European HIV Drug Resistance Workshop: epidemiology

The European HIV Drug Resistance Workshop took place in Monte Carlo, Monaco, March 29^th - 31^st 2006.

Epidemiology of HIV drug resistance

Is the prevalence of HIV drug resistance increasing over time, either as acquired or as transmitted resistant strains? The answer rather intriguingly appears to be no. For resistance acquired on treatment, there is strong evidence that the profile of resistance mutations sustained over time may be explained by the historical practice of sub-optimal sequential therapies. Similarly, for transmission of resistance, the evidence is less consistent but it would appear that we may have reached an epidemiologic plateau with levels of transmission of resistant virus stabilising in most European countries.

Resistance acquired on treatment

Epidemiological studies involving observational treatment cohorts confirm that higher prevalences of transmission can be correlated with a history of mono and dual sequential therapies. Dominique Costagliola from INSERM Paris confirms this. In a cohort of 498 heavily treatment-experienced patients with detectable viral load (i.e. above 1000 copies/ml) including 12% of patients with exposure to enfuvirtide (T-20, Fuzeon), “the observed pattern of resistance is still mainly explained by initiation of therapy with single or dual therapy”, she told the meeting.

In this group of patients, a median four mutations (inter-quartile range 1-6) relating to nucleoside reverse transcriptase inhibitors (NRTI) were detected, four (range 2-8) protease inhibitor (PI) mutations and surprisingly, none (range 0-1) associated with non-nucleoside reverse transcriptase inhibitors (NNRTI). M46IL, L90M, V82AFLST and I84V were the key primary PI mutations detected in 44%, 31%, 27% and 19% of patients respectively. Of the NRTI-associated mutations, the M184V was the most prevalent (58%) mutation, followed by T215YF (51%), followed by the thymidine analogue mutations (TAMs) M41L (50%), D67N (38%), L210W (33%), K219QE (17%) and K70R (16%). The most frequently reported NNRTI mutation was K103N. This study reported that 19% of patients had virus that showed no sensitivity to drugs from at least two antiretroviral classes, and 88% had resistance to at least one antiretroviral drug, of whom 77% had resistance to at least one NRTI, 66% to at least one PI, 50% to NNRTIs and 7% to T-20. Of concern is that the majority of these primary mutations confer cross-class resistance, making future salvage regimens significantly more challenging.

Transmission of drug-resistant HIV

The experience across Europe of drug resistance transmitted strains, whilst still a cause for concern, does confirm a stable prevalence of around 9% in newly diagnosed individuals. Large-scale, pan-European representative surveillance programmes supported by the European Union (EU), such as SPREAD and its prolific off-springs, WATCH, CAPTURE and CATCH have strengthened our knowledge of the impact of HIV treatments through the pooling of national public health data-sets. From a prospective study of 1083 newly-diagnosed individuals with HIV from 17 European countries, infected through a variety of transmission routes, comprising almost equal heterosexual and homosexual exposure (42% versus 44% respectively), this sub-study analyses noted some interesting developments.

Seventy seven per cent of all new infections in this study were male, 34% were from migrant communities and 13 different subtypes and circulating recombinant forms (CRFs) were detected: B (66%), A (9%), C (9%), G (4%), CRF02_AG (4%). Of these, 9.1% were infected with drug-resistant HIV. The prevalence of resistance associated to each class was: 5.4% (NRTI), 3% (PI), and 2.6% (NNRTI). From a public health and a prescribing perspective three findings from this studies are worthy of note:

• Only 1% of this group were infected with dual-class resistance.
• Patients originating from high-prevalence countries had a lower risk of being infected with drug-resistant strains.
• Disease progression was observed to be similar in patients infected with drug-resistant strains as compared to those with wild-type transmission.

A somewhat higher frequency of 13% of drug resistance transmission was reported in a German study of newly diagnosed individuals. The authors explain that the criteria for study selection can determine higher or lower levels of transmission in particular the inclusion of seroconverters compared to those with unknown date of infection. In a demographic study, the research revealed that those infected through heterosexual transmission (20%) were highly represented in the group with an unknown date of infection and this was true also of those infected with non-B subtypes. Both these groups differed from the seroconversion group who were more likely to report their infection earlier and had a known date of infection (most probably because it involved a recent exposure to HIV). The transmission rate of resistance in seroconverters was similar to the rate in newly-diagnosed patients (14% versus 13% respectively). This study raises an interesting issue of defining criteria for measuring time of infection against a profile of transmitted resistance. It also confirms that there is indeed a difference in testing and reporting behaviour between MSM and heterosexual groups. Its own conclusion, however, does not support a significant differential between these two groups on the impact of resistance transmission.

