CROI: Could earlier ART reduce risk of death from non-AIDS related illnesses in people with HIV?

This article is more than 16 years old. Click here for more recent articles on this topic

“Evidence suggests that HIV may well play a role in several serious non-AIDS defining events… and the use of ART (antiretroviral therapy) may well reduce the risk of some serious non-AIDS events,” said Professor Andrew Phillips of the Royal Free & University College Medical School in London, speaking on Monday morning in a plenary lecture at the Fifteenth Conference on Retroviruses and Opportunistic Infections (CROI) in Boston.

His presentation was based on an analysis of available morbidity and mortality data from a number of cohorts and large trials since the introduction of ART, which all seem to suggest that non-AIDS related illnesses are accounting for a growing proportion of the deaths in people with HIV — though even these causes of death have become less common on ART. The studies also suggest that, untreated, HIV could be increasing the risk of death from other serious conditions even at higher CD4 cell counts.

To further reduce the risk of these illnesses and death in people with HIV, he suggested increasing efforts to diagnose HIV as early as possible; research to better understand the risk factors for these non-AIDS events in people with HIV and to define ways to manage or prevent them; and finally, “we need to be looking at whether ART should to be initiated earlier in patients with CD4 counts above 500.”

Trends in death rates over time, and the reasons for those deaths

Since the introduction of highly active ART, starting around 1996, the rate of death among people with HIV has fallen sharply and data from the US HIV Outpatient Study (with over 8500 participants) show that the rate seems to keep falling over time. A very similar trend has been observed in a large cohort from the UK, with a death rate of just below 1% in 2006 among those seen for care during the year.



Relating to the heart and blood vessels.


Genes, proteins or chemicals that can act as signals for certain diseases.

cardiovascular disease

Disease of the heart or blood vessels, such as heart attack (myocardial infarction) and stroke.

AIDS defining condition

Any HIV-related illness included in the list of diagnostic criteria for AIDS, which in the presence of HIV infection result in an AIDS diagnosis. They include opportunistic infections and cancers that are life-threatening in a person with HIV.


In HIV, an individual who is ‘treatment naive’ has never taken anti-HIV treatment before.

“That’s a relatively low rate,” said Prof Phillips, “but that death rate is still well above the general population [death] rate for people of the same ages. So what are people dying from?”

Data from a French cohort suggest that even though AIDS-related conditions are still the leading cause of death, they only account for slightly over a third of all deaths in people with HIV. Other causes accounted for the rest including most commonly cancer, hepatitis C, cardiovascular disease among others.

And importantly, failure of ART does not seem to explain why most people with HIV are dying, even among the AIDS-related deaths. According to a British audit of the AIDS-related deaths in 2005, clinicians listed the situations in which the death occurred — and the most common scenario was initiating treatment too late (explaining 40% of the cases). Of note, only 5% of the deaths were due to multidrug resistant HIV leaving the person with no remaining ART options. “With the new drugs coming through, and recently licensed, it seems like this situation will continue for some time,” said Prof Phillips.

Although other causes may account for a higher proportion of the deaths of people with HIV, the introduction of ART also produced a dramatic decrease in the incidence of non-AIDS related death according to unpublished data from the EuroSida cohort.

Could HIV increase risk of serious non-AIDS conditions and death?

This suggests that, without treatment, HIV could be increasing the risk of death from certain non-AIDS related illnesses — and by implication, that ART might reduce the risk of death from those other diseases. So Prof Phillips focused on four disease areas including non-AIDS malignancies, end stage kidney disease, cardiovascular events, liver cirrhosis and death from other causes. He did not focus on the adverse effects of ART — but noted that it isn’t always easy to distinguish between the effects of HIV and the effects of ART in these cohorts.

HIV could increase the risk of non-AIDS illnesses in a number of ways. Some general mechanisms he listed that could potentially alter pathogenic processes include early loss of CD4 cells in the gastrointestinal tract, changes in the mucosal barrier allowing microbes to spread, generalised immune activation or fibrosis of lymphatic tissue.

But when it comes specifically to cancer, HIV-related immunodeficiency leads to a reduced capacity to control organisms (like certain viruses) that trigger cancerous processes; and allows infections that cause chronic inflammation that may in turn increase the risk of cancer. Or the immune system could lose the ability to recognise (and eliminate) transformed cells that develop into malignancies.

