Researchers may have caught HIV becoming more virulent

A study from Cuba has generated wide media interest because researchers have identified a particular variety of the virus which is associated with rapid post-diagnosis falls in CD4 count and progression to AIDS. In the study, every member of the still relatively small minority of people who had this viral variant had progressed to clinically defined AIDS within three years of infection. It also seems to be becoming more common in Cuba and may partly explain what appears to be an increase in the proportion of people who progress rapidly to AIDS. However, it is not a drug-resistant strain and antiretroviral therapy (ART) works just as well against it as it does against any other strain of HIV.

This viral strain, dubbed CRF19_cpx, is a recombinant: this means that its genetic code is a patchwork of gene fragments that come from other varieties of HIV. In this case, the genes of CRF19_cpx come from subtypes A, D and G, which are mainly found in Africa. There, subtype D is associated with faster progression, and A/D recombinants even more so.

The researchers believe that the ultimate cause of CRF19_cpx's unusual virulence is that its protease gene, which comes from HIV subtype D, produces an unusually potent protease enzyme that generates exceptional numbers of viral copies and rapidly alters the arms race between HIV and the immune response against it in favour of the virus.

So-called 'circulating recombinant forms' (CRFs) are not uncommon. In some countries they may even predominate, such as in Thailand, where CRF01_AE is so common it is often just called 'subtype E' (and has been associated with faster progression). CRF19_cpx has been found in Africa, but it is only in Cuba that it has been noted to be spreading.

It has also been noted that there appeared to be a trend towards faster progression to AIDS among people in Cuba diagnosed since 2007, with an estimated 13-16% being true ‘fast progressors’ (i.e. with a CD4 count below 200 cells/mm³ within three years of infection). This is higher than the 5-10% reported internationally, but is not unprecedented: a study published in December found, in Europe, the time between infection and a CD4 count below 200 cells/mm³ has halved in the last two decades. This may be driven by evolutionary pressure; because of antiretroviral therapy, HIV now has less time to jump between individuals and so only the fittest, fastest-replicating viruses will survive.

Conversely, pressures in some parts of Africa may be driving HIV towards lower virulence in that region.

The Cuban researchers only found that nine out of a minority of 52 people who had progressed to AIDS rapidly after diagnosis had CRF19_cpx, so this virus is not the cause of all cases; but it is striking that there was no-one in this small study with CRF19_cpx who was not a fast progressor. This is the first time a specific viral variant has so unambiguously been shown to be associated with rapid progression to AIDS.

The study

Between late 2007 and 2011, patients from the Pedro Kouri Institute of Tropical Medicine, Cuba’s premier public health institute, were recruited to the study if they had been diagnosed less than three years after infection and were not taking antiretroviral therapy (ART). Their CD4 count, viral load, and CD4 percentage were taken at this time. The average time between infection and diagnosis was 22 months in people who later on turned out to be slow progressors (see below) and 14 months in people who turned out to be fast progressors. This difference was not statistically significant.

When these patients returned to the hospital with an AIDS diagnosis within 36 months from the estimated time of infection, they were diagnosed as 'rapid progressors' and a second sample was taken, after which they were put on antiretroviral therapy (ART). AIDS was defined either as a CD4 count less than 200 cells/mm³, a CD4 percentage below 14%, or an AIDS-related clinical condition.

Within these three years a second sample was also taken from the slower progressors, but they were only recruited into the study as 'slow progressors' when, at the three year mark, they had not yet developed AIDS. On this second sample, a second battery of tests was performed that, measured HIV subtype, the levels of a whole panel of different cytokines (signalling proteins of the immune system), sexually transmitted infections, drug resistance, and also the so-called tropism of their virus.

This latter is a measure of the particular immune-system cells the virus likes to infect and may have profound implications for how fast people progress to an AIDS diagnosis – see more below.

On the basis of this three-year test, fifty-two of the patients were included in the study as rapid progressors, and 21 slow progressors. Both groups were recruited consecutively – i.e. every diagnosed patient within a certain timeframe who fit the criteria was included – but separately. The rapid progressors were less common than other patients, so their group took longer to recruit, though the study was not set up to calculate what proportion of the entire Pedro Kouri clinic were fast progressors.

If the second test showed that patients needed to start ART, as was the case with all fast progressors, then they were started on it at this point (Cuba’s HIV treatment guidelines at that time still had 200 cells/mm³ as the minimum recommended threshold for ART).

These two groups were compared with a third group of 22 people who met the same criteria for AIDS as the fast progressors but who had all had HIV for more than ten years, and where there had been at least six years between HIV and AIDS diagnoses.

Test results at diagnosis

The 95 patients were all similar in terms of ethnicity, gender (80% male), age at diagnosis (average 32) and sexual behaviour. The fast-progressors were less likely to be gay men and considerably less likely to report anal sex.

The CD4 count at diagnosis was 577 and 522 cells/mm³ in the chronically infected and slow-progressor groups, but it was only 276 cell/mm³ in the fast-progressors, indicating that their immune systems had suffered a particularly strong attack from HIV during acute infection.

There was a considerable difference in the first viral load measurement taken after diagnosis. In the fast-progressors the median viral load was 63,000 copies/ml, in the long-term infected it was 1600 copies/ml and in the recently diagnosed slow-progressors it was undetectable, i.e. below 50 copies/ml.

This latter finding is really interesting in itself: remember, none of these people was on ART. Have the researchers discovered a group of viral controllers?

Lead researcher Professor Anne-Mieke Vandamme of the Rega Institute in Leuven, Belgium told Aidsmap: “You are right to find that peculiar; we had the same concern. Half of the [slow-progressor] patients had an undetectable viral load, while half had a higher viral load” (up to a few thousand copies/ml).

