This regular feature on HIV and TB integration is kindly supported by the Stop TB department of the World Health Organization.
Starting ART two weeks after TB treatment reduces death risk by one-third in people with HIV
Starting antiretroviral therapy (ART) in people with TB and advanced HIV disease two weeks after starting TB therapy, rather than waiting eight weeks, significantly improves survival according to the results of the CAMbodian Early vs Late Introduction of Antiretroviral drugs (or CAMELIA) study, presented as a late breaker at the Eighteenth International AIDS Conference in Vienna.
The study randomised 661 participants with HIV and smear-positive TB to either an early-start arm or one starting HIV treatment after eight weeks. By the study’s conclusion, 59 patients had died in the early-start arm, compared to 90 in the late-start arm – a 34% difference. “The difference in mortality rate was highly significant,” said Dr Francois Xavier Blanc, who is lead investigator for the study.
At the same time, however, immune reconstitution inflammatory syndrome (IRIS) – which is frequently cited by clinicians as a reason to delay starting ART – was more common among people HIV treatment early (and was even associated with five deaths, compared to one IRIS-associated death in the delayed treatment arm). However, this did not offset the survival benefit provided by earlier ART.
Notably, the survival benefit emerged gradually over time and only became significant well after ART had begun (ART appeared to be effective in both groups, with no significant differences between virological or CD4 response).
The study was conducted in people with very advanced HIV – the median CD4 cell count was 25mm3 at the time of study inclusion. Although it is not clear that early treatment would make as significant a difference in people with less advanced HIV, the study adds significant clinical evidence to the 2010 WHO recommendations to begin ART in TB patients “as soon as possible” after starting TB treatment – and will intensify calls for integration of TB/HIV services in countries with a high burden of co-infection.
Because of the potential importance of this study, after the publication of the original version of this report on aidsmap, we queried HATIP’s advisory panel and other experts to see what they think about the study findings, and what the consequences may be for clinical practice. Their responses are incorporated into this version of the article.
In many settings, TB clinicians have wanted to delay ART until after TB treatment, or at least until after the intensive phase of treatment – which generally ends at eight weeks – because of concerns about high pill burdens, overlapping side-effects, drug interactions and IRIS. However, during this time, HIV progression may continue, perhaps even at an accelerated rate because of the TB infection, making it both more difficult to fight off TB and to ward off other HIV-related conditions.
A large proportion of people with HIV and TB die despite receiving standard TB treatment.
However, a growing body of data suggests that starting ART while still on TB treatment could improve outcomes.
For instance, almost two years ago, preliminary results of the SAPIT study found twice the number of deaths in participants in whom ART was delayed until after TB treatment was finished, compared to patients who got ART while still on TB treatment, either within the first four weeks of TB treatment, or within four weeks of concluding intensive TB treatment. The optimal time to start ART in people who have begun treatment for active TB remained unclear, however, so that study is continuing to compare those two 'integrated therapy' arms.
That study validated 2003 WHO guidelines, which recommended that ART be started in people with HIV who have begun TB treatment as soon as it is clear that TB treatment is tolerated (between two weeks and two months). The most recent version of the guidelines however, change the wording slightly: strongly recommending that TB treatment be started first, followed by ART as soon as possible afterwards (and within the first eight weeks). The guidelines note, however, that there was only evidence of "moderate quality" to support this recommendation.
The CAMELIA study
The CAMELIA study was conducted at five study sites in Cambodia by the French organisation Agence Nationale de Recherche sur la Sida et les hépatites virales (ANRS), and the US National Institute of Allergy and Infectious Diseases (the CIPRA trial), in partnership with the Cambodian Health Committee. It was designed as a superiority trial to determine the best time to start ART in people with advanced HIV disease (CD4 cell counts below 200) who had started TB treatment. Again, its two arms compared the introduction of later ART (eight weeks) vs early (two weeks). The primary endpoint was survival at the end of the trial (when the last participant completed 50 weeks of ART), with an intent-to-treat analysis.
All the participants received daily treatment with the standard TB regimen, with two months of intensive treatment (four drugs) followed by a continuation phase with rifampicin/isoniazid, and the same ART regimen (d4T/3TC/efavirenz). The first participant enrolled in January 2006 and the last one in May 2009, so the study has just been completed. The median duration of follow-up in the trial was 27 months.
