GLOBALART · Improved prediction of HIV drug resistance in different viral subtypes by bioinformatic analysis of genetic and clinical data

Improved prediction of HIV drug resistance in different viral subtypes by bioinformatic analysis of genetic and clinical data

The overall objective of the project GlobalART #246599: Improved prediction of HIV drug resistance in different viral subtypes by bioinformatic analysis of genetic and clinical data was to enhance the understanding of drug resistance development and phylogenetic evolution among non-B HIV subtypes, especially HIV1-C which is predominant in large parts of Africa and the Indian subcontinent. The main outcome of the project has been the analysis of clinical cohorts from Vietnam, India, Bangladesh and Ethiopia, four countries that have all seen a substantial scale-up of ART relatively recently but that differ in many other aspects of the HIV epidemic. In Vietnam, the circulating recombinant form CRF01_AE is the predominant type while HIV1-C constitutes the vast majority of cases in Ethiopia as well as India. Subtype C is also the most common subtype in Bangladesh, but a wider variety of different subtypes and recombinant forms are found here due to the continuous introduction of strains from abroad via migrant workers. In Vietnam, India and Ethiopia, most of the included strains fell into one or a few country-specific clades that have spread widely within the country, while the Bangladeshi sequences were much more diverse and found throughout the phylogeny. For Vietnam, one large clade contained 60 out of 63 samples from this study as well as most of the included reference strains from Vietnam, and the time of the most recent common ancestor (tMRCA) of these sequences was calculated to around 1990, which is approximately 4 years earlier than previous estimates of HIV introduction to northern Vietnam. An additional small clade represented a separate introduction around 1997. The CRF01_AE epidemic in Vietnam started among drug users in the south and then appears to have spread via this route to the north around 1990. Since then the epidemic has gone from concentrated to mature and in our phylogeny, virus transmitted via injecting drug use and sexual transmission were completely intermixed. In India, which has the third largest burden of HIV worldwide, sexual transmission is predominant and also constituted 90% of the 168 samples from four regions included in our study. Our analysis of three different genes revealed that the Indian HIV1-C epidemic has evolved largely independently from HIV1-C elsewhere. It was introduced in the early 1970’s (the tMRCA was calculated to 1971, which makes it one of the oldest epidemics outside Africa) and after an initial slow start the epidemic took off around 1980 with an exponential growth phase. The first clinical case was detected in 1986 and by the time interventions started, around 1990, the epidemic had already grown out of control. However, since 1990, the growth went in to a stationary phase and in recent years there has been a decline in the number of new infections. In Ethiopia, there is yet another story. It is the country where subtype C was first discovered and also the origin of the sub-subtype C’. The analysis is still in progress and a final tMRCA has not been calculated, but the preliminary results show distinct clusters (C and C’) indicating a high level of transmission within the country, and also a high level of recombination between these sub-subtypes. The overall HIV prevalence is high and it is known from before that HIV has been present in Ethiopia since at least the mid 1960’s, so like in India there has been plenty of time for the development of country-specific characteristics. Bangladesh is different from the other countries in that there is very little evidence of extensive transmission within the country. Except from two local clusters of HIV-1C among people who inject drugs in Dhaka, only very small clusters of 2-3 people could be detected. This could partly be caused by sampling error as it is very difficult to reach populations outside the traditional risk groups, but from the available data it seems like there is no such thing as a Bangladesh-specific variant of subtype C. Most strains do, however, belong to a large regional cluster, where cross-border transmission frequently occurs between nearby countries. In addition, different strains of HIV1-C have been introduced again and again from further afield. In spite of high levels of transactional sex and high risk behaviors among key populations, Bangladesh remains a low-prevalence country for HIV, since the concentrated epidemic among drug users in the capital has not managed to spread to the sex industry or general population. Transmitted drug resistance mutations (TDRM) in the pol-gene were investigated for the Vietnamese and Ethiopian cohorts, where the prevalence of any TDRM was 6.3% and 3.9% respectively. In Vietnam TDRMs were detected in 4/63 samples, and no sequence contained more than one. The detected mutations were L210W, L74I, V75M (NRTI) and Y181C (NNRTI). Among the Ethiopian sequences, 5/127 harbored mutations. Again, no sample had more than one mutation and the TDRMs were M46L (PI), L74V and L210W (NRTI); L100I and K101E (NNRTI). These results include only snapshots of drug resistance mutations present in supposedly treatment-naive populations about which very limited data was available. In order to study trends in resistance over time in a more systematic manner, we took advantage of the unique database InfCareHIV, which includes nearly all known HIV-patients in Sweden since the beginning of the epidemic, including all treatment details and resistance test results. The analysis included all ART-experienced patients in Sweden between 1997 and 2011 (N=6537), and showed a dramatic decline in the level of resistance towards the three major drug classes NRTI, NNRTI and PI over time. The decline was most pronounced between 2003 and 2007, but even after this the levels have gone down apart from a recent small, but worrying, increase in NNRTI resistance. Finally, the sequences from the Ethiopian and Vietnamese cohorts were also included in a large multi-center collaborative study for differentiation between recent (<1 year) and chronic (> 1 year) infection based on sequence ambiguity level. The optimal cut-off was calculated to 0.47% and this might be a useful tool for estimating the duration of infection prior to the sampling date for more accurate tMRCA calculations. However, it requires access to original chromatograms and can therefore rarely be applied to reference sequences retrieved from data base sources. The sequencing and analysis of samples from these diverse epidemics is an important step towards better understanding of the molecular epidemiology and drug resistance development of HIV in different parts of the world.

Data: CORDIS, © European Union