Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

48-Week Efficacy and Safety of Dolutegravir Relative to Commonly Used Third Agents in Treatment-Naive HIV-1–Infected Patients: A Systematic Review and Network Meta-Analysis

Abstract

Background

A network meta-analysis can provide estimates of relative efficacy for treatments not directly studied in head-to-head randomized controlled trials. We estimated the relative efficacy and safety of dolutegravir (DTG) versus third agents currently recommended by guidelines, including ritonavir-boosted atazanavir (ATV/r), ritonavir-boosted darunavir (DRV/r), efavirenz (EFV), cobicistat-boosted elvitegravir (EVG/c), ritonavir-boosted lopinavir (LPV/r), raltegravir (RAL), and rilpivirine (RPV), in treatment-naive HIV-1–infected patients.

Methods

A systematic review of published literature was conducted to identify phase 3/4 randomized controlled clinical trials (up to August 2013) including at least one third agent of interest in combination with a backbone nucleoside reverse transcriptase inhibitor (NRTI) regimen. Bayesian fixed-effect network meta-analysis models adjusting for the type of nucleoside reverse transcriptase inhibitor backbone (tenofovir disoproxil fumarate/emtricitabine [TDF/FTC] or abacavir/lamivudine [ABC/3TC]) were used to evaluate week 48 efficacy (HIV-RNA suppression to <50 copies/mL and change in CD4+ cells/µL) and safety (lipid changes, adverse events, and discontinuations due to adverse events) of DTG relative to all other treatments. Sensitivity analyses assessing the impact of NRTI treatment adjustment and random-effects models were performed.

Results

Thirty-one studies including 17,000 patients were combined in the analysis. Adjusting for the effect of NRTI backbone, treatment with DTG resulted in significantly higher odds of virologic suppression (HIV RNA<50 copies/mL) and increase in CD4+ cells/µL versus ATV/r, DRV/r, EFV, LPV/r, and RPV. Dolutegravir had better or equivalent changes in total cholesterol, LDL, triglycerides, and lower odds of adverse events and discontinuation due to adverse events compared to all treatments. Random-effects and unadjusted models resulted in similar conclusions.

Conclusion

Three clinical trials of DTG have demonstrated comparable or superior efficacy and safety to DRV, RAL, and EFV in HIV-1–infected treatment-naive patients. This network meta-analysis suggests DTG is also favorable or comparable to other commonly used third agents (ATV/r, LPV/r, RPV, and EVG/c).

Introduction

Two of the primary goals of anti-HIV therapy are to suppress plasma HIV viral replication and preserve and restore the number of circulating CD4+ T cells, the immune cells attacked by HIV [1], [2]. Highly active antiretroviral therapy (HAART) has achieved these goals for many patients, resulting in reduction of HIV-associated morbidity and prolonging survival to nearly that of the normal population [3], [4]. For treatment-naive patients, HAART typically includes a combination of two nucleoside reverse transcriptase inhibitors (NRTIs, the “backbone”) with one or more drugs from the more potent classes (the “third agent”) [1], [2]. The US Department of Health and Human Services (DHHS) and the European AIDS Clinical Society guidelines have recommended several third agents for the treatment of infection: ritonavir-boosted atazanavir (ATV/r), darunavir (DRV/r), lopinavir (LPV/r), efavirenz (EFV), cobicistat-boosted elvitegravir (EVG/c), raltegravir (RAL), and rilpivirine (RPV) [1], [2]. Of these, RPV and LPV/r are recommended as alternative regimen options by DHHS [2]. Many of these regimens have comparable efficacy but vary in dosing frequency, pill burden, drug interactions, and potential side effects.

Initial choice of therapy is central to long-term management of HIV infection as treatment switching has been associated with higher healthcare costs and increased likelihood of treatment failure [5][7]. Therefore, use of safe, well-tolerated, and effective regimens is important to allow patients to achieve long-term virologic suppression from the start of initial therapy, which may lead to improved clinical and economic outcomes including improved immune function, quality of life, and ability to control other comorbid conditions [8], [9].

Dolutegravir (DTG) has recently been approved for the treatment of HIV-1 disease in combination with other antiretroviral agents. DTG has been shown to exhibit a higher barrier to resistance compared to RAL and EVG, is dosed once daily, and has limited drug interactions including no food restrictions [10]. Three phase 3 clinical trials have shown DTG superiority to EFV [11] and DRV/r [12] and non-inferiority to RAL [13] as first-line treatment; evidence versus other guideline-recommended third agents has not yet been explored. The objective of this study is to estimate the efficacy and safety of DTG relative to other guideline-recommended agents in a Bayesian network meta-analysis (NMA). Results of this analysis will help understand comparability of DTG to all recommended agents.

Methods

Identification and selection of study data

The PubMed/MEDLINE, Embase, and Cochrane databases were systematically searched (up to August 2013) to identify randomized controlled trials (RCTs) evaluating efficacy and/or safety of ATV/r, DRV/r, DTG, EFV, EVG/c, LPV/r, RAL, or RPV in treatment-naive HIV-1 patients. PubMed and EMBASE search terms were “HIV-1 [mesh] OR HIV infections [mesh] NOT pregnancy [mesh] AND ((dolutegravir OR GSK1349572) OR (efavirenz OR Sustiva OR Stocrin OR DMP-266) OR (raltegravir OR Isentress OR MK-0518) OR (elvitegravir OR GS-9137 OR JTK-303) OR (rilpivirine OR Edurant OR TMC 278) OR (darunavir OR Prezista OR TMC-114) OR (atazanavir OR Reyataz OR BMS-232632) OR (lopinavir OR ABT-378 OR Aluviran OR Koletra OR Kaletra) OR (Atripla OR Quad OR Stribild OR Eviplera OR Complera))”. The ClinicalTrials.gov registry, US FDA summary basis of approvals, EMA EPAR scientific discussions, and references of published systematic reviews and meta-analyses were also searched for any additional data. Abstracts of the 2013 meeting of the International AIDS Society and the Interscience Conference on Antimicrobial Agents and Chemotherapy were searched to identify recent presentations. Two phase 3 studies of DTG with data available after August 2013 were also included.

Study selection was conducted by two independent researchers who performed an initial review and selection of study titles/abstracts followed by full text review and selection. Disagreements between the reviewers were resolved by consensus. Pre-specified inclusion criteria included treatment-naive patients with HIV-1 infection; studies published in English; phase 3 or 4 RCT; patients aged ≥13 years; use of at least one of the third agents of interest; and reporting at least one of the efficacy outcomes of interest after 48 weeks of treatment. Non-randomized observational studies; single-arm studies; and studies examining different dosages of the same drug, structured treatment interruptions, maintenance treatments, or treatment switching were excluded, as were publications where outcomes specific to a treatment-naive population could not be distinguished. Studies reporting outcomes such that results could not be obtained for each treatment arm individually were also excluded. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed through all phases in the study [14].

