CROI: Risk of new illnesses declines sharply after first two months on treatment in resource-limited settings

The ART-LINC and ART-CC collaborations

Large cohort studies have long been used to monitor the natural history of people with HIV and safety and effectiveness of various treatments used in clinical practice. In 2001, a pooled network of databases from Europe and North America was established, called the ART Cohort Collaboration (ART-CC), to monitor prognosis in the era of ART.

Recently, a similar network was established in resource-limited settings, called the ART-LINC Collaboration. This multinational network of HIV treatment programs in Africa, Brazil, and Asia now involves 40 centres, 20 countries and 31,000 people on ART.

One of the first uses of the database was to investigate relatively high mortality rates that had been reported post initiation of ART, particularly in the early months, in resource-limited settings. In an early comparison between both databases (ART-CC and ART-LINC), people in resource-limited settings were found to have a four-fold greater risk of death during the first month on ART, and continued to have a higher risk out to six months, compared to people on ART in high income settings. However, the causes of death were unknown.

So the current study was conducted to examine HAI during the first year on ART in people from the ART-LINC cohort, to identify factors associated with a newly recorded illness during the period, and to compare the findings from ART-LINC with those from ART-CC.

People on antiretroviral therapy (ART) in resource-limited settings experience HIV-associated illnesses (HAI) and tuberculosis (TB) at double the rate observed in high-income countries, according to a study reported at the Thirteenth Conference on Retroviruses and Opportunistic Infections in Denver.

These illnesses are most commonly diagnosed in the first couple months after initiating treatment — and may necessitate closer monitoring of people on ART during this period, according to the study’s presenter, Dr Paula Braitstein of the University of Berne, Switzerland. (To view or listen to the complete presentation, visit the conference website).

Results

In ART-LINC, there were 4540 people available for the analysis with a total of 3255 person years of follow-up. 689 HAI events were reported which averaged out to 212 events per 1000 person years. This was more than twice the rate of 82 events per 1000 person years seen in the ART-CC (which had 12,574 people, with 8,709 person years of follow-up, and 710 events).

The patterns of HAI observed were different in resource constrained settings than in high income settings. The most frequently reported HAI in ART-LINC were, in order, tuberculosis, Kaposi’s Sarcoma (KS), candidiasis, bacterial pneumonia, cryptococcus, toxoplasmosis, wasting syndrome, and mycobacterium avium disease (MAI), while in the ART-CC, MAI, KS, cytomegalovirus disease, PCP, tuberculosis, oesophageal candidiasis, encephalopathy and toxoplasmosis were more commonly reported.

To some extent, there are real regional differences in prevalence of illnesses (such as for TB), however, some of the difference may be attributable to different diagnostic techniques and approaches that can vary significantly across settings.

However, over the course of observation, there was a similar decline in the relative risk of HAI overall and individual illnesses, with the exception of bacterial infections, which actually appeared to become more common between months 4-6 in the ART-LINC cohort. Otherwise, the lion’s share of HAI and TB were observed in the first few months of treatment.

Factors associated with HAI in the first year included female sex (p=0.001) and age under 30 years (p=0.81). However, Dr Braitstein noted that a survivor bias could have explained the age-effect. As for the higher risk in women, she postulated that that the effect may have been due to adherence problems that women may face, because the greater risk was only observed over the course of the year — but not during the first three months.

The factors most significantly associated with risk of HAI were baseline CD4 cell counts, initial ART regimen and a history of opportunistic infections or TB (all p<0.001). While there was no significant difference between being on a NNRTI-based regimen or a PI-based regimen, individuals taking other three drug regimens (triple nucleoside analogs such as abacavir/AZT/3TC) had double the risk of developing a HAI. This does not bode well for those who would like to cast abacavir/AZT/3TC as an alternative regimen for resource-limited settings.

The study

The study was conducted when there were only 16 centres in the cohort. The analysis included every treatment-naive adult with a known baseline CD4 count who started triple drug ART and had at least one day of follow-up. HAI included all HIV-related and AIDS defining infections and conditions as well as TB (both pulmonary and extrapulmonary). To avoid counting pre-existing conditions, events were defined as 'newly diagnosed and/or recorded' if at least 90 days had passed since the last episode of the same illness or in the case of TB, 180 days to account for the standard length of TB treatment.

Several potential risk factors for the occurrence of HAI were considered including age, sex, baseline CD4, initial treatment regimen (non-nucleoside reverse transcriptase inhibitor [NNRTI]-based vs. protease inhibitor [PI]-based vs. other ≥3 drugs), and having a HAI or TB at or before starting ART.

Interpretation

The differences between high and low income settings may partly be explained by the most significant predictors of post-ART illnesses: having a low CD4 cell count and a history of opportunistic infections or tuberculosis before going on ART. Both factors are much more common in people with HIV in resource-limited settings, who often do not present for diagnosis and treatment until they are already very ill.

It is also impossible to distinguish between new and occult infections — or illnesses that simply hadn’t been diagnosed yet. Many of the conditions seen in the first three months could have been lurking beneath the surface, revealed and exacerbated by ART-associated immune reconstitution inflammatory syndrome (IRIS). Other infections could have occurred because the immune system hadn’t had a chance to recover yet.

In light of this, Dr Braitstein recommended more vigilant care during the first few months on ART. "Frequent monitoring and the use of prophylaxis may also improve outcomes in this very vulnerable time following the initiation of treatment," she concluded.