Among HIV-1 infected patients who achieved virologic suppression, the use of atazanavir without pharmacologic boosting is debated. We evaluated the efficacy and tolerance of maintenance therapy with unboosted atazanavir in clinical practice.
Methods and Results
This multicenter retrospective cohort study evaluated the efficacy of switching HIV-1-infected patients controlled on triple therapy to unboosted (ATV0, n = 98) versus ritonavir-boosted atazanavir (ATV/r, n = 254) +2 nucleos(t)ide reverse transcriptase inhibitors. The primary endpoint was time to virologic failure (VF, >200 copies/mL). ATV groups were compared controlling for potential confounding bias by inverse probability weighted Cox analysis and propensity-score matching. Overall and adjusted VF rates were similar for both strategies. Both strategies improved dyslipidemia and creatininemia, with less jaundice in the ATV0 group.
In previously well-suppressed patients, within an observational cohort setting, ATV0–based triple-therapy appeared as effective as ATV/r- based triple-therapy to maintain virologic suppression, even if co-administered with TDF, but was better tolerated.
Citation: Hocqueloux L, Choisy P, Le Moal G, Borsa-Lebas F, Plainchamp D, et al. (2012) Pharmacologic Boosting of Atazanavir in Maintenance HIV-1 Therapy: The COREYA Propensity-Score Adjusted Study. PLoS ONE 7(11): e49289. doi:10.1371/journal.pone.0049289
Editor: Nicolas Sluis-Cremer, University of Pittsburgh, United States of America
Received: August 2, 2012; Accepted: October 8, 2012; Published: November 9, 2012
Copyright: © 2012 Hocqueloux et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: BMS through an academic grant to Caen University, France. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: Dr. Parienti has received grant support, travel grants, and/or honoria from Boehringer Ingelheim, Abbott, BMS, MSD and Gilead. Other authors have declared that no competing interests exist.
Although most patients achieve virologic control with combined antiretroviral therapy (cART), adherence and tolerability issues (particularly metabolic and cardiovascular-associated morbidity) remain important concerns for HIV-infected patients receiving long-term treatment , . Unboosted atazanavir (ATV0) and tenofovir disoproxil fumarate (TDF) are two antiretrovirals that combine a low pill burden with good tolerability and favorable metabolic profile –. However, a pharmacological interaction occurs during co-administration of ATV and TDF, which has resulted in recommendations to continue low-dose ritonavir (RTV), which might reduce the expected benefit , , . With this background, not one of the randomized studies demonstrating the non-inferiority of a maintenance strategy with ATV0-based triple therapy was conducted in patients receiving TDF , , . Recently, two cohort studies have demonstrated the durability and safety of maintenance with ATV0-based triple therapy co-administered with TDF, but none included direct comparison with a RTV-boosted ATV regimen (ATV/r) , .
This study compared the efficacy of a switch from ATV/r to ATV0, with or without TDF, as a maintenance strategy in pre-treated patients on successful cART.
Patients and Methods
The COREYA study (COhort with REYAtaz) was a retrospective analysis of a large prospective multicentric cohort of HIV-1-infected subjects in 5 teaching hospitals in France. Patients had virological suppression (plasma HIV RNA <50 cp/mL) while receiving a triple therapy (2 nucleoside reverse transcriptase inhibitors [NRTI] +1 boosted protease inhibitor [PI], 2 NRTI +1 non-nucleoside reverse transcriptase inhibitor [NNRTI] or 3 NRTI), and had switched to ATV-based triple therapy, with or without ritonavir (RTV), between 2004 and 2011. Baseline was defined as the day ATV0 or ATV/r was initiated; there was no restriction on the use of TDF.
In this retrospective study, all participants provided informed written consent for the anonymous use of their clinical and biological data for biomedical research at the time their data were entered in the electronic database. The COREYA study was approved by the Institutional Review Board at the Côte de Nacre University Hospital, Caen, France (A10-D07-VOL.10, applicable to all French sites) on June 2010 and waived written informed consent, given the retrospective and observational nature of the current study. In addition, the anonymous use of the electronic database has been approved by the CNIL (Commission National de l’Informatique et des Libertés: http://legimobile.fr/fr/cnil/dec/aut/rech/2011/DR-2011-239/).
