Zoe Fox1, and Jens D. Lundgren,1 for the MaxCmin1 Trial Groupa
1Hvidovre University Hospital and 2Rigshospitalet, Copenhagen, 3Odense University Hospital, Odense, and 4Aarhus University Hospital, Aarhus, Denmark; 5St. Thomas Hospital, 6Roche Products Limited, and 7Royal Free Hospital, London, United Kingdom; 8Hospital J. M. Ramos Mejia, 9Fundacíon Huesped, 10F. J. Muniz (FUNDAI), and 11Helios Salud, Buenos Aires, Argentina; 12Università Vita e Salute, Ospedale San Rafaele, Milan, Italy; 13Centre Hospitalier Universitaire Saint-Pierre, Brussels, Belgium; 14Ullevål University Hospital, Oslo, Norway; 15Kantonsspital, St. Gallen, Switzerland
The Journal of Infectious Diseases 2003;188:635-642
This trial assessed the rate of virological failure at 48 weeks in adult human immunodeficiency virus (HIV) type 1infected patients assigned indinavir/ritonavir (Idv/Rtv; 800/100 mg 2 times daily) or saquinavir/ritonavir (Sqv/Rtv; 1000/100 mg 2 times daily) in an open-label, randomized (1 : 1), multicenter, phase 4 design.
Three hundred six patients began the assigned treatment. At 48 weeks, virological failure was seen in 43 (27%) of 158 and 37 (25%) of 148 patients in the Idv/Rtv and Sqv/Rtv arms, respectively. The time to virological failure did not differ between study arms (P = .76). When switching from randomized treatment was counted as failure, this was seen in 78 of 158 patients in the Idv/Rtv arm, versus 51 of 148 patients in the Sqv/Rtv arm (P = .009).
A switch from the randomized treatment occurred in 64 (41%) of 158 patients in the Idv/Rtv arm, versus 40 (27%) of 148 patients in the Sqv/Rtv arm (P = .013). Sixty-four percent of the switches occurred because of adverse events. A greater number of treatment-limiting adverse events were observed in the Idv/Rtv arm, relative to the Sqv/Rtv arm.
In conclusion, Rtv-boosed Sqv and Idv were found to have comparable antiretroviral effects in the doses studied.
Patients randomized to receive Sqv/Rtv were allowed to change from the Sqv soft-gel formulation (Fortovase; Roche) to the hard-gel formulation (Invirase; Roche) without this being considered a switch from the assigned treatment.
Nine patients switched from Idv/Rtv to Sqv/Rtv, and 4 patients switched from Sqv/Rtv to Idv/Rtv. There was a significantly higher percentage of patients in the Idv/Rtv arm (41%) than in the Sqv/Rtv arm (27%) who prematurely switched from the assigned treatment (P = .013). Twenty-two patients reduced the dose of the assigned treatment during follow-up (21 in the Idv/Rtv arm and 1 in the Sqv/Rtv arm).
Per study protocol, the primary population for analysis was the intention-to-treat/exposed (ITT/e) population, including all randomized patients who had taken at least 1 dose of the assigned treatment. This analysis is also termed the "ITT switch included" analysis. In the other protocol-stipulated analysis, switching from the assigned treatment constituted failure (ITT/e/switch = failure [ITT/e/s]). The higher discontinuation rate in the Idv/Rtv arm resulted in a significantly higher virological failure rate in this arm in the ITT/e/s analysis (P = .009, log rank test).
At week 48, 203 (68%) of 306, 155 (51%) of 306, and 186 (93%) of 201 patients had a VL of <50 copies/mL in the ITT/e, ITT/e/s, and during-treatment analyses, respectively. Only when switching from the assigned treatment was counted as having a VL of >50 copies/mL (ITT/e/s) was a significant difference observed, with more patients in the Sqv/Rtv arm having a suppressed VL at week 48 (57% vs 46%,actual percentages not presented in published article but reported by Roche press release, and looked the same by visual observation of graph in publication).
Of the patients exposed to the study medication, 100 (33%) of 306 experienced at least 1 AE of grade 3 or 4 (65 [41%] in the Idv/Rtv arm vs. 35 [24%] in the Sqv/Rtv arm; P = .001, 2 test). Of these, the treating physician assessed the relationship to the assigned treatment as being at least possible in 46 (29%) in the Idv/Rtv arm versus 19 (13%) in the Sqv/Rtv arm (P = .001, 2 test). There was a significant difference between the 2 study groups in the distribution by organ system of AEs grade 3 and 4, with a higher number of renal, dermatological, and gastrointestinal side effects in the Idv/Rtv arm (data not shown).