The study enrolled 661 subjects, 332 randomised to early ART, 322 to late. Even though all subjects had smear-positive TB, some were not culture-positive. In fact, there were 38 subjects who were culture-negative, plus 12 with Nontuberculous mycobacteria (NTM) in the early treatment arm – compared to 31 culture-negative and four with NTM in the late arm.
Patient characteristics at baseline were similar. However, “these patients were extremely sick”, said Dr Blanc. The median CD4 cell count at inclusion was 25 cells/mm3, the median viral load was 5.6 log and the body mass index (BMI) was below 17, indicating severe wasting.
About 90% of the subjects had pulmonary disease. MDR-TB was rare, found in about 2% of the participants, and there was no difference between the arms. However, drug sensitivity tests were unavailable for about 68 patients.
After a follow-up time of 712.4 person-years, there were 59 deaths in the early arm, a mortality rate of 8.28 (95% confidence interval (CI) 6.42 to 10.69). In the late arm, after 653.7 days follow-up time, there were 90 deaths for a mortality rate of 13.77 (95% CI 11.20 to 16.93). There was a significant reduction in mortality in the early arm (p= 0.02). Only 12 patients (1.8%) were lost to follow-up, and fewer than 2% missed clinic visits.
Kaplan-Meier curves showed the survival over time. As noted earlier, the difference in survival between the two arms became more marked with time. At week 50, the survival probability was 86.1% (95% CI 81.8 to 89.4) in the early arm, vs 80.7% (95% CI 76.0 to 84.6) in the late arm (p= 0.07).
At week 100, the survival probability in the early arm was 82.6% (95% CI 78 to 86.4), but in the late arm it had dropped to 73% (95% CI 67.7 to 77.6) (p = 0.006). At week 150, survival was more or less stable in the early arm, but it continued to decline in the late arm.
In a multivariate analysis, several factors (apart from being assigned to the late arm) were associated with an increased risk of mortality, including BMI, low Karnosky scores, having both pulmonary and extra-pulmonary disease, and having an infection with a non-tuberculous mycobacterium (these do not necessarily respond to TB treatment – but notably, there were more of these in the early treatment arm).
Not surprisingly, having MDR-TB was highly associated with mortality with an adjusted hazard ratio of 8.02 (4.00 to 16.07), p <0.001.
“We also observed that IRIS was much more frequent in the early arm, nearly 2.5-fold in the early arm versus the late arm, but most of the time it was easy to manage. No patients received corticosteroids,” said Dr Blanc, adding that IRIS tended to emerge two to three weeks after ART started (in both arms). “We need to be prepared to face that and manage that,” he said.
Although there were six deaths associated with IRIS in the trial, Dr Stephen Lawn of the Desmond Tutu HIV Centre told HATIP, “I think it can be a tough call to distinguish deaths from TB IRIS vs deaths with TB IRIS...and it is important to note that those with the highest risk of TB IRIS are the ones with the highest pre-existing mortality risk. The number of TB IRIS deaths reported in the early arm (approx 4.5%) is consistent with the systematic analysis by Muller et al in Lancet Infect Dis 2010” (see references).
The response to ART was quite impressive in both arms. At week 50, more than 95% had undetectable viral loads and this has remained constant over the course of the study (this however, is an on-treatment analysis). Likewise, CD4 cells increased by a median of 114 cells over baseline, and have continued rising for participants with longer follow-up. There was a non-significant difference in median CD4 cell counts after week 102 (median 230 on early ART and 201 on late ART at week 150). Note, given that the median CD4 cell count at entry was 25, many patients were still at risk for a number of HIV-related complications for quite some time despite being on treatment, and would continue to require cotrimoxazole prophylaxis.
Another point Dr Blanc stressed was that “These patients were extremely adherent, despite their very poor condition at enrollment.”
“We can speculate that initiating ART two weeks after the onset of TB treatment could potentially save 150,000 of the 450,000 annual HIV-TB related deaths,” Dr Blanc said in conclusion.