The primary endpoint was time to virological failure (single plasma HIV RNA >200 cp/mL; VF>200). Plasma HIV RNA >50 cp/mL (VF>50) was also analyzed. Secondary endpoints were safety (fasting lipids, renal function and hyperbilirubinemia), CD4+ reconstitution and T cell residual activation levels, pharmacokinetics, and reasons for change of treatment strategy other than VF.
Data collected at baseline included: demographics, medical and antiretroviral treatment history, genotype (where available for patients with a history of VF), CD4+ cell count nadir and plasma HIV RNA viral load (PVL) zenith before treatment, hepatitis B or C co-infection, bilirubinemia and serum creatinine levels, fasting cholesterol and triglycerides.
Follow-up data included: CD4+ cell counts, PVL, bilirubinemia and serum creatinine levels, fasting cholesterol, triglycerides, genotype at failure (if any) and change in cART regimen. Blood samples were taken 3 to 4 times a year, as recommended by national guidelines. When available, T cell activation status and plasma ATV trough concentrations (ATV Ctrough) data were collected.
The activation markers (CD38 and HLA-DR) were analyzed on gated CD4+ and CD8+ T cells and T cell activation levels were mainly defined as the percentages of CD4+ and CD8+ T cells expressing both CD38 and HLA-DR. T cell activation was measured in freshly collected, EDTA-anticoagulated whole blood and analyzed by flow cytometry (Becton-Dickinson FACS Canto II). Study patients were compared with a group that had undetectable PVL under NNRTI-based triple-therapy and HIV-uninfected healthy donors (personal unpublished data). All treated groups were matched for age, sex ratio, nadir CD4+ count, highest PVL before cART and duration with undetectable PVL.
Measurements of ATV Ctrough were performed using high-performance liquid chromatography. Only samples collected 24±2 hours after the last intake of ATV0 or ATV/r were considered.
All genotypes were (re-)interpreted using the ANRS 2011 algorithm (http://www.hivfrenchresistance.org/2011/Algo-2011.pdf).
With a rate for the primary endpoint of 4 per 100 patient-years in the ATV/r group, median accrual and follow-up times of 1 and 2 years, respectively, a one-sided alpha of 0.05, a power of 80%, a true hazard ratio of 1.0 and a proportion of ATV/r (control group) of 66.6%, and a hazard ratio non-inferiority margin of 3.0, the calculated sample size was 241.
Data were expressed as mean ± standard deviation, or median (interquartile range) and percentage. Differences between patients switching to ATV0 or ATV/r were compared using Chi-square or Fisher exact tests for categorical variables and two-tailed, unpaired t test or Wilcoxon rank sum test for continuous variables.
The association between VF and treatment strategy was evaluated by the hazard ratio and its 95% confidence interval (CI), with a value >1 signifying an increased risk of virologic failure in the ATV0 group. Analysis using propensity scores was conducted to limit inclusion biases. Potential indication or ‘channeling’ biases were adjusted for by developing a propensity score for switching to one versus another strategy . Two different marginal structural models were used , : inverse probability weighting (IPWT) and matching on the propensity score. The primary endpoint was modeled by IPWT Cox model, after checking the proportionality assumption which was met. Second, a one-to-one greedy 5 to 1 digit technique to match one control (ATV/r group) by one case (ATV0 group) nested within the overall cohort was performed. In this matched sample, baseline characteristics included in the propensity score were compared between cases and controls by paired-tests . The probability of endpoint was then modeled in a Cox model with robust sandwich variance estimators including the ATV group as explanatory factor. A P value <0.05 was considered significant and all P values were two-tailed. Statistical analyses were performed using SAS statistical software, version 9.1 (SAS Institute Inc., Cary, NC).
A total of 352 patients were included, 98 of whom switched to ATV0 and 254 to ATV/r. Overall, most patients (191/352, 54%) switched from lopinavir/r-based cART. The main reasons for switching to ATV were to reduce pill burden (218/352, 62%) and side effects (91/352, 26%). Sixty-one patients (17%) had a history of VF; genotyping was available in 65%, and resistance-encoding mutations for ≥1 drug in the NRTI, NNRTI and PI classes were found in 53%, 53% and 24% of the available sequences, respectively. There were significant differences between the two groups (boosted and unboosted) at baseline (Table 1). Factors significantly associated with a switch to ATV without ritonavir rather than a switch to ATV with ritonavir were female sex, asymptomatic HIV infection (according to the CDC classification), higher CD4 cells nadir, high number of previous lines of treatment, prior treatment with 3 NRTI, longer time with undetectable plasma viral load, higher CD4 cell count and more dyslipidemia as the reason for the switch. Propensity score matching limited these differences (as shown in Table 1). The mean follow-up was 2.2±1.7 years, which was similar in the two groups.
