 |
|
|
| |
Projection to 2010 of trends in multi-drug resistant HIV in infectious gay men in the UK
|
| |
| |
Andrew Phillips, UCL, London
13th CROI, feb 2006, Denver
Poster 649
Background
Prevalence of transmitted resistance in ART-naive individuals has tended to remain stable or fall in recent years in North America and Europe, but in the UK a rise has been reported. Prevalence of MDRHIV in new infections has remained low at below 2%. The future trend will likely largely depend on changes in prevalence of MDR-HIV in the infectious pool, but such trends are not straightforward to predict. For example, the extent to which transmitted resistance persists as majority virus is uncertain, as is the potential role of superinfection of already infected individuals with resistant strains.
Future introduction of new agents active against strains of HIV resistant to existing drugs must also be considered. To make projections that are informed by the best state of knowledge of all the relevant underlying processes and to evaluate the impact of different factors, a model is required. Here, we use a stochastic computer simulation model of HIV (HIV Synthesis) to re-construct the
UK epidemic among gay men, and project to 2010.
Phillips' Conclusion
The projections from our model suggest that the prevalence of MDR-HIV
in the infectious pool in gay men in the UK is unlikely to increase
over the next 5 years, essentially due to the low and decreasing
proportion of the infectious pool who are on ART.
Note from Jules: for this analysis Philips assumes infectious virus is defined at viral load >1000 c/mo, and then it appears from what I see in Table 2 he projects 9.1% of Projected % with MDR-HIV in 2010 with >50 c/ml and 4.8% will be >5000 cp/ml. If I got this correctly I don't this this applies in the USA. Philips says "Our model predicts that a substantial proportion of patients with MDR-HIV in 2010 will have viral load <1000 copies/mL". As I can see he also assumes less HIV+ people will be undiagnosed, but -- in the chart immediately below he assumes most HIV-diagnosed people will be off ART. He also reports the observed % with resistance I presume currently by 6 years on HAART, overall 30%, and 3-4% with triple-class resistance. It appears to me that in Table 2 that he assumes 5.1% are "Projected % with MDR-HIV in 2010" among transmitted mutations. I think he presumes resistance mutations will not persist (I question how he evaluates that mutations won't persist, by which assay, a standard or sensitive assay) & further presumes this will reduce possibility for transmission of mutations. I question these assumptions, or at least his presumption that these are true. In addition risk for transmission appears to be perhaps greatest soon after infection when viral load may be at its highest level, which would increase risk for transmission & if resistance is present, transmission of resistance.
Results
The reconstructions suggest that the size of the infectious pool of gay men grew before 1995, but has declined since (Figure 2) (although information on
new infections is limited so the number of new infections in the 2000-2005
period could be under-estimated). Crucially, the proportion of the pool who
are on ART has decreased, due to the increasing success of ART in reducing
viral load. The proportion of the infectious pool carrying MDR-HIV in their
majority virus, which had grown after the initial use of the NNRTI and PI
classes to around 13% in 2002, shows a decline after that time (Figure 3).
This pattern is seen in both the base model, in which it is assumed that
transmitted resistance mutations persist indefinitely in majority virus (Figure
3(a)) and in an adapted model in which such mutations are assumed not to
persist in majority virus beyond 3 months after infection (in the absence of
ART which selects for the mutations; Figure 3(b)). The percentage with at
least one resistance mutation to any of the three classes reached a value
between 30-35% in 1994 and remains at that level when resistance mutations
are assumed to persist (Figure 3(a)), but shows a decrease when transmitted
mutations are assumed not to persist (Figure 3(b)). Of those in the infectious
pool with MDR-HIV in majority virus, the model suggests 94% were on ART in
2005 and 88% will be in 2010. Further, less than 3% of those with MDR-HIV
were infected with MDR-HIV, the remainder all acquiring the mutations on
ART.
