Treatment of type 1 HIV (HIV-1) with combination antiretroviral therapy (cART) consisting of two nucleoside/nucleotide analogues plus either a nonnucleoside reverse transcriptase inhibitor (NNRTI) or a protease inhibitor reduces morbidity and mortality and currently represents the standard of care for this infection [1–4].
The combination of abacavir (ABC)/lamivudine (3TC) plus atazanavir (ATV) boosted with ritonavir (RTV) (ATV/r) is an effective initial protease inhibitor-containing regimen that can reduce viral load. Although the addition of low-dose RTV to ATV increases the plasma exposure of ATV yielding an effective cART regimen with rare resistance observed after treatment failure, the potential for adverse effects, including lipid alterations and potential future metabolic complications, increases [5–7]. As ATV is a once daily (q.d.), well tolerated protease inhibitor that is also approved for use without RTV boosting, there is interest among practitioners to determine whether removing RTV from this boosted protease inhibitor regimen results in similar efficacy but improved safety profiles through improved tolerability, a reduction in lipids, and the potential to minimize long-term metabolic adverse events.
An induction–simplification treatment strategy is well suited to answer this question by providing potent initial virologic suppression followed by simplification to a standard cART regimen, which optimizes tolerability and adherence, minimizes short and long-term toxicity, and reduces drug–drug interactions.
The current study (Atazanavir, Ritonavir, Induction with Epzicom Study or ARIES) evaluates the safety and efficacy of ATV/r administered q.d. for 36 weeks followed by randomization (1: 1) to a simplification regimen of ATV q.d. or continuation of ATV/r q.d., each in combination with the ABC/3TC fixed-dose combination (FDC) q.d. for an additional 48 weeks in ART-naive, HIV-1-infected, human leukocyte antigen (HLA)-B*5701-negative patients. Unlike other NRTIs, which require the use of ATV/r due to drug–drug interactions, ABC/3TC may be administered with unboosted ATV, making it an ideal backbone to evaluate in this induction–simplification treatment strategy. A planned interim analysis of the nonrandomized, induction period (baseline to week 36) demonstrated viral load suppression (HIV RNA <50 copies/ml) in 80% of patients by 36 weeks . Results of the randomized, simplification period (weeks 36–84) are presented in the current analysis.
Patients from the United States, Canada, and Puerto Rico were eligible for enrollment into the initial 36-week induction period if they were infected with HIV, at least 18 years of age, ART-naive, had a plasma HIV RNA level of at least 1000 copies/ml with any CD4+ lymphocyte count, and had an appropriate screening of viral genotype. Patients were excluded if they were positive for either the HLA-B*5701 allele or hepatitis B surface antigen, had medical conditions that could compromise their safety or interfere with drug absorption, required use of prohibited medications, or had protocol-specified abnormal laboratory values or a creatinine clearance of less than 50 ml/min as estimated by the Cockcroft–Gault equation.
Following the 36-week induction period, patients were eligible for randomization, provided all the following criteria were met: plasma HIV RNA levels were below 50 copies/ml on two consecutive study assessments prior to week 36, plasma HIV RNA level was below 50 copies/ml at the study assessment immediately preceding week 36 (randomization visit), and patients did not meet the definition of virologic failure. Virologic failure during the initial induction period was defined as failure to achieve plasma HIV RNA level below 400 copies/ml by week 30 or confirmed HIV RNA rebound of at least 400 copies/ml after achieving plasma HIV RNA level below 400 copies/ml.
The study was approved by the ethics review boards at each of the participating centers and conducted in accordance with Good Clinical Practice. All patients provided written informed consent.