Median percentage change from baseline compared to week 48 in fasting total cholesterol, low-density lipoprotein (LDL) cholesterol, and total triglyceride levels in the intention-to-treat/exposed analysis: total cholesterol- 17% SQV/rtv vs 8% IDV/rtv; LDL cholesterol- 18% SQV/rtv vs 3% IDV/rtv; total triglycerides- 22% SVQ/rtv vs 9% IDV/rtv.
Equivalence was observed for efficacy, whereas Idv/Rtv lead to an increased risk of treatment-limiting AEs and AEs of grade 3 and/or 4. As a consequence of the safety profile of Idv/Rtv, fewer patients continued to receive this treatment throughout the 48 weeks, leading to differences in the efficacy analyses, in which continuation with study medication influence the outcome. In addition, Idv/Rtv was found to cause a higher risk of elevating blood levels of lipids and bilirubin.
Cohort studies have shown that, among human immunodeficiency virus (HIV)infected patients beginning highly active antiretroviral therapy (HAART) and achieving suppression of HIV-1 RNA to levels below detection, the annual rate of virological rebound is 15%. The main reasons for the failure of HAART are treatment-limiting toxicity, adherence problems, virological failure, and low potency of the drugs. Other studies have shown long-term virus suppression to be dependent on safety, good adherence, and high plasma concentrations of antiretroviral drugs. Ritonavir (Rtv) boosting (i.e., Rtv in doses of 50200 mg), in combination with other protease inhibitors (PIs), results in higher plasma concentration of these other PIs. This is due to inhibition of the P450CYP3A4 enzyme system in the intestine and liver and, possibly, inhibition of P-glycoprotein efflux. Other benefits of Rtv boosting are a reduction in the number of doses, from 3 times daily (t.i.d.) to 2 times daily (b.i.d.), fewer restrictions on food intake, and a lower pill burden, which is associated with better adherence. All these factors have been associated with a better treatment outcome.
PATIENTS, MATERIALS, AND METHODS
This is a randomized (1 : 1), phase 4, open-label, multicenter trial involving 28 sites in 13 countries. A heterogeneous population was enrolled, including patients who were PI naive, PI intolerant, or for whom PI therapy had failed. PI-experienced patients with prior use of either of the study drugs were not precluded from participation; however, only patients with an equal chance of benefit from and/or risk of development of treatment-related AEs to the 2 study PIs at the time of screening could be randomized. This assessment was made by site physicians, and the final decision was made by the trial physician at the Copenhagen HIV Programme (CHIP) on the basis of ART history, prior virological and clinical failure, and available resistance tests. Before randomization, the treating physician decided the concomitant use of at least 2 nucleoside reverse-transcriptase inhibitors (NRTIs) and/or non-NRTIs (NNRTIs).
Patients randomized to receive Sqv/Rtv were allowed to change from the Sqv soft-gel formulation (Fortovase; Roche) to the hard-gel formulation (Invirase; Roche) without this being considered a switch from the assigned treatment. During the trial, modification of the randomized treatment was allowed in the case of virological failure or treatment-limiting toxicities. If available, dose reduction was performed on the basis of therapeutic-drug monitoring. Of note, patients experiencing virological failure, according to the protocol's definition, were allowed to continue receiving the assigned treatment at the discretion of the treating physician.
Definition of virological, immunological, and clinical failure.
For patients entering the study with a VL of <200 copies/mL, virological failure was defined as a VL of 200 HIV-1 RNA copies/mL. For patients entering the study with a VL of 200 copies/mL, virological failure was defined as any increase in HIV-1 RNA load of 0.5 logs and/or a VL of 50,000 HIV-1 RNA copies/mL at week 4, 5000 copies/mL at week 12, or 200 copies/mL at week 24 or thereafter. All cases of suspected virological failure were confirmed by a second VL determination performed at least 2 weeks later. Once reconfirmed, the time of virological failure was defined as the time of the first VL measurement that met the failure criteria.
Immunological failure was defined as a decrease in the CD4 cell count of >50% from the baseline level, provided that the baseline CD4 cell count was >150 cells/L. For patients with a baseline CD4 cell count of 100150 cells/L, immunological failure was defined as a CD4 cell count of <50 cells/L and, for patients with baseline CD4 cell count of <100 cells/L, immunological failure was defined as a CD4 cell count of <25 cells/L.
Power calculation and statistics.
The trial was powered to show equivalence between the study arms, with an 80% chance that the 95% confidence interval (CI) for the difference in virological failure rates would exclude a difference of >15% in either direction. This was based on a sample size of 150 patients/arm and an underlying failure rate of 20% in both arms.
