HIV may be becoming less fit as it adapts to the immune system

HIV, at least in some parts of the world, may be developing a lower replicative capacity as it adapts to variations in the human immune system, studies in southern Africa and elsewhere suggest.

Philip Goulder of the University of Oxford told the AIDS Vaccine conference last month that competition between HIV and certain varieties of human HLA (human leukocyte antigen) genes may be contributing to a diminution in HIV virulence, a lower community viral load, and an increased proportion of ‘elite controllers’ in the population.

Goulder remarked that these changes seemed to be happening surprisingly fast, and that in some populations the introduction of antiretroviral therapy would also tend to reduce the fitness of the HIV that was still circulating.

HLA subtype and viral fitness

Goulder began by noting that it has been known for some time that people with certain variants of HLA tended to develop lower viral loads than average if they acquired HIV.

The HLA genes are the human version of the major histocompatibility complex (MHC), a set of proteins present in all vertebrates, which helps the body distinguish between its own tissues and foreign invaders. They sit on the surface of cells and display ‘epitopes’, short sections from protein molecules inside the cell. If the immune system does not recognise these epitopes as belonging to the self, it will mount a response and attack the foreign invader. Some HLA variants are better at displaying epitopes that stimulate a strong immune response than others.

If HLAs are too efficient, the result can be auto-immune disease and a high susceptibility to allergies. It is interesting that the HLA variant most strongly associated with a lower HIV viral load and slower disease progression, B*5701, is also associated with hypersensitivity (allergy) to the drug abacavir (Ziagen, also in Kivexa/Epzicom).

B*5701 is not the only HLA subtype associated with lower viral load. Most other varieties of the broader B*57 type are also associated with slower progression, as are B*58, B*51 and to a lesser extent B*31, 38, 42 and 44. Certain other subtypes are associated with higher viral loads and faster HIV progression.

One of Goulder’s population studies, of people with HIV in Durban, South Africa, showed that subtypes B*5701, 5702, and 5801 were associated with at least a threefold (0.5 log) lower viral load than the average for the population, and B*5703 with a fivefold reduction (0.7 log).

HLA escape mutations

HIV can develop resistance mutations to HLA as it does to drugs. These change the viral proteins into ones the HLA molecule cannot ‘grab hold of’ and display on the cell surface, and thus the virus becomes less visible to the immune system. However, these viral protein changes may also exact a fitness cost by causing viral replication turn over more slowly.

The first sign that this might be changing the make-up of the HIV in a population came from Japan, which has an unusually high proportion of people with the B*51 HLA variant (20% as opposed to 10% in Oxford and 1% in Durban). It also has an unusually high proportion of HIV that carries a specific, fitness-reducing HLA-resistance mutation in HIV called I135X. Three-quarters of people with HIV carry this mutation in Japan versus 40% in Oxford and 20% in Durban. Goulder wondered if this might be part of the reason Japan had maintained a particularly low HIV prevalence (less than 0.1%) despite having a developed gay scene.

The relative scarcity of HIV in Japan makes it hard to study. Instead, Goulder looked at viral loads and HLA variants in the very high-prevalence cities of Durban and Gaborone in Botswana.

The HIV epidemic is older in Botswana than it is in Durban – it reached peak prevalence in about 2000, which Durban did ten years later. The average viral load in untreated people is also lower: 15,350 copies/ml (4.2 logs) compared with 29,350 copies/ml (4.5 logs) in Durban. This is despite the fact that the average CD4 count is 50 cells/mm³ lower in Gaborone, indicating that people on average have been living with HIV for longer – which would also suggest a higher viral load.

The same survey found a higher prevalence of HIV HLA ‘escape’ mutations in Gaborone than Durban – 46% of people in Gaborone had HIV that the HLA B*57 and B*58 molecules could not display, compared to 38% in Durban.

These mutations had real effects. As we said above, in Durban, people with the B*57 and B*58 variants of HLA had lower viral loads than average. In Gaborone, they did not: their viral loads were identical to the rest of the population and in the case of the B*5702 variant, they actually had viral loads 0.3 logs higher than average. In other words, the HIV in Gaborone had evolved into a form which was essentially unaffected by more efficient HLA types. HIV in Gaborone was also less fit: in a test-tube assay, it replicated 11% more slowly than the Durban virus.

Recent drops in viral fitness – and more elite controllers

In Durban, there was evidence that a reduction in viral fitness might be happening right now. Goulder compared viral loads in female patients in the largest HIV clinic there in 2002-5 and 2012-13 (he studied people accessing antenatal care, as their viral load and treatment status was better documented). Excluding women who were on antiretroviral therapy (ART) or had been on ARV drugs for pregnancy less than three months previously, and women with CD4 counts less than 350 who might be on undocumented ART, the average viral load in women was 13,550 copies/ml (4.13 log) in 2002-5 and 5750 copies/ml (3.76 log) in 2012-13, and this difference was significant (p = 0.0013).

There also appeared to be an unusually high proportion of women in the population who were controlling HIV spontaneously. Twenty-four of the first 239 women whose viral loads were tested (10%) had a viral load under 50 copies/ml off ART. This compares with a prevalence of no more than 0.5% in most populations studied. Other explanations – false-positive HIV tests, women taking ‘black market’ or informal  ART, differences in viral load assays – were ruled out, and furthermore the average CD4 count in the women with undetectable viral loads was 890 cells/mm³ – essentially normal and indicative of viral control.

Co-factors in reduced transmission and prevalence

HLA adaptation is not the only reason HIV fitness might be declining. Goulder pointed out that the fittest virus was also not necessarily the one that got transmitted most often. This is because untreated people with high viral loads tend to die quicker so have less time in which to pass on their HIV. Modelling work by the team at Imperial College, London has established that the ‘optimal’ viral load in terms of the efficiency of HIV transmission is 4.52 logs (33,000 copies/ml).

Goulder added that HLA escape mutations and higher mortality in people with high viral loads might now be joined by a third reason viral load and fitness might fall over time. In a population where ART is introduced in stages, the sickest people with the highest viral loads will get treated first, and then the next most sick, and so on. This will tend to remove the fast progressors, who have a high viral load ‘set point’, from the untreated population first.  

ART will only tend to have this effect in populations where a lot of HIV is passed on in chronic infection: in populations where most people diagnosed are treated, most HIV will be passed on by the undiagnosed – who by definition will not be on ART.

Nonetheless, he added, falls in HIV prevalence and rises in life expectancy in Africa and some other regions might not only be caused by behavioural changes or by putting increasing proportion of people on treatment. 

“It appears there are more factors that are driving viral fitness down than we thought,” he remarked.