Three researchers independently abstracted data from the final selection of studies into a structured Microsoft Access database and data were reconciled for accuracy. The Effective Public Health Practice Project Quality Assessment, a quality assessment tool, was used to assess selection bias, study design, confounders, blinding, data collection methods, and withdrawals and dropouts [15].

Data analysis

Efficacy outcomes analyzed were virologic suppression of HIV RNA<50 copies/mL (intention-to-treat [ITT] populations, Missing/Non-Completers = Failure) and CD4+ cell change from baseline (ITT). On the basis of FDA guidance to industry [16], the following algorithms for virologic suppression were considered comparable: FDA Snapshot-50, confirmed virologic response-50, Time to Loss of Virologic Response-50, and HIV RNA<50 copies/mL. Safety outcomes analyzed were total cholesterol (TC), high-density lipoprotein (HDL), low-density lipoprotein (LDL), and triglyceride (TG) changes from baseline, adverse events (AEs; all grades due to any reason), and discontinuations due to AEs.

A Bayesian NMA framework was used to generate estimates of relative treatment outcomes [17]. This approach statistically combines the data from all clinical trials within an integrated analysis to generate a pooled estimate of the relative treatment effect of each intervention compared to all others. Models were programmed and executed using WinBUGS version 1.4.3 [18].

Treatment effects for virologic suppression, AEs, and discontinuation outcomes are estimated as odds ratios (OR) of DTG relative to a comparator. Relative CD4+ cell change and lipid changes are estimated as the mean “difference of difference” from baseline to week 48. Uncertainty around point estimates is measured by the 95% credible interval (CrI), which indicates that the outcome estimate falls within the given range with 95% probability. Credible intervals of ORs not including 1 and CrIs of mean differences not including 0 are considered “statistically significant.” Homogeneity of virologic suppression, CD4+ cell change, and discontinuation treatment effects were assessed by Q statistic (chi-square test) for pairs of third-agent treatment comparisons with three or more available studies.

Differential NRTI backbone effects independent of the third agent on treatment efficacy and lipid changes have been observed in the literature [19][21]. In an effort to more accurately estimate the independent effect of the third agents of interest, we included statistical adjustment for the type of NRTI backbone within the meta-analysis models (details presented in Appendix S1). Backbones were categorized into three groups: tenofovir disoproxil fumarate/emtricitabine (TDF/FTC), abacavir/lamivudine (ABC/3TC), and all other (including investigator “choice”), as no evidence was found to support distinction among other backbone regimens. Backbone regimen adjustment was possible in the analysis for virologic suppression, CD4+ change and lipids, but not AEs and discontinuations due to an insufficient number of studies and no strong clinical relevance in the case of discontinuation.

Fixed-effect models for all outcomes were chosen based on the deviance information criterion and the presence of only one study for many pairs of treatment comparisons. Limited data to estimate random-effects model parameters have been noted to lead to poor estimation of the width of the distribution of intervention effects [22]. To evaluate the robustness of the overall conclusions on the choice of model selection, backbone-unadjusted and random-effects model results are also presented (see Appendix S2 for random-effects model results).

Consistency of the modeled outcomes with observed trial data from studies not including EFV was evaluated as a measure of model validity. Results were considered consistent if the outcomes for the comparisons reported in the trials were similar to the same comparisons estimated from the model [23]. For binary outcomes, such as virologic suppression, consistency was measured by the ratio of the ORs of the direct and indirect estimates. For other continuous outcomes the difference of the mean changes from baseline between the indirect and direct estimates were calculated. If the 95% CrI for these values did not include 1 or 0 for the 2 measures, respectively, model results were considered inconsistent.

Results

A total of 1163 unique title/abstracts were screened from all search sources, where 176 records were selected for full text review and 54 publications representing 45 unique clinical trials were selected for data abstraction. After data abstraction, 23 articles were excluded from the meta-analysis, including 17 that did not present HIV cohorts representative of the general population [24][40] and 6 with incomparable virologic suppression definitions (i.e., did not define ITT/PP population and/or treatment of non-completers/missing data) [41][46]. Ultimately, 31 RCTs were included into the meta-analysis, representing data from 17,000 HIV-infected patients (Figure 1) with 26 reporting virologic suppression data, 28 CD4+ cell change, 20 TC, 19 HDL, 17 LDL, 17 TG, 11 AEs, and 18 studies reporting discontinuation due to AEs.

thumbnail
Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Flow Chart.

PubMed/MEDLINE, Embase, and Cochrane databases were searched to identify randomized controlled trials evaluating efficacy and/or safety of ATV/r, DRV/r, DTG, EFV, EVG/c, LPV/r, RAL, or RPV in treatment-naive HIV-1–infected patients. Records were screened by independent researchers, who selected study titles and abstracts for full text review. Following several rounds of exclusion based on multiple criteria, 31 trials and publications were selected for subsequent analysis. *Additional records were identified via ClinicalTrials.gov, the Food and Drug Administration (FDA), scientific discussions of the European Medicines Agency (EMA)/European Public Assessment Reports (EPAR), and third-agent package inserts. Each of these were found to be included in initial search records and noted as such. Reasons for exclusion at time of full text review: non-randomized trial; Phase 1/Phase 2 trials; patient population age <13 years; outcomes not of interest; trial duration <12 weeks; and out-of-network comparator. 34 publications were matches to ClinicalTrials.gov registry results (NCTs) to ensure comprehensive extraction of all available data pertaining to outcomes of interest.

https://doi.org/10.1371/journal.pone.0105653.g001

EFV was the most prevalent treatment arm included in the studies (n = 20), followed by ATV/r (n = 9), LPV/r (n = 8), DRV/r (n = 3), DTG (n = 3), RPV (n = 3), EVG/c (n = 2), and RAL (n = 2). Studies were found to be generally similar with respect to age and baseline clinical characteristics (Table S1) [11][13], [19], [20], [46][71]. The majority of patients were male (mean, 79.6%; range, 53.3%–93.1%) with mean age ranging from 29 to 40 years. Average baseline CD4+ cell count in the studies ranged from 150 to 396 cells/µL and log10 HIV RNA levels ranged from 4.52 to 5.41 copies/mL. All but one study included more than 50 patients per treatment arm and only 8 of the 31 included less than 100 (range of 31–465 patients). No statistically significant heterogeneity among treatment effects was identified for the EFV-RPV (p = 0.78; 3 studies) and EFV-LPV/r (p = 0.13; 3 studies) comparisons, the only comparisons associated with more than 2 studies.