Per study protocol, the primary population for analysis was the intention-to-treat/exposed (ITT/e) population, including all randomized patients who had taken at least 1 dose of the assigned treatment. This analysis is also termed the "ITT switch included" analysis. In the other protocol-stipulated analysis, switching from the assigned treatment constituted failure (ITT/e/switch = failure [ITT/e/s]). In both analyses, patients who withdrew consent, who were lost to follow-up, or who died constituted failure, and the time of failure was the time of the event (whichever came first). Some patients withdrew their consent during follow-up but permitted reporting of laboratory data measured as part of their routine care. For these patients, withdrawn consent did not constitute (virological) failure.
No differences were observed at baseline, with regard to medical history, demographic, clinical, and laboratory parameters, or exposure to ART prior to baseline. Patients were primarily white (84%) men (78%) who engaged in homosexual- or bisexual-risk behavior (49%) with a median age of 39 years. The median CD4 cell count nadir was 110 cells/L (interquartile range [IQR], 40205 cells/L), the median CD4 cell count was 277 cells/L (IQR, 137450 cells/L), and the median VL was 3.9 log10 copies/mL (IQR, 1.75.1 log10 copies/mL); 39% of patients had a baseline VL of <400 copies/mL, and 30% had had a prior clinical AIDS-defining disease. At enrollment, 25% of patients were ART naive, 14% were ART experienced but PI naive, and 61% were PI experienced.
Complete week-48 follow-up data were available for 285 (93%) of the 306 patients who initiated the assigned treatment, 202 (66%) of whom continued to receive the assigned treatment. No difference was seen between the 2 study arms in the rate of patients lost to follow-up (7%). The 104 patients who prematurely switched from the assigned treatment did so primarily because of nonfatal, clinical AEs (n = 67). Among the 104 patients, no significant differences at the P = .05 level were observed between the study arms in the proportion of patients who switched treatment regimens, who received a mono or dual PI-, NNRTI-, or abacavir-based HAART regimen at week 48, or who discontinued treatment for any reason. Nine patients switched from Idv/Rtv to Sqv/Rtv, and 4 patients switched from Sqv/Rtv to Idv/Rtv. There was a significantly higher percentage of patients in the Idv/Rtv arm (41%) than in the Sqv/Rtv arm (27%) who prematurely switched from the assigned treatment (P = .013, 2 test). This difference was driven by patients who discontinued the randomized treatment because of a nonfatal, clinical AE (28% of patients assigned to Idv/Rtv arm vs. 15% in the Sqv/Rtv arm; P = .004, 2 test). Of the nonfatal, clinical AEs that led to patients' switching from the assigned treatment, 66% were of grade 1 or 2. More renal, skin and hair, and gastrointestinal AEs were observed in patients in the Idv/Rtv arm (data not shown). Twenty-two patients reduced the dose of the assigned treatment during follow-up (21 in the Idv/Rtv arm and 1 in the Sqv/Rtv arm).
Virological, immunological, and clinical outcome.
The primary efficacy outcome, rate of virological failure, was seen in 77 (25%) of 306 patients, with no difference between the study arms (P = .76, log rank test). The median VL at the time of failure was 2279 copies/mL, slightly higher in the Idv/Rtv arm (3857 copies/mL) than in the Sqv/Rtv arm (881 copies/mL) (P = .40). The difference between the 2 study arms in the proportion of patients experiencing virological failure was 2.2% (95% CI, -2.8% to 7.2%), with a higher proportion of protocol-defined virological failures in the Idv/Rtv arm. Using a Farrington-Manning equivalence test, we found sufficient evidence at the 5% level of significance to claim that the difference in success rates between the 2 treatments is <15% (P < .0048).
At week 48, 203 (68%) of 306, 155 (51%) of 306, and 186 (93%) of 201 patients had a VL of <50 copies/mL in the ITT/e, ITT/e/s, and during-treatment analyses, respectively. Only when switching from the assigned treatment was counted as having a VL of >50 copies/mL (ITT/e/s) was a significant difference observed, with more patients in the Sqv/Rtv arm having a suppressed VL at week 48 (57% vs 46%, reported separately by roche press release but confirmed by visual observation of graph in published article).
The median fasting baseline lipid values were as follows: for total cholesterol, 4.7 mmol/L in the Idv/Rtv arm and 4.8 mmol/L in the Sqv/Rtv arm (normal range, 3.46.2 mmol/L); for LDL cholesterol, 3.1 mmol/L in the Idv/Rtv arm and 3.2 mmol/L in the Sqv/Rtv arm (normal range, 1.73.2 mmol/L); and, for total triglyceride, 1.6 mmol/L in the Idv/Rtv arm and 1.7 mmol/L in the Sqv/Rtv arm (normal range, 0.52.3 mmol/L). Significantly higher lipid elevations were seen in the Idv/Rtv arm, compared with the Sqv/Rtv arm, at weeks 4 and 48 (ITT/e analysis). These differences were even more pronounced when the actual median changes from baseline, rather than the median percentage change, were considered (data not shown). Similar differences were seen when restricting the analysis to patients who continued to receive their trial medication (data not shown).