Table 1. Baseline characteristics and main outcomes for patients switching to unboosted (ATV0) versus boosted (ATV/r) atazanavir-based regimens in the overall population and in the propensity-matched subgroups.doi:10.1371/journal.pone.0049289.t001
Primary and Secondary Virologic Endpoints
Overall, VF>200 occurred in 3/98 (3.1%) and 13/254 (5.1%) patients in the ATV0 and ATV/r groups, respectively (p = 0.39) (Figure 1). In the IPWT-analysis, risk of VF>200 was not statistically different (HR 0.85, 95%CI [0.38–1.92], p = 0.70); the upper boundary of the 95% CI was below the pre-defined non-inferiority margin for the HR. Corresponding values for VF>50 were 3/98 (3.1%) and 27/254 (10.6%; p = 0.025), which translated to a significant difference favouring ATV0 over ATV/r in the IPWT-analysis (HR 0.39, 95%CI [0.19–0.80], p = 0.011).
Figure 1. Kaplan-Meier curves of time without viral failure (VF, >200 cp/mL) in patients switching to to unboosted (ATV0) versus boosted (ATV/r) atazanavir-based regimens in the overall population.doi:10.1371/journal.pone.0049289.g001
For analyses in the propensity-matched subgroups, 3/72 (4.1%) and 3/72 (4.1%) patients experienced VF>200 (HR 0.95, 95%CI [0.25–3.61], p = 0.94) in the ATV0 and ATV/r groups, respectively (Figure 2); results for the VF>50 analysis were similar. Overall, and in the ATV0 group, there was a trend for less VF in patients receiving TDF.
Figure 2. Kaplan-Meier curves of time without viral failure (VF, >200 cp/mL) in patients switching to to unboosted (ATV0) versus boosted (ATV/r) atazanavir-based regimens in propensity-matched subgroups.doi:10.1371/journal.pone.0049289.g002
Other Secondary Endpoints
During follow-up, CD4+ count increased from baseline (p<0.0001), with both ATV0 and ATV/r groups showing similar immune reconstitution (p = 0.68). There was no difference between the ATV0, ATV/r and NNRTI groups in terms of residual T cell activation levels. HIV-uninfected donors (n = 21) had much lower T cells activation levels than all three treated groups (p = 0.002 and p = 0.02, respectively).
In the ATV0 group, the mean Ctrough (n = 66) was 200±176 ng/mL (36% of values were >200 ng/mL, that is the recommended threshold to be above). Of note, there was no difference in ATV0 Ctrough between patients with or without TDF in the backbone (200±188 vs 205±153, respectively, p = 0.64). There was no association between ATV0 Ctrough and VF (216 if VF, 199 if no VF, p = 0.43). Patients with ATV0 Ctrough <200 ng/mL (n = 42) had similar VF>200 rate than patients with ATV0 Ctrough >200 ng/mL (3% vs 12%, p = 0.21, Log-Rank test). In the ATV/r group Ctrough (n = 30) values were much higher (949±1401, p<0.0001) than in the ATV0 group.
LDL-cholesterol (n = 153, −0.29±0.95 mmol/L, p<0.001) and triglycerides (n = 173, −0.54±1.26, p<0.001), but not HDL cholesterol (n = 166, 0.01±0.34, p = 0.79), decreased from baseline in both groups, with no statistically significant difference between groups. There was also a significant decrease in creatinemia (n = 337, −3.8±15 µmol/L, p = 0.001). No significant difference was found between the patients according to the ATV group, or the presence or absence of TDF. Hyperbilirubinemia grade 3–4 was less frequent in ATV0 than in ATV/r group (9% vs 30%, p<0.0001).