HIV has also been associated with a range of kidney disorders, such as HIV-associated nephropathy (which can be reversed by ART). But there is also a link with other pathologies such as immune complex glomerulonephritis and there is a high prevalence of proteinuria and raised creatinine levels (markers of kidney damage) that are associated with high HIV RNA levels and low CD4 cell counts. Proteinurea and elevated creatinine have both been associated with all causes of mortality in people with HIV (not just deaths due to kidney disease).

HIV infection has also been associated with some negative changes in biomarkers (or potential biomarkers) for cardiovascular disease such as HDL-cholesterol depletion, inflammation (as measured by IL-6 and C-reactive protein), endothelial activation/dysfuntion (as measured by VCAM/ICAM), activation of coagulation (as measured by D-dimer). Some of these chances seem to be at least partly reversed by ART.

Liver fibrosis also occurs more rapidly in people with hepatitis B or C virus (HBC and HCV) when they are coinfected with HIV though the mechanism for this isn’t entirely clear.

But 25 years into the HIV epidemic, and more than ten years since the introduction of ART, there is a large body of data that could show whether HIV is indeed increasing the risk of developing and dying from these serious conditions. Prof Phillips reviewed three types of evidence including 1) studies comparing of the risks between HIV-infected and uninfected people, 2) studies that looked at whether the risk of these conditions was associated with viral load and CD4 cell counts 3) and randomised trials that might show whether ART had an impact on these non-AIDS events.

The risk of serious non-AIDS events between people with HIV and people without HIV

There are certain limitations in making comparisons between people who are HIV-positive and those who are not because the groups may differ in other ways, besides HIV infection, that could confound the analysis and in ways that would be extremely difficult to adjust for. Dr Phillips mentioned smoking as one obvious example, but there are a host of others, such as the very factors that made some people more vulnerable to becoming HIV infected in the first place, and which could put people at a higher risk of other illnesses as well (economic or social situations, poor self-esteem, access to health services, etc). In addition, he noted that each non-AIDS condition has its own set of risk factors that could act differently in HIV infected people. Finally, in some of the studies, the HIV-infected people in the cohorts could have been taking ART, which could impact the data.

That being said, a recent meta-analysis, comparing the incidence of cancers in 444,172 people with HIV with that in 31,977 immune suppressed transplant patients versus the general population found that there was a significantly higher risk in both people with HIV and the transplant patients for 20 out of 28 cancers examined. Since these populations could be assumed to be quite different, the findings strongly suggest that these cancers are linked with immunodeficiency. A couple of studies last year also found an increased risk of lung cancer in people with HIV, independent of smoking.

The data on kidney disease isn’t as clear. In a study of US Veterans last year that looked at the risk of end stage renal disease (ESRD) in patients according to whether they were black or white patients, and according to HIV status. HIV appeared to have no significant impact on ESRD in the white patients. But black patients were at twice the risk of ESRD compared to whites, and in this population HIV appeared to increase the risk substantially (after adjustment for a range of factors).

Four major studies have found an increased risk of cardiovascular disease among HIV-positive patients compared to people without HIV. However, in one of the studies, this association was only observed in people who had been on a protease inhibitor for 18 months; and in another study, the link was only observed in younger patients.

As for liver disease, one study has reported that among 4855 men and boys with haemophilia (and probably HCV) 1218 who were HIV-infected had a markedly higher cumulative risk of death from liver disease within 25 years. Similar findings were reported for HBV in the Multicenter AIDS Cohort Study (MACS).

Findings to be presented on Wednesday this week at the conference, will highlight the risk of death from any cause in ART-naïve people with CD4 cell counts higher than 350 compared to the general population.

The association of CD4 and viral load with the risk of serious non-AIDS events

The data from two large collaborations, including the DAD collaboration (which contains 10 cohorts including CPCRA, EuroSida and others following over 33,000 people over time) and the CASCADE collaboration (which is in subjects with known dates of seroconversion — though Prof Phillips was only looking at the ART-naïve subjects) consistently show that the risk of death, whether from non-AIDS causes or all causes, increases with decreasing CD4 cell counts. At the higher CD4 counts, the rates of death from all causes and non-AIDS causes are virtually identical because, Prof Phillips said “virtually all the deaths were from non-AIDS causes.”

The association generally holds true for each of the four conditions. For instance, there is a significantly reduced risk of non-AIDS malignancy associated with higher CD4 cell count in the DAD and Cascade collaborations, and in pooled data from the First trial (a study comparing different starting strategies) though not in the SMART study. Further data from the Aquitaine Cohort supporting this conclusion were presented later at CROI.

