High-resolution molecular epidemiology and evolutionary history of HIV-1 subtypes in Albania

Abstract

Background: HIV-1 epidemic in Western Europe is largely due to subtype B. Little is known about the HIV-1 in Eastern Europe, but a few studies have shown that non-B subtypes are quite common. In Albania, where a recent study estimated a ten-fold increase of AIDS incidence during the last six years, subtype A and B account for 90% of the know infections.

Methodology/principal findings: We investigated the demographic history of HIV-1 subtype A and B in Albania by using a statistical framework based on coalescent theory and phylogeography. High-resolution phylogenetic and molecular clock analysis showed a limited introduction to the Balkan country of subtype A during the late 1980s followed by an epidemic outburst in the early 1990 s. In contrast, subtype B was apparently introduced multiple times between the mid-1970s and mid-1980s. Both subtypes are growing exponentially, although the HIV-1A epidemic displays a faster growth rate, and a significantly higher basic reproductive number R(0). HIV-1A gene flow occurs primarily from the capital Tirane, in the center of the country, to the periphery, while HIV-1B flow is characterized by a balanced exchange between center and periphery. Finally, we calculated that the actual number of infections in Albania is at least two orders of magnitude higher than previously thought.

Conclusions/significance: Our analysis demonstrates the power of recently developed computational tools to investigate molecular epidemiology of pathogens, and emphasize the complex factors involved in the establishment of HIV-1 epidemics. We suggest that a significant correlation exists between HIV-1 exponential spread and the socio-political changes occurred during the Balkan wars. The fast growth of a relatively new non-B epidemic in the Balkans may have significant consequences for the evolution of HIV-1 epidemiology in neighboring countries in Eastern and Western Europe.

Figure 1. Maximum likelihood phylogenetic analysis of HIV-1A pol sequences.

Branch lengths were estimated with the best fitting nucleotide substitution model according to a hierarchical likelihood ratio test , and were drawn in scale with the bar at the bottom indicating 0.1 nucleotide substitutions per site. One * along a branch represents significant statistical support for the clade subtending that branch (p<0.001 in the zero-branch-length test and bootstrap support >75%). The color of each tip branch represents the country (or geographic region) of origin of sequence corresponding to that branch, according to the legend in the figure. A. Phylogenetic tree of 31 HIV-1A strains from Bulgaria and 122 subtype A reference sequences downloaded from the Los Alamos HIV database. The tree was rooted using two HIV-1B strains as outgroup. The arrow indicates the most recent common ancestor (MRCA) of the Albanian monophyletic clade. B. Phylogenetic tree of 21 HIV-1B strains from Bulgaria and 46 subtype B reference sequences downloaded from the Los Alamos HIV database. The tree was rooted using two HIV-1A strains as outgroup. Solid boxes highlight statistically supported clades of Albanian sequences.

Figure 2. Likelihood mapping of HIV-1A and B pol sequences.

Each dot represents the likelihoods of the three possible unrooted trees for a set of four sequences (quartets) selected randomly from the data set (see Materials and Methods): dots close to the corners or the sides represent, respectively, tree-like, or network-like phylogenetic signal in the data. The central area of the likelihood map, highlighted by a smaller blue triangle inside the map, represents star-like signal. The percentage of dots in the central area is given at the basis of each map. A. Likelihood mapping of 10,000 random quartets of HIV-1A (left) and HIV-1B (right) Albanian sequences. B. Likelihood mapping of 10,000 random quartets of HIV-1A (left) and HIV-1B (right) Albanian+reference sequences downloaded from the Los Alamos HIV databases.

Figure 3. Bayesian skyline plots of HIV-1 subtypes in Albania.

Non-parametric curves of HIV-1 effective population size (effective number of infections, Ne) over time in Albania were estimated from pol gene sequences employing a Bayesian framework. Genetic distances were transformed into a timescale of years taking into account the different sampling times of the viral strains and by enforcing a relaxed molecular clock model (see Materials and Methods). Solid lines indicate median, and 95% upper and lower high posterior density (HPD) estimates of Ne, according to the color legend to the right; dotted blue lines are parametric estimate of Ne(t) according to the exponential demographic model. Broken lines between the top and bottom panels highlight the time window common to the two epidemics. Top panel. HIV-1 subtype B. Each arrow indicates the estimated origin of the corresponding Albanian clade (indicated by the roman numeral) in the tree in Figure 1B. The first arrow from the left indicates the time of the MRCA. 95% HPD intervals are given in parenthesis below each estimate. Bottom panel. HIV-1 subtype A. The arrow indicates the estimated origin of the MRCA of the Albanian clade shown by the tree in Figure 1A. 95% HPD intervals are given in parenthesis below each estimate.

Figure 4. Estimates of HIV-1A and HIV-1B basic reproductive number (R0) in Albania.

R0 estimates for different average duration of infection (D) were inferred from the Bayesian estimates of population growth rate (see materials and Methods). A Black bar indicates the 95% high posterior density intervals of the estimate. An * on top of the bar indicate a statistically significant difference (paired t-test, p<0.01) between R0 estimates for two different subtypes.

Figure 5. Phylogeographic mapping of HIV-1A and B epidemic in Albania.

A. The bubblegrams show the frequency of gene flow (migrations) in Albania to/from different geographic areas and the country capital Tirane. The surface of each circle is proportional to the percentage of observed migrations in the ML genealogy (supplemental Figure S3). Migrations were inferred with a modified version of the Slatkin and Maddison algorithm , for the HIV-1A (top panel) and HIV-1B (bottom panel) subtype from the maximum likelihood inferred genealogies given in Figure 1 and 2, respectively. B. Detailed mapping of HIV-1A (indicated by the red arrows) and HIV-1B (indicated by the blue arrows) gene flow among different Albanian geographic areas.