|Home | About | Journals | Submit | Contact Us | Français|
Despite high mortality rates in tuberculosis patients with HIV co-infection, there is continued controversy on when to initiate antiretroviral therapy (ART) in these patients.
We conducted an open-label randomized controlled trial in Durban, South Africa to determine optimal timing of ART initiation in relation to TB treatment. Acid-fast bacilli (AFB) smear positive tuberculosis patients with HIV infection and CD4+ counts <500 cells/mm3 (n=642) were randomized to one of two integrated treatment arms (ART initiation during tuberculosis treatment) or to a sequential treatment arm (ART initiation upon tuberculosis treatment completion). Participants received standard tuberculosis therapy, cotrimoxazole prophylaxis and once daily didanosine, lamivudine and efavirenz ART regimen. The primary endpoint was all-cause mortality.
This analysis compares data from the sequential treatment arm and the combined integrated treatment arms up to 1 September 2008, when the Safety Monitoring Committee recommended halting the sequential treatment arm. Demographic, clinical and laboratory characteristics at baseline and adverse event rates during follow-up were similar in the study arms. Mortality was 56% lower (hazard ratio: 0.44; 95% Confidence Interval: 21% to 75%; p = 0.003) in the integrated arm (5.4 per 100 person-years (25 deaths; n=429)) compared to sequential arm (12.1 per 100 person-years (27 deaths; n=213)). Mortality rates were lower regardless of CD4+ count level.
Initiating ART during tuberculosis treatment in AFB positive patients with HIV co-infection and CD4+ counts <500 cells/mm3 significantly improves survival and provides further impetus for the integration of tuberculosis and AIDS services.
In 2007, it was estimated that there were about 33 million HIV infected people1 and 9.2 million new tuberculosis cases globally 2. The two diseases are closely intertwined and the number of co-infected patients continues to grow rapidly3. Tuberculosis is the most common presenting opportunistic disease4 and cause of mortality in AIDS patients in developing countries5. Notwithstanding effective tuberculosis chemotherapy, in the presence of HIV, tuberculosis is associated with substantially higher case fatality rates6 and is also the most common notified cause of death in South Africa7. In 2007, South Africa had an estimated 5.3 million people infected with HIV and 341,165 reported tuberculosis cases, of whom approximately 73% were HIV co-infected8.
The optimal timing for initiation of antiretroviral therapy (ART) in HIV-tuberculosis co-infected patients remains unclear. Current guidelines are based on observational studies and expert opinion9. Despite World Health Organisation (WHO) guidelines supporting concomitant treatment of both diseases and urging more aggressive management10, ART initiation is often deferred until completion of TB treatment because of concerns related to potential drug interactions between rifampicin and some classes of antiretrovirals11, immune reconstitution inflammatory syndrome (IRIS)12, 13, overlapping side effects14, high pill burden and programmatic challenges15.
The SAPiT (Starting Antiretroviral therapy at three Points in Tuberculosis) trial was designed to determine the optimal time to initiate ART in HIV-tuberculosis co-infected patients receiving tuberculosis treatment.
The SAPiT trial (protocol number: CAPRISA 003) was conducted at the CAPRISA eThekwini HIV-tuberculosis clinic. This adjoins one of the largest ambulatory (out-patient) tuberculosis facilities in South Africa, the Prince Cyril Zulu Communicable Disease Centre (PCZCDC) in Durban.
The South African National Tuberculosis Control Programme guidelines16 stipulatethat first episode tuberculosis be treated with a fixed drug combination of rifampicin, isoniazid, ethambutol and pyrazinamide dosed according to pre-treatment weight for a 2 month intensive phase. Thereafter, the 4-month continuation phase comprises a fixed-drug combination of isoniazid and rifampicin. Patients with a past history of tuberculosis receive a 60 day intensive phase which includes streptomycin, followed by a 100 day continuation phase. Patients were routinely offered clinic-based directly observed therapy by clinic-based nurses. Some patients selected community-based supervisors, heads of households, and treatment supporters in workplaces who supervised and recorded the taking of medication.