It’s not really clear that his calculation (a one-third reduction in death) was appropriately applied here, as many of the annual TB deaths in people with HIV occur in people not receiving treatment at all. But if everyone with advanced HIV disease received appropriate and timely diagnosis and received TB treatment followed by ART two weeks afterwards, the survival benefit could potentially be expected to be even greater. However, this would require far better integration of TB/HIV services than exists presently.
The study has only recently completed, so the analysis should be treated as preliminary, and some questions still remain unanswered. For example, was there a faster time to TB culture conversion in people on early ART?
“From the point of view of TB control programmes, they will be interested to know whether early ART improves TB-specific treatment outcomes, ie. 6 months after starting anti-TB treatment is the treatment success rate increased and the case fatality lowered,” said Professor Anthony Harries, Senior Adviser to International Union Against Tuberculosis and Lung Disease.
Another study also presented at AIDS2010 could have bearing on this question. Dr Gabriel Chamie reported on the microbiologic, radiological and clinical treatment outcomes in 223 coinfected Ugandans with CD4 cells above 350 who were randomised to start ART 2-4 weeks after starting TB treatment versus those who only were put on ART if their CD4 cells fell to below 250. To cut a long story short - there were no TB treatment failures and no differences in culture conversion or other responses per arm. Dr Chamie concluded that six months on ART may have been too short a time to see differences in immunological responses to m.TB. But if people whose immune systems are somewhat more intact can’t mount these responses within six months, it might be even more difficult to get a rapid TB-specific immune response people with a baseline CD4 cell counts averaging around 25.
If ART doesn’t dramatically improve TB-specific treatment outcomes, the survival benefit could be more associated with treating HIV. However, it is curious that in CAMELIA, simply starting ART a mere six weeks earlier produced a survival benefit that doesn’t seem to be explained by differences in virological or CD4 cell responses between the arms.
CAMELIA’s results are in contrast to a study by Tavuka et al, which was presented at the 2nd South African TB Conference and included over 1600 subjects and found no significant difference in the risk of death if antiretroviral (ARV) treatment was initiated within two months of beginning TB treatment or more than two months after beginning TB treatment. About 10% of the subjects died in each arm. Notably, this study involved people with higher CD4 cell counts and follow-up was only until 12 months after initiating TB treatment.
Part of what is most interesting about the CAMELIA study, however, is that ART didn’t so much prevent more early deaths as cumulative deaths over time. From the Kaplan-Meier curves, while differences in survival emerge gradually after the first few months of TB treatment, in the late ART arm, it appears as though about half of the deaths occur after the time TB therapy would have been completed, and about a third of the deaths occurred more than 50 weeks after starting TB treatment. The difference between the two arms only becomes significant at some point after 50 weeks.
“In SAPIT the mortality difference also opened out greatly after the completion of TB treatment,” Mark Harrington of the Treatment Action Group told HATIP. In fact, the survival difference in that study starts to become more marked 10 to 12 months after TB treatment was initiated.
“It would be good to stratify death in those on early ART and later ART by 2-months of TB treatment, 6-months of TB treatment (ie at the end), 12-months – the Kaplan Meier will show it but it may also be useful to look at these set points in time,” said Professor Harries.
Another question is whether there were other clinical changes on ART that might have been prognostic of better outcomes, such as, was there less weight loss in those on early ART? One thing that both Tavuka et al and the CAMELIA study noted was that low BMI was highly associated with poor survival. Prof Harries has observed this as well.
“Our studies in Malawi showed that low BMI in TB patients was strongly associated with early 1-month mortality (before ART) and I am not sure theoretically that ART could fix this in time,” he told HATIP.
Perhaps early ART is primarily protecting against the greater constitutional consequences of active TB in people with HIV. Even if TB is successfully treated, without ART during active TB disease, people might end up more wasted, weaker and frail — and more susceptible to succumbing to other infections and HIV related conditions. This may be a particular danger in very advanced HIV disease. As a member of the audience at the conference pointed out, it isn’t clear that you would see a similar survival benefit, or IRIS frequency, in people with higher CD4 cell counts.
Even so, the CAMELIA data seem to be a profound illustration that people with advanced HIV need ART as soon as possible for their HIV disease.
