Molecular Analysis of Human Immunodeficiency Virus Type 1 (HIV-1)–Infected Individuals in a Network-Based Intervention (Transmission Reduction Intervention Project): Phylogenetics Identify HIV-1–Infected Individuals With Social Links

Molecular Analysis of Human Immunodeficiency Virus Type 1 (HIV-1)–Infected Individuals in a... Abstract Background The Transmission Reduction Intervention Project (TRIP) is a network-based intervention that aims at decreasing human immunodeficiency virus type 1 (HIV-1) spread. We herein explore associations between transmission links as estimated by phylogenetic analyses, and social network–based ties among persons who inject drugs (PWID) recruited in TRIP. Methods Phylogenetic trees were inferred from HIV-1 sequences of TRIP participants. Highly supported phylogenetic clusters (transmission clusters) were those fulfilling 3 different phylogenetic confidence criteria. Social network–based ties (injecting or sexual partners, same venue engagement) were determined based on personal interviews, recruitment links, and field observation. Results TRIP recruited 356 individuals (90.2% PWID) including HIV-negative controls; recently HIV-infected seeds; long-term HIV-infected seeds; and their social network members. Of the 150 HIV-infected participants, 118 (78.7%) were phylogenetically analyzed. Phylogenetic analyses suggested the existence of 13 transmission clusters with 32 sequences. Seven of these clusters included 14 individuals (14/32 [43.8%]) who also had social ties with at least 1 member of their cluster. This proportion was significantly higher than what was expected by chance. Conclusions Molecular methods can identify HIV-infected people socially linked with another person in about half of the phylogenetic clusters. This could help public health efforts to locate individuals in networks with high transmission rates. HIV-1, phylogenetics, transmission links, social ties, PWID A sharp increase in human immunodeficiency virus type 1 (HIV-1) diagnoses among people who inject drugs (PWID) in Athens, Greece, was observed between late 2010 and early 2011 [1–3]. The number of HIV-1 diagnoses peaked in 2012 and decreased afterward. Initial responses included high needle-and-syringe and opioid substitution programs coverage and a scale-up of HIV testing and counseling rates among PWID in the Athens metropolitan area. A large “seek, test, treat, and retain” program (ARISTOTLE) between 2012 and 2013 also significantly contributed to the reduction of HIV incidence among PWID [4, 5]. In addition, a social network–based intervention (Transmission Reduction Intervention Project [TRIP]) was conducted between 2013 and 2015 targeting mostly PWID in Athens [6, 7]. TRIP aimed at preventing HIV incidence by identifying recently HIV-infected individuals and their social networks in which transmissions are more likely to occur. The project recruited and tested social network members of recently infected individuals to identify more people who had been recently infected and therefore were more infectious. TRIP also distributed community alerts targeting individuals socially close to a new infection and made efforts to increase linkage to care for those testing HIV positive [6]. Social network–based data have been traditionally collected by interviews, field observation, and HIV testing techniques to make inferences about pathways of HIV spread. Molecular approaches have also been used in this field [8–13]. Of course, both social network–based and molecular approaches have limitations. For example, social network–based studies fail to describe the entire network due to incomplete recruitment of its members. Moreover, although social network–based studies include some information about the risk behavior of their members, in most cases detailed description of how HIV transmission occurs between the network members is not straightforward [14]. Molecular methods, on the other hand, provide an estimate of viral linkage, which in turn is a proxy of an actual HIV transmission pathway within a population [15], but they cannot infer who infected whom within the inferred transmission networks, nor do they tell us anything about the uninfected members of the network or the community [14]. The combination of social and molecular methods could fill gaps in how a risk network is structured. In an effort to combine social and molecular epidemiology (hereby named social phylogeny), the current article studies patterns of viral spread among PWID and their social contacts recruited in TRIP, Athens, and explores associations between transmission links, as estimated by phylogenetic analyses, and social networks. METHODS Participants and Data People eligible to participate in TRIP were those aged ≥18 years, able to answer the study questionnaire, and qualified for 1 of its 3 arms. Recruitment for each arm began with seeds and controls and was based on HIV testing results, HIV recency test results, and HIV testing history [6, 7]. The primary arm consisted of recently HIV-infected seeds (infected in the past 6 months) and their networks, the second arm included long-term HIV-infected controls who were also used as seeds for network recruitment (LCSs), and the third arm consisted of negative controls [6, 7]. Seeds were asked to elicit names of members of their networks and to help recruit them to be tested for HIV. Network members of recently HIV-infected seeds and LCSs were recruited for 2 steps. Recruitment stopped at second-degree network members unless a recently infected individual was identified and then a new 2-step recruitment process began. There was no social network recruitment for the negative controls. A total of 356 individuals (90.2% PWID) participated in TRIP between June 2013 and July 2015. One hundred fifty PWID were infected with HIV-1. Of these, viable samples for sequencing were obtained from 118 individuals (2 sequences from 1 individual) (Figure 1). Sequences were available in the protease and partial reverse transcriptase regions of the virus genome (pol gene) (873 nucleotides). Figure 1. View largeDownload slide Unrooted phylogenetic tree of human immunodeficiency virus type 1 sequences isolated from Transmission Reduction Intervention Project participants (N = 119) estimated by the RAxML (version 8) program. The names of subtypes, circulating recombinant forms, and unique recombinant forms are shown at the tips of the tree. Highly supported phylogenetic (transmission) clusters are marked in different colors. Numbers at the nodes show bootstrap support values. Abbreviations: CRF, circulating recombinant form; LTN, local transmission network; PWID, people who inject drugs; URF, unique recombinant form. Figure 1. View largeDownload slide Unrooted phylogenetic tree of human immunodeficiency virus type 1 sequences isolated from Transmission Reduction Intervention Project participants (N = 119) estimated by the RAxML (version 8) program. The names of subtypes, circulating recombinant forms, and unique recombinant forms are shown at the tips of the tree. Highly supported phylogenetic (transmission) clusters are marked in different colors. Numbers at the nodes show bootstrap support values. Abbreviations: CRF, circulating recombinant form; LTN, local transmission network; PWID, people who inject