Design and interventions
ARIES is a randomized (1: 1), open-label, noninferiority study that investigates the safety and efficacy of an induction–simplification strategy using two q.d. regimens containing a dual nucleoside, plus a protease inhibitor. After successfully completing 36 weeks of initial ART containing ABC sulfate/3TC FDC (600 mg/300 mg, Epzicom or Kivexa; GlaxoSmithKline, London, UK) plus ATV (300 mg, Reyataz; Bristol Myers Squibb, New York City, New York, USA) plus RTV (100 mg, Norvir; Abbott Laboratories, Abbott Park, Illinois, USA), and meeting eligibility criteria, patients were stratified on their baseline viral load (<100 000 or ≥100 000 copies/ml) and randomized (1: 1) to either continue the induction regimen or switch to a simplified regimen containing ABC/3TC plus ATV (400 mg). Randomization was achieved when sites called a centralized voice response system and received their treatment allocation based on a randomized sequence generated by the study statistician. Patients with suspected hypersensitivity reaction (HSR) to ABC were allowed to remain in the study after substitution of zidovudine/3TC twice daily (150 mg/300 mg, Combivir; GlaxoSmithKline) for ABC/3TC. These patients had ABC permanently discontinued and underwent an ABC skin patch test to provide immunologic confirmation of ABC HSR .
Procedures and assessments
Pretreatment evaluations included a clinical assessment and laboratory evaluations (plasma HIV RNA, T-lymphocyte subsets, hepatitis B and C serologies, HLA-B*5701 determination, hematology, clinical chemistries, serum pregnancy test, and viral genotype). On-study evaluations included clinic visits at day 1 (baseline) and at weeks 2, 4, 8, 12, 24, 30, 36, 40, 48, 60, 72, 84, and withdrawal. At each visit, samples for HIV RNA, CD4+/CD8+ lymphocyte subsets, clinical chemistries, and hematology were collected and analyzed. Pregnancy testing was performed at baseline, weeks 36, 84, and whenever pregnancy was suspected. Samples for viral genotypic resistance testing were collected at all visits and were analyzed at baseline and time of failure for patients with virologic failure. HIV RNA concentrations were measured by the Roche Cobas Amplicor HIV-1 Monitor or Ultrasensitive Monitor Test (Roche Diagnostic Systems, Branchburg, New Jersey, USA). Adverse events, concurrent medications, and HIV-associated conditions were also assessed at each visit. Adverse events were graded using the 2004 Division of AIDS toxicity grading scale. All suspected ABC HSR events were reported as serious adverse events (SAEs). Viral genotypes and phenotypes were performed at Monogram Biosciences Inc. (South San Francisco, California, USA). Genotypic mutations were defined according to International AIDS Society-USA Guidelines (December 2008) . All other laboratory tests were performed by Quest Diagnostics (Van Nuys, California, USA). Smoking status and diabetes status were assessed at baseline and week 84 for Framingham Risk Score calculation .
The study was sponsored by GlaxoSmithKline. The sponsor developed the study design with input from investigators and analyzed the data. All authors had full access to the data and vouch for the accuracy and completeness of the data and analyses. The manuscript was written and approved by all of the authors, each of whom contributed to drafts and revisions.
The primary efficacy endpoint was the proportion of patients with HIV RNA level below 50 copies/ml at week 84 by the time to loss of virologic response (TLOVR) algorithm. Secondary endpoints include the proportion of patients with HIV RNA level below 50 copies/ml at week 36; proportion of patients with HIV RNA level below 400 copies/ml; changes in HIV RNA and CD4+ cell counts; time to virologic failure; change in fasting lipids, insulin, and glucose; and change in genotypic and phenotypic resistance at virologic failure. Safety endpoints include incidence of grades 2–4 adverse events, treatment-related and treatment-limiting adverse events, SAEs, and the incidence of suspected ABC HSR. Virologic failure during the randomized simplification period (weeks 36–84) was defined as confirmed rebound in HIV RNA level of at least 400 copies/ml after achieving below 400 copies/ml. Results of the primary efficacy endpoint and secondary endpoints occurring at week 84 are presented. Results of the 36-week induction period have previously been published .
