Modified Directly Observed Antiretroviral Therapy Compared With Self-administered Therapy in Treatment-Naïve HIV-1–Infected Patients:  A Randomized Trial FREE

The success of modern antiretroviral therapy for the treatment of human immunodeficiency type 1 (HIV-1) infection remains highly dependent on strict adherence to the regimen.^1- 3 Although rates of adherence in the developing world have recently been reported to be higher than those in the developed world,⁴ challenges to adherence remain in both settings.⁵ Unfortunately, relatively few interventions have been demonstrated to be effective.⁶

The most consistently associated barriers to adherence have been forgetting to take doses,⁷ depression,^7- 10 active substance use,^9,11- 12 lack of social support,¹³ interrupted access to medications,¹⁴ and medication toxic effects.¹⁵ Directly observed therapy (DOT) strategies can address many but not all barriers to adherence.¹⁶ However, the results of randomized trials of DOT strategies to enhance antiretroviral therapy outcomes have been mixed.^17- 19 This heterogeneity may have been due to inadequacy of sample size, differences in DOT implementation, and the populations studied. In addition, prior studies did not assess whether DOT entrained adherence behavior after DOT was discontinued.

We modified DOT (mDOT) to be implemented only on weekdays and to focus on only 1 drug in the regimen. Our objectives were to test whether mDOT over the initial 6 months of a first antiretroviral regimen would result in improved virological outcomes after 6 months and whether it would entrain good adherence behavior after it was stopped.

We conducted an open-label, 3-arm randomized clinical trial comparing a once-daily lopinavir/ritonavir soft gel capsule (sgc) (800 mg/200 mg regimen) via self-administration with either once-daily lopinavir/ritonavir sgc (800 mg/200 mg) via mDOT or twice-daily lopinavir/ritonavir sgc (400 mg/100 mg) via self-administration. The study aimed to answer 2 separate questions with a common comparison arm (once-daily self-administered therapy). We report herein the results of the comparison between the mDOT and self-administration strategies using once-daily lopinavir/ritonavir.

The regimens all included emtricitabine (200 mg once daily) and either extended-release stavudine (100 mg once daily) or tenofovir (300 mg once daily). At the outset (October 2002), the regimens exclusively included extended-release stavudine, an investigational formulation that is not commercially available. In March 2004, participants were allowed to choose tenofovir as well, either as part of the initial treatment or as a switch from stavudine during the study if desired by the patient and health care provider. Randomization to treatment strategy was stratified by screening plasma HIV-1 RNA level of 100 000 copies/mL with dynamic balancing by main institution and assigned via permuted blocks in a 2:2:1 ratio. Modified DOT had the smaller sample size based on different primary end point considerations than for the once-daily vs twice-daily lopinavir/ritonavir comparison. Allocation was performed using an AIDS Clinical Trials Group (ACTG) centralized-computer system. Owing to the nature of mDOT, there was no masking of assignment.

Individuals were eligible if they had a plasma HIV-1 RNA level higher than 2000 copies/mL, were treatment naïve (≤7 days of prior antiretroviral treatment), with any CD4 lymphocyte count, weighed at least 40 kg, were older than 13 years, and were willing and able to participate in mDOT. Modified DOT was defined as the observed and documented taking of lopinavir/ritonavir at least 5 days per week (Monday through Friday) and was conducted by any person in a medically related field (eg, nurse, pharmacist, social worker) who was not a relative or close friend. Doses could be observed at any location. Participants were provided with a limited extra supply of medications for weekends, holidays, and emergencies. The choice of observer had to be approved by the site and could include site staff. The intervention had the following 2 phases: (1) active mDOT for the initial 24 weeks and (2) post-mDOT phase of self-administered therapy for weeks 24 to 48. The study did not prescribe actions to be taken by the observer or the site if nonadherence ensued. All participants provided informed consent, and the study was approved by the committees on the protection of human subjects or ethics boards at each site.