Overall, discontinuation for other reasons than VF occurred in 113/352 patients (32%), after a mean 1.5 years. Main causes were: clinical tolerability issues and patients’ request (n = 78), physician’s decision to change strategy (n = 9), pregnancy (n = 8), dyslipidemia (n = 3), regimen simplification (n = 3), death (n = 1) and miscellaneous (n = 11). The discontinuation rate differed markedly between groups, with a much lower rate for ATV0 versus ATV/r, even after adjustment for propensity score (Table 1).
This study suggests that, in previously well-controlled patients, ATV0 is an effective alternative to ATV/r for maintaining long-term viral suppression, even in treatment-experienced patients and in the presence of TDF. In this setting, ATV0 is better tolerated than ATV/r.
For the main analysis, we chose to define VF as >200 cp/mL rather than >50 cp/mL because the lower threshold may categorize an unacceptably high number of patients who ultimately re-suppress to <50 cp/mL without a change in ART as having VF , . The results show that the failure rate at 1, 2 and 3 years using the 200 cp/mL threshold is remarkably low in patients with ATV0 (0, 4 and 4% respectively) and similar to published figures , , . In one study, using a VF threshold of <400 cp/mL, the failure rate was <2%, <3% and <5% at 1, 2 and 3 years, respectively . In this group, the only risk factor for failure was co-infection with hepatitis C, and the authors suggested a link between substance abuse and poor compliance . In our study, the rate of co-infection was low and patients had good control prior to the switch to ATV0 (almost 3 years). In addition, the reason patients in our study switched to ATV0 (to reduce side effects or lower pill burden) is likely to perpetuate a high adherence rate.
It is noteworthy that our patients frequently had ATV0 Ctrough below the recommended threshold without an increased risk of VF. Poor adherence to ATV0 is very unlikely in the context of patients with long-term viral suppression, good immunologic reconstitution and low immune activation. This suggests that, in patients with long-term viremia suppression, the current recommended cut-off of >200 ng/mL may be overestimated.
Overall, TDF is associated with trend toward a better virological control. In the context of co-administration with ATV0, which does not allow complete coverage of the day in more than half of patients, the long half-life of TDF may offer an advantage over abacavir. Moreover, in our study, TDF did not have a negative impact on the pharmacokinetics of ATV0, as there was no more under-dosing than in the group without TDF. Our results are similar to those reported by Calcagno et al but contradict previously reported pharmacokinetic studies , . In highly pre-selected patients within a cohort study, we did not observe this well described pharmacokinetic interaction , however our study was not designed to assess such observations. Therefore, this result should be interpreted with caution and therapeutic drug monitoring may be warranted on a case-by-case basis.
ATV is associated with a good metabolic profile and low levels of insulin-resistance. Indeed, a recent study reported that ATV, even boosted, is not associated with an excess risk for coronary artery disease or stroke . The use of unboosted ATV avoids ritonavir-associated toxicity and also reduces ATV systemic, which has been shown to decrease the risk of dyslipidemia, renal impairment and jaundice , , ,  and also perhaps nephrolithiasis . It is interesting to note in our study that the risk of ATV0 discontinuation for causes other than VF was significantly lower than in the ATV/r group. These findings suggest that ATV0 is better tolerated than ATV/r. The reduction of side effects is likely to be one of the main causes of the durability of ATV0-based regimen, as it has been demonstrated that side effects are strongly associated with decreased average adherence , leading to VF .
The main limitation of this study is its retrospective design, which can introduce important biases, particularly channelling bias. For example, the use of abacavir has been associated with an increased risk of myocardial infarction in cohort studies despite multivariable adjustments ,  but not in a carefully matched case-control study  or in randomized controlled trials . In our study, the ATV/r boosted group was clearly disadvantaged at baseline, combining markers of poor prognosis (more previous AIDS-defining illnesses, lower CD4+ counts, etc). This might explain the better secondary virologic outcome (VF<50) observed in the ATV0-based group despite the use of the sophisticated IPWT adjustment. We attempted to overcome this by incorporation of an analysis based on propensity score matching subgroup nested in our cohort and therefore excluding patients too different to be matched. The latter analysis did not confirm the better virologic outcome (VF<50) in the ATV0 group. More importantly, our results appear to be consistent with current literature, including from randomized controlled trials , .
In conclusion, maintenance of patients on successful cART for ≥1 year with ATV0-based triple therapy is associated with low rates of VF and discontinuation without VF compared with ATV/r. In this setting, the use of TDF had no deleterious impact on ATV0 Ctrough and virologic efficacy.