A sample size of 374 patients (187 patients per arm) were needed to provide 90% power to establish the noninferiority of ATV plus ABC/3TC compared with ATV/r plus ABC/3TC, assuming an 85% success rate in each group at week 84. Noninferiority was defined as a two-sided 95% confidence interval (CI) that excluded differences as large as 12% in the direction of inferiority of the ATV group.
The primary population for efficacy analyses was the intent-to-treat exposed (ITT-E) population, which included all enrolled patients exposed to at least one dose of study medication. Other analyses for proportion of patients achieving HIV RNA levels below the lower limits of detection (<50 and <400 copies/ml) included observed and missing/discontinuation equal to failure (MD = F).
According to the TLOVR algorithm, responders were patients with confirmed viral load below 50 (400) copies/ml who had not yet met any nonresponder criterion. Nonresponders were patients who never achieved confirmed HIV RNA level below 50 (400) copies/ml, prematurely discontinued study or study medication for any reason, had confirmed rebound to at least 50 (400) copies/ml, or had an unconfirmed HIV RNA level of at least 50 (400) copies/ml at their week 84 study visit. The ITT-E observed analysis included all observed data, whereas the ITT-E, MD = F analysis required patients with missing data or data collected after discontinuation of randomized study medication to be considered failures.
The primary safety population included all randomized patients who consumed at least one dose of study drug.
Patient disposition and baseline characteristics
Of the 515 patients who were initially enrolled in the study between March 2007 and August 2007, a total of 419 patients were randomized at week 36 after the nonrandomized induction period and included in the ITT-E population. Ninety percent (379/419) of randomized patients in the overall ITT-E population completed the study from baseline to week 84 (Fig. 1).
Overall, baseline characteristics of the randomized population were similar for both treatment arms, with the exception of a greater proportion of patients with Centers for Disease Control and Prevention classification C in the unboosted ATV arm compared with the ATV/r arm (16 versus 8%).
No statistical test was performed to adjust for the numerical imbalance of that baseline covariate. The majority of patients were men (84%) and had low cardiovascular risk as defined by Framingham risk assessment (90%). Patients had a median baseline HIV RNA level of 5.05 log10 copies/ml (53% had baseline HIV RNA ≥100 000 copies/ml) and a median baseline CD4+ cell count of 200 cells/μl (see Table 1).
After randomization at week 36, 86% (181/210) of patients in the ATV group successfully maintained HIV RNA level below 50 copies/ml at week 84 using a TLOVR analysis versus 81% (169/209) patients in the ATV/r arm. The two-sided 95% CI (−1.75 to 12.48%) stratified by baseline RNA excluded the predefined noninferiority margin of −12%, thus demonstrating the noninferiority of ATV to ATV/r. These results were robust with multiple sensitivity analyses (MD = F and observed analyses) (Fig. 2).
When assessing the below 400 copies/ml endpoint by TLOVR analysis, 92 versus 86% of patients in the ATV and ATV/r arms, respectively, achieved HIV RNA level below 400 copies/ml (95% CI 0.44–12.22, P = 0.036).
Virologic response by baseline viral load strata demonstrated that 85% (84/99) versus 79% (76/96) of patients in the baseline HIV RNA level below 100 000 copies/ml stratum experienced virologic success as analyzed by the TLOVR algorithm (HIV RNA <50 copies/ml), whereas 87% (97/111) versus 82% (93/113) of patients met this endpoint in the baseline HIV RNA level of at least 100 000 copies/ml stratum for the ATV and ATV/r treatment groups, respectively.
The immunologic response was notable during the 84-week period with a median increase in CD4+ cell counts of 240 and 259 cells/μl and a median value at week 84 of 434 and 469 cells/μl for the ATV and ATV/r arms, respectively.
A total of eight (2%) patients met the per-protocol definition of virologic failure (confirmed HIV RNA rebound ≥400 copies/ml after achieving plasma HIV RNA <400 copies/ml) from weeks 36 to 84 (ATV, 1; ATV/r, 7). At failure, no major protease inhibitor treatment-emergent mutations were detected. Only two treatment-emergent minor protease polymorphisms, G16G/E and K20K/R, were detected in two patients (single polymorphism in viruses isolated in each patient) in the ATV/r arm. Virus from all eight patients exhibiting virologic failure remained susceptible to all protease inhibitors.