From 28 June 2005 to 11 July 2008, HIV-infected (by two rapid HIV tests) and AFB smear positive (by auramine and Ziel-Niehlsen stain) patients, who were 18 years or older were recruited. To be included, patients had to be independently confirmed at the Department of Medical Microbioloy, Nelson R Mandela School of Medicine, to have AFB smear positive disease, initiated on the standard tuberculosis treatment regimen at the PCZCDC, have a CD4+ count <500 cells/mm3 at screening and have no clinical contraindications to initiation of ART. Female participants were required to agree to use contraception while on efavirenz.
The SAPiT study is a three arm, open-label, randomized controlled trial. After providing written informed consent, qualifying HIV-tuberculosis co-infected patients were randomized to one of three treatment arms in a 1:1:1 ratio in permuted blocks of six or nine with no stratification using sealed envelopes; Arm 1 - ART to be initiated within 4 weeks of starting tuberculosis treatment (early integrated treatment arm), Arm 2 - ART to be initiated within 4 weeks of completing the intensive phase of tuberculosis treatment (late integrated treatment arm) and Arm 3 - ART to be initiated within 4 weeks after completing tuberculosis treatment (sequential treatment arm).
All participants received adherence counselling, cotrimoxazole prophylaxis, and the same once daily ART regimen; didanosine (250mg if weight <60kg and 400mg if weight >60kg), lamivudine (300mg) and efavirenz (600mg). ART adherence was assessed monthly by pill counts (pills issued minus pills returned as a percentage of anticipated pill consumption). Regardless of trial arm assignment, participants could be initiated on ART at any time by PCZCDC clinicians, study clinicians or their personal physicians at their discretion.
Follow-up visits for safety and clinical status monitoring were scheduled monthly for 24 months. Adverse events were graded using the Division of AIDS (NIAID/NIH) Table for Grading Adult and Pediatric Adverse Events, version 1.0, 28 December 2004. CD4+ count using the FACS Calibur flow cytometer (Becton Dickinson, Franklin Lakes NJ, USA), HIV RNA (Roche Cobas Amplicor HIV-1 Monitor v1.5) were performed at screening, randomization and 6 monthly thereafter, while radiological changes and sputum conversion were monitored at screening, end of the intensive phase of tuberculosis treatment, one month before the end of tuberculosis treatment and whenever clinically indicated. Secondary endpoints included tolerability, toxicity, HIV RNA, tuberculosis outcomes and IRIS. IRIS was defined as a paradoxical deterioration in clinical status after ART initiation without another attributable cause.
Following a review on 1 September 2008, almost two months after completion of enrolment, the Safety Monitoring Committee recommended during a planned interim analysis of the data that all participants in the sequential treatment arm be initiated on ART as soon as possible but continue in follow-up to study completion. The Committee also recommended continuation of the two integrated treatment arms with no changes. The participants in the sequential arm were contacted within a week of the Committee’s meeting and almost all were initiated on ART within a month. We present data up to 1 September 2008, comparing the sequential treatment arm to the combined early and late integrated treatment arms, hereinafter referred to as the integrated treatment arm.
The target sample size was 649, including anticipated loss to follow–up, based on 80% power and an alpha of 0.05 to detect a 60% reduction in mortality with a predicted mortality rate of 10% in the worst arm. All analyses were by intention-to-treat. The primary outcome was analyzed using Kaplan-Meier curves and the log-rank test. The time on study was calculated as the time from randomization to death, termination from the trial or 1 September 2008, whichever occurred first. Poisson approximations were used for mortality rate confidence intervals. Proportional hazards regression models were used to adjust for confounders. Fisher’s exact test was used for categorical data and unpaired t-tests or the Wilcoxon two-sample test for continuous data.
The trial (NCT00398996) was approved by the University of KwaZulu-Natal’s Biomedical Research Ethics Committee (E 107/05) and the Medicines Control Council (MCC ref: 20060157).