Study visits were scheduled to occur at 4, 8, 16, 24, 32, 40, and 48 weeks after randomization. Each visit included clinical assessments and laboratory testing, including measurement of plasma HIV-1 RNA levels (Roche UltraSensitive HIV-1 Monitor assay [Roche Molecular Systems, Pleasanton, California], run at a central laboratory). Plasma HIV-1 RNA assays were batched through visit week 8 but were performed in real-time subsequently. An additional safety visit was scheduled for visit week 2. Alcohol and substance use were assessed using self-reports, and excessive alcohol use was defined as averaging more than 2 drinks per day or binge drinking.²⁰ Participants were to follow this schedule regardless of whether they were taking their originally randomized study regimen.

The primary efficacy end point, virologic success, was defined as lack of virologic failure through week 24. Virological failure was defined as a confirmed HIV-1 RNA level of 200 copies/mL or higher after 2 consecutive measurements of an HIV-1 RNA value lower than 200 copies/mL (confirmed viral relapse) or a confirmed HIV-1 RNA level of 200 copies/mL or higher at or after visit week 24 without prior confirmed viral relapse. Secondary end points included virologic success through week 48, time to virologic failure, change in CD4 lymphocyte count from baseline at visit week 24 and 48, grade 3 or 4 sign, symptom, or laboratory measure that was at least 1 grade higher than baseline on the National Institute of Allergy and Infectious Diseases Division of AIDS (DAIDS) toxicity scale,²¹ clinical events (AIDS-defining illnesses or death), adherence, and genotypic resistance. Adherence was measured using electronic monitors (Medication Event Monitoring System [MEMS]; Aardex Corp, Zug, Switzerland), which recorded the date and time of the bottle opening on the lopinavir/ritonavir bottles and by self-report using a standardized ACTG questionnaire.²² The adherence data from the MEMS were not provided to the participants, site staff, or study team. Attitudes toward mDOT were assessed in the mDOT participants at the outset and end of the active mDOT phase via a questionnaire developed for this study. Viral genotypic resistance testing (Monogram Biosciences, San Francisco, California) was performed on pairs of samples from all participants with virologic failure at any time during follow-up: one stored at baseline and a second obtained at the time of virologic failure. Resistance was defined based on the Stanford algorithm version 4.2.6 (http://hivdb.stanford.edu/index.html).

The primary analysis assessed whether the probability of virologic success was higher for mDOT compared with self-administered therapy through week 24 using Kaplan-Meier estimates from the survival distribution for the time to virologic failure and a 1-sided Z test for a difference in the probabilities of 0.075, taking an intent-to-treat approach. A 1-sided test was chosen a priori because we were only interested in whether mDOT was better than self-administered therapy for virologic success. All other comparisons were 2-sided. Time to virologic failure was defined as the time from randomization to the date of the initial failure sample. Participants whose last plasma HIV-1 RNA sample was an unconfirmed failure were considered virologic failure at the time of the last sample. For participants who missed visit week 24, if both visit week 16 and 32 samples had HIV-1 RNA levels higher than 200 copies/mL, the failure date was imputed at week 24. For participants without virologic failure, the failure time was censored at the last date their plasma HIV-1 RNA was measured. Secondary intent-to-treat analyses compared the probability of sustained virological success through week 48 using week 48 Kaplan-Meier estimates and 2-sided 95% confidence intervals around the difference in probability of virologic success. Time-to-event distributions were compared with a log-rank test that was stratified by screening plasma HIV-1 RNA level (<100 000 vs ≥100 000 copies/mL). Modification of the effect of the intervention by the mDOT phase was assessed in a Cox proportional hazards model of virologic success over the 48 weeks with an interaction term for before study week 24 (active mDOT) or after study week 24 (post-mDOT).

Adherence to lopinavir/ritonavir therapy was calculated separately for each individual over visit weeks 0 to 24 and weeks 24 to 48 using the following electronic monitor data:

(Number of Bottle Openings ÷ Number of Days of Observation) × 100%.