“We could not find any reason for that; we double checked the data and that is just how it was in our 21 consecutively recruited patients. We think it might be due to some viral subtypes being unquantifiable with the test we used."

In the whole group of 95 patients, 25 had subtype B, six subtype C, 29 B/G recombinants and other mixtures, seven a recombinant called CRF18_cpx, and nine the CRF19_cpx recombinant. All those nine were recently diagnosed and fast progressors.

There was a much bigger viral load at diagnosis in patients who turned out to have the CRF19_cpx subtype; in the six patients with this subtype whose viral load was measured at diagnosis, their viral load averaged over a million copies/ml.

Testing three years after infection

The 73 patients who were recently diagnosed were then evaluated prospectively, meaning that their viral load, CD4 count and CD4 percentage were measured again at the three-year mark. The chronically infected were evaluated so that their viral load and CD4 count and percentage at the test nearest to their reaching an AIDS-defining criterion (i.e. around the year 2008) was included.

At this point of AIDS diagnosis there was no difference in the viral load between the fast-progressors and the chronically infected, who had viral loads averaging 50,000 copies/ml; the viral load in patients with CRF19-cpx was not significantly higher either. The slow progressors had a viral load ten times lower – 5000 copies/ml – but no longer undetectable, and this was not statistically different from other patients. The average CD4 count in the fast progressors at this time was 189 cells/mm³, in the chronically infected 212 cells/mm³ and in the slow progressors 501 cells/mm³.

The researchers measured a number of different cytokines – soluble immune-modulating molecules indicative of inflammation. They found very low levels of most of them in the slow-progressor group, and high levels of them in both groups with AIDS-defining criteria. The only cytokine where there was a significant difference between the fast-progressor and chronically infected group was that the former had higher levels of the cytokine called RANTES. This is particularly interesting as RANTES is a natural blocker of HIV infection, upon which the CCR5 inhibitors such as maraviroc were based.

The researchers found that oral candidiasis was the only one of a range of sexually (or sometimes-sexually) transmitted infections associated with being a fast progressor – while 23% of this group had oral candidiasis, none of the chronically infected with AIDS had it and less than 5% of the slow progressors.

As we already said, HIV viral subtype varied by group. Slow progressors either had subtype B (42%), one of several different B/G recombinant viruses (37%) or a recombinant called CRF18_cpx (11%) – in the remainder, subtype could not be determined. In the chronically infected with AIDS, the proportions were roughly similar except for a few subtype C cases too. In the fast progressors, 17% of them were, as already noted, the only people who had the CRF19_cpx virus. In addition, far fewer had subtype B – again, only 17%. About a third of them had the B/G recombinant virus and 10% subtype C. So while this hardly indicates that the CRF19 virus was the cause of all fast progression, it might indicate that its recent expansion into the population might be linked to an apparent increase in fast progression.

They also mentioned the tropism of the virus. Different varieties of HIV prefer different immune-system cells to infect, and this so-called tropism measures whether HIV prefers to link to a cell receptor called CCR5, a different one called CXCR4, or either one. Tropism is complex – see this essay for a good explanation. CCR5-tropic viruses are easily transmitted and proliferate well, but cause relatively limited immune-system damage and slow progression, whereas CXCR4-tropic viruses (and mixed-tropism ones) are rarely transmitted, replicate less well, but cause devastating immune damage. Part of the causation of AIDS is that HIV tends to mutate into versions that can use the CXCR4 receptor within the body and once a certain proportion of virus becomes CXCR4-tropic, very fast immune decline results.

In this study, none of the slow progressors had virus that exclusively used CXCR4; this compared with 16% of fast progressors and 19% with chronic AIDS. Conversely, 71% of slow progressors had virus that exclusively used CCR5, compared with 50% of the chronically infected and 46% of fast progressors.

These tropism differences were not statistically significant. However when CRF19-infected patients were compared to the other fast progressors, the difference in coreceptor use was highly significant; most of CRF19_spx viruses were using exclusively CXCR4, even though the researchers found that the part of the CRF19_cpx virus genome responsible for binding to cells was from HIV subtype A, which is not associated with rapid switching to CXCR4 tropism.

However, while viral load at the three-year sampling mark was not significantly different between the groups, if the production of just one component of HIV, its protease protein, was measured, the researchers found that patients with CRF19_cpx had protease fitness levels nearly a hundred times higher than others. The gene for protease in the recombinant virus does come from the more virulent subtype D. Protease is responsible for part of the process of manufacturing new viral particles, so this could explain CRF19_cpx’s higher viral load.

Possible conclusions

The researchers' hypothesis is this: CRF19_cpx initially replicates at very high levels – as that six-million-copy viral load shows. At this time it will be CCR5-tropic and infecting cells with that receptor. But the body will manufacture huge amounts of the protein RANTES as a defence against cellular invasion – as shown by its higher levels in people with CRF19_cpx.

This blocks CRF19_cpx from infecting any more cells via this route. So it is forced early on to change to a CXCR4-tropic virus. This then reproduces more slowly than before – hence the lack of difference in viral load at the three-year mark – but is much more pathogenic.

That is the hypothesis, anyway. But as the researchers point out, they did not measure the entire gene sequence of CRF19_cpx, so it could have other characteristics they missed.

Nonetheless – even though it is still very much a minority variant in a minority of people – this is the first time a variety of HIV has been found to be exclusively associated with people who progress rapidly to AIDS. There is no evidence CRF19_cpx is more susceptible to becoming drug resistant, and ART should work with it just as well. But this is a reminder of HIV’s genetic adaptability, and that the timespan between infection between HIV infection and symptomatic AIDS may be shorter than we think in some people.