A total of 642 HIV-tuberculosis co-infected patients were enrolled; 429 in the integrated treatment arms and 213 in the sequential treatment arm (Fig 1). At baseline, participants in integrated and sequential treatment study arms were comparable in terms of age, CD4+ count and HIV RNA levels (Table 1).
On 1 September 2008, 338 (52.6%) participants were still in active follow-up, 52 (8.1%) had died during follow-up, 134 (20.9%) had completed follow-up, and 56 (8.7%) had been terminated prior to study completion. Of the 62 (9.7%) participants regarded as lost to follow-up (9.8% in the integrated and 9.9% in the sequential treatment arm); 35 were known to be alive while the clinical status of remaining 27 (12 in the sequential treatment arm) was unknown. The median duration of follow-up in the trial was 12.1 months (interquartile range: 6.1–21.6 months).
The median duration of tuberculosis treatment was similar in participants who completed tuberculosis treatment, 210 days (n=271) in the integrated treatment arm and 207 days (n=137) in the sequential arm. At the time of analysis 102 participants in the integrated treatment arm and 48 participants in the sequential treatment arm were still on tuberculosis treatment.
Of the 350 integrated treatment arm participants who initiated ART, 338 did so during tuberculosis treatment. On average, participants in the integrated treatment arm started ART 70 days (standard deviation (SD):72.0) after the start of tuberculosis treatment. Of the 100 sequential treatment arm participants who initiated ART, 7 did so during tuberculosis treatment. ART was initiated on average 260 days (SD: 70.9) after tuberculosis treatment initiation. Thus participants in the sequential treatment arm initiated ART, on average, 190 days later than participants in the integrated treatment arm.
There were 25 deaths in the integrated treatment arm with a mortality rate of 5.4 per 100 person-years compared to 27 deaths in the sequential treatment arm with a mortality rate of 12.1 per 100 person-years (HR=0.44, p=0.003) (Fig 2, Table 2). After adjusting for baseline WHO status (Stage 4 vs 3), CD4+ count, age, sex, past history of tuberculosis, extra-pulmonary tuberculosis and baseline HIV RNA, the hazard ratio (integrated versus sequential) was 0.43 (95% CI: 0.25 to 0.77, p=0.004).
Information on death was based on hospital chart notes (n=28), death certificates (n=1) or two independent verbal reports of death (n=23). Based on chart notes and death certificate (n=29), the causes of death in the integrated treatment arm were: tuberculosis (including meningitis), 2; respiratory distress or Pneumocystis jirovecii pneumonia, 6; and one each of metabolic acidosis, cardiomyopathy and motor vehicle accident. Those in the sequential treatment arm were: tuberculosis (including meningitis), 6; respiratory distress or Pneumocystis jirovecii pneumonia, 3; and one each of: meningitis, gastroenteritis, renal failure, hepatic failure and glioma. The cause of death was unclear in the chart notes of four participants.
Baseline CD4+ count independently predicted mortality in both treatment arms. Mortality was lower in the integrated treatment arm in all CD4+ count strata (Table 2). Baseline CD4+ count was similar in both arms. There was no interaction between CD4+ count and treatment arm (p=0.57).
ART adherence by pill count was 97.2% and 97.6% in the integrated and sequential treatment arms, respectively. Tuberculosis treatment outcomes were similar in both study arms for first episode and retreatment cases (Table 3). The proportion with suppressed HIV RNA level (Table 3) was higher in the integrated compared to sequential treatment arm at 12 months post-enrolment (90.0% vs 77.8%;, respectively p=0.006). However, proportion with suppressed HIV RNA 6 months after ART initiation was similar in both arms (Table 3).
IRIS was experienced by 12.4% (53/429, 95% CI, 9.5 to 15.9) participants in the integrated treatment arm and 3.8% (8/213, 95% CI, 1.8 to 7.5) in the sequential treatment arm (p<0.001). None of the deaths were assessed to be IRIS-related. There were 140 (30 per 100 person-years) and 71 (32 per 100 person-years) non-IRIS grade 3 and 4 adverse events reported in the integrated and sequential treatment arms respectively (p=0.69).