Days on which the participant was instructed not to take the medication (eg, owing to toxic effects) were excluded from the calculation. Percentage adherence and CD4 lymphocyte count change from baseline were compared between treatment strategies using the Wilcoxon rank sum test. Self-reported adherence was assessed with a Cochran-Mantel-Haenszel test. Changes in attitudes toward mDOT were compared with a signed rank test. Analyses were performed using SAS version 9 (SAS Institute Inc, Cary, North Carolina) and S-plus version 6 (Insightful Corp, Seattle, Washington) software.

An independent study monitoring committee was impaneled by the ACTG and consisted of investigators not directly involved in the study. The committee performed 3 reviews, with blinded efficacy comparisons presented only at the first review soon after study initiation. The stopping guideline was based on the method of Lan and DeMets²³ with an O’Brien-Fleming spending function.²⁴ Reported P values are nominal.

The sample size calculation required consideration of the other aim of the study. The aim comparing once-daily vs twice-daily lopinavir/ritonavir required 150 participants each. Thus, we targeted enrollment of 75 participants for the mDOT strategy to have 81% power to detect a difference of at least 0.075 between mDOT and once-daily self-administered therapy and probabilities of virologic success at week 24 of 0.91 and 0.70 for the 2 respective strategies.

The study enrolled 403 participants into the 3 strategies. A total of 243 participants were randomized—161 to self-administered therapy and 82 to mDOT. Enrollment began in October 2002 at 23 US sites (232 participants) and 1 site in Johannesburg, South Africa (11 participants), and ended in January 2005, with the last follow-up occurring in January 2006. Baseline characteristics appeared to be well-balanced between the strategies (Table 1). Figure 1 displays the flow of participants. All participants initiated treatment with their assigned lopinavir/ritonavir–based regimen, and those randomized to mDOT all started mDOT. There was no significant difference in time to premature study discontinuation (P = .59).

A higher proportion of mDOT participants remained on their originally assigned lopinavir/ritonavir dose schedule by week 24 compared with self-administered therapy participants (n = 69 [84%] vs n = 126 [78%], respectively). However, there was no statistically significant difference in time to discontinuation of the originally assigned lopinavir/ritonavir dose schedule (P = .38). By week 48, 48 mDOT participants (73%) and 73 self-administered therapy participants (68%) remained on their assigned lopinavir/ritonavir dose schedule, and the time to discontinuation through week 48 was likewise not significantly different (P = .35).

Sixty-one participants (74%) received mDOT through visit week 24. Adherence to mDOT was high, with 62 of the 82 mDOT participants (76%) attending more than 80% of the intended mDOT visits and only 9 (11%) missing more than 30% of the intended visits. Intended mDOT visits were missed at least once without prior arrangement with the site by 32 participants (39%). Fifty-four of the mDOT participants (68%) considered mDOT to be either somewhat or very helpful, based on the mDOT attitude questionnaire. At the end of the active mDOT phase, the majority of the evaluable participants (n = 53 [83%]) considered mDOT to be either somewhat or very helpful. After comparing attitudes at the end of the active mDOT phase with baseline in the 61 participants with available data at both time points, we found that the attitude toward mDOT improved significantly (P = .004).

Figure 2 displays the virological outcomes. The proportion of participants with virologic success by week 24 was higher in mDOT (7 failures; Kaplan-Meier estimate, 0.91 [95% CI, 0.81-0.95]) than self-administered therapy (23 failures; Kaplan-Meier estimate, 0.84 [95% CI, 0.77-0.89]). However, the difference of 0.07 (1-sided 95% lower confidence bound, −0.01) was not large enough to declare superiority based on an a priori–specified absolute difference of 0.075, nor did the lower 95% confidence bound exclude no difference (P = .56). Thus, mDOT cannot be declared superior to self-administered therapy in this treatment-naïve population.

