Our results based on a phylogeographic study of a large number of sequences sampled from 16 countries in Europe and Israel provided important clues about HIV-1 subtype B spatial diffusion across Europe. Notably according to the findings of phylogenetic analyses, viral lineages sampled from all countries except Poland, Austria, Luxembourg and Portugal, showed low levels of grouping according to the geographic origin. For most countries, we identified small networks of local transmission, but to a different extent in each country, along with sequences showing no particular geographic clustering. Such a pattern suggests that the subtype B epidemic in most countries was introduced by several founders, some of them causing subsequent local dispersal, while others lead to dead end infections. We should note that under the conditions of our study, we cannot estimate the percentage of infections occurring within local transmission networks, since we don't have sufficient covering per country.
Poland's epidemic dispersal is quite different. Based on the high number of viral lineages coalescing to a common origin within the country, we suggest that the epidemic is the result of a few migrations of the virus successfully spreading within the local population. This pattern is consistent with a main viral dispersal through IDU networks associated with extensive local epidemics. Monophyletic HIV epidemics have been described among IDUs for other European countries, as well, including also non-B subtypes strains [12, 13, 26, 27, 42–45].
For Austria, Poland and Luxembourg we identified more extensive local transmission networks than for the other European countries. Similarly HIV local networks have been described for Canada, Greece and the UK [46–50].
According to the epidemiological data, most of the subtype B infections newly diagnosed during 2002–2004, occurred locally. The geographic distribution by means of the viral evolutionary history, the phylogeography, on the other hand, revealed high levels of viral dispersal. Both observations are not necessarily in contradiction. Rather, they suggest that most of the migration identified through phylogeography may date from earlier in the transmission chain, and that the pre-existing complexity of the epidemic (multiple sources of introduction from diverse localities) is the main reason for the continuous extensive geographical dispersal across the viral phylogeny. Particularly if there are multiple founders, subsequent infections will be dispersed, across the viral phylogeny, according to the geographic origin of the founders' source. This is in accordance with previous findings about multiple introductions of the subtype B infection through sexual intercourses or IDU across Europe [13, 47, 49, 51, 52].
In addition to epidemic dispersal patterns, our study provided important findings about HIV-1 subtype B major sources and targets for migratory events, as well as localities with bidirectional viral dispersion.
In particular, Greece, Portugal, and Spain attract many travellers and tourists, especially from Central Europe, thus suggesting that HIV dispersal from Southern to Central Europe may, at least in part, occur by travellers infected during their stay in Southern Europe http://epp.eurostat.ec.europa.eu/portal/page/portal/product_details/publication?p_product_code=KS-DS-08-001.
For countries classified among the HIV migratory targets (Austria, Belgium and Luxembourg) the epidemic was mainly imported due to the high HIV mobility to these countries. According to the epidemiological information, the highest rate of imported infections from other European countries occurs in Luxembourg. Moreover, the fact that 13% of the population of Luxembourg is of Portuguese origin provides a plausible explanation for the migratory pathway from Portugal http://www.migrationinformation.org/datahub. Another significant pathway was tracked from Italy to Austria, in accordance with the high influx from Italy during recent years http://www.migrationinformation.org/datahub. Denmark provided migratory target from another Scandinavian country (Sweden) but also from Spain. This is in accordance with epidemiological findings that a percentage of HIV subtype B infections in Denmark originated from Sweden and Spain.
Additionally we identified several countries showing bidirectional migration. Notably, for the Netherlands 6 significant pathways were detected from and to the same localities. The Netherlands is among the countries in Europe with the most diverse geographical origin among newly diagnosed patients, confirmed by the high percentage of non-Dutch individuals among the newly HIV-infected patients during 2003–2004 [32, 53]. Moreover, because of its policies, the Netherlands attracts foreign drug users and male homosexuals, two populations known to be at higher risk for HIV infection .
Migratory pathways inferred through viral phylogenies cannot be directly validated by other sources of information (epidemiological figures, mobility and immigration information, tourism, etc), because these data are not stratified by subtype. Moreover, due to the high mobility of population within Europe and the complexity of the epidemic spread, information about the locus of infection for an individual doesn't necessarily match with the geographic origin of the source. On the other hand, phylogenetic analysis of viral sequences provides a realistic approach for the reconstruction of HIV transmission chains or networks [36, 46, 47, 49, 54–56], therefore suggesting that statistical phylogeography is appropriate for inferring the spatial dispersal of a viral epidemic.
Given the high complexity of the epidemic, dense sampling is needed in order to accurately reconstruct the spatial characteristics of the subtype B infections in Europe. This provides one of the limitations of this study; on the other hand however the analysis of our dataset, which is the largest available at the time of analysis, provides for a first time a description of the geographic distribution of viral lineages as well as the significant migrations of HIV subtype B across Europe, by means of viral phylogenies. Dense sampling for each locality would be ideal for such purposes; however limited availability of sequences for several countries, as well as computation time provide as the major limitations for such a study.
We paid special attention to representativeness of our data. The prospective SPREAD collection strategy (data from 2002–2004) was specifically designed to avoid such a bias , while the retrospectively collected CATCH data (1996–2002) were sampled as part of national surveillance studies designed to investigate the transmission of drug resistance or as part of the standard clinical practice of baseline sequencing for all newly diagnosed cases in each participating center . For most countries where national data were available, the data were a rather good representation of the national epidemic.
In conclusion, HIV-1 subtype B phylogeographies provide a new insight for the first time into the pathways of spatial diffusion and virus migration across Europe. HIV-1 subtype B was each time introduced from multiple sources and subsequently spread locally, but the pattern is not uniform across Europe. The countries grouped into sources (Greece, Portugal, Serbia and Spain) and sinks (Austria, Belgium and Luxembourg) of virus migration, as well as countries with significant bidirectional migration (Denmark, Germany, Italy, Israel, Norway, the Netherlands, Sweden, Switzerland and the UK). The only exception was Poland where a significant number of sequences fell within a monophyletic cluster. These results suggest that mobility of the virus matches mobility of the host, such that in order to reduce further spread of the epidemic, prevention measures should not only be directed towards national populations, but also towards migrants, travellers and tourists who are the major sources and targets of HIV dispersal.