English language writing support was provided by Nicola Ryan of inScience Communications. This assistance was funded by Bristol-Myers-Squibb Inc. The authors thank Fabien Chaillot for administrative support.
This work was presented in part at the 13th European AIDS Conference, Belgrade, Serbia, October 12–15, 2011 [Abstract PE7.5/8].
Conceived and designed the experiments: LH JJP. Performed the experiments: JJP LH PC GLM FBL DP EL TP XT YY. Analyzed the data: JJP. Contributed reagents/materials/analysis tools: JJP LH PC GLM FBL DP EL TP XT YY. Wrote the paper: LH JJP.
- 1. Panel on Antiretroviral Guidelines for Adults and Adolescents. Guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents. Department of Health and Human Services. January 10, 2011; 1–166. (n.d.). Available: http://aidsinfo.nih.gov/contentfiles/AdultandAdolescentGL.pdf. Accessed 2 April 2011.
- 2. Yeni P (2010) Prise en charge médicale des personnes infectées par le VIH. Paris: La documentation française.
- 3. Ghosn J, Carosi G, Moreno S, Pokrovsky V, Lazzarin A, et al. (2010) Unboosted atazanavir-based therapy maintains control of HIV type-1 replication as effectively as a ritonavir-boosted regimen. Antivir Ther (Lond) 15: 993–1002 doi:10.3851/IMP1666.
- 4. Squires KE, Young B, Dejesus E, Bellos N, Murphy D, et al. (2010) Similar efficacy and tolerability of atazanavir compared with atazanavir/ritonavir, each with abacavir/lamivudine after initial suppression with abacavir/lamivudine plus ritonavir-boosted atazanavir in HIV-infected patients. AIDS 24: 2019–2027 doi:10.1097/QAD.0b013e32833bee1b.
- 5. Valantin MA, Bittar R, de Truchis P, Bollens D, Slama L, et al. (2010) Switching the nucleoside reverse transcriptase inhibitor backbone to tenofovir disoproxil fumarate + emtricitabine promptly improves triglycerides and low-density lipoprotein cholesterol in dyslipidaemic patients. J Antimicrob Chemother 65: 556–561 doi:10.1093/jac/dkp462.
- 6. Friis-Møller N, Thiébaut R, Reiss P, Weber R, Monforte AD, et al. (2010) Predicting the risk of cardiovascular disease in HIV-infected patients: the data collection on adverse effects of anti-HIV drugs study. Eur J Cardiovasc Prev Rehabil 17: 491–501 doi:10.1097/HJR.0b013e328336a150.
- 7. Lang S, Mary-Krause M, Cotte L, Gilquin J, Partisani M, et al. (2010) Increased risk of myocardial infarction in HIV-infected patients in France, relative to the general population. AIDS 24: 1228–1230 doi:10.1097/QAD.0b013e328339192f.
- 8. Taburet A-M, Piketty C, Chazallon C, Vincent I, Gérard L, et al. (2004) Interactions between atazanavir-ritonavir and tenofovir in heavily pretreated human immunodeficiency virus-infected patients. Antimicrob Agents Chemother 48: 2091–2096 doi:10.1128/AAC.48.6.2091-2096.2004.
- 9. Gatell J, Salmon-Ceron D, Lazzarin A, Van Wijngaerden E, Antunes F, et al. (2007) Efficacy and safety of atazanavir-based highly active antiretroviral therapy in patients with virologic suppression switched from a stable, boosted or unboosted protease inhibitor treatment regimen: the SWAN Study (AI424–097) 48-week results. Clin Infect Dis 44: 1484–1492 doi:10.1086/517497.
- 10. Giuntini R, Martinelli C, Ricci E, Vichi F, Gianelli E, et al. (2010) Efficacy and safety of boosted and unboosted atazanavir-containing antiretroviral regimens in real life: results from a multicentre cohort study. HIV Med 11: 40–45 doi:10.1111/j.1468-1293.2009.00740.x.
- 11. Pavie J, Porcher R, Torti C, Medrano J, Castagna A, et al. (2011) Efficacy and safety of a switch to unboosted atazanavir in combination with nucleoside analogues in HIV-1-infected patients with virological suppression under antiretroviral therapy. J Antimicrob Chemother 66: 2372–2378 doi:10.1093/jac/dkr316.