CD4 lymphocyte count improvements were similarly distributed, with mDOT having a median increase from baseline to visit week 24 of 136 cells/μL (25th-75th percentile, 83-192 cells/μL) and self-administered therapy having a median increase of 130 cells/μL (25th-75th percentile, 60-210 cells/μL), and were not statistically significantly different (P = .46).

In secondary analyses to week 48, the proportion with virologic success was lower for mDOT (20 failures; Kaplan-Meier estimate; 0.72 [95% CI, 0.61-0.81]) than for self-administered therapy (33 failures; Kaplan-Meier estimate, 0.78 [95% CI, 0.70-0.84]). The 95% confidence interval on the difference of −0.18 to 0.07 did not exclude a difference of zero. However, in analysis of time to virologic failure, the hazard ratio for mDOT vs self-administered therapy differed by mDOT study phase, that is, the effect of the mDOT differed between the active mDOT phase and the post-mDOT phase (test for interaction, P = .04). During the active mDOT phase, the hazard ratio of virologic failure for the self-administered therapy was 1.79 (95% CI, 0.77-4.17) times that of mDOT. During the post-mDOT phase, the hazard ratio of virologic failure for self-administered therapy was 0.54 (95% CI, 0.25-1.16) times that of mDOT. Although the treatment strategies cannot be considered different from each other over initiation to week 24 or from week 24 to week 48, the potential effect of mDOT was different over these periods, with the active mDOT phase favoring mDOT and the post-mDOT phase favoring self-administered therapy.

We also compared the rate of developing an AIDS-defining illness or dying between the treatment strategies. By week 24, 1 participant (1%) receiving mDOT developed an AIDS-defining illness or death compared with 11 (7%) receiving the self-administered therapy. Incident events in self-administered therapy included 2 cases of Kaposi sarcoma, 2 cases of esophageal candidiasis, 1 case of wasting syndrome, and 1 case each of cytomegalovirus retinitis, toxoplasmic encephalitis, progressive multifocal leukoencephalopathy, disseminated Mycobacterium avium infection, and pulmonary tuberculosis, as well as 1 death from metastatic small cell carcinoma. The incident event in the mDOT group was cytomegalovirus colitis. The time to new AIDS-defining illness or death was significantly different, favoring mDOT through week 24 (P = .05) but not through week 48 (P = .19).

Adherence was high for both strategies, as assessed by both electronic monitors and self-report. At week 24, for the 80 mDOT participants (98%) with available electronic monitor data, median adherence for all days on study was 93% (25th-75th percentile, 78%-96%), and for the 149 self-administered therapy participants (93%) with available data, median adherence was 92% (25th-75th percentile, 80%-96%) (P > .50). Similarly, at week 48, for the 62 participants (76%) receiving mDOT with available data, median adherence was 92% (25th-75th percentile, 2%-98%), and for the 116 participants (72%) receiving self-administered therapy, median adherence was 91% (25th-75th percentile, 0%-99%) (P > .50). Comparing self-reported adherence at week 24, we found that a larger proportion of mDOT participants reported perfect adherence than did self-administered therapy participants (72% vs 62%), but at week 48, a larger proportion of self-administered therapy participants reported perfect adherence than did mDOT participants (72% vs 58%). However, these differences were not statistically significant (P = .13 and .25, respectively).

Table 2 gives grade 3 or 4 signs, symptoms, or laboratory abnormalities that were 1 grade above baseline and present in more than 2% of participants. There were no statistically significant differences in the time to these events between treatment strategies (log-rank P value, .62 through week 24 and .96 through week 48).

Table 3 gives the proportion of virologic failures with protease and reverse transcriptase mutations at baseline and the proportion with new mutations that were not present at baseline. Among the 58 virologic failures (22 mDOT and 36 self-administered therapy), 56 (97%) had specimens available for genotypic antiretroviral resistance testing. New protease mutations were invariably seen in those with other protease mutations at baseline. New reverse transcriptase mutations were seen in 8 self-administered therapy participants (23%) with virologic failure and 6 mDOT participants (27%) with virologic failure. New reverse mutations invariably included M184V. The emergence of new resistance mutations was neither clinically nor statistically significantly different between treatment strategies.

