When to start ART in patients co-infected with TB: results from two trials presented at CROI

Nathan Geffen, Community Media Trust

The complex question of the optimal time to start antiretroviral therapy (ART) in HIV-positive patients co-infected with TB was the subject of two important presentations at a plenary session of the 18th Conference on Retroviruses and Opportunistic Infections. Diane Havlir presented the findings of ACTG’s Stride trial and Salim Abdool Karim presented the results of the SAPIT trial. [1, 2]


Havlir and colleagues conducted an open label international trial to confirm their hypothesis that in patients starting TB treatment, ART initiated within 2 weeks (immediate initiation) could reduce mortality and morbidity compared to patients starting ART within 8-12 weeks (early initiation).

Patients were randomised to immediate (405 patients) and early (401 patients) ART treatment arms. Patients had confirmed or presumed TB and a CD4 cell count <250 cells/mm3. The ART regimen for 97% of patients was efavirenz and tenofovir/emtracitabine. The TB treatment was country approved. Nearly half of the patients had confirmed TB with a median CD4 count of 77 cells/mm3 (IQR: 36-145).

The primary endpoints of the study were all cause–mortality and new AIDS-defining illnesses by 48 weeks. No data is available for after 48 weeks as patients were not followed beyond this endpoint. Secondary endpoints were safety, CD4, HIV RNA changes, TB IRIS and TB outcomes.

The proportion of AIDS or death between the immediate (12.9%) and the early (16.1%) arms were not statistically significant (p=0.45). However, a pre-specified analysis that considered patients with a CD4 cell count <50 cells/mm3 found that the proportion of patients with AIDS or death was 26.6% in the early arm compared to 15.5% in the immediate arm and this was significant (p=0.02). The majority of AIDS or death events occurred within the first 24 weeks after randomisation, with the highest proportion amongst the early arm with a CD4 count <50 cells/mm3. There was barely any difference in endpoints in the higher CD4 strata (11.5% versus 10.3% for the immediate versus early arms respectively; p=0.67).

Thirteen cases of cryptococcal disease made it the most common AIDS primary endpoint (n=63) The next two most common AIDS illness included oesophageal candidiasis (n=12) and Kaposi’s Sarcoma (n=11).

There were 31 deaths in the immediate arm versus 37 deaths in the early arm. TB was the largest contributor to deaths (21 out of 68 patients). During question time, Havlir was asked if the 14 TB-related deaths on the immediate arm versus seven in the early arm were indicative of fatal TB IRIS. She pointed out that these cases were reviewed and that they were a consequence of TB progressing in contrast to patients with IRIS who get better and then become ill again.

Twenty-one cases of AIDS-related deaths were reported and 16 non-AIDS related deaths (respiratory, renal and hepatic disease being most common). The frequency of TB IRIS in the immediate arm was 11% and 5% in the early arm (p=0.002).

In summary the investigators found that immediate ART did not overall reduce AIDS-defining disease overall and death compared to early ART, but for patients with CD4 counts < 50 cells/mm3 immediate ART reduced AIDS and mortality. Grade 3 or 4 toxicities, HIV RNA suppression rates or CD4 increase did not differ between the arms. TB IRIS was higher in the Immediate arm although it did not increase mortality. They concluded that in patients with CD4 counts <50 cells/mm3 ART should be started within two weeks.


Salim Abdool Karim and colleagues conducted a 3-armed open label trial called SAPiT. The sequential arm, in which patients first completed their TB treatment course and then initiated ART was stopped by the DSMB due to the significantly higher mortality in that arm. We have previously reported on this aspect of the trial. [3] This report is confined to the results of the remaining two arms.

HIV-positive patients with smear-positive TB and CD4 counts <500 cells/mm3 were randomised into an early integrated therapy arm (214 enrolled and ART initiated within four weeks of starting TB treatment) and a late integrated therapy one (215 enrolled and ART initiated within four weeks of completing the intensive phase of TB treatment). Baseline characteristics for age, gender and CD4 count were similar in both arms.

All participants attended the TB-DOTS programme at eThekwini Clinic in Durban and the study’s primary endpoints were death and AIDS defining illness.