At virologic failure, only one of these eight virologic failure patients had HIV with detectable treatment-emergent NRTI mutations. HIV from this patient, who was randomized to the ATV group, developed a treatment emergent mutation (M184M/I/V) at failure, which resulted in reduced susceptibility to 3TC. At baseline, virus from this patient also had a detectable NRTI drug resistance-associated mutation (K219K/Q). The site reported that the patient was noncompliant and had missed doses of drug in the weeks preceding the determination of virologic failure.
At the time of initial failure, each patient's healthcare provider was queried as to whether the patient had experienced any illness, interruption in therapy, or immunization that could account for the sudden increase in viral load. Site correspondence indicated that nonadherence with the dosing regimen or therapy interruption were considered a factor for five of the eight patients who met virologic failure criteria, and so may have contributed to the reason for withdrawal indicated by the investigator on the case report form noted in Fig. 1. For some patients in the ATV/r arm who met per protocol virologic failure definitions, noncompliance rather than virologic failure was cited by the investigator as the reason for withdrawal, although these patients also met protocol-defined virologic failure criteria. Not all patients who met virologic failure criteria were withdrawn from the study. As per protocol, patients who met virologic failure criterion with HIV RNA level below 2000 copies/ml at confirmation of virologic failure could choose to remain on study, and three patients meeting this criterion have remained on study after week 84 with subsequent viral resuppression.
Adverse events of at least moderate severity and considered by the investigator to be related to treatment over the full course of the study (baseline to week 84) occurred in 30% (63/210) and 33% (70/209) of patients in the ATV and ATV/r arms, respectively. Hyperbilirubinemia, diarrhea, and nausea were most commonly reported and were the only events that occurred with an overall frequency of at least 3% (Table 2). During the randomized period from weeks 36 to 84, when RTV was excluded in the simplification arm, 10% (22/210) of patients in the ATV group and 14% (29/209) of patients in the ATV/r group experienced a grades 2–4 treatment-related adverse event, most commonly hyperbilirubinemia [ATV, 4% (9/210); ATV/r, 10% (20/209)]. Few events (2%, 8/419) led to study discontinuation.
As previously reported, suspected hypersensitivity to ABC occurred in 0.8% (4/515) of patients during the initial nonrandomized induction period (baseline to week 36) and no additional cases were reported through week 84 .
Treatment-related SAEs were uncommon (2%). There was one fatal event reported during the initial nonrandomized induction period, a 45-year-old white men developed Castleman's disease followed by hepatic and renal failure leading to death. None of these events were attributed to study drug by the investigator . No myocardial infarctions (MIs) were reported through 84 weeks.
Overall, patients in the ATV arm appeared to demonstrate a more favorable median fasting lipid profile from weeks 36 to 84 compared with those in the ATV/r arm (Table 3). The reduction in median fasting triglyceride levels was greater in the ATV arm (week 36, 163 mg/dl; week 84, 123 mg/dl) compared with the ATV/r (week 36, 160 mg/dl; week 84, 153 mg/dl) arm during the randomization phase. Fasting total and low-density lipoprotein-cholesterol increases were below National Cholesterol Education Program (NCEP) guidelines for treatment for both groups; fasting triglyceride increases were below NCEP guidelines for the ATV arm by week 84.
This study enrolled a large, racially diverse, predominantly male population with relatively advanced HIV disease. More than half of the randomized population (53%) had HIV RNA level of at least 100 000 copies/ml and half had CD4+ cell counts below 200 cells/μl at baseline.