drugs; URF, unique recombinant form. Social network–based studies raise ethical concerns as participants are interviewed about sensitive issues and are asked to name contacts who have not yet consented to this disclosure of information. It should be noted, however, that the identity of the index person is never revealed, nor is information about contacts conveyed back to the index person. In addition, similar studies have been conducted in the past with multiple benefits for HIV prevention at both the individual and the public health level, and without causing any harm to the participants [16, 17]. Finally, the US Centers for Disease Control and Prevention also recommends social network–based approaches in routine partner services for HIV infection [18]. TRIP was approved by the institutional review boards of the National Development and Research Institutes in New York City and of the Hellenic Scientific Society for the Study of AIDS and Sexually Transmitted Diseases in Athens. All experiments were performed in accordance with relevant guidelines and regulations. All participants provided written informed consent. The accession numbers of the HIV-1 sequences generated on the purpose of the study are MH157288–MH157404. Social Networks Social networks were constructed according to questionnaire-based interviews, recruitment links, and field/venue observations. A first-degree social contact was an individual who had sex or had injected with a TRIP participant, or an individual who had sex or had injected in the presence of that participant. Second-degree network members were linked to first-degree contacts, but not the index. Social Network Visualizer version 1.9 was used to visualize social networks. Specifically, we used the degree centrality measure, which quantifies how many ties a node has to other nodes in the network and is considered a measure of social link activity. HIV-1 Sequence Subtyping Subtypes were identified for 119 sequences using the online automated HIV-1 subtyping tool COMET version 0.2 (Context-based Modeling for Expeditious Typing) (https://comet.lih.lu/). Subtyping results were further confirmed by performing phylogenetic analysis using 247 globally sampled sequences as references, which were representative of all known HIV-1 subtypes and circulating recombinant forms (CRFs), and were available on the Los Alamos HIV-1 sequence database (http://www.hiv.lanl.gov). Sequences from PWID sampled during an outbreak (2011–2015) in Athens were also used as references [19]. Moreover, we tested for the presence of recombination in all unclassified sequences using the RDP4 program [20], and also the bootscanning approach as implemented in the SimPlot version 3.5.1 program [21]. Sequence alignment was performed on MEGA (Molecular Evolutionary Genetics Analysis) version 5.2 using the MUSCLE algorithm, and alignments were manually edited according to the encoded reading frame [22]. Tree visualization and annotation were done using the FigTree version 1.4 program [23]. Phylogenetic (Transmission) Cluster Identification We performed phylogenetic analyses to identify the presence of highly supported transmission clusters (phylogenetic clusters) among our sequences. Specifically, phylogenetic analysis was conducted on (1) the total number of sequences (N = 158) by using all the available samples per participant (n = 118), and (2) the total number of sequences by using only 1 sample per participant (n = 118) with 1 exception. Specifically, 2 sequences for the same participant sampled at different time points fell within different local transmission networks (LTNs), suggesting superinfection. In case >1 sequence was available per participant, we used the oldest sample in which HIV was present. Phylogenetic trees were estimated from the underlying nucleotide sequences using different methods: (1) We analyzed our data by using the approximate maximum likelihood (ML) method under the generalized time reversible (GTR+cat) model of nucleotide substitution model including a Γ-distributed rate of heterogeneity among sites as implemented in FastTree version 2.1 and RAxML version 8 programs [24, 25]; (2) we also performed the analysis by neighbor-joining method with ML distances using the GTR model including a Γ distribution rate heterogeneity among sites as implemented in the PAUP*4.0 program [26]; (3) finally, we analyzed our data by using the Bayesian method with the GTR substitution model with Γ distributed rate, as implemented in MrBayes version 3.2.2 software [27]. The Markov chain Monte Carlo (MCMC) ran for 10 × 105 generations (burn-in was set to 10%), with 4 chains per run, and with MCMC sampling every 1000 steps. Each MCMC run checked for convergence using Tracer version 1.6. Highly supported phylogenetic clusters (transmission clusters) were defined as these consisted of ≥2 sequences, and with (1) Shimodaira-Hasegawa and bootstrap values greater than 0.90 and 70%, respectively, for phylogenetic trees estimated by ML and neighbor-joining method; and (2) a posterior probability >0.80 for phylogenetic trees estimated by the Bayesian method (phylogenetic confidence criterion). Only phylogenetic clusters fulfilling all 3 of these criteria were considered as highly supported. Significance Testing of the Proportion of Individuals With Social Ties That Were Found Within the Transmission Clusters We used social network–based ties to estimate the proportion of individuals found within the transmission (phylogenetic) clusters who also had social ties. Our purpose was to identify whether this percentage was significantly higher than what was expected by chance. Therefore, we performed a random shuffling of taxa (the set of viral sequences analyzed by phylogenetic analysis) at the tips (tips correspond to the external nodes or “leaves” of the phylogenetic tree) of the ML tree, estimated by the RAxML (version 8) program, to simulate a population with random mixing given the inferred phylogeny. In a random mixing population, an infected individual would have an equal probability of transmitting the virus to any other individual. We created 100 reshuffling replicates of the ML tree (we randomly reshuffled the tips 100 times) using Mesquite version 3.01 [28]. Then, for each reshuffled tree, we estimated the proportion of individuals within clusters who had social ties with at least 1 member of their cluster as is expected to occur under the assumption of random mixing. Finally, we tested whether the estimated proportion of individuals within clusters, who had social ties with at least 1 member of their cluster, differed significantly from the distribution of the expected proportions, against the null hypothesis of complete mixing (panmixis hypothesis). RESULTS HIV-1 Subtyping Results Subtyping analysis showed that 81.5% (n = 97) of the sequences clustered within previously identified PWID LTNs in the Athenian outbreak (CRF14_BG, CRF35_AD, subtype A, and subtype B) [1, 29]. The unique recombinant forms of HIV-1 were also identified, consisting of partial sequences from previously identified PWID LTNs (n = 13 [11.0%]). The rest of the sequences (n = 9 [7.5%]) were tied to non-PWID transmission clusters (subtype A, CRF56_cpx, B/D recombinant form). Subtyping results are presented in detail in Table 1. The proportion of sequences in the