During the conference, Dr Kevin de Cock of the US Centers for Disease Control asked if it would make more sense to treat both conditions at once. Dr Blanc pointed out that there are often very real operational challenges to starting HIV treatment earlier – such as the timing of getting a patient with TB tested for HIV and giving them their results.
“I think it is reasonable to have a two week delay: 72% of our patients did not know their status a month before going into the study,” said Dr Blanc.
The results created quite a stir on the same day activists in a marched through the conference centre in a 'cough-in', carrying coffins and signs that said “No more people with HIV dying from TB”. In the question period and during a later session on TB and HIV co-infection, and in subsequent correspondence with HATIP, various experts discussed the implications of the findings.
“I think that overall these data are very important and very credible. These provide strong support for the 2010 WHO guidelines (which is very reassuring!) and also entirely consistent with the Zolopa study in patients who were also very immunocompromised but with non-TB illnesses. We clearly need further data from other settings and at different CD4 counts but all the data are pointing in the same direction,” Dr Lawn told HATIP, but he added:
“There is an important exception, though: when the OI is intra-cranial. Immediate ART conferred no survival benefit among patients with TB meningitis in Vietnam (2 weeks vs 2 months) (Torok et al ICAAC 2009). And immediate ART vs >10 week delay was associated with much higher mortality in patients with cryptococcal meningitis in Zimbabwe (Makadadzange et al CID 2010). When these serious OIs involved vital structures within the confines of the cranium, then the consequences of IRIS are likely to be far more severe. Thus, I think early ART is good except for patients with CNS TB or crypto[coccal meningitis]. More data are needed here,” he told HATIP.
Professor Di Gibb of the UK’s Medical Research Council also told HATIP she believes “knowing more about how easy IRIS was to treat ...or might be if the CD4 is higher…outside major centres would be important for programmes.”
“This is a wonderful study that complements the results of the SAPIT study. I want to point out that in the SAPIT study, the CD4 counts were substantially higher, and the rate of IRIS was about 12% in those who started ART on TB drugs and there was no mortality attributable to IRIS. So I think that the IRIS issue – not that it’s not important – it should not be a barrier to initiating ART in people with HIV and TB,” said Gerry Friedland of Yale University, speaking at the conference.
“It makes it very compelling now more than ever, to do the testing as well as initiate ART,” said Dr Wafaa El-Sadr of Columbia University and ICAP in a TB/HIV session just after the CAMELIA findings were presented. “I’m wondering whether we should reconceptualise the treatment of TB in patients with HIV, and frame it to them as 'ART is part of treating their TB' rather than presenting it as treating two diseases. I think there’s a compelling need to change the way we are presenting ART to TB patients.”
“If the statement “the earlier the better” cannot scientifically be made as yet, but among experts we agree that it is highly probable, then common clinical and public health sense and a patient’s perspective needs to be taken into account as well,” Eric van Praag of Family Health International in Tanzania told HATIP.
“If the patient is ambulant and pill load is understood and acceptable and there is easy access to the TB/HIV clinic, and a home-based care programme is guaranteed for follow-up of side effects/ IRIS, then we should advocate for a start at 2 weeks. In situations where patients are not that stable clinically, or have very low CD4 counts, or live far or isolated from the clinic, or are not really ready as yet, we should be more prudent, manage other conditions, and nutrition, and try linking up with home-based care or outreach. In other words, we need to have room for individualisation of clinical management,” he concluded.
Blanc FX et al. Significant enhancement in survival with early (2 weeks) vs. late (8 weeks) initiation of highly active antiretroviral treatment (HAART) in severely immunosuppressed HIV-infected adults with newly diagnosed tuberculosis. Eighteenth International AIDS Conference, Vienna, abstract THLBB106, 2010.
Müller M et al. Immune reconstitution inflammatory syndrome in patients starting antiretroviral therapy for HIV infection: a systematic review and meta-analysis. Lancet Infect Dis 2010 Apr;10(4):251-61.
View abstract and slides from this session on the official conference website.
