The present study is the first in-depth description of HIV-1 molecular epidemiology in Central America based on a large number of samples collected in six of the seven Central American countries. The findings show that subtype B is predominant in the region, which is congruent with reports from neighboring countries such as Mexico, Colombia, and Venezuela (45
), as well as smaller reports from Central America (14
). Interestingly, however, while HIV-1 group M subtype B strains worldwide usually intermix in a star-like phylogeny, indicative of the panmictic structure of the epidemic, strains from Central America appeared to be highly compartmentalized. The phylogenetic analysis showed that the majority of the current subtype B cases in Central America appeared to originate from a single early introduction (95% HPD 1955 to 1977). Furthermore, four additional statistically supported monophyletic clades consisting of at least 10 Central American sequences appeared to have evolved independently, suggesting the existence of discrete subepidemics within different countries, which originated from separate introductions. The fact that the largest clade included 62% of the strains may suggest that of the possible multiple subtype B introductions in the region, one was very successful and gave rise to a regional epidemic, particularly affecting Honduras and El Salvador. The origin of the major Central American clade occurred relatively soon (median estimate, 1966) after the emergence of the most recent common ancestor of HIV-1 group M subtype B (median estimate, 1957), which may explain in part the early compartmentalization of the epidemic. It is important to note, however, that our median estimate of HIV-1 group M subtype B TMRCA (1957), based on pol
sequences, is 1 decade earlier than current estimates (1966) based on env
). This may be due, at least in part, to the higher star-like signal (phylogenetic noise) in the pol
gene, resulting in an unrealistically deep divergence date, although the 95% HPD intervals (1944 to 1968) of our estimates are still overlapping with the ones reported in the literature. The existence of independent monophyletic clades also implies the presence of separate transmission networks for subtype B spread within the Central American region. Therefore, our data may provide valuable information for targeted intervention and prevention.
The five Central American clades appeared to belong to the U.S./pandemic clade of HIV-1 group M subtype B (31
), but there was no clear pattern clustering subtype B strains from Central America with specific sequences from the rest of the world. Interestingly, our analyses did show that some strains from Haiti were closely related to strains belonging to the major Central American clade, although statistical support was not significant, suggesting a possible common ancestor between these two epidemics. In addition, the Bayesian trees showed Haitian sequences basal to the HIV-1 group M subtype B phylogeny, in agreement with the results of Gilbert et al. (31
). It is also important to note that, in spite of the observed phylogenetic noise in the pol
data sets analyzed in the present work, most of the Central American sequences still clustered within a highly supported monophyletic clade, which strengthens the main finding that a single introduction is responsible for the majority of the current cases in the region.
The observation that Central American clades and strains may have originated in the U.S. or Haitian strains is expected and in agreement with known early epidemiological data, as well as other phylogenetic analyses (31
), geographical proximity, known tourism, and legal and illegal immigration patterns. However, given the low resolution in the phylogenetic trees and the unbalanced sampling from other American countries, no firm conclusions can be drawn and the route of the earliest HIV-1 transmissions in Central America remains to be investigated.
We found evidence of TDR in 7.5% of the 417 treatment-naive patients from the six Central American countries, which is a moderate level according to WHO criteria (48
). This is the first comprehensive report of TDR in Central America. The results have to be interpreted in light of the history of antiretroviral therapy in the countries, which initially often involved uncontrolled monotherapy and dual therapy with antiretroviral drugs that were sold illegally or sent from patients' relatives in the United States, whereas in more recent years, almost all therapy is provided as combination therapy through national treatment programs.
Some limitations of our study should be mentioned. First, we cannot exclude the possibility that the sampled population is not fully representative of all HIV-1 cases in the six Central American countries included in the study because the samples represent only a fraction of all HIV infections in the region. Furthermore, sampling was unbalanced between countries. It is possible that a more balanced sampling would show additional clades and/or that clades II to V are more widely spread across Central America than suggested by our analyses. Consequently, while our study is the most comprehensive investigation of HIV-1 spread in Central America to date, the full picture will require additional sampling from several other countries, including Guatemala, which was not sampled at all. It is also possible that the coalescence times for clades II to V would be more similar to those of clade I (i.e., earlier TMRCAs) had they been more thoroughly sampled. Another limitation was the use of partial, rather than full, genome sequences, which could underestimate the proportion of recombinant strains and possibly may limit the possibilities to resolve completely the deepest portions of the trees. However, our preliminary results suggest that the subtypes and clustering of Honduran strains are similar in analyses based on env V3 sequences.
In conclusion, this is the first comprehensive study of how and when HIV-1 has entered and spread in Central America, which is a region with a substantial HIV/AIDS burden. HIV-1 group M subtype B predominates, although sporadic non-B strains were identified. Phylogenetic and molecular clock analysis showed one major well-supported monophyletic cluster compatible with a single early subtype B introduction accounting for most current cases, as well as a subsequent expansion into regional subepidemics, which deserves further investigation in order to understand the ecological factors driving subtype B's successful emergence and dissemination in Central America.