Elite controllers may pay a high price for their low viral load

Hospitalisations due to cardiovascular disease and psychiatric illness more common
This article is more than 10 years old. Click here for more recent articles on this topic

About one in 200 people with HIV maintains an undetectable viral load and high CD4 counts without having to take antiretroviral therapy (ART). These so-called ‘elite controllers’ have long been the targets of research into how their immune systems control HIV replication, and whether treatments such as therapeutic vaccines could induce the same state in other HIV-positive people, enabling them to stop taking ART.

A recent US study, however, has found that the health of elite controllers (ECs) is in some ways poorer than the health of other people with HIV with CD4 counts over 350 cells/mm3 – and even of some people with detectable viral loads.

Specifically, the study found that over the six-year study, ECs on average spent more than twice as many nights in hospital than people on ART, and about 50% more than people not on ART and with detectable viral loads. After adjusting for various differences between the EC and non-EC population, the study still found that ECs were 56% more likely to be admitted to hospital than people with undetectable viral loads on ART. While ECs had lower hospital admission rates for some conditions, these were outweighed by considerably higher admission rates for cardiovascular disease and for psychiatric conditions.


elite controllers

A small subset of people living with HIV who are able to control HIV replication in the absence of antiretroviral treatment for an unusually long period of time. Definitions vary, but an elite controller is usually defined as a person whose viral load has remained below 50 copies. However, because HIV continues to replicate even in elite controllers, ART is recommended for elite controllers who have declining CD4 counts or who develop HIV-related complications. Elite controllers and viraemic controllers are members of a larger group known as HIV controllers. Around half of HIV controllers can also be described as long-term non-progressors.

multivariate analysis

An extension of multivariable analysis that is used to model two or more outcomes at the same time.

exclusion criteria

Defines who cannot take part in a research study. Eligibility criteria may include disease type and stage, other medical conditions, previous treatment history, age, and gender. For example, many trials exclude women who are pregnant, to avoid any possible danger to a baby, or people who are taking a drug that might interact with the treatment being studied.

detectable viral load

When viral load is detectable, this indicates that HIV is replicating in the body. If the person is taking HIV treatment but their viral load is detectable, the treatment is not working properly. There may still be a risk of HIV transmission to sexual partners.

chronic obstructive pulmonary disease (COPD)

Chronic obstructive pulmonary disease (COPD) is a chronic inflammatory lung disease that causes obstructed airflow from the lungs. Symptoms include difficulty breathing, cough, mucus (sputum) production and wheezing. It is caused by long-term exposure to irritating gases or particulate matter, most often from tobacco smoking (active or passive).

This study, which is the first study to look at clinical outcomes in an adequately large group of ECs, has two implications. Firstly, it raises the question of whether ECs should be put on ART – or on other medications – to address their specific raised health risks. Secondly, as mentioned above, ECs have been seen as a model for immune control of HIV without ART, but the study brings into question whether they are the right model to aim for in trying to achieve a ‘functional cure’ for HIV (one in which people still have HIV but can stay off therapy for prolonged periods).

The study

The study investigated hospitalisations in people with HIV in a large clinical cohort called the HIV Research Network, which covered eleven hospital-based HIV clinics throughout the US, nine of which had data about the reason for patients’ hospital admissions.

Elite controllers were defined as people who had had at least three consecutive viral load measurements below the limit of detection (usually 50 copies/ml) over a period of at least one year while not taking ART.

The study initially identified 188 elite controllers in a population of 34,354 patients (0.56%) but some people were excluded. This was largely due to the fact that patients with CD4 counts below 350 cells/mm3 were excluded: since elite controllers tend to have high CD4 counts, it would be misleading to compare their health with that of all other people with HIV, some of whom would have low CD4 counts and AIDS-related conditions. In 22 cases ECs were excluded because they were found to have spent short periods on ART.

This left 149 elite controllers studied in a population of 23461 patients (0.64%) but because ECs visit hospital less often for monitoring, they only represented 0.3% of clinical data recorded. The researchers calculated that, taking into account people not in clinical care, their study indicated that a minimum of 0.43% of US people with HIV are ECs.

There were differences between the ECs and other patients. Half of the ECs were women, compared with just over a quarter of patients in general (women tend to have lower viral loads than men, and it was already known that more ECs are women); and 58% of ECs were of black ethnicity compared with 41% of patients in general. The average CD4 count was 778 cells/mm3 in ECs, and about 500 copies/ml in the other patient groups.

The researchers compared hospitalisation rates in these people with 4709 people on ART and with undetectable viral loads, 7998 people with viral loads between 50 and 1000 copies/ml, many of whom were on ART, and 10,605 people with viral loads over 1,000 copies/ml, most of them not on ART.


The hospitalisation rate in people who were undetectable on ART was 10.5 per 100 patient-years. In elite controllers it was 22.3 per 100 patient-years. Even people with detectable viral loads had lower rates of hospitalisation than ECs: in people with viral loads under 1000 copies/ml it was 12.6 per 100 patient-years and in people with viral loads of 1000 copies/ml or more it was 16.9 per 100 patient-years.

