In the present study we describe the expanded molecular surveillance and epidemiological assessment of the HIV-1 epidemic in Bulgaria. Our results extend those reported previously by our group that initially identified multiple subtypes in Bulgaria, which we proposed were sustained by viral inflow from European countries, USA and Africa 
. In addition to our earlier results, current national data from ongoing HIV and AIDS surveillance suggest that the HIV/AIDS incidence in Bulgaria remains low overall. To obtain a broader picture of the HIV-1 epidemic in Bulgaria we analyzed new pol
sequences from 125 HIV-1-infected persons and combined these with 77 pol
sequences from patients reported in our previous study, altogether representing about 20% of infections identified to date. Our study shows that the HIV-1 epidemic in Bulgaria has several key characteristics relative to the genetic composition of circulating viruses in the country, including a high genetic diversity of more than 15 different HIV-1 subtypes, CRFs and URFs, two new sub-subtype variants, and the introduction of seven extremely rare HIV-1 forms, including CRF05_DF, CRF04_cpx CRF36_cpx, CRF14_BG, and a plethora of different URFs that are distinct from all reference sequences. The data suggest the presence of local networks in certain risk groups with specific HIV-1 genotypes 
, an unequal subtype distribution among different risk groups, with most MSM infected with subtype B, and the lack or low prevalence of certain genotypes in specific groups, and the fluctuation of different HIV-1 genotypes overtime and in certain risk groups.
In our current study we found a broader genetic heterogeneity of HIV-1 circulating in Bulgaria than that previously reported by our group 
. Most infections were with the major HIV-1 subtypes followed by a variety of CRFs, URFs and sub-subtypes. Although subtype B was the most prevalent genotype in both studies and represents about half the HIV-1 infections in Bulgaria, our results differ from those in Western Europe where subtype B comprises the vast majority of infections 
. Our results show that subtype B has been introduced into Bulgaria multiple times, but also with local expansion in certain populations. In contrast to our previous study 
, the second largest genotype in Bulgaria is now CRF01_AE, followed by CRF02_AG. We also identified seven new complex circulating forms not seen before in Bulgaria and which are typically found in Cameroon, Libya, Singapore, Thailand, Saudi Arabia, Greece, Cyprus, Portugal, Spain, or France. Notably, two of these CRFs have been associated with outbreaks in IDUs in Greece (CRF04_cpx and CRF14_BG) 
, but only one of these genotypes (CRF14_BG) was present in an IDU in Bulgaria and may represent a recent infection. The others were present in non-IDU populations (three heterosexual and one blood product recipient) suggesting that these rare variants may be spreading outside of IDUs. We also identified nine possible new URFs circulating in Bulgaria that may be the result of the mixing of so many genotypes in the population similar to the natural history of HIV-1 in Africa. These URFs were found to be associated with persons reporting travel abroad who may be facilitating the increase of viral diversity in Bulgaria. All of the URFs identified in our study, including the clusters of A-like and F-like sequences, will require further characterization using complete genomes to determine their genetic composition and final classification as was done for subtype K and the A and F sub-subtypes 
The distribution of HIV-1 subtypes in our study population varied by age, sex, geography, and risk exposure. The greatest number of infections and broadest HIV-1 diversity occurred in major cities where most immigrants tend to live. We also observed the largest number of HIV subtypes and hence the greatest amount of HIV diversity in persons over 45 years old despite their being fewer total infections in this age group. Persons infected by heterosexual or IDU transmission had the greatest variety of subtypes, compared to MSM (p<0.0001). Furthermore, we also observed that closely related viral clades are increasing in circulation in geographically restricted IDU subgroups. The HIV-1 CRF02_AG seen in one IDU subgroup appears to have been transmitted within separate transmission clusters. Both women and IDUs were more likely to be infected with subtypes CRF01_AE and CRF02_AG, though the trend in IDUs may be due to an overall increase in IDU infections over time in Bulgaria. In contrast, with the exception of two persons, all MSM were infected with subtype B. Our results are consistent with several studies from different European countries that found a predominance of subtype B in MSM 
. There is also recent independent evidence that subtype B in Bulgaria may be phylogenetically related to subtype B strains identified in Russia, Greece and some other neighboring Balkan countries 
. In addition, our phylogenetic and epidemiologic analyses showed an absence or reduced prevalence of certain viral strains in specific populations. For example, neither MSM nor individuals infected by blood transfusion were infected with HIV-1 CRF01_AE, which was the second most common genotype in Bulgaria. These findings suggest a pattern of independent and limited transmission of some viral clades within certain population groups without viral exchange between them. The high level of HIV-1 diversity and wide distribution among various risk groups and demographic classifications are likely to influence various aspects of the epidemic in Bulgaria, including patient treatment, disease progression, laboratory monitoring, and vaccine development.