Figure 2 displays the network of identified treatment comparisons included in the meta-analysis. Every study did not report every outcome (Table S1), and thus networks for individual outcome analyses varied. All studies included in the analysis examined at least one third agent of interest. “Connector” third agents (ATV, saquinavir-boosted ritonavir [SQV/r], fosamprenavir-boosted ritonavir [FPV/r], and nelfinavir [NFV]) were also included when 2 or more trials were identified comparing those agents to 2 or more treatments of interest. Inclusion of such connector treatments is recommended by published guidelines [72] since it provides additional indirect evidence. Trials with treatment arms examining different backbone NRTI regimens in combination with the same third agent were included in backbone-adjusted analyses.

thumbnail
Figure 2. Network of treatment comparisons contained within the identified clinical trials.

The major classes of third agents studied in the selected trials are indicated along the perimeter of the figure: NNRTIs, green; integrase inhibitors and PIs, purple; connectors, yellow. Black lines connecting each of the treatments of interest (red dots) represent a publication or clinical trial containing those two agents. Connector agents are drugs identified in 2 or more trials, and which were compared to 2 or more treatments of interest; connector agents are also members of the PI class. ATV = atazanavir; ATV/r = ritonavir-boosted atazanavir; DTG = dolutegravir; DRV/r = ritonavir-boosted darunavir; EFV = efavirenz; EVG/c = cobicistat-boosted elvitegravir; FPV/r = ritonavir-boosted fosamprenavir; LPV/r = lopinavir-boosted ritonavir; NFV = nelfinavir; NNRTI = non-nucleoside reverse transcriptase inhibitor; PI = protease inhibitor; RAL = raltegravir; RPV = rilpivirine; SQV/r = ritonavir-boosted saquinavir.

https://doi.org/10.1371/journal.pone.0105653.g002

Virologic suppression and CD4+ cell count change

Mean odds of virologic suppression (HIV RNA<50 copies/mL) were significantly higher for DTG than ATV/r, DRV/r, EFV, LPV/r, and RPV (Figure 3a). Backbone-unadjusted ORs of DTG were similar but slightly lower than the adjusted model results for all comparators (which affected the significance of treatment difference versus RPV). Similar to virologic suppression, DTG was estimated to have significantly higher mean CD4+ cell increases than ATV/r, DRV/r, EFV, LPV/r, and RPV (Figure 3b). The relative increase in CD4+ count was highest for DTG compared to EFV (37.9 cells/µL [95% CrI: 20.5,55.39]). Model results that were unadjusted for the NRTI backbone generated higher mean increases for DTG relative to all comparators, which resulted in DTG gaining significance compared to EVG/c. Random-effects model results were similar (see Appendix S2). Pooled estimates of the absolute probability of achieving virologic suppression and absolute mean CD4+ changes at week 48 are shown in Table 1.

thumbnail
Figure 3. Comparison of immunologic endpoints with dolutegravir versus third agents of interest.

(A) Odds ratio [95% CrI] for virologic suppression (HIV RNA<50 c/mL) of DTG compared with other third agents. Odds ratio values greater than 1 indicate the comparison favors DTG; CrI intervals that do not contain 1 are considered statistically significant. (B) Mean [95% CrI] CD4+ cell increase with dolutegravir versus third agents of interest. Crls of mean differences that do not contain 0 are considered statistically significant. ATV = atazanavir; ATV/r = ritonavir-boosted atazanavir; Crl = credible interval; DTG = dolutegravir; DRV/r = ritonavir-boosted darunavir; EFV = efavirenz; EVG/c = cobicistat-boosted elvitegravir; FPV/r = ritonavir-boosted fosamprenavir; LPV/r = lopinavir-boosted ritonavir; NFV = nelfinavir; RAL = raltegravir; RPV = rilpivirine; SQV/r = ritonavir-boosted saquinavir. *Indicates statistically significant comparison.

https://doi.org/10.1371/journal.pone.0105653.g003

thumbnail
Table 1. Estimated probability of virologic suppression and absolute CD4+ cell count change from baseline.

https://doi.org/10.1371/journal.pone.0105653.t001

Lipid changes

DTG had significantly lower associated TC, HDL, and LDL increases (Figure 4) relative to ATV/r, DRV/r, EFV, EVG/c, and LPV/r, with the exception of DRV/r and HDL change. DTG was not significantly different than RAL or RPV in any of these lipid outcomes. Models unadjusted for the NRTI backbone resulted in slightly higher relative mean increases for DTG. Conversely, HDL changes for DTG improved, achieving insignificance rather than being significantly lower compared to ATV/r and EVG/c (and statistically improved compared with RPV).

thumbnail
Figure 4. Relative changes in cholesterol and triglyceride parameters for dolutegravir versus third agents of interest.

Mean changes (mg/dL [95% Crl] in lipid levels with DTG compared with other third agents are shown for (A) total cholesterol (TC), (B) HDL cholesterol, (C) LDL cholesterol, and (D) triglycerides. In all cases, Crls of mean differences that do not include 0 are considered statistically significant. ATV = atazanavir; ATV/r = ritonavir-boosted atazanavir; DTG = dolutegravir; DRV/r = ritonavir-boosted darunavir; EFV = efavirenz; EVG/c = cobicistat-boosted elvitegravir; FPV/r = ritonavir-boosted fosamprenavir; HDL = high-density lipoprotein; LDL = low-density lipoprotein; LPV/r = lopinavir-boosted ritonavir; NFV = nelfinavir; RAL = raltegravir; RPV = rilpivirine; SQV/r = ritonavir-boosted saquinavir; TC = total cholesterol. *Indicates statistically significant comparison.

https://doi.org/10.1371/journal.pone.0105653.g004

Lower mean increases in TG were associated with DTG compared with DRV/r and LPV/r. Increases in TG were similar across all other comparisons except for RAL, for which higher mean TG increases were observed. The difference between the results for the model adjusted for NRTI versus the unadjusted model was smaller for TG than for the other lipids measured, although unadjusted results were associated with smaller uncertainty intervals, generating significantly lower and higher TG increases compared to ATV/r (−0.3 [−16.4, −0.2]) and RPV (12.5 [3.2,21.8]), respectively. Random-effects model results for lipids outcomes were similar to the fixed-effects model results (Appendix S2).