PWID LTNs was similar to that in the total population of HIV-1 infected PWID (CRF14_BG: 437 [50.0%]; CRF35_AD: 139 [15.9%]; subtype B: 116 [13.3%]; and subtype A: 54 [6.2%]) sampled during the outbreak (1 January 2011–31 October 2014) [29]. Discordances in the subtyping results obtained with COMET and manual phylogenetic analysis were found for 20 of 119 sequences (16.8%). Differences were found for some unique recombinant forms and CRF14_BG classified as subtype G in COMET. Table 1. Human Immunodeficiency Virus Type 1 Subtype Distribution and Clustering Pattern for Sequences Isolated From People Who Inject Drugs Transmission Subnetwork  HIV-1 Subtype/CRF  No. of Sequences (%)  No. of Clustered Sequences (%)  No. of Clusters  Range of Clusters  Individuals Within Clusters With Social Ties  PWID LTNs    A  9 (7.6)  7 (21.9)  2  2–5  2    B  17 (14.3)  6 (18.8)  2  3  4    CRF14_BG  53 (44.5)  15 (46.9)  7  2–3  4    CRF35_AD  18 (15.1)  2 (6.2)  1  2  2    Subtotal  97 (81.5)  30 (93.8)  12  2–5  12  Non-PWID LTNs    A  7 (5.9)  …  …  …  …    B/D recombinant  1 (0.8)  …  …  …  …    CRF56_cpx  1 (0.8)  …  …  …  …    Subtotal  9 (7.5)  …  …  …  …  URFs including LTNs    13 (11.0)  2 (6.2)  1  2  2    Total  119 (100)  32 (100)  13  2–5  14  Transmission Subnetwork  HIV-1 Subtype/CRF  No. of Sequences (%)  No. of Clustered Sequences (%)  No. of Clusters  Range of Clusters  Individuals Within Clusters With Social Ties  PWID LTNs    A  9 (7.6)  7 (21.9)  2  2–5  2    B  17 (14.3)  6 (18.8)  2  3  4    CRF14_BG  53 (44.5)  15 (46.9)  7  2–3  4    CRF35_AD  18 (15.1)  2 (6.2)  1  2  2    Subtotal  97 (81.5)  30 (93.8)  12  2–5  12  Non-PWID LTNs    A  7 (5.9)  …  …  …  …    B/D recombinant  1 (0.8)  …  …  …  …    CRF56_cpx  1 (0.8)  …  …  …  …    Subtotal  9 (7.5)  …  …  …  …  URFs including LTNs    13 (11.0)  2 (6.2)  1  2  2    Total  119 (100)  32 (100)  13  2–5  14  Abbreviations: CRF, circulating recombinant form; HIV-1, human immunodeficiency virus type 1; LTN, local transmission network; PWID, people who inject drugs; URF, unique recombinant form. View Large Table 1. Human Immunodeficiency Virus Type 1 Subtype Distribution and Clustering Pattern for Sequences Isolated From People Who Inject Drugs Transmission Subnetwork  HIV-1 Subtype/CRF  No. of Sequences (%)  No. of Clustered Sequences (%)  No. of Clusters  Range of Clusters  Individuals Within Clusters With Social Ties  PWID LTNs    A  9 (7.6)  7 (21.9)  2  2–5  2    B  17 (14.3)  6 (18.8)  2  3  4    CRF14_BG  53 (44.5)  15 (46.9)  7  2–3  4    CRF35_AD  18 (15.1)  2 (6.2)  1  2  2    Subtotal  97 (81.5)  30 (93.8)  12  2–5  12  Non-PWID LTNs    A  7 (5.9)  …  …  …  …    B/D recombinant  1 (0.8)  …  …  …  …    CRF56_cpx  1 (0.8)  …  …  …  …    Subtotal  9 (7.5)  …  …  …  …  URFs including LTNs    13 (11.0)  2 (6.2)  1  2  2    Total  119 (100)  32 (100)  13  2–5  14  Transmission Subnetwork  HIV-1 Subtype/CRF  No. of Sequences (%)  No. of Clustered Sequences (%)  No. of Clusters  Range of Clusters  Individuals Within Clusters With Social Ties  PWID LTNs    A  9 (7.6)  7 (21.9)  2  2–5  2    B  17 (14.3)  6 (18.8)  2  3  4    CRF14_BG  53 (44.5)  15 (46.9)  7  2–3  4    CRF35_AD  18 (15.1)  2 (6.2)  1  2  2    Subtotal  97 (81.5)  30 (93.8)  12  2–5  12  Non-PWID LTNs    A  7 (5.9)  …  …  …  …    B/D recombinant  1 (0.8)  …  …  …  …    CRF56_cpx  1 (0.8)  …  …  …  …    Subtotal  9 (7.5)  …  …  …  …  URFs including LTNs    13 (11.0)  2 (6.2)  1  2  2    Total  119 (100)  32 (100)  13  2–5  14  Abbreviations: CRF, circulating recombinant form; HIV-1, human immunodeficiency virus type 1; LTN, local transmission network; PWID, people who inject drugs; URF, unique recombinant form. View Large Phylogenetic (Transmission) Clusters Phylogenetic analysis of the total number of sequences (n = 158) using all samples per participant (n = 118) showed that 2 different sequences sampled from the same participant at different time points clustered in 2 distinct LTNs (subtype A and CRF14_BG), suggesting superinfection. Subsequent phylogenetic analysis of the total number of sequences, using only 1 sample per participant for 117 participants and 2 samples for the 1 participant with sequences from 2 distinct LTNs (n = 119), revealed the existence of 13 highly supported phylogenetic clusters (transmission clusters) containing 2–5 individuals each. In total, these clusters consisted of 32 sequences (27.0%) (Figure 1). For the participant with sequences belonging to distinct LTNs, we analyzed both sequences. Hypothesis Testing We explored the potential association between transmission clusters, as estimated by phylogenetic analysis, and social network–based ties. We found that 7 of 13 phylogenetic clusters included 14 individuals (of 32 individuals found within clusters [43.8%]) who also had a first-degree social tie with at least 1 member of their phylogenetic cluster (Figure 2A). Further analysis revealed that phylogenetic clusters included 17 individuals (of 32 [53.1%]) who also had first-degree (n = 14) or second-degree (n = 3) social ties with at least 1 member of their phylogenetic cluster (Figure 2B). Then, we tested if the estimated proportion of individuals within clusters who had a first-degree social tie with at least 1 member of their phylogenetic cluster (43.8%) differed significantly from the distribution of the expected proportions under the null hypothesis of random mixing (panmixis hypothesis) (Figure 3). Our analysis revealed strong evidence for an observed proportion that was significantly higher than what was expected by chance (Figure 3). Figure 2. View largeDownload slide Social networks of individuals (n = 32) found within phylogenetic (transmission) clusters visualized by Social Network Visualizer (version 1.9): A, First-degree social ties. B, First- and second-degree social ties. Nodes are marked in different colors according to individual’s phylogenetic (transmission) cluster. Dotted circles in red represent the number of ties a node has to other nodes in its network. The nodes with the higher number of social ties are placed closer to the center of the circle. The degree of centrality measure quantifies the number of ties for node to others and it is often considered a measure of social tie activity. The number of ties is proportional to centrality. Diamond shape was used for individuals who had no social ties with other members of their phylogenetic (transmission) cluster. Figure 2. View largeDownload slide Social networks of individuals (n = 32) found within phylogenetic (transmission) clusters visualized by Social Network Visualizer (version 1.9): A, First-degree social ties. B, First- and second-degree social ties. Nodes are marked in different colors according to individual’s phylogenetic (transmission) cluster. Dotted circles in red represent the number of ties a node has to other nodes in its network. The nodes with the higher number of social ties are placed closer to the center of the circle. The degree of centrality measure quantifies the number of ties for node to others and it is often considered a measure of social tie activity. The number of ties is proportional to centrality. Diamond shape was used for individuals who had no social ties with other members of their phylogenetic (transmission) cluster. Figure 3. View largeDownload slide