ARIES was designed as a treatment strategy study to evaluate the use of unboosted ATV in combination with ABC/3TC in patients who have achieved initial virologic suppression (HIV RNA <50 copies/ml) on boosted ATV plus ABC/3TC. Not surprisingly, the 84-week results demonstrated the antiviral efficacy of ATV plus ABC/3TC compared with ATV/r plus ABC/3TC based on the proportion of patients achieving plasma HIV RNA level below 50 copies/ml at week 84 (TLOVR analysis), thus establishing the noninferiority of unboosted ATV to boosted ATV/r after an initial 36-week induction period with ATV/r plus ABC/3TC. Using the below 400 copies/ml threshold, not only was noninferiority established in the TLOVR analysis but superiority of the ATV arm over ATV/r based on the 95% CI and the P value was also demonstrated. Furthermore, virologic response rates (RR) for the below 50 copies/ml endpoint stratified by baseline plasma HIV RNA level (<100 000 and ≥100 000 copies/ml) revealed no differences in RR for either treatment arm in the 419 patients who were randomized at week 36 with undetectable HIV RNA.
These findings are consistent with a smaller study (InduMa) conducted in Europe, which also demonstrated the noninferiority of ATV to ATV/r in patients who achieved HIV viral suppression with ATV/r plus two NRTIs. The patients maintained a similar level of virological suppression through 48 weeks after switching to unboosted ATV .
Few patients [ATV, 1/210 (<1%); ATV/r, 7/209 (3%)] met the protocol definition of virologic failure during the randomization period (weeks 36–84); a total of 22 patients met the criteria for virologic failure over the entire study period (baseline to week 84). For the majority of patients who met the definition of virologic failure during the randomized period, a therapy interruption or nonadherence/compliance issue was reported by the patient's healthcare provider. Only one patient developed reduced susceptibility to any study drug (3TC) at failure; virus from this patient also selected a treatment-emergent NRTI mutation (M184M/I/V). This same patient also had evidence of NRTI drug resistance at baseline (K219K/Q), although it is not possible to say whether the appearance of the M184M/I/V mixture at virologic failure was the result of de-novo selection or outgrowth of archived virus containing this mutation that emerged under selection pressure. No virus isolated from patients experiencing virologic failure selected any major HIV protease mutation and all isolates remained susceptible to protease inhibitors at failure.
Both regimens were generally well tolerated during the 84-week study period, with most events being of mild-to-moderate intensity. Thirty percent of patients in the ATV arm and 33% of patients in the ATV/r arm were reported to have a grades 2–4 treatment-related adverse event. Treatment-related SAEs were few and with the exception of suspected ABC HSR, no single event occurred in more than one patient through week 84.
Mild increases in fasting lipids were seen, which were generally consistent with those seen in other studies using ATV/r [13,14]. As expected, dropping the RTV from the regimen did result in improved lipid and bilirubin levels for those patients on the simplification arm.
Cardiovascular risk is increasingly becoming an important consideration in the care of HIV-infected patients. Most (90%) patients in ARIES were classified as low risk (≤10%, 10-year coronary heart disease risk) at baseline for developing a cardiovascular-related outcome based on Framingham Score. Furthermore, there were no MIs reported during the study period.
Several limitations of this study merit consideration. The applicability of these results is limited to an ART-naive population who achieve virological suppression (HIV-RNA <50 copies/ml) on ABC/3TC plus ATV/r and, therefore, extrapolating these findings to other patient populations may not be appropriate. The use of unboosted ATV may offer less forgiveness than a boosted protease inhibitor regimen, as plasma concentrations of the protease inhibitor are lower and are more susceptible to patient adherence issues. Furthermore, because of the potential for drug–drug interactions, this type of induction–simplification treatment strategy is restricted to certain NRTI backbones such as ABC/3TC that can be used in combination with unboosted ATV.
Overall, these results demonstrate that a regimen of ATV plus ABC/3TC is potent and well tolerated in patients who have achieved initial suppression on an induction regimen and represents a viable treatment simplification strategy.
This study was supported by GlaxoSmithKline.