Predicting transmission of drug resistant HIV

A dynamic model to explain the phenomenon of transmission and prediction of viral evolution was reported at this meeting by Sally Blower, a biological mathematician at the University of California, Los Angeles (UCLA). Dr Blower began by describing a modelling technique that conceptualises epidemic dynamics. The model designed to differentiate between drug-susceptible and drug-resistant pathogens, defines the evolution of resistance as fundamentally dynamic with potential impact at a population level. The calculations involving 33 equations allow for individual as well as interacting variables such as viral fitness, transmitted, acquired or “amplified resistance”, which they describe as the consequence of multi-drug resistance (MDR) accumulated as a function of sequential therapies. An interesting feature of this model is the construction of possible evolutionary trajectories based on two distinct scenarios; if drug-resistant strains remain as fit as wild-type (WT) or if resistant virus is less fit than WT.

Assuming that drug-resistant strains are as equally fit as WT, the model suggests that the WT scenario will mostly likely persist. However, if the resistant virus is less fit, then there could be viral equilibrium, that is, the equal possibility that the WT and drug-resistant strains could co-exist. The model further suggests that HIV epidemics should be viewed as an “amplification cascade of drug-resistant strains”, meaning that the fitness of the virus is the driving force in population resistance evolution. Most importantly, the results from this study would suggest that the impact of drug-resistance strains has been to decrease the severity of HIV epidemics. The strength of Dr Blower’s presentation is reinforced by similar modelling strategies presented by Dr Blower at last year’s workshop from which she predicted the transmission of drug-resistant strains would stabilise around 9-10% in the developed world.

A number of albeit theoretical concerns were highlighted by delegates in response to Dr Blower’s presentation: for example, if one drug-resistant strain were to be selected, which viral strain would that be? Secondly, since there is evidence of clustering of infection with transmitters likely to repeatedly transmit how would behavioural characteristics be addressed in such modelling studies? Dr Blower confirmed that behaviour such as non-adherence to treatment has already been included but for the analysis of clusters of infection, a stochastic model in high-risk groups may be more appropriate. Other differences that could impact on the outcomes of modelling strategies were highlighted by Mounir Ait-Khalid from GlaxoSmithKline (GSK) such as inequalities in treatment, lack of optimised therapy or inadequate clinical management support for example in resource-poor countries. In such circumstances Dr Blower noted, the predictions would be even more calamitous and may compound the epidemic.

Overview

In an interesting overview on the epidemiology of HIV resistance in Europe confirming much of the above, Luc Perrin from Geneva highlighted additional methodological caveats to collating resistance data within and across national boundaries. There are as yet no standard mechanisms for reporting either transmitted or acquired resistance, he asserted, so direct comparisons may not always be possible. The techniques by which these data are generated, largely from academic or clinical cohorts are not always consistent with the criteria for data collection appropriate for epidemiological studies. National data collated for pan-European analyses from research networks such as SPREAD and CATCH will improve this situation. SPREAD for example uniformly collects data for national HIV-epidemics and employs standardised methods for data analyses. By systematically stratifying cross-country information, these projects are already delivering meaningful results from sub-set analysis.

Dr Perrin went on to support the premise that levels of transmitted drug resistance are stabilising across Europe and that there may be little variability observed between European countries. He also confirmed the hypothesis that single mutations may be rapidly lost, but MDR virus could persist for much longer; he described a patient from his own clinic, a 35 year old MSM infected with MDR whose virus had evolved to lose only the single M184V mutation. For patients with non-B viral subtypes he noted, they were more likely to report NRTI than PI-associated resistance, perhaps still reflecting the limited access or time on treatment. Finally, Dr Perrin raised a public health alert, noting that 30% of new infections continue to be transmitted by individuals with primary HIV infection with implications for resourcing stronger testing and health education campaigns.

Which mutations are most likely to be transmitted?

Evidence of consecutive transmission of two-class resistance in two untreated patients came from Anne-Mieke Vandamme’s team in Leuven, Belgium. Using population sequencing and phylogenetic analyses, they reported a case of an individual diagnosed in 2001 with clade B virus with mutations at RT 210W, 215Y, PR 20R, 36I, 54V, 71V, 88D and 90M. In 2005 his partner was subsequently diagnosed and his viral strain displayed the same profile at RT with the 210W but as previously reported in other studies, a reversion at 215 to 215S and a similar profile in the protease region but with the absence of mutation at 90M. Phylogenetic testing supported the linkage between these two individuals and revealed three distinctive clusters dispersed with variants and highlighting a mixture detected at position 90 (ML) at a later time from the donor. The significance of this finding is three-fold. It represents the first report of a consecutive transmission-chain of dual-class resistance in untreated patients, secondly, it reinforces the long-term persistence of transmitted dual-resistance and finally, it provides evidence of a dynamic acquisition and loss of primary PI mutations in the absence of therapy.