AEs and discontinuation due to AEs

Odds of experiencing an AE were significantly lower for DTG compared to ATV/r, EFV, and LPV/r (Table 2). Odds of discontinuation due to AEs were significantly lower with dolutegravir than with all treatments except RAL and RPV. RE model results showed no significant difference in odds of AEs between DTG and any other comparator and odds of discontinuation due to AEs were lower for DTG relative to only ATV/r, EFV, and LPV/r (Appendix S2).

thumbnail
Table 2. Mean odds ratio (95% CrI) of AEs and discontinuation due to AEs.

https://doi.org/10.1371/journal.pone.0105653.t002

Model validation

Consistency was assessed for changes in degree of virologic suppression (measured by levels of HIV RNA) and levels of CD4+ T cells (cells/µL) and lipids measures by comparing modeled estimates from the network meta-analysis with the non-EFV– compared data reported directly from the studies (Table 3) [73]. Agreement was found between RCT and model estimates for all 3 measurements, with the consistency measures including 0 (for continuous CD4+ T-cell count and lipid changes) and 1 (for virologic suppression), indicating consistency between the model findings and the direct clinical trial data.

thumbnail
Table 3. Difference (95% CrI) between direct clinical trial data and indirect model estimates.

https://doi.org/10.1371/journal.pone.0105653.t003

Discussion

This Bayesian meta-analysis estimated efficacy and safety outcomes of DTG relative to eight first-line treatment options, providing comparative evidence to other recommended third agents that had not been assessed in randomized clinical trials. Thirty-one RCTs including 14 treatments and approximately 17,000 treatment-naive HIV-1 patients were included in the analysis. Results indicated DTG was similar to or superior to nearly all comparators of interest in every outcome. The only exceptions were: 1) HDL change, where ATV/r, EFV, EVG/c, and LPV/r demonstrated greater increases and 2) backbone-unadjusted models of TC, LDL, and TG changes, where RPV resulted in significantly lower lipid increases than DTG, though backbone-adjusted model results were not significantly different.

Results of this analysis compare to those of a smaller meta-analysis published in 2011 prior to the introduction of RPV, EVG/c, and DTG [74]. Vieira and colleagues [74] included seven studies of EFV, LPV/r, ATV/r, DRV/r, FPV/r, and RAL within a random-effects Bayesian meta-analysis to conclude that all studied treatments have similar virologic suppression efficacy at 48 weeks and that only RAL had greater improvement in CD4+ cell count at week 48 compared to EFV, which was also observed in our study. The current analysis includes data from Vieira and colleagues [74] plus 24 additional trials, which were added in part due to the inclusion of three newer third agents (8), the inclusion of connector treatments (10), and backbone adjustment, which allowed for inclusion of trials examining two arms with the same third agent (4).

Inclusion of studies of so-called connector treatments is recommended by the UK guidelines for evidence synthesis under some circumstances [72] but is not very commonly applied within NMAs, in part because NMAs are used to examine the relative outcomes of all relevant comparators, thus reducing the likelihood of other comparators that are not of interest. However, for the treatment of HIV, the universe of available therapies is larger than the set of guideline-recommended treatments, as newer options with greater potency, tolerability, and convenience have replaced older treatments as preferred first-line options. Although connectors were not strictly necessary in this analysis to generate a connected network, inclusion of these trials added trial data that strengthened the estimates between treatments of interest. The disadvantage of adding these treatments is the increased risk of inconsistency among the trial comparisons, but this was not observed within our model (Table 3).

We have also included statistical adjustment for the NRTI backbone regimens used in each treatment combination. This adjustment can be considered a meta-regression with the backbone category as the covariate. This feature has not been included in other published meta-analyses of HIV treatment, as most clinical trials examine two or more third agents in combination with the same NRTI backbone (or investigator choice of backbone regimen). With such trials, backbone adjustment is not necessary because NMA calculations use the relative difference between treatment arms, so the effect of the third agent independent of the NRTI backbone is the model outcome. In the case of this analysis, one study examined DTG+ABC/3TC compared to EFV+TDF/FTC. A backbone-unadjusted NMA comparison for this study would not isolate the treatment effects of DTG and EFV, necessitating the use of the NRTI backbone covariate. To provide additional information to estimate the backbone coefficients, trials comparing the same third agent with different backbones were also included. Results of these analyses indicate that backbone agents are less influential in the probability of virologic suppression, but may have a larger impact on CD4+ cell count change and lipid outcomes.

Random-effects meta-analyses tend to generate larger uncertainty intervals than fixed-effects models, which could impact conclusions of statistical significance when making comparisons among the treatments. Larger uncertainty with random-effects models was also observed in this analysis. Some comparisons with the random-effects models resulted in no significant difference between DTG and comparator where there had been significance in the fixed-effects model.

As with any scientific research, statistical significance between treatments for any clinical endpoint may not necessarily imply clinical significance of the observed effects. For virologic suppression, official guidance documents, such as the FDA guidance to industry on the development of drugs for the treatment of HIV-1 infection [16], provide explicit guidelines clinical trials must satisfy to prove non-inferiority/superiority (e.g., requiring a non-inferiority margin of 10–12 percent), and these limits can be used to imply clinical and statistical significance. However, such explicit recommendations are not available for all clinical endpoints. For CD4+ cell count, although it is predictive of disease progression [75][77] the clinical impact and significance of a <50 cells/mm3 difference in CD4+ cell recovery between two treatments (as reported in this analysis) is unknown, and has yet to be established in long-term follow-up.

DTG had lower rates of discontinuation due to adverse events compared to most of the comparators in this analysis. Integrase inhibitors have established a reputation as a class of drugs with a low rate of discontinuation that is supported by long-term follow-up results from the STARTMRK study [78]. Two of the most recently approved third agents (DTG and RPV) have shown a lower rate of discontinuations due to adverse events than their comparator EFV [11], [79], [80]. Results from this NMA align with these conclusions.

NMA methodology is subject to limitations typical to any meta-analysis as well as to some unique limitations. Notably, the results obtained represent the statistical aggregation of data from the network pool. Thus, meta-analysis results should be consistent with but are not exactly equal to any individual RCT. Results of a given meta-analysis also depend on the quality and comparability of its collection of studies. In HIV, large-scale phase 3/4 studies are generally homogeneous, and the methodologies used to conduct the included studies were consistent (Table S1). To ensure comparability of specific data inputs, only data meeting specific definitions of the virologic suppression outcome and of the algorithm for treatment of missing data were included in the analysis.

The majority of trials were similar in most study and patient characteristics, limiting any bias from potential treatment effect modifiers, such as baseline HIV RNA levels (average log10 HIV RNA levels ranged from 4.52–5.41 copies/mL). However, some variation existed between the studies in the average baseline CD4+ cell count, which ranged from 150 to 396 cells/µL. Hence, a secondary analysis was conducted including baseline viral load and CD4+ cell count as covariates, but no significant impact was found on the treatment effects.