The SAPiT trial showed that initiation of ART during tuberculosis treatment in patients with sputum AFB smear positive tuberculosis and HIV-infection with CD4+ counts <500 cells/mm3 reduced mortality by 56% (95% CI, 21% to 75%). The mortality rate rose from 5.4 per 100 person-years to 12.1 per 100 person-years when initiation of ART was delayed until completion of tuberculosis treatment. The period between completion of tuberculosis treatment and ART initiation is important; a considerable number of the deaths in the sequential arm occurred during this time (Fig 2). Once initiated, however, ART was associated with comparable high levels of viral suppression in both treatment arms, which are similar to those observed in other AIDS treatment programmes in South Africa17.
Mortality in HIV-tuberculosis co-infected patients is known to be high despite effective tuberculosis treatment5. Observational studies among HIV-tuberculosis co-infected patients, have indicated that ART initiation during tuberculosis treatment improves treatment outcomes. A meta-analysis of 6,934 patients from five hospitals in Madrid, showed a significant 63% improvement in survival in patients initiated on ART during tuberculosis therapy18. In Thailand, an analysis of 1003 patients demonstrated a 20-fold higher mortality rate in patients not receiving ART with tuberculosis treatment.19 Another Thai review of 626 patients found a HR of 0.17 for death in patients who commenced ART during tuberculosis treatment20. Although these studies show a consistent association between ART and survival in co-infected patients, the SAPiT trial is the first randomized controlled trial to corroborate these retrospective observational data.
Mortality was 46% lower (p=0.04) in the integrated treatment arm participants with CD4+ counts <200 cells/mm3. While the number of deaths is small in the sub-group with CD4+ counts between 200 and 500, the trend of lower mortality in the integrated treatment arm was present. This finding has treatment guideline and policy implications. Current WHO guidelines on treatment of HIV-tuberculosis co-infection recommends deferment of ART initiation until completion of tuberculosis treatment in WHO stage 3 patients with CD4+ counts >200 cells/mm3 10. Our findings suggest that this guideline be expanded to include co-treatment of HIV and tuberculosis in patients with CD4+ counts <500 cells/mm3.
There is increasing evidence to suggest that earlier ART initiation in HIV-infected patients even without tuberculosis is associated with improved outcomes21–23. In an analysis of 8,362 asymptomatic patients with HIV infection in the United States and Canada22, mortality was 69% lower in patients who initiated ART at CD4+ counts between 350 and 500 cells/mm3. Similarly, data from 18 prospective cohort studies have shown that deferring ART was associated with higher rates of AIDS and death than starting therapy when the CD4+ count was >350 cells/mm3 23.
However, major concerns regarding the early initiation of ART during tuberculosis treatment are the potential mitigating effects of the increased risk of IRIS, additive toxicities and potential adverse effect on tuberculosis treatment outcomes. The study found similar rates of grade 3 and 4 adverse events in the two treatment arms and similar tuberculosis outcomes. Since many of the deaths occurred after tuberculosis treatment completion, the tuberculosis treatment providers were unaware of the benefits of tuberculosis and HIV co-treatment. Although the IRIS incidence rate was significantly higher in the integrated treatment arm, this was not unexpected. IRIS has been associated with earlier ART initiation in co-infected patients24, 25. The IRIS incidence rate in the integrated treatment arm was similar to that observed in other developing country cohorts24, 26 including a retrospective analysis of hospitalised Thai patients receiving both ART and tuberculosis treatment that found that 12.6% (21/167) experienced an IRIS event26. However, none of the deaths in the SAPiT trial, where information on cause of death was available, were considered attributable to IRIS. It is reassuring that recent studies of tuberculosis-associated IRIS indicate that this complication is rarely fatal and that severe episodes can be successfully managed with prednisone27. Thus, the concern about increasing the likelihood of IRIS must be tempered by the survival benefit shown in our study. Nevertheless, the paradoxical deterioration in the clinical status is sufficiently common to warrant close clinical monitoring in the first few months of ART initiation in patients co-infected with tuberculosis.