Proportions of individuals within phylogenetic (transmission) clusters with social ties. The red dot corresponds to the estimated proportion of individuals within transmission clusters who had first-degree social ties with at least 1 other member of their cluster (43.8%). The distribution of the proportion was estimated after 100 reshuffling rounds at the tips of the maximum likelihood tree (panmixis hypothesis). Figure 3. View largeDownload slide Proportions of individuals within phylogenetic (transmission) clusters with social ties. The red dot corresponds to the estimated proportion of individuals within transmission clusters who had first-degree social ties with at least 1 other member of their cluster (43.8%). The distribution of the proportion was estimated after 100 reshuffling rounds at the tips of the maximum likelihood tree (panmixis hypothesis). DISCUSSION Social network–based extensions of traditional contact tracing methods are promising public health interventions to prevent HIV transmissions [7, 30]. TRIP aimed to reduce HIV transmission by identifying recently HIV-infected individuals and tracing their social networks where transmissions occur at high rates [6]. Early identification and treatment of recently infected individuals are expected to have benefits at both the individual and the public health level [31, 32]. An important question, however, is the extent to which phylogenetically based clusters are related to social network–based ties. This relationship can affect the efficiency and value of social network–based interventions to reduce HIV incidence. We found, using a combination of molecular and social network–based tracing methods, that a high proportion of individuals (43.8%) in phylogenetic clusters also have first-degree social ties with at least 1 member of their cluster. This means that almost half of the individuals with viral linkage are also socially linked, suggesting that during an HIV outbreak among PWID, transmissions occurred preferentially within socially linked individuals and that social network–based methods helped us locate them. To our knowledge, this is one of the few studies to combine molecular and social network–based data. This is of public health importance, especially for HIV prevention in highly vulnerable, marginalized populations (PWID, transgender people, undocumented migrants, racial/ethnic groups, homeless people, and, in some countries, gay and bisexual men). In a previous study, Wertheim et al estimated the proportions of named partners infected with genetically closely related viruses and vice-versa (ie, the proportion of individuals for whom viruses were closely related) and they were linked by naming their partners in New York [12]. They found that 53% of individuals with genetically similar viruses were also linked based on their reports about the partners they have. Their estimates were close to ours (43.8% and 53.1% for first- and second-degree of social ties, respectively). Wertheim et al estimated transmission clusters based on pairwise comparisons that fell below a genetic threshold [12]. In the current study, we selected all sequences within well-defined phylogenetic clusters to belong to transmission clusters. The approach that Wertheim et al [12] used is probably better suited to identify linked pairs, especially among sexual partners; however, for studying an HIV outbreak situation, where genetically identical viruses have spread rapidly among PWID, choosing the appropriate genetic threshold is challenging [33]. In the absence of previously well-established threshold for genetic distances, we defined transmission clusters based on bootstrap/Shimodaira-Hasegawa/posterior support that probably corresponds to PWID infected from a common source rather than those with direct transmission links. The high similarity of the genetic sequences analyzed in the current study might be also due to the rather high conservation of the gene that was sequenced (partial pol). In another study related to the outbreak investigation among PWID in Indiana, Campbell et al inferred transmission clusters based on a combined analysis of epidemiological and molecular data [33]. For those reported high-risk contacts, 34% were found as potential transmission partners. This proportion is close to our estimate, although different approaches have been used. Based on these findings, it seems that HIV transmissions occur frequently among persons with measurable social ties, which highlights the importance of social network research for HIV prevention and epidemiology. Another group of studies was based on the respondent-driven sample (RDS) strategy, which limits the number of people recruited per individual and therefore has certain limitations for the description of social networks. The benefits of social network–based tracing approaches over RDSs are that they allow the recruitment of an unlimited number of individuals reporting social ties to an index case and that they collect data on, and allow analysis of, social and risk linkages of a given participant with people other than the ones who gave them a coupon or whom they gave a coupon to. In previous studies using molecular methods, RDS was performed mostly among men who have sex with men (MSM) and female sex workers. Lepej et al suggested that 5 of 12 (42%) MSM recruited with RDS belonged to a single phylogenetic cluster; however, they were not directly linked and 3 of them belonged to different RDS chains [34]. In a similar analysis in El Salvador, among 34 MSM who belonged to transmission clusters, 8 (24%) were recruited in the same RDS chain [35]. On the other hand, Fujimoto et al showed that viral transmissions do not occur between the socially connected young adult black MSM [36]. Additionally, phylogenetic studies have been used extensively to study characteristics of transmission networks between different transmission risk and racial groups and across diverse geographic locations and between subpopulations [9, 10, 37–50]. As of now there are no consensus criteria of phylogenetic confidence (eg, bootstrap or posterior probability support) for the definition of phylogenetic clusters [10, 51]. Evidence from a previous study suggests that the selection of a stricter definition of phylogenetic clusters (higher bootstrap values in combination with lower genetic distance thresholds) resulted in higher proportions of viral linkage from individuals with recent-phase transmissions [52]. We chose less strict phylogenetic criteria to avoid potential bias toward identifying recently infected individuals within clusters. We chose, however, to perform our analysis using different methods and 3 criteria for clustering (Shimodaira-Hasegawa, bootstrap, and posterior probability support). Phylogenetic analyses have certain limitations for identifying transmission clusters—for example, that clustering inferences depend on sampling coverage, which is the proportion of individuals for whom viral sequences are available [14, 15]. In our study, the sampling coverage was 79% (118 of 150 participants who were infected with HIV-1). Moreover, phylogenetic resolution depends on the number of nucleotides analyzed and also on the levels of genetic diversity within the alignment. Given these limitations, it has been suggested that phylogenetic