Statistically significant heterogeneity was not identified for available comparisons, although it must be noted that heterogeneity tests are known to have low power to detect differences when informed by a small sample of studies [81]. Only 2 comparisons were informed by 3 trials; all remaining comparisons were based on either 1 or 2 trials. Direct and indirect RCT comparisons were available for several treatment pairs and no significant differences were found between the 2, suggesting consistency within the evidence network.

Although the scope of this analysis was limited to comparative clinical effectiveness, decision makers are increasingly using cost-effectiveness as a criterion for selection of optimal treatment strategies. Cost-effectiveness analyses of DTG have been conducted elsewhere [82], [83] and provide evidence weighing the price of DTG against its clinical advantages. To quantify these advantages relative to comparators, NMAs have become increasingly used to understand the overall clinical efficacy and safety of new treatments within the landscape of currently available options, especially when comparative RCTs including all options are impractical. The results presented herein demonstrate that the efficacy and tolerability of DTG is at least comparable to, if not better than, other recommended front-line options for the treatment of HIV-1 infection.

Supporting Information

Table S1.

Study characteristics and outcome data. Patient demographics; viral load, CD4+ cell count, and percent of patients with viral suppression (<50 c/mL); baseline cholesterol measurements (LDL, HDL, TC, TG); and adverse events for the trials included in this meta-analysis.

https://doi.org/10.1371/journal.pone.0105653.s001

(DOC)

Acknowledgments

All listed authors meet the criteria for authorship set forth by the International Committee for Medical Journal Editors. The authors wish to acknowledge Varun Ektare and Sonam Mehta of Pharmerit International for assistance with extraction of data from the studies, and Clint Smith for providing editorial assistance during the development of this manuscript.

Previous Presentation

This work was presented as a poster (#PE7/7) at the 14th European AIDS Conference, October 16–19, 2013; Brussels, Belgium.

Author Contributions

Conceived and designed the research: DAP SJS WYT LS JWL RC SP KAG RRC JS GN. Study identification: DAP SJS WYT LS JWL RC KAG RRC JS GN. Data analysis and extraction: DAP SJS WYT LS. Contributed to the writing of the manuscript: DAP SJS WYT LS. Review/editing of the manuscript: JWL RC SP KAG RRC JS GN.