We acknowledge several limitations of our study. Using all-cause mortality as the primary endpoint might underestimate the potential impact of integrated HIV-tuberculosis treatment on tuberculosis and HIV-specific mortality. Since we were not able to obtain reliable information on the causes of all deaths in the trial, we were not able to estimate the impact on deaths related only to tuberculosis and HIV. As our trial only included patients who had AFB smear positive disease and were diagnosed and treated in an ambulatory tuberculosis clinic setting, results may not be directly generalizable to all forms and severity of tuberculosis. Since a retrospective analysis of 549 AIDS patients with extra-pulmonary tuberculosis showed that the introduction of HAART significantly improved survival28, early ART initiation may have similar benefits in extra-pulmonary tuberculosis patients. While we have no reason to believe that our findings do not apply to sputum smear negative tuberculosis this needs empiric confirmation. Another limitation was the delay in ART initiation after tuberculosis treatment completion in the sequential treatment arm due to clinical grounds (often due to raised liver enzymes) or missed visits. Further, the question of when during the course of tuberculosis treatment, ART should be initiated awaits completion of the study.
In summary, the findings from this study provide compelling evidence of the benefit of initiation of ART during tuberculosis treatment in patients with HIV disease. They support recommendations by the WHO and others for the integration of tuberculosis and HIV care. Health care providers within countries and funders should consider monitoring the proportion of HIV-tuberculosis co-infected patients who are initiated on ART. Such a metric could serve as an impetus for HIV-tuberculosis integration and serve as a key indicator of the success of such programs. This would require that all newly diagnosed tuberculosis patients be offered an HIV test; those who are HIV positive be offered a CD4+ count and those with a CD4+ count <500 cells/mm3 be initiated on ART during tuberculosis treatment. If implemented in South Africa alone, it is estimated that such a strategy could lead to approximately 150,000 more HIV-tuberculosis co-infected patients initiating ART during tuberculosis treatment annually and an estimated 10,000 deaths averted each year.
CAPRISA was established as part of the Comprehensive International Program of Research on AIDS (CIPRA) (grant # AI51794) from the US National Institutes of Health. The US President’s Emergency Plan for AIDS Relief (PEPfAR) funded the care of all the participants in the trial. The Global Fund to fight AIDS, Tuberculosis and Malaria funded the cost of the drugs used in the trial. The research infrastructure to conduct this trial, including the data management, laboratory and pharmacy cores were established through the CIPRA grant. We gratefully acknowledge the participants in the study.
The study was designed by SSAK, WES, GF, AGray, and QAK. Data was gathered by, KN, NP, MK, SB, GN, JP, TG and AS. Data was analysed by AG. SSAK, KN, NP, and AG vouch for data and analysis. KN, SSAK, CB and AG wrote the paper and SSAK decided to publish the paper. We also thank Professor Willem Sturm of the Nelson R Mandela School of Medicine for the confirmatory tuberculosis testing; Dr Surie Chinnapa and Sr Jeanne Liebertrau of the Prince Cyril Zulu Communicable Disease Centre; Dr Gray Handley, Dr Ed Tramont, Dr Rod Hoff, Dr Sandi Lerhman and Dr Richard Hafner of the Division of AIDS at the NIH; Dr Gavin Churchyard, Dr Douglas Taylor, and Dr Mark Weaver for serving on the SAPiT Safety Monitoring Committee; Ms Anushka Naidoo, the on-site study pharmacist; members of the Community Advisory Board; Ms Nomapando Barnabas of the CAPRISA Community Programme; Ms Natasha Samsunder and Mr Keith Coetzee for laboratory analysis; Ms Nonhlanhla Yende for statistical support; Ms Irene van Middelkoop for data management and all the members of the CAPRISA 003–SAPiT trial team.
The authors of this manuscript have no financial interests or conflicts of interest to disclose. All staff employed by CAPRISA are required to sign a confidentiality agreement when they join the organisation.
An overview of the results were presented the Conference on Retroviruses and Opportunistic Infections Montreal, February 2009