trees represent a proxy of an actual HIV transmission pathway [15]. The design of TRIP and the recent nature of the HIV-1 outbreak among PWID in Athens, Greece, which started approximately between late 2010 and early 2011 [3], suggest that the high proportion of individuals with social ties within their phylogenetic clusters was probably due to the proximity in time of interviews and transmissions. Specifically, TRIP recruitment took place between June 2013 and July 2015, and among the 150 HIV-infected participants, 45 (30%) were recently infected. Social networks are complex and evolve over time as individuals modify their relationships or their risk behavior [15]. We show here that, under the conditions of our study, phylogenetically based transmission pathways were significantly associated with social networks traced relatively close to transmission dates. Specific conditions under which this association exists in different epidemics and populations remain to be elucidated. A limitation of this work is the high genetic similarity of sequences from the study population (which may be due to the short period of time since the start of the outbreak). To overcome this limitation, we selected as transmission (phylogenetic) clusters only those highly supported by 3 different criteria for clustering. Moreover, we tested if the proportion of those within transmission clusters with social ties is higher than panmixis by simulating the null hypothesis across the optimized topology (tree). The proportion of individuals within transmission clusters and, consequently, the fraction of those with social ties might be higher if more genetically diverse genomic regions were used. Finally, our analysis benefits from the combination of detailed social network–based and molecular data to explore whether HIV transmission pathways follow social ties. This is important to understand the role of social networks in HIV transmission and consequently to identify critical parameters of the epidemic (eg, populations at higher risk to transmit). Molecular analyses can be used to trace past transmission events and, when combined with social data, can be valuable for public health interventions. Notes Financial support. The study was supported by the National Institute on Drug Abuse (award number DP1 DA034989) and the Hellenic Scientific Society for the Study of AIDS and Sexually Transmitted Diseases. Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed. Presented in part: 13th International Conference on Molecular Epidemiology and Evolutionary Genetics of Infectious Diseases, Antwerp, Belgium, May 2016. References 1. Paraskevis D, Nikolopoulos G, Tsiara Cet al.   HIV-1 outbreak among injecting drug users in Greece, 2011: a preliminary report. Euro Surveill  2011; 16. pii: 19962. Google Scholar CrossRef Search ADS PubMed  2. Pharris A, Wiessing L, Sfetcu Oet al.   Human immunodeficiency virus in injecting drug users in Europe following a reported increase of cases in Greece and Romania, 2011. Euro Surveill  2011; 16. pii: 20032. Google Scholar PubMed  3. Paraskevis D, Paraschiv S, Sypsa Vet al.   Enhanced HIV-1 surveillance using molecular epidemiology to study and monitor HIV-1 outbreaks among intravenous drug users (IDUs) in Athens and Bucharest. Infect Genet Evol  2015; 35: 109– 21. Google Scholar CrossRef Search ADS PubMed  4. Hatzakis A, Sypsa V, Paraskevis Det al.   Design and baseline findings of a large-scale rapid response to an HIV outbreak in people who inject drugs in Athens, Greece: the ARISTOTLE programme. Addiction  2015; 110: 1453– 67. Google Scholar CrossRef Search ADS PubMed  5. Sypsa V, Psichogiou M, Paraskevis Det al.   Rapid decline in HIV incidence among persons who inject drugs during a fast-track combination prevention program after an HIV outbreak in Athens. J Infect Dis  2017; 215: 1496– 505. Google Scholar PubMed  6. Friedman SR, Downing MJJr, Smyrnov Pet al.   Socially-integrated transdisciplinary HIV prevention. AIDS Behav  2014; 18: 1821– 34. Google Scholar CrossRef Search ADS PubMed  7. Nikolopoulos GK, Pavlitina E, Muth SQet al.   A network intervention that locates and intervenes with recently HIV-infected persons: the Transmission Reduction Intervention Project (TRIP). Sci Rep  2016; 6: 38100. Google Scholar CrossRef Search ADS PubMed  8. Friedman SR, Ompad DC, Maslow Cet al.   HIV prevalence, risk behaviors, and high-risk sexual and injection networks among young women injectors who have sex with women. Am J Public Health  2003; 93: 902– 6. Google Scholar CrossRef Search ADS PubMed  9. Paraskevis D, Nikolopoulos GK, Magiorkinis G, Hodges-Mameletzis I, Hatzakis A. The application of HIV molecular epidemiology to public health. Infect Genet Evol  2016; 46: 159– 168. Google Scholar CrossRef Search ADS PubMed  10. Grabowski MK, Redd AD. Molecular tools for studying HIV transmission in sexual networks. Curr Opin HIV AIDS  2014; 9: 126– 33. Google Scholar CrossRef Search ADS PubMed  11. Brenner BG, Wainberg MA. Future of phylogeny in HIV prevention. J Acquir Immune Defic Syndr  2013; 63( Suppl 2): S248– 54. Google Scholar CrossRef Search ADS PubMed  12. Wertheim JO, Kosakovsky Pond SL, Forgione LAet al.   Social and genetic networks of HIV-1 transmission in New York City. PLoS Pathog  2017; 13: e1006000. Google Scholar CrossRef Search ADS PubMed  13. Friedman SR, Curtis R, Neaigus A, Jose B, Des Jarlais DC. Social networks, drug injectors’ lives, and HIV/AIDS . New York: Kluwer/Plenum, 1999. 14. Vasylyeva TI, Friedman SR, Paraskevis D, Magiorkinis G. Integrating molecular epidemiology and social network analysis to study infectious diseases: towards a socio-molecular era for public health. Infect Genet Evol  2016; 46: 248– 55. Google Scholar CrossRef Search ADS PubMed  15. Delva W, Leventhal GE, Helleringer S. Connecting the dots: network data and models in HIV epidemiology. AIDS  2016; 30: 2009– 20. Google Scholar CrossRef Search ADS PubMed  16. Friedman SR, Neaigus A, Jose Bet al.   Sociometric risk networks and risk for HIV infection. Am J Public Health  1997; 87: 1289– 96. Google Scholar CrossRef Search ADS PubMed  17. Woodhouse DE, Rothenberg RB, Potterat JJet al.   Mapping a social network of heterosexuals at high risk for HIV infection. AIDS  1994; 8: 1331– 6. Google Scholar CrossRef Search ADS PubMed  18. Centers for Disease Control and Prevention. Recommendations for partner services programs for HIV infection, syphilis, gonorrhea, and chlamydial infection. MMWR Recomm Rep  2008; 57: 1– 83; quiz CE1-4. 19. Paraskevis D, Nikolopoulos G, Fotiou Aet al.   Economic recession and emergence of an HIV-1 outbreak among drug injectors in Athens metropolitan area: a longitudinal study. PLoS One  2013; 8: e78941. Google Scholar CrossRef Search ADS PubMed  20. Martin DP, Lemey P, Lott M, Moulton V, Posada D, Lefeuvre P. RDP3: a flexible and fast computer program for analyzing recombination. Bioinformatics  2010; 26: 2462– 3. Google Scholar CrossRef Search ADS PubMed  21. Lole KS, Bollinger RC, Paranjape RSet al.   Full-length human immunodeficiency virus type 1 genomes from subtype C-infected seroconverters in India, with evidence of intersubtype recombination. J Virol  1999; 73: 152– 60. Google Scholar PubMed  22. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol Biol Evol  2011; 28: 2731– 9. Google Scholar CrossRef Search ADS PubMed  23. Drummond AJ, Suchard MA, Xie D, Rambaut A. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Mol Biol Evol  2012; 29: 1969– 73. Google Scholar CrossRef Search ADS PubMed  24. Stamatakis A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics  2014; 30: 1312– 3. Google Scholar CrossRef Search ADS PubMed  25. Price MN, Dehal PS, Arkin AP. FastTree 2–approximately maximum-likelihood trees for large alignments. PLoS One  2010; 5: e9490. Google Scholar CrossRef Search ADS PubMed  26. Wilgenbusch JC, Swofford D. Inferring evolutionary trees with PAUP. Curr Protoc Bioinformatics  2003; chapter 6, unit 6.4. Google Scholar PubMed  27. Ronquist F, Teslenko M, van der Mark Pet al.   MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol  2012; 61: 539– 42. Google Scholar CrossRef Search ADS PubMed  28. Maddison WP, Maddison DR. Mesquite: a modular system for evolutionary analysis. Version 3.10. 2018. Available at http://mesquiteproject.org. 29. Kostaki E, Magiorkinis G, Psichogiou Met al.   Detailed molecular surveillance of the HIV-1 outbreak among people who inject drugs (PWID) in Athens during a period of four years. Curr HIV Res  2017; 15: 396– 404. Google Scholar PubMed  30. Green N, Hoenigl M, Chaillon Aet al.   Partner services in adults with acute and early HIV infection. AIDS  2017; 31: 287– 93. Google Scholar CrossRef Search ADS PubMed  31. Cohen MS, Chen YQ, McCauley Met al.   HPTN 052 Study Team. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med  2011; 365: 493– 505. Google Scholar CrossRef Search ADS PubMed  32. Lundgren JD, Babiker AG, Gordin Fet al.   Initiation of antiretroviral therapy in early asymptomatic HIV infection. N Engl J Med  2015; 373: 795– 807. Google Scholar CrossRef Search ADS PubMed  33. Campbell EM, Jia H, Shankar Aet al.   Detailed transmission network analysis of a large opiate-driven outbreak of HIV infection in the United States. J Infect Dis  2017; 216: 1053– 62. Google Scholar CrossRef Search ADS PubMed  34. Lepej SZ, Vrakela IB, Poljak M, Bozicevic I, Begovac J. Phylogenetic analysis of HIV sequences obtained in a respondent-driven sampling study of men who have sex with men. AIDS Res Hum Retroviruses  2009; 25: 1335– 8. Google Scholar CrossRef Search ADS PubMed  35. Dennis AM, Murillo W, de Maria Hernandez Fet al.   Social network-based recruitment successfully reveals HIV-1 transmission networks among high-risk individuals in El Salvador. J Acquir Immune Defic Syndr  2013; 63: 135– 41. Google Scholar CrossRef Search ADS PubMed  36. Fujimoto K, Coghill LM, Weier CAet al.   Short communication: Lack of support for socially connected HIV-1 transmission among young adult black men who have sex with men. AIDS Res Hum Retroviruses  2017; 33: 935– 40. Google Scholar CrossRef Search ADS PubMed  37. Chan PA, Hogan JW, Huang Aet al.   Phylogenetic investigation of a statewide HIV-1 epidemic reveals ongoing and active transmission networks among men who have sex with men. J Acquir Immune Defic Syndr  2015; 70: 428– 35. Google Scholar CrossRef Search ADS PubMed  38. Poon AF, Joy JB, Woods CKet al.   The impact of clinical, demographic and risk factors on rates of HIV transmission: a population-based phylogenetic analysis in British Columbia, Canada. J Infect Dis  2015; 211: 926– 35. Google Scholar CrossRef Search ADS PubMed  39. Kharsany AB, Buthelezi TJ, Frohlich JAet al.   HIV infection in high school students in rural South Africa: role of transmissions among students. AIDS Res Hum Retroviruses  2014; 30: 956– 65. Google Scholar CrossRef Search ADS PubMed  40. Avila D, Keiser O, Egger Met al.   Swiss HIV Cohort Study. Social meets molecular: combining phylogenetic and latent class analyses to understand HIV-1 transmission in Switzerland. Am J Epidemiol  2014; 179: 1514– 25. Google Scholar CrossRef Search ADS PubMed  41. Middelkoop K, Rademeyer C, Brown BBet al.   Epidemiology of HIV-1 subtypes among men who have sex with men in Cape Town, South Africa. J Acquir Immune Defic Syndr  2014; 65: 473– 80. Google Scholar CrossRef Search ADS PubMed  42. Robertson AM, Garfein RS, Wagner KDet al.   Proyecto El Cuete IV and STAHR II. Evaluating the impact of Mexico’s drug policy reforms on people who inject drugs in Tijuana, B.C., Mexico, and San Diego, CA, United States: a binational mixed methods research agenda. Harm Reduct J  2014; 11: 4. Google Scholar CrossRef Search ADS PubMed  43. Volz EM, Frost SD. Inferring the source of transmission with phylogenetic data. PLoS Comput Biol  2013; 9: e1003397. Google Scholar CrossRef Search ADS PubMed  44. Volz EM, Ionides E, Romero-Severson EO, Brandt MG, Mokotoff E, Koopman JS. HIV-1 transmission during early infection in men who have sex with men: a phylodynamic analysis. PLoS Med  2013; 10: e1001568; discussion e1001568. Google Scholar CrossRef Search ADS PubMed  45. Lin H, He N, Zhou Set al.   Behavioral and molecular tracing of risky sexual contacts in a sample of Chinese HIV-infected men who have sex with men. Am J Epidemiol  2013; 177: 343– 50. Google Scholar CrossRef Search ADS PubMed  46. Levy I, Mor Z, Anis Eet al.   Men who have sex with men, risk behavior, and HIV infection: integrative analysis of clinical, epidemiological, and laboratory databases. Clin Infect Dis  2011; 52: 1363– 70. Google Scholar CrossRef Search ADS PubMed  47. Lewis F, Hughes GJ, Rambaut A, Pozniak A, Leigh Brown AJ. Episodic sexual transmission of HIV revealed by molecular phylodynamics. PLoS Med  2008; 5: e50. Google Scholar CrossRef Search ADS PubMed  48. Brenner BG, Roger M, Moisi DDet al.  ; Montreal PHI Cohort and HIV Prevention Study Groups. Transmission networks of drug resistance acquired in primary/early stage HIV infection. AIDS  2008; 22: 2509– 15. Google Scholar CrossRef Search ADS PubMed  49. Vasylyeva TI, Liulchuk M, Friedman SRet al.   Molecular epidemiology reveals the role of war in the spread of HIV in Ukraine. Proc Natl Acad Sci U S A  2018; 115: 1051– 6. Google Scholar CrossRef Search ADS PubMed  50. Paraskevis D, Kostaki E, Nikolopoulos GKet al.   Molecular tracing of the geographical origin of human immunodeficiency virus type 1 infection and patterns of epidemic spread among migrants who inject drugs in Athens. Clin Infect Dis  2017; 65: 2078– 84. Google Scholar CrossRef Search ADS PubMed  51. Hassan AS, Pybus OG, Sanders EJ, Albert J, Esbjörnsson J. Defining HIV-1 transmission clusters based on sequence data. AIDS  2017; 31: 1211– 22. Google Scholar CrossRef Search ADS PubMed  52. Marzel A, Shilaih M, Yang WLet al.  ; Swiss HIV Cohort Study. HIV-1 transmission during recent infection and during treatment interruptions as major drivers of new infections in the Swiss HIV Cohort Study. Clin Infect Dis  2016; 62: 115– 22. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journal of Infectious Diseases Oxford University Press

Molecular Analysis of Human Immunodeficiency Virus Type 1 (HIV-1)–Infected Individuals in a Network-Based Intervention (Transmission Reduction Intervention Project): Phylogenetics Identify HIV-1–Infected Individuals With Social Links

Loading next page...
 
/lp/ou_press/molecular-analysis-of-human-immunodeficiency-virus-type-1-hiv-1-cFMCYdKNmn
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, e-mail: journals.permissions@oup.com.
ISSN
0022-1899
eISSN
1537-6613
D.O.I.