Analysis of longitudinal epidemiologic and demographic data with HIV-1 genotype can provide important information regarding infection dynamics during different stages of the epidemic in Bulgaria. Using this strategy by year of HIV/AIDS diagnosis we found a dissimilar rate of introduction and spread of different HIV-1 clades over time during the history of the epidemic in Bulgaria. Some clades persist, such as subtype B, while others like CRF01_AE and subtype C have declined, and other clades appear to be emerging, like CRF02_AG. We also found that there was a significant increase in the number of infections in MSM and IDUs with a concomitant decline in heterosexual and vertical infections. Some of these observed changes in the dynamics of HIV-1 infection in Bulgaria can be explained by the initial introduction of specific HIV-1 clades in certain populations spreading afterwards within populations at increased risk for infection. We also found that during the early years of the epidemic HIV-1 was most likely found in Bulgarians returning from abroad and who then transmitted HIV-1 via heterosexual contact. However, in recent years the significant increase in transmission of certain HIV-1 subtypes to and among IDUs and MSM is most likely due to increased risky behavior and spread of subtypes within these groups but also coincided with a decrease in heterosexual infections.
Our results also suggest that the infection dynamics and HIV-1 genetic diversity of the HIV-1 epidemic in Bulgaria are fluctuating. There are a few factors that seem relevant to help explain this phenomenon. Firstly, the central geographic location of Bulgaria at the crossing point between Western Europe, Eastern Europe, and the Middle East facilitates the introduction of HIV-1 into and beyond Bulgaria. In addition, an increase of West African immigrants to Europe to escape war, oppression and poverty has likely helped increase the HIV-1 diversity in the country 
. The spread of HIV-1 within high risk groups like IDUs and MSM from 2006–2009 supports the changing epidemiological dynamics of HIV-1 infection which could indicate an increase of these populations in Bulgaria or increases in risky behavior or both. There has been a dramatic reduction and inversion of the socio-economic status of the Bulgarian population from a planned to a market economy after 1989 associated with high unemployment rates of about 20% in 2001 prior to the period of our study and a subsequent fall in the gross domestic product by 50%, which may lead individuals to substance abuse with an increased risk for infection (www.google.com/publicdata
). An underestimation of MSM transmission in the early years of the epidemic due to stigma associated with reporting of homosexual contacts could also explain the observed increase in MSM infections. Thus, control of the HIV-1 epidemic in these populations will require targeted interventions in conjunction with continuous and increased surveillance efforts of specific at-risk groups, especially IDUs and MSM and their contacts. These public health initiatives will be crucial to understanding the details of the epidemiology of HIV-1 in Bulgaria and to assess the efficacy of prevention strategies and control of the epidemic.
The cross-sectional design of our study, which included convenience specimens, collected as patients were diagnosed or came into clinics for care, may not truly represent the other 80% of reported cases in Bulgaria. The effect of this possible sampling error in our analyses is unknown but may influence genotype distribution over time and in different populations. For example, we determined that MSM were overrepresented and IDUs were underrepresented in our two studies combined which may affect the trend interpretations in these risk groups. Increased HIV-1 testing and surveillance, especially in IDUs, should help to overcome this potential bias and improve the accuracy of understanding the epidemic in Bulgaria and the proposed targeted interventions. Also, our phylogenetic results are based on a single genomic region, in some cases genotypes were only identified from limited numbers of patients. Thus, confirmation of the rare genotypes identified herein may require sequence analysis of additional regions or complete genomes and their detection in larger numbers of persons. Any potential association of infection with a geographic location is based on self-reporting and may be affected by recall or other biases and thus may require analysis of larger HIV-1 datasets from the purported country where the transmission was reported to confirm the origin. Minor differences between our results and those of our previous study most likely represent the use of more stringent criteria for genotype classification and the use of a larger set of reference sequences that included more CRFs in the current study.
In conclusion, we found a broad level of HIV-1 genetic diversity in Bulgaria, including most major subtypes and CRFs and the identification of novel URFs and sub-subtypes, whose composition fluctuates in different populations and during different phases of the epidemic. Our data also suggest the extremely dynamic nature of the Bulgarian epidemic is characterized by an unequal distribution of different HIV-1 genotypes among high risk populations. These findings emphasize the need for sustained and focused molecular epidemiological surveillance to identify transmission links that can be targeted by prevention strategies to control the HIV-1 epidemic in Bulgaria.