We re-evaluated the prevalence of MDR-HIV in 2010 and the change between
2005-2010 (4.7% and -3.2%, respectively, in the base model; Figure 3), after
varying assumptions in the model (Table 2). This resulted in a poorer fit of
one or more aspects of the model to observed data so they are less plausible
scenarios. Nevertheless, the projected stable or decreasing prevalence of
MDR-HIV held in each case, except the unlikely scenario in which adherence
levels decrease between 2005-2010.
Comments by Philips
Our predictions of decreasing prevalence of MDR-HIV in the infectious pool do not appear to rely on the anticipated development of new drugs; the prevalence is predicted to be relatively stable even if no such new drugs become available. The model assumes that regimens from 2003 owards are compiled based on resistance testing. If such tests are not done or ignored then some patients with
three active drugs available might remain with viral load >1000 copies/mL, resulting in us under-estimating the prevalence of resistance (and possibly MDR-HIV) in the infectious pool.
Our findings relate to the infectious pool and should not necessarily be taken to imply that fewer people will harbour MDR-HIV. Our model predicts that a substantial proportion of patients with MDR-HIV in 2010 will have viral load <1000 copies/mL.
The prevalence of MDR-HIV that has been observed in newly infected patients in the Europe and North America is considerably lower than the prevalence in the infectious pool of up to 13% we estimated. There are several possible reasons for this lack of correspondence between the prevalence of resistance in the infectious pool and that in newly infected persons. First, HIV with resistance mutations may be less transmissible than virus without, and this may depend on which mutations are present. Second, there may be differences in risky
sexual behaviour between those with resistance mutations and those without. Third, mutations in virus transmitted may revert back or be outgrown by other sub-species present in the infecting inoculum and hence not persist in majority virus in the new host. Fourth, infectiousness probably depends on the actual viral load level and there may be differences in viral load between those with those with resistance mutations and those without. Further, ART may reduce the
risk of transmission independently of viral load.
In our model we assumed the prevalence of resistance at infection for a person infected at a given time was as observed from studies carried out at that time (1-2%). When in a sensitivity analysis we instead assumed that the prevalence of mutations for new infections was exactly that in the infectious pool at the time, there was a much higher prevalence for 2010 (19.6% vs. 4.7% in base model) but still no increase for 2005-2010. Despite the above reasons for discrepancies between the prevalence of resistance in the infectious pool and in new infections, the direction of trends in the prevalence of resistance in new infections will likely follow those in the infectious pool, unless the factors influencing the lack of correspondence are themselves altering appreciably with time. We need to increase our understanding of the factors responsible for the above discrepancies and to examine whether these are changing over time.
Brief Description of the HIV Synthesis Model
-- Dependencies in the model are illustrated in Figure 1 below.
-- Attempts to reconstruct the population of people infected with HIV who live(d) in the UK to 2010
-- Tracks the infection and effect of ART in 3 month periods.
-- Fixed Variables:-Risk group, gender, HBV status, HCV status, fixed tendency to adhere to ART, transmitted mutations
-- Variables tracked over time:- Calendar time, in UK, diagnosed, viral load / max viral load, CD4 count / min CD4 count, specific drugs (current use / previous / stopped due to toxicity or failure / resistance), current toxicity, current adherence to ART, current risk of mutation arising and surviving, presence of resistance
mutations (majority virus / anywhere) (nucleosides # TAMS, 65, 74, 184, NNRTI mutation yes/no, PR 30, 33, 46, 48, 50 [V & L], 82, 84, 90), risk of AIDS / CDC B / death from HIV / death from non-HIV
-- Model fits relatively closely to a range of observed surveillance and clinical data (Table 1).
Initial Assumptions
-- Transmitted mutations persist in majority virus
-- Infectious defined at viral load > 1000 cps/mL
-- 1200 new infections per year
-- Diagnosis rates will remain constant to 2010
-- No acquisition of resistance by superinfection
-- MDR means at least one resistance mutation for each class
|
|
| |
| |
|
|
|
|
|