References

  1. 1. European AIDS Clinical Society (2013) Guidelines, Version 7.0. Available: http://www.eacsociety.org/Portals/0/Guidelines_Online_131014.pdf. Accessed 2014 Jun 24.
  2. 2. Department of Health and Human Services (2014) Guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents. Available: http://aidsinfo.nih.gov/contentfiles/lvguidelines/adultandadolescentgl.pdf. Accessed 2014 Jun 24.
  3. 3. Hogg R, Althoff K, Samji H, Cescon A, Modur S, et al.. (2013) Increases in life expectancy among treated HIV-positive individuals in the United States and Canada, 2000–2007. 7th IAS Conference on HIV Pathogenesis, Treatment and Prevention. Abstract TUPE260.
  4. 4. Rodger AJ, Lodwick R, Schechter M, Deeks S, Amin J, et al. (2013) Mortality in well controlled HIV in the continuous antiretroviral therapy arms of the SMART and ESPRIT trials compared with the general population. AIDS 27: 973–979.
  5. 5. Fong R, Cheng AC, Vujovic O, Hoy JF (2013) Factors associated with virological failure in a cohort of combination antiretroviral therapy-treated patients managed at a tertiary referral centre. Sex Health 10: 442–447.
  6. 6. Martin S, Foley K, Baser O (2007) Incremental medical costs associated with increased changes in HAART regimens in a US patient sample. 4th IAS Conference on HIV Pathogenesis, Treatment and Prevention. Abstract WEPEB047.
  7. 7. Solem C, Snedecor S, Khachatryan A, Nedrow K, Tawdrous M, et al. (2014) Cost of treatment in a US commercially-insured HIV-1 population. PLoS One 9: e98152.
  8. 8. Mannheimer S, Friedland G, Matts J, Child C, Chesney M (2002) The consistency of adherence to antiretroviral therapy predicts biologic outcomes for human immunodeficiency virus-infected persons in clinical trials. Clin Infect Dis 34: 1115–1121.
  9. 9. Paterson DL, Swindells S, Mohr J, Brester M, Vergis EN, et al. (2000) Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med 133: 21–30.
  10. 10. Fernandez-Montero JV, Barreiro P, Labarga P, De Mendoza C, Soriano V (2014) Dolutegravir, abacavir and lamivudine as HIV therapy. Expert Opin Pharmacother 15: 1051–1057.
  11. 11. Walmsley SL, Antela A, Clumeck N, Duiculescu D, Eberhard A, et al. (2013) Dolutegravir plus abacavir-lamivudine for the treatment of HIV-1 infection. N Engl J Med 369: 1807–1818.
  12. 12. Clotet B, Feinberg J, Lunzen J, Khuong-Josses M, Antinori A, et al. (2014) Once-daily dolutegravir versus darunavir plus ritonavir in antiretroviral-naive adults with HIV-1 infection (FLAMINGO): 48 week results from the randomised open-label phase 3b study. Lancet 383: 2222–2231.
  13. 13. Raffi F, Rachlis A, Stellbrink HJ, Hardy WD, Torti C, et al. (2013) Once-daily dolutegravir versus raltegravir in antiretroviral-naive adults with HIV-1 infection: 48 week results from the randomised, double-blind, non-inferiority SPRING-2 study. Lancet 381: 735–743.
  14. 14. Moher D, Liberati A, Tetzlaff J, Altman DG (2009) Group P (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA Statement. Open Med 3: e123–130.
  15. 15. Dias S, Welton NJ, Sutton AJ, Ades AE (2013) Evidence synthesis for decision making 1: introduction. Med Decis Making 33: 597–606.
  16. 16. U.S. Department of Health and Human Services (2013) Guidance for Industry: Human Immunodeficiency Virus-1 Infection: Developing Antiretroviral Drugs for Treatment. Available: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM355128.pdf. Accessed 2014 Jun 24.
  17. 17. Lu G, Ades AE (2004) Combination of direct and indirect evidence in mixed treatment comparisons. Stat Med 23: 3105–3124.
  18. 18. Lunn DJ TA, Best N, Spiegelhalter D (2000) Win-BUGS – a Bayesian modelling framework: concepts, structure, and extensibility. Stat Comput 10: 325–337.
  19. 19. Post FA, Moyle GJ, Stellbrink HJ, Domingo P, Podzamczer D, et al. (2010) Randomized comparison of renal effects, efficacy, and safety with once-daily abacavir/lamivudine versus tenofovir/emtricitabine, administered with efavirenz, in antiretroviral-naive, HIV-1-infected adults: 48-week results from the ASSERT study. J Acquir Immune Defic Syndr 55: 49–57.
  20. 20. Daar ES, Tierney C, Fischl MA, Sax PE, Mollan K, et al. (2011) Atazanavir plus ritonavir or efavirenz as part of a 3-drug regimen for initial treatment of HIV-1. Ann Intern Med 154: 445–456.
  21. 21. Behrens G, Maserati R, Rieger A, Domingo P, Abel F, et al. (2012) Switching to tenofovir/emtricitabine from abacavir/lamivudine in HIV-infected adults with raised cholesterol: effect on lipid profiles. Antivir Ther 17: 1011–1020.
  22. 22. Higgins JPT, Green S (2011) Cochrane Handbook for Systematic Reviews of Interventions, Version 5.1.0. Available: http://www.cochrane.org/handbook. Accessed: 2014 Jun 24.
  23. 23. Dias S, Welton NJ, Sutton AJ, Ades AE (2013) Evidence synthesis for decision making 5: the baseline natural history model. Med Decis Making 33: 657–670.
  24. 24. Berenguer J, Gonzalez J, Ribera E, Domingo P, Santos J, et al. (2008) Didanosine, lamivudine, and efavirenz versus zidovudine, lamivudine, and efavirenz for the initial treatment of HIV type 1 infection: final analysis (48 weeks) of a prospective, randomized, noninferiority clinical trial, GESIDA 3903. Clin Infect Dis 47: 1083–1092.
  25. 25. Bonnet M, Bhatt N, Baudin E, Silva C, Michon C, et al. (2013) Nevirapine versus efavirenz for patients co-infected with HIV and tuberculosis: a randomised non-inferiority trial. Lancet Infect Dis 13: 303–312.
  26. 26. Campbell TB, Smeaton LM, Kumarasamy N, Flanigan T, Klingman KL, et al. (2012) Efficacy and safety of three antiretroviral regimens for initial treatment of HIV-1: a randomized clinical trial in diverse multinational settings. PLoS Med 9: e1001290.
  27. 27. Dejesus E, Mills A, Bhatti L, Conner C, Storfer S (2011) A randomised comparison of safety and efficacy of nevirapine vs. atazanavir/ritonavir combined with tenofovir/emtricitabine in treatment-naive patients. Int J Clin Pract 65: 1240–1249.
  28. 28. Dore GJ, Cooper DA, Pozniak AL, DeJesus E, Zhong L, et al. (2004) Efficacy of tenofovir disoproxil fumarate in antiretroviral therapy-naive and -experienced patients coinfected with HIV-1 and hepatitis B virus. J Infect Dis 189: 1185–1192.
  29. 29. Lockman S, Hughes M, Sawe F, Zheng Y, McIntyre J, et al. (2012) Nevirapine- versus lopinavir/ritonavir-based initial therapy for HIV-1 infection among women in Africa: a randomized trial. PLoS Med 9: e1001236.
  30. 30. Maggiolo F, Ripamonti D, Ravasio L, Gregis G, Quinzan G, et al. (2003) Outcome of 2 simplification strategies for the treatment of human immunodeficiency virus type 1 infection. Clin Infect Dis 37: 41–49.
  31. 31. Mankhatitham W, Luaengniyomkul A, Manosuthi W (2012) Lipid profile changes in Thai HIV and tuberculosis co-infected patients receiving non-nucleoside reverse transcriptase inhibitors-based antiretroviral therapy. J Med Assoc Thai 95: 163–169.