Requiring tuberculin skin tests before implementing IPT leads to avoidable and substantial number of TB cases
By Theo Smart
Delaying a course of isoniazid preventive therapy (IPT) to prevent active tuberculosis (TB) because a person with HIV hasn’t yet had a tuberculin skin test (TST), or received their results, leaves the individual at very high risk of developing active TB, according to results from the THRio study team, presented on Monday at the Eighteenth International AIDS Conference in Vienna.
Although those with positive TSTs had the highest risk of TB in the study, approximately 60% of the TB cases occurred in those who were TST unknowns, and the rates were also high in those who were waiting to have their TSTs read.
“It’s taking too long for people to get a TST, and among those who have a TST, even for those who have a positive result, it’s taking them a long time to get put onto IPT,” said Dr Jonathan Golub of Johns Hopkins Center for Tuberculosis and the Consortium to Respond Effectively to the AIDS Tuberculosis Epidemic (CREATE).
A course of isoniazid preventive therapy (IPT) has been shown to reduce the incidence of active TB in people with HIV – especially among those who are TST-positive. A positive TST can demonstrate whether someone has a latent TB infection. If latent TB is treated with a course of isoniazid – usually for at least six months – the risk of subsequently developing active TB is more than halved in people with HIV.
In Brazil, it has long been policy that people with HIV receive an annual tuberculin skin test and, if they are positive, that they be given IPT in order to reduce the incidence of TB disease. However there has been extremely poor uptake of this recommendation – partly because of the difficulty in implementing TSTs.
The THRio study’s object was to see if widespread implementation of the policy would result in reduced prevalence of TB in the HIV-infected population in Rio de Janiero, Brazil. To find out, the study team conducted a cluster-randomised trial in 29 public HIV clinics in Rio. (ART has been freely available since 1996 to any person with HIV in Brazil once they are medically eligible for treatment.)
The clinics were randomised to the date on which IPT services would begin, over a phased implementation. Primary outcomes in the study were the TB rates before and after the intervention, measured at the clinic level among all the TST-eligible patients.
The intervention involved training the HIV clinic staff (two clinics every other month) to screen properly for active TB in their clients with HIV, and properly perform a TST and provide IPT in those without a TB history, who had not yet taken IPT or who had a prior positive skin test. A six-month course of IPT was then given to anyone with a positive TST but without active TB, and to any contact of someone with active TB.
Before the intervention, data were abstracted from clinic records including variables such as data of HIV diagnosis, ART history, CD4 and viral load. The intervention began at two clinics in September 2005, and new clinics were enrolled every two months until January 2008, with follow-up until December 2009.
Training improves TST performance, but not well enough
After the intervention, the number of TSTs performed quadrupled (from 1505 to 6125), and the number of patient with a positive skin test increased fourfold (though the proportion of those testing positive remained constant at 13.4 to 13.55%). Prior the intervention, over 60% of the patients had not received a TST over a two-year period, but after the intervention about 50% had received one in the last six months. Likewise, the time to TST and to IPT both improved after the intervention.
“However, we still believe that it is taking too long to get skin-tested and onto preventive therapy,” said Dr Golub. “So we looked at what would happen if we removed TST from the picture, if we could provide IPT to all patients regards of their TST status.”
So the study team compared TB risk by TST and IPT status. TB-free and IPT-naive patients were analysed from HIV diagnosis until TB diagnosis, death, or censoring at last follow-up, excluding those with a very short follow-up time – less than 60 days after their HIV diagnosis. Cox proportional hazards models compared TB risk across TST and IPT categories, stratified by ART.
The study included around 12,000 patients with 49,169 person-years (PY) of follow-up. Over the course of follow-up, there were 1083 incident TB cases (incidence rate (IR)=2.2/100PY); 95% confidence interval (CI):2,1 to 2.3). Males had a somewhat higher incidence rate than women, 2.4/100PY (95% CI 2.2 to 2.4 vs 1.9/100PY (95% CI 1.7 to 2.1). Among those with CD4 cells below 200mm3, the incidence rate was 3.1/100PY (95% CI 2.8 to 3.4) but people with higher CD4 cell counts had lower incidence rates.