The authors would like to thank the patients, study investigators, study coordinators, and the GlaxoSmithKline study monitoring group for their contributions to this study. We also thank Bristol-Myers Squibb for providing Reyataz for this study. We also thank the GlaxoSmithKline study team members: S. Bhuta, M. Bomar, J. Bond, C. Brothers, M. Daniel, N. Figliola, V. Garay, M. Gartland, T. Lai, Q. Liao, M.A. Moore, P. Patel, D. Percival, D. Raimonde, J. Royall, S. Skerget, M. Schultz, I. Song, B. Wine, and M. Vourvahis.
ARIES Study Investigators: B. Akil, J. Applebaum, J. Baril, N. Bellos, C. Cohen, P. Cook, M. Cuenca, D. Berger, I. Brar, C. Brinson, F. Carpio-Cedraro, E. DeJesus, R. Dretler, J. Duggan, R. Elion, J. Gathe, E. Godofsky, R. Greenberg, R. Hao, K. Henry, A. Khalsa, J. Kort, P. Kumar, P. Lackey, A. LaMarca, C. Lucasti, C. McDonald, P. McLeroth, I. Melendez-Rivera, A. Mestre, J. Morales-Ramirez, D. Murphy, R. Nahass, C. Newman, W. O'Brien, P. O'Keefe, E. Oldfield, E. Overton, D. Pearce, M. Potter, A. Rachlis, M. Ramgopal, B. Rashbaum, F. Rhame, G. Richmond, J. Rodriguez, P. Salvato, A. Sanchez, L. Santiago, J. Sarria, P. Sax, S. Schneider, R. Scott, A. Scribner, G. Sepulveda, G. Simon, D. Siraj, J. Slim, L. Sloan, C. Small, G. Smith, K. Squires, K. Tashima, P. Tebas, M. Thompson, J. Torees, V. Trivedi, T. Vanig, S. Walmsley, D. Ward, W. Weinberg, M. Weinert, B. Young.
The ClinicalTrials.gov registration number for the ARIES study (EPZ108859) was NCT00440947.
K.E.S. received consultancy fees, research funding, or both from Abbott, Boehringer-Ingelheim, Bristol-Myers Squibb, Gilead Sciences, GlaxoSmithKline, Koronis, Merck, Pfizer, Schering-Plough, Tibotec, and Tobira. B.Y. has received consultancy fees, speaking honoraria, research funding, or all from Bristol-Myers Squibb, Cerner Corporation, Gilead Sciences, GlaxoSmithKline, Hoffman-LaRoche, Merck & Co., Monogram Biosciences, Pfizer, and Vertex Pharmaceuticals. E.D. has received consultancy fees, speaking honoraria, research funding, or all from Abbott, Achillion, Avexa, Boehringer-Ingelheim, Bristol-Myers Squibb, Gilead Sciences, GlaxoSmithKline, Hoffman-LaRoche, Merck & Co., Pfizer, Schering-Plough, Taimed, Tobira, Tibotec, and Vertex and Virco. N.B. has received consultancy fees, speaking honoraria, and research support from Abbott, GlaxoSmithKline, and Tiobtec. D.M. has received consultancy fees, speaking honoraria, or both from Abbott, Boehringer-Ingelheim, Bristol-Myers Squibb, Gilead, Pfizer, and Tibotec. H.Z., L.G.P., L.L.R., P.G.W., and M.S.S. were all employees of GlaxoSmithKline at the time of writing.
The sponsor developed the study design with input from prospective investigators and analyzed the data. Substantial contributions to study conception, design, analysis, and interpretation of the data were made by K.E.S., B.Y., H.Z., L.G.P., L.L.R., P.G.W., and M.S.S. Substantial contributions to acquisition of data and critical review of the manuscript were made by K.E.S., B.Y., E.D., N.B., and D.M. All authors had full access to the data and vouch for the accuracy and completeness of the data and analyses. The manuscript was written and approved by all of the authors, each of whom contributed to drafts and revisions.