This large randomized trial assessing the virological effect of directly observed antiretroviral therapy in a uniformly treated HIV-infected antiretroviral-naïve population failed to demonstrate a clear benefit. Although a 7% higher rate of virologic success with mDOT bordered on an a priori threshold of a clinically significant difference of 7.5%, the potential benefit suggested by the results is modest. The observed difference is consistent with a true difference of −1% or higher (based on 95% confidence interval). Interestingly, after mDOT was discontinued, there was some evidence that any potential benefit was not sustained. The virological findings are supported by the adherence results, which likewise did not show significant differences at week 24. While DOT may provide adherence benefits when it is ongoing, our study suggests that good adherence behavior is not entrained by 24 weeks of DOT. This finding makes mDOT administered by health professionals less attractive as an intervention.

There was a suggestion of a protective effect of mDOT on new AIDS-defining events and deaths. This intriguing finding could possibly be explained in part by the lower, though not statistically significant, rate of virologic failure for mDOT over the initial weeks of the study. In addition, perhaps the increased early consistent contact with health professionals conferred an as yet unexplained clinical benefit.¹⁶ Given that the number of events was small, this finding would need to be confirmed in other studies of variations of DOT strategies to conclude that mDOT reduces the incidence of clinical events.

Comparing our findings with other randomized trials testing mDOT interventions may help identify elements of mDOT programs with future promise. Wohl et al¹⁸ studied a mixed treatment-naïve and treatment-experienced population without selection for risk factors for nonadherence in whom community workers administered DOT for 6 months. Similar to our study, they found that DOT did not confer a significant advantage in either virologic suppression or adherence.¹⁸ Macalino et al¹⁹ compared a daily mDOT strategy with self-administered therapy in a randomized trial of 87 active substance users followed up for only 3 months. Unlike our study, they found a marginally significant benefit, although secondary analyses noted that the entire benefit was conferred on experienced patients in whom prior antiretroviral regimens had failed. The clear difference between studies is that ours did not include treatment-experienced patients and suggests that implementing mDOT in an unselected treatment-naïve population may be providing an intervention to many who do not actually need it. Altice et al¹⁷ conducted a trial of DOT in injection drug users and found an advantage with respect to both virologic suppression and CD4 lymphocyte count changes. We had too few injection drug users to make robust conclusions about this subpopulation.

Our study has several strengths compared with other trials of mDOT strategies in HIV. First, the population was entirely antiretroviral therapy–naïve and treated with a uniformly potent antiretroviral regimen. Second, we observed participants both during and after mDOT, allowing us to assess any lasting benefit of the intervention. Third, we measured adherence using an electronic monitor, a technique considered to be the most sensitive for nonadherence and the most accurate.²⁵ However, this study has several potential limitations. First, we studied an experimental formulation of stavudine, which may have selected for a more adherent population. Our findings might have demonstrated a greater benefit in a less-adherent population with potentially more to gain from mDOT. Second, our regimen included soft-gel lopinavir/ritonavir, which has been replaced by tablets. Yet, our findings are unlikely to have been substantially different with newer regimens because of the high rate of adherence and success with self-administered therapy. Third, the intervention did not include a transition phase of tapering of mDOT to allow the participants to develop a set of adherence strategies to replace mDOT. This addition might have improved outcomes for mDOT between weeks 24 and 48. Fourth, the relatively low AIDS-defining illness and death rates and the fact that this comparison was a secondary objective limit our ability to make firm conclusions about clinical benefits of mDOT.

Although mDOT was appealing and accepted by the participants, the small effect and apparent need for sustained mDOT to maintain the benefit should limit enthusiasm for wide implementation. However, our findings do not exclude the possible utility of mDOT in populations with a high potential for nonadherence, such as those with a history of treatment failure. If tested in those populations, it may be important to avoid the abrupt discontinuation of mDOT, allowing instead a transition period for the gradual introduction of self-administered therapy skills. Furthermore, if refined and retested in the future, mDOT strategies should consider incorporating adherence skills into the design to increase the sustainability of the intervention's effect.