The provision of IPT reduced the incidence rate of TB from 2.3/100PY (95% CI 2.2 to 2.4) down to 1.5/100PY (95% CI 1.2 to 1.9) – a drop of 35%. ART also reduced the risk from an incidence rate of 3.4/100PY (95% CI 3.1 to 3.7) among those not on ART to 1.7/100PY (95% CI 1.6 to 1.8) on ART – a drop of 50%.
Results were then stratified for all the patient by TST status. The background incidence rate remains the same at 2.2/100PY, but the incidence rate for those who were TST-positive who did not go onto IPT was extremely high at 14.9/100PY (95% CI 12.8 to 17.5). However, when those TST-positive people were given IPT, there was a reduction of around 90%, with an incidence rate of 1.6/100PY (95% CI 1.3 to 2.1).
Among TST negatives, there were low rates of TB whether given IPT or not (consistent with last year’s findings from the BOTUSA trial).
However, the group that the study team was most interested in were the TST unknowns (who were never tested). Among this population, IPT achieved a 50% reduction in TB incidence, going from an incidence rate of 2.5/100PY (95% CI 2.3 to 2.7) for those not on IPT down to 1.2/100PY (95% CI 0.7 to 20). (It should be noted that far fewer in this group got IPT, however).
Then the researchers stratified this group of ‘unknowns’ based upon whether they were on ART. Giving IPT to people who were taking ART reduced the incidence of TB by about 35% (from 1.7/100PY (95% CI 1.6 to 2.0) to 1.1/100PY (95% CI 0.6 to 2.0, although these confidence intervals overlap). Among the ‘unknowns’ not on ART, there was an approximately 60% reduction in those who received IPT from 3.8/100PY (95% CI 3.4 to 4.3) to 1.4/100PY (95% CI 0.4 to 4.3) – although again these confidence intervals overlap, probably because few people not on ART had been put on IPT.
But what would happen if the TST were not performed (or at least didn’t need to be performed for a person to get IPT)?
The researchers looked at the data on at all the people who were eligible for a TST from the time the intervention began at the clinic, and followed them until they received IPT, received a TB diagnosis, died, or the follow-up ended. What they found was a rate of 1.5/100PY (95% CI 1.3 to 1.7) for the group overall. The rates were higher for those who were on follow-up for 6 months or 12 months.
“What this means is that if we had given IPT to these patients instead of making them wait to get a TST and act upon that result into for their TST, we could have prevented a large amount of TB at an earlier time,” said Dr Golub.
Theresearchers then looked at people who were eligible for a TST, and followed them until one was performed (and a result was received). Among those who were categorised as 'unknown' and then tested positive, the incidence rate was extremely high, at 22.0/100PY (95% CI 17.3 to 27.6). The incidence was very low among those who tested TST-negative but was twice as high among those who remained unknown, at 1.9/100PY (95% CI 1.6 to 2.3). These results again underscore that TB cases could be prevented by earlier IPT among TST unknowns.
This hasn’t been the first time that a study has shown that requiring TSTs acts as a barrier to IPT implementation. WHO’s new guidelines on IPT, a new version of which will be released in a couple of months, do not require that TSTs are performed, because they are technically difficult in many settings. But even in a middle-income country like Brazil, which should have the capacity to perform TSTs, this study shows that the requirement still impedes IPT implementation. According to Golub, there are a number of policy implications.
“We could eliminate TST and give IPT to all HIV-infected patients. This would save a lot of money, it would be much more convenient, much easier to conduct at the clinics.” He said that although this might mean over-treating TST-negatives, IPT has low toxicity.
Another option would be to improve the provision of TST, but that is after all what this study had been trying to do, with some limited success. However, a final option might offer a compromise, which is to give IPT to all people with HIV until a TST can be performed and read. In settings with lower burdens of TB, it may then be possible to discontinue IPT in someone who is TST-negative. However, the wisdom of doing this in a setting with a higher burden of TB, such as South Africa, and where people may be exposed to TB in clinic settings, is uncertain.
Golub JE et al. Value of the tuberculin skin testing for isoniazid preventive therapy for HIV-infected patients. Eighteenth International AIDS Conference, abstract MOAB0305, Vienna, 2010.
Jonathan E Golub's abstract is available on the official conference website.