  32. 32. Manosuthi W, Sungkanuparph S, Tantanathip P, Lueangniyomkul A, Mankatitham W, et al. (2009) A randomized trial comparing plasma drug concentrations and efficacies between 2 nonnucleoside reverse-transcriptase inhibitor-based regimens in HIV-infected patients receiving rifampicin: the N2R Study. Clin Infect Dis 48: 1752–1759.
  33. 33. Molina JM, Clumeck N, Redant K, Rimsky L, Vanveggel S, et al. (2013) Rilpivirine vs. efavirenz in HIV-1 patients with baseline viral load 100,000 copies/ml or less: week 48 phase III analysis. AIDS 27: 889–897.
  34. 34. Nelson M, Amaya G, Clumeck N, Arns da Cunha C, Jayaweera D, et al. (2012) Efficacy and safety of rilpivirine in treatment-naive, HIV-1-infected patients with hepatitis B virus/hepatitis C virus coinfection enrolled in the Phase III randomized, double-blind ECHO and THRIVE trials. J Antimicrob Chemother 67: 2020–2028.
  35. 35. Nishijima T, Takano M, Ishisaka M, Komatsu H, Gatanaga H, et al. (2013) Abacavir/lamivudine versus tenofovir/emtricitabine with atazanavir/ritonavir for treatment-naive Japanese patients with HIV-1 infection: a randomized multicenter trial. Intern Med 52: 735–744.
  36. 36. Sierra-Madero J, Villasis-Keever A, Mendez P, Mosqueda-Gomez JL, Torres-Escobar I, et al. (2010) Prospective, randomized, open label trial of efavirenz vs lopinavir/ritonavir in HIV+ treatment-naive subjects with CD4+<200 cell/mm3 in Mexico. J Acquir Immune Defic Syndr 53: 582–588.
  37. 37. Uy J, Yang R, Wirtz V, Sheppard L, Farajallah A, et al. (2011) Treatment of advanced HIV disease in antiretroviral-naive HIV-1-infected patients receiving once-daily atazanavir/ritonavir or twice-daily lopinavir/ritonavir, each in combination with tenofovir disoproxil fumarate and emtricitabine. AIDS Care 23: 1500–1504.
  38. 38. Walmsley S, Avihingsanon A, Slim J, Ward DJ, Ruxrungtham K, et al. (2009) Gemini: a noninferiority study of saquinavir/ritonavir versus lopinavir/ritonavir as initial HIV-1 therapy in adults. J Acquir Immune Defic Syndr 50: 367–374.
  39. 39. Wester CW, Thomas AM, Bussmann H, Moyo S, Makhema JM, et al. (2010) Non-nucleoside reverse transcriptase inhibitor outcomes among combination antiretroviral therapy-treated adults in Botswana. AIDS 24 Suppl 1S27–36.
  40. 40. Soriano V, Arasteh K, Migrone H, Lutz T, Opravil M, et al. (2011) Nevirapine versus atazanavir/ritonavir, each combined with tenofovir disoproxil fumarate/emtricitabine, in antiretroviral-naive HIV-1 patients: the ARTEN Trial. Antivir Ther 16: 339–348.
  41. 41. Gulick RM, Ribaudo HJ, Shikuma CM, Lalama C, Schackman BR, et al. (2006) Three- vs four-drug antiretroviral regimens for the initial treatment of HIV-1 infection: a randomized controlled trial. JAMA 296: 769–781.
  42. 42. Gallant JE, Staszewski S, Pozniak AL, DeJesus E, Suleiman JM, et al. (2004) Efficacy and safety of tenofovir DF vs stavudine in combination therapy in antiretroviral-naive patients: a 3-year randomized trial. JAMA 292: 191–201.
  43. 43. Ortiz R, Dejesus E, Khanlou H, Voronin E, van Lunzen J, et al. (2008) Efficacy and safety of once-daily darunavir/ritonavir versus lopinavir/ritonavir in treatment-naive HIV-1-infected patients at week 48. AIDS 22: 1389–1397.
  44. 44. Pinola M, Lazzarin A, Antinori A, Carosi G, Di Perri G (2010) Lopinavir/ritonavir+ tenofovir dual therapy versus lopinavir/ritonavir-based tripl e therapy in HIV-infected antiretroviral naïve subjects: The Kalead Study. J Antivir Antiretrovir 2: 56–62.
  45. 45. van Leth F, Conway B, Laplume H, Martin D, Fisher M, et al. (2004) Quality of life in patients treated with first-line antiretroviral therapy containing nevirapine and/or efavirenz. Antivir Ther 9: 721–728.
  46. 46. Walmsley S, Bernstein B, King M, Arribas J, Beall G, et al. (2002) Lopinavir-ritonavir versus nelfinavir for the initial treatment of HIV infection. N Engl J Med 346: 2039–2046.
  47. 47. Malan DR, Krantz E, David N, Wirtz V, Hammond J, et al. (2008) Efficacy and safety of atazanavir, with or without ritonavir, as part of once-daily highly active antiretroviral therapy regimens in antiretroviral-naive patients. J Acquir Immune Defic Syndr 47: 161–167.
  48. 48. Gallant JE, DeJesus E, Arribas JR, Pozniak AL, Gazzard B, et al. (2006) Tenofovir DF, emtricitabine, and efavirenz vs. zidovudine, lamivudine, and efavirenz for HIV. N Engl J Med 354: 251–260.
  49. 49. Riddler SA, Haubrich R, DiRienzo AG, Peeples L, Powderly WG, et al. (2008) Class-sparing regimens for initial treatment of HIV-1 infection. N Engl J Med 358: 2095–2106.
  50. 50. Smith KY, Weinberg WG, Dejesus E, Fischl MA, Liao Q, et al. (2008) Fosamprenavir or atazanavir once daily boosted with ritonavir 100 mg, plus tenofovir/emtricitabine, for the initial treatment of HIV infection: 48-week results of ALERT. AIDS Res Ther 5: 5.
  51. 51. Puls RL, Srasuebkul P, Petoumenos K, Boesecke C, Duncombe C, et al. (2010) Efavirenz versus boosted atazanavir or zidovudine and abacavir in antiretroviral treatment-naive, HIV-infected subjects: week 48 data from the Altair study. Clin Infect Dis 51: 855–864.
  52. 52. Mills AM, Nelson M, Jayaweera D, Ruxrungtham K, Cassetti I, et al. (2009) Once-daily darunavir/ritonavir vs. lopinavir/ritonavir in treatment-naive, HIV-1-infected patients: 96-week analysis. AIDS 23: 1679–1688.
  53. 53. Vrouenraets SM, Wit FW, Fernandez Garcia E, Moyle GJ, Jackson AG, et al. (2011) Randomized comparison of metabolic and renal effects of saquinavir/r or atazanavir/r plus tenofovir/emtricitabine in treatment-naive HIV-1-infected patients. HIV Med 12: 620–631.
  54. 54. Molina JM, Andrade-Villanueva J, Echevarria J, Chetchotisakd P, Corral J, et al. (2008) Once-daily atazanavir/ritonavir versus twice-daily lopinavir/ritonavir, each in combination with tenofovir and emtricitabine, for management of antiretroviral-naive HIV-1-infected patients: 48 week efficacy and safety results of the CASTLE study. Lancet 372: 646–655.
  55. 55. Sax PE, DeJesus E, Mills A, Zolopa A, Cohen C, et al. (2012) Co-formulated elvitegravir, cobicistat, emtricitabine, and tenofovir versus co-formulated efavirenz, emtricitabine, and tenofovir for initial treatment of HIV-1 infection: a randomised, double-blind, phase 3 trial, analysis of results after 48 weeks. Lancet 379: 2439–2448.
  56. 56. DeJesus E, Rockstroh JK, Henry K, Molina JM, Gathe J, et al. (2012) Co-formulated elvitegravir, cobicistat, emtricitabine, and tenofovir disoproxil fumarate versus ritonavir-boosted atazanavir plus co-formulated emtricitabine and tenofovir disoproxil fumarate for initial treatment of HIV-1 infection: a randomised, double-blind, phase 3, non-inferiority trial. Lancet 379: 2429–2438.
  57. 57. Smith KY, Patel P, Fine D, Bellos N, Sloan L, et al. (2009) Randomized, double-blind, placebo-matched, multicenter trial of abacavir/lamivudine or tenofovir/emtricitabine with lopinavir/ritonavir for initial HIV treatment. AIDS 23: 1547–1556.