No difference between the study groups was seen in hematological, renal, or hepatic laboratory parameters, except for bilirubin levels, which were 10 and 11 mol/L at baseline in the Idv/Rtv and Sqv/Rtv arms, respectively (normal range, 422 mol/L). In the Sqv/Rtv arm, the bilirubin level did not change over time, whereas, in the Idv/Rtv arm, it increased to 20 mol/L at week 4, followed by a decline to 15 mol/L at week 48.
The heterogeneous study population included introduces a serious limitation, because the trial would not have sufficient power to describe the outcome within each of the subgroups included if the outcomes of the treatments were affected by the stage of HIV infection or treatment. To address this limitation, multivariate models of the key efficacy outcomes were developed. Of importance, patients entering the trial with a VL of >400 copies/mL had a significantly increased risk of experiencing protocol-defined virological failure and of not achieving virological suppression (<50 or <400 copies/mL) at week 48, compared with patients who were virologically suppressed at baseline. However, being antiretroviral or PI naive at the time of enrollment did not independently affect the risk of a poor virological outcome. The hazards ratios for the comparison of virological failure in the Idv/Rtv versus Sqv/Rtv arm were comparable in univariate and multivariate models adjusting for other variables. To further elucidate whether baseline characteristics may have influenced the efficacy outcomes, 2 substudies are currently investigating genotypic-resistance mutations at baseline and at the time of virological failure and single-nucleotide polymorphisms in the multidrug resistance1 locus of stored peripheral blood mononuclear cells. In addition, one substudy is investigating efficacy and safety in relation to trough levels (Cmin) of the study PIs at weeks 4 and 48.
In the analysis in which switching from the assigned treatment is equal to virological failure or lack of virological suppression, Sqv/Rtv tended to have superior virological activity than did Idv/Rtv. This result was to be expected, because a higher proportion of patients in the Idv/Rtv arm switched from the randomized treatment. The trial was not designed and did not have the statistical power to test whether there were differences in risk of protocol-defined immunological and clinical failures between the 2 study arms. No formal statistical analysis of these efficacy parameters was appropriate, because of the low number of such failures observed.
Finally, the efficacy and safety outcome of patients randomized to receive Idv/Rtv is comparable with data from the recently completed BEST trial. The BEST trial randomized patients receiving a stable regimen, including Idv (800 mg t.i.d.), to either continue this regimen or switch to Idv/Rtv (800/100 mg b.i.d.). In the present trial, few patients were receiving IDV (800 mg t.i.d.) at the time of screening; hence, the patients who received Idv/Rtv in the 2 trials are not directly comparable.
In the present trial, 21 (13%) of 158 patients in the Idv/Rtv arm reduced the Idv dose. The present trial was not designed to evaluate whether this strategy lowered the risk of AEs or affected the efficacy of the treatment, nor was the sample sufficiently large for formal testing of these important questions. A randomized trial should be done to evaluate whether Idv/Rtv in lower dosing has a more favorable AE profile and maintained virological efficacy, compared with either Idv/Rtv (800/100 mg b.i.d.) or other commonly used Rtv-boosted PI regimens prior to the introduction of other Idv/Rtv dosing regimens in routine care.
Compared with patients in the Sqv/Rtv arm, patients in the Idv/Rtv arm had significant increases from baseline in total cholesterol, LDL cholesterol, and triglyceride levels at weeks 4 and 48. Other drugs (NNRTIs and stavudine) that could potentially influence these parameters were well balanced between the 2 groups at baseline. Therefore, these findings suggest that Idv/Rtv affects the lipid metabolism adversely, relative to Sqv/Rtv. Because the same Rtv dosing was used in both arms, it is likely that it is the Idv component that causes lipid levels to increase. However, another possibility is that the Rtv metabolism is affected differently by Idv, compared with Sqv. These mechanisms will be explored further by correlating drug levels at weeks 4 and 48 with lipid changes. Differences between PIs boosted by the same Rtv dosing has not been observed previously, whereas it was reported recently that lopinavir/Rtv lead to greater elevation of lipid levels, compared with nelfinavir .
In conclusion, we have found that, in this open-label study of a heterogeneous patient populationreflecting the real-life situationSqv/Rtv has antiretroviral effects comparable to those of Idv/Rtv in the doses studied. We observed more treatment-limiting AEs in the Idv/Rtv arm, relative to the Sqv/Rtv arm, and found that more patients in the Sqv/Rtv arm remained virologically suppressed at week 48, probably because of a better toxicity profile.