Both arms had similar rates of AIDS defining illness or death with 18 deaths in the early arm and 19 in the late arm. The Incidence Rate Ratio (IRR) was 0.89 (95% CI: 0.44 to 1.79; p=0.73).

When the results were stratified for CD4 count of <50 cells/mm3, a 68% reduction of AIDS or death was found in the early and this approached significance (IRR: 0.32 [0.07-1.13], p=0.06). For participants with CD4 counts >50 cells/mm3 no discernable differences in AIDS/Death were noted (IRR: 1.51 [0.61-3.95], p=0.34).

In patients with CD4 counts <50 cells/mm3, the reduction in AIDS/death in the early arm overshadowed the 5-fold higher risk of IRIS (95% CI; IRR 4.7 [1.5-19.6]; p=0.01) and the increasing trend in drug switches.

HIV suppression was greater than 90% after 18 months irrespective of CD4 status. Similarly TB treatment was successfully completed in about 80% of patients with no significant differences across groups.

On the other hand in patients with CD4 counts > 50 cells/mm3 there was an IRIS rate of 15.8 person years and 7.2 in the late arm; CI: 95%; IRR: 2.2 [1.1-4.5]; p=0.02) and this was significant. Rates of drug switches were 7/100 patient years in the early arm and 1 in the late arm (CI: 95%; IRR: 6.8 [0.8-551.1]; p=0.04). The lower rates of IRIS and drug switches in the late therapy arm indicated a slight benefit to starting ART during the continuation phase of TB treatment in patients with CD4 counts >50 cells/mm3.


The main results of both studies were similar. Immediate ART is warranted in patients with CD4 counts <50 cells/mm3 while ART for patients with CD4 counts >=50 cells/mm3 can be postponed until the continuous phase of TB treatment, but not beyond.

These studies have implications for some guidelines in high incidence countries. For example, he South African Guidelines for Antiretroviral Therapy in Adults state with respect to patients co-infected with TB state:

“[For patients with] CD4 count <200 cells/mm3: commence ART after it is clear that the patient’s TB symptoms are improving and that TB therapy is tolerated. The suggested time period to commence ART is between 2 and 8 weeks after starting TB therapy.

CD4 count 200 – 350 cells/mm3: delay ART until after the intensive phase of TB therapy (2 months) unless the patient has other serious HIV-related illness. The longer delay before commencing ART in this group is recommended to reduce the risk of shared toxicity (as the patient will then only be on fewer TB drugs) and to reduce the risk of the immune reconstitution inflammatory syndrome (see below).

CD4 count >350 cells/mm3: defer ART.”

When the guidelines are updated, the drafters will have to consider:

  • Whether the recommendations be modified to explicitly state that patients with CD4 counts <50 cells/mm3 must be started immediately on ART, or does the current framing sufficiently cover that?
  • Whether the recommendations be modified to initiate ART to patients with CD4 counts of 50-350 cells/mm3 only after the intensive phase?

The answers to these questions are not clear.

While the SAPIT trial included patients with CD4 counts <500 cells/mm3, there is not yet enough data on patients with CD4 counts of 350 to 500 cells/mm3 to justify a change in guidelines. Hopefully the START and TEMPRANO trials, scheduled to complete in 2015 and 2013 respectively, will help answer this question. [5, 6]


  1. Havlir D et al. 2011. International randomized trial of immediate vs early ART in HIV+ patients treated for TB: ACTG 5221 STRIDE study. 18th Conference on Retroviruses and Opportunistic Infections, 27 Feb to 2 March, Boston.
  2. Abdool Karim S et al. 2011. Optimal timing of ART during TB therapy: findings of the SAPiT trial. 18th Conference on Retroviruses and Opportunistic Infections, 27 Feb to 2 March, Boston.
  3. Geffen N. 2009. HIV and TB from CROI. HTB South. April 2009.
  4. Southern African HIV Clinicians Society and Department of Health. 2008. Guidelines for antiretroviral treatment in adults.
  5. INSIGHT. Strategic Timing of Antiretroviral Treatment (START).
  6. ANRS. Early antiretroviral treatment and/or early isoniazid prophylaxis against tuberculosis in HIV-infected adults (ANRS 12136 TEMPRANO).

Links to other websites are current at date of posting but not maintained.