  58. 58. Yeni P, Cooper DA, Aboulker JP, Babiker AG, Carey D, et al. (2006) Virological and immunological outcomes at 3 years after starting antiretroviral therapy with regimens containing non-nucleoside reverse transcriptase inhibitor, protease inhibitor, or both in INITIO: open-label randomised trial. Lancet 368: 287–298.
  59. 59. Jemsek JG, Arathoon E, Arlotti M, Perez C, Sosa N, et al. (2006) Body fat and other metabolic effects of atazanavir and efavirenz, each administered in combination with zidovudine plus lamivudine, in antiretroviral-naive HIV-infected patients. Clin Infect Dis 42: 273–280.
  60. 60. Eron J Jr, Yeni P, Gathe J Jr, Estrada V, DeJesus E, et al. (2006) The KLEAN study of fosamprenavir-ritonavir versus lopinavir-ritonavir, each in combination with abacavir-lamivudine, for initial treatment of HIV infection over 48 weeks: a randomised non-inferiority trial. Lancet 368: 476–482.
  61. 61. Echeverria P, Negredo E, Carosi G, Galvez J, Gomez JL, et al. (2010) Similar antiviral efficacy and tolerability between efavirenz and lopinavir/ritonavir, administered with abacavir/lamivudine (Kivexa), in antiretroviral-naive patients: a 48-week, multicentre, randomized study (Lake Study). Antiviral Res 85: 403–408.
  62. 62. Aberg JA, Tebas P, Overton ET, Gupta SK, Sax PE, et al. (2012) Metabolic effects of darunavir/ritonavir versus atazanavir/ritonavir in treatment-naive, HIV type 1-infected subjects over 48 weeks. AIDS Res Hum Retroviruses 28: 1184–1195.
  63. 63. Montaner JS, Schutz M, Schwartz R, Jayaweera DT, Burnside AF, et al. (2006) Efficacy, safety and pharmacokinetics of once-daily saquinavir soft-gelatin capsule/ritonavir in antiretroviral-naive, HIV-infected patients. MedGenMed 8: 36.
  64. 64. Andersson LM, Vesterbacka J, Blaxhult A, Flamholc L, Nilsson S, et al. (2013) Lopinavir/ritonavir, atazanavir/ritonavir, and efavirenz in antiretroviral-naive HIV-1-infected individuals over 144 weeks: an open-label randomized controlled trial. Scand J Infect Dis 45: 543–551.
  65. 65. Squires K, Lazzarin A, Gatell JM, Powderly WG, Pokrovskiy V, et al. (2004) Comparison of once-daily atazanavir with efavirenz, each in combination with fixed-dose zidovudine and lamivudine, as initial therapy for patients infected with HIV. J Acquir Immune Defic Syndr 36: 1011–1019.
  66. 66. Cohen C (2013) STaR study: single tablet regimen rilpivirine/emtricitabine/tenofovir DF has non-inferior efficacy compared to efavirenz/emtricitabine/tenofovir DF and improves patient reported outcomes. 7th IAS Conference on HIV Pathogenesis, Treatment and Prevention. Abstract TUPE284.
  67. 67. Lennox JL, DeJesus E, Lazzarin A, Pollard RB, Madruga JV, et al. (2009) Safety and efficacy of raltegravir-based versus efavirenz-based combination therapy in treatment-naive patients with HIV-1 infection: a multicentre, double-blind randomised controlled trial. Lancet 374: 796–806.
  68. 68. Cohen CJ, Andrade-Villanueva J, Clotet B, Fourie J, Johnson MA, et al. (2011) Rilpivirine versus efavirenz with two background nucleoside or nucleotide reverse transcriptase inhibitors in treatment-naive adults infected with HIV-1 (THRIVE): a phase 3, randomised, non-inferiority trial. Lancet 378: 229–237.
  69. 69. DeJesus E, McCarty D, Farthing CF, Shortino DD, Grinsztejn B, et al. (2004) Once-daily versus twice-daily lamivudine, in combination with zidovudine and efavirenz, for the treatment of antiretroviral-naive adults with HIV infection: a randomized equivalence trial. Clin Infect Dis 39: 411–418.
  70. 70. DeJesus E, Herrera G, Teofilo E, Gerstoft J, Buendia CB, et al. (2004) Abacavir versus zidovudine combined with lamivudine and efavirenz, for the treatment of antiretroviral-naive HIV-infected adults. Clin Infect Dis 39: 1038–1046.
  71. 71. Bartlett JA, Johnson J, Herrera G, Sosa N, Rodriguez A, et al. (2006) Long-term results of initial therapy with abacavir and Lamivudine combined with efavirenz, amprenavir/ritonavir, or stavudine. J Acquir Immune Defic Syndr 43: 284–292.
  72. 72. Dias S, Sutton AJ, Ades AE, Welton NJ (2013) Evidence synthesis for decision making 2: a generalized linear modeling framework for pairwise and network meta-analysis of randomized controlled trials. Med Decis Making 33: 607–617.
  73. 73. Bucher HC, Guyatt GH, Griffith LE, Walter SD (1997) The results of direct and indirect treatment comparisons in meta-analysis of randomized controlled trials. J Clin Epidemiol 50: 683–691.
  74. 74. Vieira MC, Kumar RN, Jansen JP (2011) Comparative effectiveness of efavirenz, protease inhibitors, and raltegravir-based regimens as first-line treatment for HIV-infected adults: a mixed treatment comparison. HIV Clin Trials 12: 175–189.
  75. 75. de Wolf F, Spijkerman I, Schellekens PT, Langendam M, Kuiken C, et al. (1997) AIDS prognosis based on HIV-1 RNA, CD4+ T-cell count and function: markers with reciprocal predictive value over time after seroconversion. AIDS 11: 1799–1806.
  76. 76. Phillips AN, Lundgren JD (2006) The CD4 lymphocyte count and risk of clinical progression. Curr Opin HIV AIDS 1: 43–49.
  77. 77. Goujard C, Bonarek M, Meyer L, Bonnet F, Chaix ML, et al. (2006) CD4 cell count and HIV DNA level are independent predictors of disease progression after primary HIV type 1 infection in untreated patients. Clin Infect Dis 42: 709–715.
  78. 78. Rockstroh JK, DeJesus E, Lennox JL, Yazdanpanah Y, Saag MS, et al. (2013) Durable efficacy and safety of raltegravir versus efavirenz when combined with tenofovir/emtricitabine in treatment-naive HIV-1-infected patients: final 5-year results from STARTMRK. J Acquir Immune Defic Syndr 63: 77–85.
  79. 79. Nelson MR, Elion RA, Cohen CJ, Mills A, Hodder SL, et al. (2013) Rilpivirine versus efavirenz in HIV-1-infected subjects receiving emtricitabine/tenofovir DF: pooled 96-week data from ECHO and THRIVE Studies. HIV Clin Trials 14: 81–91.
  80. 80. Molina JM, Cahn P, Grinsztejn B, Lazzarin A, Mills A, et al. (2011) Rilpivirine versus efavirenz with tenofovir and emtricitabine in treatment-naive adults infected with HIV-1 (ECHO): a phase 3 randomised double-blind active-controlled trial. Lancet 378: 238–246.
  81. 81. Huedo-Medina TB, Sanchez-Meca J, Marin-Martinez F, Botella J (2006) Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychol Methods 11: 193–206.
  82. 82. Despiégel N, Anger D, Martin M, Monga N, Cui Q, et al. (2014) Assessment of real-life predictive power of a new cost-effectiveness model in HIV: antiretroviral analysis by Monte Carlo individual simulation (ARAMIS-DTG). Value Health 17: A273.
  83. 83. Despiégel N, Anger D, Martin M, Gilchrist K, Camejo RR (2013) Validation of life expectancy estimates from the ARAMIS cost-effectiveness model in HIV - a corroboration exercise using CEPAC, the gold standard economic model. 14th HIV European AIDS Conference.