10.1093/infdis/jiy239
Publisher site
See Article on Publisher Site

Abstract

Abstract Background The Transmission Reduction Intervention Project (TRIP) is a network-based intervention that aims at decreasing human immunodeficiency virus type 1 (HIV-1) spread. We herein explore associations between transmission links as estimated by phylogenetic analyses, and social network–based ties among persons who inject drugs (PWID) recruited in TRIP. Methods Phylogenetic trees were inferred from HIV-1 sequences of TRIP participants. Highly supported phylogenetic clusters (transmission clusters) were those fulfilling 3 different phylogenetic confidence criteria. Social network–based ties (injecting or sexual partners, same venue engagement) were determined based on personal interviews, recruitment links, and field observation. Results TRIP recruited 356 individuals (90.2% PWID) including HIV-negative controls; recently HIV-infected seeds; long-term HIV-infected seeds; and their social network members. Of the 150 HIV-infected participants, 118 (78.7%) were phylogenetically analyzed. Phylogenetic analyses suggested the existence of 13 transmission clusters with 32 sequences. Seven of these clusters included 14 individuals (14/32 [43.8%]) who also had social ties with at least 1 member of their cluster. This proportion was significantly higher than what was expected by chance. Conclusions Molecular methods can identify HIV-infected people socially linked with another person in about half of the phylogenetic clusters. This could help public health efforts to locate individuals in networks with high transmission rates. HIV-1, phylogenetics, transmission links, social ties, PWID A sharp increase in human immunodeficiency virus type 1 (HIV-1) diagnoses among people who inject drugs (PWID) in Athens, Greece, was observed between late 2010 and early 2011 [1–3]. The number of HIV-1 diagnoses peaked in 2012 and decreased afterward. Initial responses included high needle-and-syringe and opioid substitution programs coverage and a scale-up of HIV testing and counseling rates among PWID in the Athens metropolitan area. A large “seek, test, treat, and retain” program (ARISTOTLE) between 2012 and 2013 also significantly contributed to the reduction of HIV incidence among PWID [4, 5]. In addition, a social network–based intervention (Transmission Reduction Intervention Project [TRIP]) was conducted between 2013 and 2015 targeting mostly PWID in Athens [6, 7]. TRIP aimed at preventing HIV incidence by identifying recently HIV-infected individuals and their social networks in which transmissions are more likely to occur. The project recruited and tested social network members of recently infected individuals to identify more people who had been recently infected and therefore were more infectious. TRIP also distributed community alerts targeting individuals socially close to a new infection and made efforts to increase linkage to care for those testing HIV positive [6]. Social network–based data have been traditionally collected by interviews, field observation, and HIV testing techniques to make inferences about pathways of HIV spread. Molecular approaches have also been used in this field [8–13]. Of course, both social network–based and molecular approaches have limitations. For example, social network–based studies fail to describe the entire network due to incomplete recruitment of its members. Moreover, although social network–based studies include some information about the risk behavior of their members, in most cases detailed description of how HIV transmission occurs between the network members is not straightforward [14]. Molecular methods, on the other hand, provide an estimate of viral linkage, which in turn is a proxy of an actual HIV transmission pathway within a population [15], but they cannot infer who infected whom within the inferred transmission networks, nor do they tell us anything about the uninfected members of the network or the community [14]. The combination of social and molecular methods could fill gaps in how a risk network is structured. In an effort to combine social and molecular epidemiology (hereby named social phylogeny), the current article studies patterns of viral spread among PWID and their social contacts recruited in TRIP, Athens, and explores associations between transmission links, as estimated by phylogenetic analyses, and social networks. METHODS Participants and Data People eligible to participate in TRIP were those aged ≥18 years, able to answer the study questionnaire, and qualified for 1 of its 3 arms. Recruitment for each arm began with seeds and controls and was based on HIV testing results, HIV recency test results, and HIV testing history [6, 7]. The primary arm consisted of recently HIV-infected seeds (infected in the past 6 months) and their networks, the second arm included long-term HIV-infected controls who were also used as seeds for network recruitment (LCSs), and the third arm consisted of negative controls [6, 7]. Seeds were asked to elicit names of members of their networks and to help recruit them to be tested for HIV. Network members of recently HIV-infected seeds and LCSs were recruited for 2 steps. Recruitment stopped at second-degree network members unless a recently infected individual was identified and then a new 2-step recruitment process began. There was no social network recruitment for the negative controls. A total of 356 individuals (90.2% PWID) participated in TRIP between June 2013 and July 2015. One hundred fifty PWID were infected with HIV-1. Of these, viable samples for sequencing were obtained from 118 individuals (2 sequences from 1 individual) (Figure 1). Sequences were available in the protease and partial reverse transcriptase regions of the virus genome (pol gene) (873 nucleotides). Figure 1. View largeDownload slide Unrooted phylogenetic tree of human immunodeficiency virus type 1 sequences isolated from Transmission Reduction Intervention Project participants (N = 119) estimated by the RAxML (version 8) program. The names of subtypes, circulating recombinant forms, and unique recombinant forms are shown at the tips of the tree. Highly supported phylogenetic (transmission) clusters are marked in different colors. Numbers at the nodes show bootstrap support values. Abbreviations: CRF, circulating recombinant form; LTN, local transmission network; PWID, people who inject drugs; URF, unique recombinant form. Figure 1. View largeDownload slide Unrooted phylogenetic tree of human immunodeficiency virus type 1 sequences isolated from Transmission Reduction Intervention Project participants (N = 119) estimated by the RAxML (version 8) program. The names of subtypes, circulating recombinant forms, and unique recombinant forms are shown at the tips of the tree. Highly supported phylogenetic (transmission) clusters are marked in different colors. Numbers at the nodes show bootstrap support values. Abbreviations: CRF, circulating recombinant form; LTN, local transmission network; PWID, people who inject drugs; URF, unique recombinant form. Social network–based studies raise ethical concerns as participants are interviewed about sensitive issues and are asked to name contacts who have not yet consented to this disclosure of information. It should be noted, however, that the identity of the index person is never revealed, nor is information about contacts conveyed back to the index person. In addition, similar studies have been conducted in the past with multiple benefits for HIV prevention at both the individual and the public health level, and without causing any harm to the participants [16, 17]. Finally, the US Centers for Disease Control and Prevention also recommends social network–based approaches in routine partner services for HIV infection [18]. TRIP was approved by the institutional review boards of the National Development and Research Institutes in New York City and of the Hellenic Scientific Society for the Study of AIDS and Sexually Transmitted Diseases in Athens. All experiments were performed in accordance with relevant guidelines and regulations. All participants provided written informed consent. The accession numbers of the HIV-1 sequences generated on the purpose of the study are MH157288–MH157404. Social Networks Social networks were constructed according to questionnaire-based interviews, recruitment links, and