February 8, 2006
The prevalence of multidrug-resistant HIV (MDR-HIV) in the United Kingdom is unlikely to increase within the next five years -- at least among gay, HIV-infected men with a moderate viral load, according to a theoretical model constructed by a British researcher. The model suggests, however, that the occurrence of any type of resistance is likely to increase.
It's very unusual to see single-author abstracts at the Conference on Retroviruses and Opportunistic Infections. This one, by Andrew Phillips of University College, London, uses sophisticated statistical methods to model the frequency of multidrug resistance over time among gay men in the United Kingdom.
The presence of MDR-HIV in the population poses challenges not only for the clinical management of people with MDR-HIV, but also because there is a risk that people with MDR-HIV may transmit their MDR-HIV to others. The likelihood and type of primary or transmitted drug resistance has implications for baseline resistance testing of treatment-naive patients, as well as the selection of initial antiretroviral regimens.
Soon after the advent of highly active antiretroviral therapy (HAART), an alarming rise in primary MDR-HIV was observed in developed countries. But studies conducted between 2000 and 2005 have generally suggested a stable or even declining prevalence of MDR-HIV among new cases of infection, typically at or below 2%.1 This apparent trend reversal raises an intriguing two-part question: How big a problem is MDR-HIV today, and is it likely to become a much bigger one in the future?
Dr. Phillips built a stochastic computer simulation model of the proportion of the "infectious pool" of gay men in the United Kingdom who, at any one time, have MDR-HIV. Phillips' "infectious pool" consists of all those at risk for transmitting HIV to others; based on published research, Phillips defines this group as those with an HIV viral load greater than 1,000 copies/mL.2
The logic behind Phillips' simulation model is as follows: The chance of any individual man becoming sexually infected with MDR-HIV depends on the relative infectivity of the MDR strains, and on the man's probability of having sex with partners who harbor MDR-HIV. This in turn depends on the proportion of people at risk for transmitting HIV who already have MDR-HIV, and any differences in sexual risk behavior between them and those who do not have MDR-HIV.
Many variables influence the probability of coming into contact with a partner who has MDR-HIV. There are, for example, a large number of people with HIV who are undiagnosed -- although in contrast to the United States, this number is estimated to be falling rapidly in the United Kingdom. Of those who are known to have HIV, most are not currently on HAART, particularly as the treatment pendulum has swung toward delaying initiation of therapy. But because treatment has improved over the years, an increasing proportion of those who are on treatment have a viral load under 1,000, and are not likely to be infectious. Others, particularly those with poor adherence to antiretrovirals, may be much more so.
Dr. Phillips had to separately model these pools of people -- undiagnosed, antiretroviral-therapy naive, currently on antiretroviral therapy, or previously treated but now off antiretroviral therapy -- and then model their interactions with one another. His model was both retroactive and forward-reaching: He went back prior to 1990, then built the model going forward through 2010, assigning probabilities for many different events (e.g., time until switching antiretroviral therapy regimen) based on the best available published data. The key assumptions were as follows:
Phillips tested the model by asking it to predict a number of epidemiological parameters, such as the national median CD4+ cell count at diagnosis and the prevalence of different resistance mutations. He found that his model fit the actual data very well.
The model was then used to reconstruct changes in the infectious pool of gay men since the introduction of antiretrovirals in 1989. It found that the proportion of people in the infectious pool who are on antiretroviral therapy increased until 1996; then, with the advent of HAART, the proportion decreased again, since HAART has been so successful at reducing viral load that it effectively removes most treated patients from the infectious pool (i.e., drives viral load to less than 1,000 copies/mL). This finding has important implications for transmitted resistance, since it suggests that even if many patients develop resistance, most have too little circulating virus to transmit it.
The model's main -- and most surprising -- conclusion is that the prevalence of MDR-HIV in the infectious pool is unlikely to increase, and in fact will probably decrease, over the next few years. The model predicts a gradual drop from a prevalence of 4.7% today to only 1.5% in 2010. The model shows that the fraction of people in the infectious pool carrying any resistance mutations climbed quickly after the introduction of antiretrovirals, plateaued after 1994 at about 30% to 35%, and remained there through 2010. But triple-class MDR first appears in the infectious pool around 1998, climbs sharply to about 13% by 2002, and then declines gradually through 2010. If the model is varied slightly to assume that transmitted mutations disappear three months after acute infection, the proportion with any mutations actually falls considerably after peaking in the late '90s, while the curve for MDR-HIV is hardly affected at all.
Since any model like this critically depends on a wide range of input statistics and assumptions, Phillips performed a sensitivity analysis, varying some of the key assumptions to see if they change the model's predictions. Most variations make very little difference: for example, the annual rate of new infections changes, or the CD4+ cell count threshold at which antiretroviral therapy is typically begun is moved down to 200 cells/mm3, or the viral load threshold for infectivity is assumed to be less than 50 copies/mL or less than 5,000 copies/mL, rather than less than 1,000, copies/mL. The only two scenarios in Phillips' model under which the proportion of MDR-HIV would increase modestly by 2010 are if:
However, both of these occurrences seem unlikely.
There are a couple of other ways in which the model could be proven wrong, however. For instance, the model assumes that in situations of regimen failure, new regimens will be chosen based on resistance testing. If they are not, the result could be many more patients with resistant virus and viral loads not suppressed to less than 1,000.
Finally, the one situation in which we might see the proportion of MDR-HIV drastically increase is if the rate of transmitted resistance matched the estimated proportion of MDR-HIV in the infectious pool. In other words, we see today much less primary transmitted MDR-HIV than would be predicted based on its amount in the infectious pool. There may be several possible reasons for this, among them:
Were this situation to change, however, the proportion of MDR-HIV could sharply rise.
The Bottom Line
The main conclusion from Phillips' study is that the prevalence of MDR-HIV in the infectious pool of gay men in the United Kingdom is unlikely to increase over the next five years. This is driven mostly by the fact that the most infectious people are not on antiretroviral therapy, while those who are on therapy tend to be successfully suppressed to a low viral load.
However, clinicians should remain mindful of a few caveats:
While we do not yet have to dread an immediately looming epidemic of MDR-HIV, Phillips' study should not falsely reassure us about the ongoing problem of transmitted resistance -- a problem that complicates a clinician's decision about baseline resistance testing and initial therapeutic regimens.
|Please note: Knowledge about HIV changes rapidly. Note the date of this summary's publication, and before treating patients or employing any therapies described in these materials, verify all information independently. If you are a patient, please consult a doctor or other medical professional before acting on any of the information presented in this summary. For a complete listing of our most recent conference coverage, click here.|
The content on this page is free of advertiser influence and was produced by our editorial team. See our content and advertising policies.