There are less data (and fewer deaths) for kidney disease. The DAD collaboration and First both show a significantly higher risk of ESRD with lower CD4 cell counts but it is not significant in SMART.

The four studies show less of an association between CD4 cell counts and cardiovascular disease (a trend but the confidence intervals are wide). In the DAD collaboration that includes a very large number of events (including non-fatal events), the hazard ratio might only lie just below 1. However, these findings could be confounded by the fact that some drugs for HIV have also been associated with a higher rate of heart disease (including, at this conference, abacavir and ddI).

There is fairly strong evidence of an increasing risk of liver disease/death with lower CD4 cell counts, except in the First trial (where there were only 14 events and the confidence intervals were wide).

There is also a relation between decreasing CD4 cell counts and non-fatal hospitalisations due to AIDS-defining events reported in the Aquitaine cohort between 2000 and 2004. However, at CD4 cell counts above 200 almost all the hospitalisations were due to non-AIDS events.

There are fewer data on the association of viral load and serious non-AIDS events. SMART looked at the risk of events with a viral load > 400 versus ≤400 and when the events were pooled together, there was a significantly reduced risk of serious non-AIDS events with a suppressed viral load, even when the data were adjusted for the most recent CD4 cell count. The evidence was fairly consistent for each of the four conditions with the exception of non-AIDS malignancies.

The impact of effective ART on the risk of serious non-AIDS events

“The real test is: if you can induce a change in viral load and CD4 count, can you see a resulting change in the incidence of serious non-AIDS condition?” said Prof Phillips. Although no large study has yet addressed this specifically, some hints can be gleaned from the SMART trial.

The design of SMART and its findings ( including extended follow-up presented at this conference) are described elsewhere on aidsmap. Essentially, this was a treatment interruption study involving thousands, but — at least right after the start of the trial — one arm was on ART and the other (the intermittent arm) was not. Over the course of the trial, most participants had CD4 cell counts over 200, and 80% of the time, the CD4 cell counts were over 350.

The study found that there was a higher hazard ratio of serious non-AIDS events off of ART (1.6 for all events combined which was statistically significant). The findings were relatively consistent for the individual conditions, though not every one was significant on its own.

“Of the 85 deaths that occurred in SMART, only 7 (8%) were from AIDS causes, so this emphasizes again that in this higher CD4 cell count range, morbidity and mortality is dominated by non-AIDS events,” said Prof Phillips.

477 of the patients in ART were either treatment naïve or off of ART for over 6 months, which allowed for a comparison deferred ART versus immediate ART. Although the number of events were too small to be able to definitively conclude anything (12 vs 2), the hazard ratio of 7.02 (95% confidence interval 1.57-31.4) was significant with a p-value of 0.01.

Another analysis of the SMART data to presented later at the conference will examine how biomarkers such as IL-6 and d-dimer differed between the two arms, and could prove useful in explaining these differences in clinical endpoints.

Conclusion and recommendations

“On balance, the evidence suggests that HIV may well play a role in several serious non-AIDS defining events,” said Prof Phillips. It’s possible that starting ART much earlier could reduce these events..

In addition to earlier HIV diagnosis, Prof Phillips urged further research into the risk factors for these events in both people on and off ART, and “in order to do that, we need to standardise diagnostic criteria and data collection methods.”

In order to see whether ART should be started earlier, Prof Phillips recommended conducting a trial comparing ART initiation in people with CD4 cell counts above 500 versus waiting until CD4 cell counts are around 350. He believes such a study would be justified because:

  • Serious non-AIDS illnesses are relatively common at higher CD4 cell counts
  • The SMART study suggests that the risk/benefit of ART favours benefit
  • The virological benefit of ART is more durable than expected when the drugs first became available
  • Early use of therapy might be cost effective, especially if associated with a reduction in transmission risk
  • Such a study would provide a basis for identifying the usefulness of biomarkers of the risk of these non-AIDS conditions, and provide insights into pathogenesis

“Research is needed to provide a basis for defining models of care for people with HIV which take into account the risk of all serious conditions,” said Prof Phillips. “Research into mechanisms by which HIV affects risk of non-AIDS conditions is needed, and it may help us understand more about the causes of such conditions outside of HIV.”

The START trial, which will evaluate early versus deferred treatment could prove to be an important resource for answering these questions and the pilot phase will begin in just a few months.


Phillips A. Morbidity and mortality in the HAART era. Fifteenth Conference on Retroviruses and Opportunistic Infections, Boston, 2008.

Prof Phillip’s full lecture and slides containing the studies he referenced can be found on the conference website.