After adjusting for gender, race and some other variables such as health insurance status, in multivariate analysis ECs spent 77% more time in hospital than patients undetectable on ART. The difference between ECs and patients with viral loads over 1000 copies/ml became non-significant, but the cardiovascular disease risk was still raised in ECs compared with patients with detectable HIV.

For some categories of illness, being an elite controller was an advantage. For instance, 24% of hospital admissions were accounted for by non-AIDS-defining infections, which would mean things like bacterial sepsis; but in ECs non-AIDS-defining infections accounted for only 2.7% of admissions. The researchers comment that the reasons for this are unknown and that more research is needed to see what immune responses protected against these infections.

However these were more than outweighed by the fact that there were more than twice as many admissions for cardiovascular disease (CVD) in ECs than there were in other patients:  CVD admissions accounted for 31% of the total in ECs compared with 13.5% in other patients.

In multivariate analysis, the risk of admission for CVD was 3.2 times greater in ECs than in people with undetectable viral loads on ART, 2.9 times greater than in people with viral loads between 50 and 1000 copies/ml, and 2.7 times greater than in people with viral loads of 1000 copies/ml or more.

It was already known that ECs have higher levels of certain inflammatory proteins in their blood and more hardening of the arteries, so the finding of more CVD in ECs at least had a mechanism to explain it.

What is less easy to explain is that ECs, in multivariate analysis, had a nearly four times greater rate of admission for psychiatric conditions than patients undetectable on ART.  

In an editorial comment on the study, Maile Karris and Richard Haubrich of the University of California, San Diego speculate that the same inflammatory processes that cause CVD could also cause inflammation in the central nervous system; the fact that the psychiatric admission rate in people was three times higher in people with viral loads over 1000 copies/ml than people on ART with undetectable viral loads could also support this idea. But this is just a hypothesis, and Karris and Haubrich point out that other possible causes of psychiatric illnesses, such as drug use and psychiatric history, were not controlled for.

It also initially looked as if admissions for lung disease were higher in ECs, with 22% of admissions in ECs for pulmonary conditions compared with 4.8% overall. However the researchers found that one particular EC patient had had 21 admissions for asthma and chronic obstructive pulmonary disease (COPD) during the study. If this patient was excluded, the rate of admission for pulmonary disease was similar in ECs to other patients. Omitting this patient also reduced the 77% higher rate of admissions in general in ECs down to 56% higher, though this was still significant, and the higher rate of CVD in ECs was unaffected.  

Limitations and implications

One major confounder in this study could have been smoking. Conscious of this, Crowell and colleagues did a post-hoc analysis where they compared smoking status in all the ECs with a randomised selection of other patients. They did find that 82% of ECs had ever smoked compared with 68% of patients with undetectable viral loads on ART, but do not consider this difference big enough to explain a tripling of the risk of CVD admissions, and one might have expected more COPD in ECs if differences in smoking rates were a significant confounder.

Karris and Haubrich draw attention to a number of other imitations in this study. Risk factors for CVD like cholesterol, obesity and high blood pressure were not evaluated. Also, because ECs who are not ill may have no reason to visit HIV hospital clinics from one year to the next, it is possible that the study may have preferentially selected ECs with pre-existing health conditions.

Because of these limitations they conclude that there is no evidence yet for prescribing ART to ECs in general, though in the primary paper Crowell and colleagues suggest that ECs could be given statins and other precautions against CVD.

In an unrelated commentary on ECs, Leslie Cockerham and Hiroyu Hatano of the University of California, San Francisco suggest that ART could benefit them by bringing down the levels of inflammatory proteins and activated CD8 cells in their system.

Although there does not seem to be one specific type of immune response associated with being an elite controller, Cockerham and Hatano draw a general 'EC profile' of more HIV-specific CD8 cells with stronger responses, higher T-cell activation in general, genes that produce cells with stronger intrinsic antiviral defences, and more sensitive immune responses in natural killer cells.  

These are the kind of cellular immune responses that some protective and therapeutic vaccines have aimed to stimulate – essentially trying to turn more people into ECs.

Cockerham and Hatano, however, argue that ECs are not the best model for a functional cure, given that “natural long-term virological control seems to be coming at an immunologic and/or clinical cost…and a cure should require a disease-free (and not just a treatment-free) state.”


Crowell TA et al. Elite controllers are hospitalized more often than persons with medically controlled HIV. Journal of Infectious Diseases, early online publication. doi: 10.1093/infdis/jiu809. 2014.

Karris MY and Haubrich RH. Antiretroviral therapy in the elite controller, justified or premature? Journal of Infectious Diseases, early online publication. doi: 10.1093/infdis/jiu812. 2014.

Cockerham LR and Hatano H. Elite control of HIV: is this the right model for a functional cure? Trends in Microbiology, early online publication. doi: http://dx.doi.org/10.1016/j.tim.2014.11.003. 2014.

This report is also available in Russian.