Correspondence: Robert Gross, MD, MSCE, Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine, 804 Blockley Hall, 423 Guardian Dr, Philadelphia, PA 19104-6021 (grossr@mail.med.upenn.edu).

Accepted for Publication: March 31, 2009.

Author Contributions: Drs Gross, Tierney, Andrade, Flexner, and Mildvan and Ms Lalama had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Gross, Andrade, Eshleman, Flanigan, Salomon, Reisler, Hogg, Flexner, and Mildvan. Acquisition of data: Tierney, Lalama, Santana, Salomon, Wiggins, Flexner, and Mildvan. Analysis and interpretation of data: Gross, Tierney, Andrade, Lalama, Rosenkranz, Eshleman, Flanigan, Santana, Salomon, Reisler, Flexner, and Mildvan. Drafting of the manuscript: Gross, Tierney, Lalama, Eshleman, Flanigan, Reisler, Flexner and Mildvan. Critical revision of the manuscript for important intellectual content: Gross, Tierney, Andrade, Lalama, Rosenkranz, Flanigan, Santana, Salomon, Reisler, Wiggins, Hogg, Flexner, and Mildvan. Statistical analysis: Gross, Tierney, Lalama, and Rosenkranz. Obtained funding: Flexner and Mildvan. Administrative, technical, and material support: Eshleman, Reisler, Wiggins, Hogg, Flexner, and Mildvan. Study supervision: Andrade, Flanigan, Santana, Flexner, and Mildvan.

Financial Disclosure: Dr Gross has received grant support from Bristol-Myers Squibb and has received grant support from Abbott for research unrelated to this study. Dr Andrade has served on an advisory board for Abbott. Dr Flanigan has received grant support from Abbott and Gilead for research unrelated to this study. Dr Santana has served on advisory boards for Gilead, Tibotec, Roche, Bristol-Myers Squibb, and Merck regarding antiretroviral medications and has received research grants from Gilead, Tibotec, Roche, and Merck. Dr Flexner has received honoraria for talks given at meetings sponsored in part by Abbott Laboratories, the manufacturers of lopinavir/ritonavir, and has served as a paid Scientific Advisory Board member for Bristol Myers-Squibb, the manufacturer of Stavudine. Dr Mildvan has received grant support from Abbott, Bristol-Myers Squibb, Boehringer-Ingelheim, Merck, Pfizer, Roche, Schering Plough, and Tibotec for research unrelated to this study and has received honoraria from Bristol-Myers Squibb, Boehringer-Ingelheim, GlaxoSmithKline, and Tibotec.

Funding/Support: This work was supported by the AIDS Clinical Trials Group via cooperative agreements with the investigators (as listed below) as well as by grants K08 MH01584 and U18 HS016946 (Dr Gross), U01 AI68636 and U01 AI068634 (Dr Tierney and Ms Lalama), U01 AI069472 (Dr Flanigan), and U01AI46370 (Dr Mildvan) from the National Institutes of Health. This study was sponsored by the National Institute of Allergy and Infectious Diseases' Division of AIDS and conducted by the AIDS Clinical Trials Group. Abbott Laboratories, Bristol-Myers Squibb, and Gilead Pharmaceuticals provided the medications and additional funding.

Role of the Sponsor: The industry supporters monitored the development of the protocol and provided input into the design. They also reviewed earlier drafts of the manuscript prior to submission and suggested modifications. The decision to incorporate sponsors' and supporters' suggestions was exclusively the purview of the authors.

Previous Presentation: This work was presented in part at the 14th Conference on Retroviruses and Opportunistic Infections; February 28, 2007; Los Angeles, California.

Additional Contributions: We thank the study participants and the mDOT observers for their contribution and acknowledge the study team members and the site personnel who participated.