- 12. D’Agostino RB Jr (1998) Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 17: 2265–2281. doi: 10.1002/(sici)1097-0258(19981015)17:19<2265::aid-sim918>3.0.co;2-b
- 13. Robins JM, Hernán MA, Brumback B (2000) Marginal structural models and causal inference in epidemiology. Epidemiology 11: 550–560. doi: 10.1097/00001648-200009000-00011
- 14. Austin PC, Grootendorst P, Anderson GM (2007) A comparison of the ability of different propensity score models to balance measured variables between treated and untreated subjects: a Monte Carlo study. Stat Med 26: 734–753 doi:10.1002/sim.2580.
- 15. Aldous JL, Haubrich RH (2009) Defining treatment failure in resource-rich settings. Curr Opin HIV AIDS 4: 459–466 doi:10.1097/COH.0b013e328331dea5.
- 16. Ribaudo H, Lennox J, Currier J, et al. Virologic failure endpoint definition in clinical trials: is using HIV-1 RNA threshold <200 copies/mL better than <50 copies/mL? An analysis of ACTG studies. Presnted at: 16th Conference on Retroviruses and Opportunistic Infections; Feb 8–11, 2009; Montreal, Canada.
- 17. Calcagno A, Bonora S, Tettoni MC, D’Avolio A, Perri GD, et al. (2009) Tenofovir coadministration is not associated with lower unboosted atazanavir plasma exposure in the clinical setting. J Acquir Immune Defic Syndr 52: 431–432 doi:10.1097/QAI.0b013e3181b08f09.
- 18. D’Arminio Monforte (n.d.) D’Arminio Monforte A, Reiss P, Ryom L, et al. Atazanavir-containing ART Is not associated with an increased risk of cardio- or cerebro-vascular events: The D:A:D study. Presented at: 19th Conference on Retroviruses and Opportunistic Infections; 5–8 March, 2012; Seattle, Washington, USA.
- 19. Ryom L, D:A:D Study Group. Exposure to ARV and the risk of renal impairment among HIV+ persons with normal baseline renal function: the D:A:D Study. Presented at: 19th Conference on Retroviruses and Opportunistic Infections; 5–8 March, 2012; Seattle, Washington, USA.
- 20. Ickovics JR, Cameron A, Zackin R, Bassett R, Chesney M, et al. (2002) Consequences and determinants of adherence to antiretroviral medication: results from Adult AIDS Clinical Trials Group protocol 370. Antivir Ther (Lond) 7: 185–193.
- 21. Parienti J-J, Ragland K, Lucht F, de la Blanchardière A, Dargère S, et al. (2010) Average adherence to boosted protease inhibitor therapy, rather than the pattern of missed doses, as a predictor of HIV RNA replication. Clin Infect Dis 50: 1192–1197 doi:10.1086/651419.
- 22. Sabin CA, Worm SW, Weber R, Reiss P, El-Sadr W, et al. (2008) Use of nucleoside reverse transcriptase inhibitors and risk of myocardial infarction in HIV-infected patients enrolled in the D:A:D study: a multi-cohort collaboration. Lancet 371: 1417–1426 doi:10.1016/S0140-6736(08)60423-7.
- 23. Obel N, Farkas DK, Kronborg G, Larsen CS, Pedersen G, et al. (2010) Abacavir and risk of myocardial infarction in HIV-infected patients on highly active antiretroviral therapy: a population-based nationwide cohort study. HIV Med 11: 130–136 doi:10.1111/j.1468-1293.2009.00751.x.
- 24. Lang S, Mary-Krause M, Cotte L, Gilquin J, Partisani M, et al. (2010) Impact of individual antiretroviral drugs on the risk of myocardial infarction in human immunodeficiency virus-infected patients: a case-control study nested within the French Hospital Database on HIV ANRS cohort CO4. Arch Intern Med 170: 1228–1238 doi:10.1001/archinternmed.2010.197.
- 25. Ding X, Andraca-Carrera E, Cooper C, Miele P, Kornegay C, et al. (2012) No Association of Abacavir Use with Myocardial Infarction: Findings of an FDA Meta-analysis. J Acquir Immune Defic Syndr. Available: http://www.ncbi.nlm.nih.gov/pubmed/22932321. Accessed 20 September 2012.