field/venue observations. A first-degree social contact was an individual who had sex or had injected with a TRIP participant, or an individual who had sex or had injected in the presence of that participant. Second-degree network members were linked to first-degree contacts, but not the index. Social Network Visualizer version 1.9 was used to visualize social networks. Specifically, we used the degree centrality measure, which quantifies how many ties a node has to other nodes in the network and is considered a measure of social link activity. HIV-1 Sequence Subtyping Subtypes were identified for 119 sequences using the online automated HIV-1 subtyping tool COMET version 0.2 (Context-based Modeling for Expeditious Typing) (https://comet.lih.lu/). Subtyping results were further confirmed by performing phylogenetic analysis using 247 globally sampled sequences as references, which were representative of all known HIV-1 subtypes and circulating recombinant forms (CRFs), and were available on the Los Alamos HIV-1 sequence database (http://www.hiv.lanl.gov). Sequences from PWID sampled during an outbreak (2011–2015) in Athens were also used as references [19]. Moreover, we tested for the presence of recombination in all unclassified sequences using the RDP4 program [20], and also the bootscanning approach as implemented in the SimPlot version 3.5.1 program [21]. Sequence alignment was performed on MEGA (Molecular Evolutionary Genetics Analysis) version 5.2 using the MUSCLE algorithm, and alignments were manually edited according to the encoded reading frame [22]. Tree visualization and annotation were done using the FigTree version 1.4 program [23]. Phylogenetic (Transmission) Cluster Identification We performed phylogenetic analyses to identify the presence of highly supported transmission clusters (phylogenetic clusters) among our sequences. Specifically, phylogenetic analysis was conducted on (1) the total number of sequences (N = 158) by using all the available samples per participant (n = 118), and (2) the total number of sequences by using only 1 sample per participant (n = 118) with 1 exception. Specifically, 2 sequences for the same participant sampled at different time points fell within different local transmission networks (LTNs), suggesting superinfection. In case >1 sequence was available per participant, we used the oldest sample in which HIV was present. Phylogenetic trees were estimated from the underlying nucleotide sequences using different methods: (1) We analyzed our data by using the approximate maximum likelihood (ML) method under the generalized time reversible (GTR+cat) model of nucleotide substitution model including a Γ-distributed rate of heterogeneity among sites as implemented in FastTree version 2.1 and RAxML version 8 programs [24, 25]; (2) we also performed the analysis by neighbor-joining method with ML distances using the GTR model including a Γ distribution rate heterogeneity among sites as implemented in the PAUP*4.0 program [26]; (3) finally, we analyzed our data by using the Bayesian method with the GTR substitution model with Γ distributed rate, as implemented in MrBayes version 3.2.2 software [27]. The Markov chain Monte Carlo (MCMC) ran for 10 × 105 generations (burn-in was set to 10%), with 4 chains per run, and with MCMC sampling every 1000 steps. Each MCMC run checked for convergence using Tracer version 1.6. Highly supported phylogenetic clusters (transmission clusters) were defined as these consisted of ≥2 sequences, and with (1) Shimodaira-Hasegawa and bootstrap values greater than 0.90 and 70%, respectively, for phylogenetic trees estimated by ML and neighbor-joining method; and (2) a posterior probability >0.80 for phylogenetic trees estimated by the Bayesian method (phylogenetic confidence criterion). Only phylogenetic clusters fulfilling all 3 of these criteria were considered as highly supported. Significance Testing of the Proportion of Individuals With Social Ties That Were Found Within the Transmission Clusters We used social network–based ties to estimate the proportion of individuals found within the transmission (phylogenetic) clusters who also had social ties. Our purpose was to identify whether this percentage was significantly higher than what was expected by chance. Therefore, we performed a random shuffling of taxa (the set of viral sequences analyzed by phylogenetic analysis) at the tips (tips correspond to the external nodes or “leaves” of the phylogenetic tree) of the ML tree, estimated by the RAxML (version 8) program, to simulate a population with random mixing given the inferred phylogeny. In a random mixing population, an infected individual would have an equal probability of transmitting the virus to any other individual. We created 100 reshuffling replicates of the ML tree (we randomly reshuffled the tips 100 times) using Mesquite version 3.01 [28]. Then, for each reshuffled tree, we estimated the proportion of individuals within clusters who had social ties with at least 1 member of their cluster as is expected to occur under the assumption of random mixing. Finally, we tested whether the estimated proportion of individuals within clusters, who had social ties with at least 1 member of their cluster, differed significantly from the distribution of the expected proportions, against the null hypothesis of complete mixing (panmixis hypothesis). RESULTS HIV-1 Subtyping Results Subtyping analysis showed that 81.5% (n = 97) of the sequences clustered within previously identified PWID LTNs in the Athenian outbreak (CRF14_BG, CRF35_AD, subtype A, and subtype B) [1, 29]. The unique recombinant forms of HIV-1 were also identified, consisting of partial sequences from previously identified PWID LTNs (n = 13 [11.0%]). The rest of the sequences (n = 9 [7.5%]) were tied to non-PWID transmission clusters (subtype A, CRF56_cpx, B/D recombinant form). Subtyping results are presented in detail in Table 1. The proportion of sequences in the PWID LTNs was similar to that in the total population of HIV-1 infected PWID (CRF14_BG: 437 [50.0%]; CRF35_AD: 139 [15.9%]; subtype B: 116 [13.3%]; and subtype A: 54 [6.2%]) sampled during the outbreak (1 January 2011–31 October 2014) [29]. Discordances in the subtyping results obtained with COMET and manual phylogenetic analysis were found for 20 of 119 sequences (16.8%). Differences were found for some unique recombinant forms and CRF14_BG classified as subtype G in COMET. Table 1. Human Immunodeficiency Virus Type 1 Subtype Distribution and Clustering Pattern for Sequences Isolated From People Who Inject Drugs Transmission Subnetwork  HIV-1 Subtype/CRF  No. of Sequences (%)  No. of Clustered Sequences (%)  No. of Clusters  Range of Clusters  Individuals Within Clusters With Social Ties  PWID LTNs    A  9 (7.6)  7 (21.9)  2  2–5  2    B  17 (14.3)  6 (18.8)  2  3  4    CRF14_BG  53 (44.5)  15 (46.9)  7  2–3  4    CRF35_AD  18 (15.1)  2 (6.2)  1  2  2    Subtotal  97 (81.5)  30 (93.8)  12  2–5  12  Non-PWID LTNs    A  7 (5.9)  …  …  …  …    B/D recombinant  1 (0.8)  …  …  …  …    CRF56_cpx  1 (0.8)  …  …  …  …    Subtotal  9 (7.5)  …  …  …  …  URFs including LTNs    13 (11.0)  2 (6.2)  1  2  2    Total  119 (100)  32 (100)  13  2–5  14  Transmission Subnetwork  HIV-1 Subtype/CRF  No. of Sequences (%)  No. of Clustered Sequences (%)  No. of Clusters  Range of Clusters  Individuals Within Clusters With Social Ties  PWID LTNs    A  9 (7.6)  7 (21.9)  2  2–5  2    B  17 (14.3)  6 (18.8)  2  3  4    CRF14_BG  53 (44.5)  15 (46.9)  7  2–3  4    CRF35_AD  18 (15.1)  2 (6.2)  1  2  2    Subtotal  97 (81.5)  30 (93.8)  12  2–5  12  Non-PWID LTNs    A  7 (5.9)  …  …  …  …    B/D recombinant  1 (0.8)  …  …  …  …    CRF56_cpx  1 (0.8)  …  …  …  …    Subtotal  9 (7.5)  …  …  …  …  URFs including LTNs    13 (11.0)  2 (6.2)  1  2  2    Total  119 (100)  32 (100)  13  2–5  14  Abbreviations: CRF, circulating recombinant form; HIV-1, human immunodeficiency virus type 1; LTN, local transmission network; PWID, people who inject drugs; URF, unique recombinant form. View Large Table 1. Human Immunodeficiency Virus Type 1 Subtype Distribution and Clustering Pattern for Sequences Isolated From People Who Inject Drugs Transmission Subnetwork  HIV-1 Subtype/CRF  No. of Sequences (%)  No. of Clustered Sequences (%)  No. of Clusters  Range of Clusters  Individuals Within Clusters With Social Ties  PWID LTNs    A  9 (7.6)  7 (21.9)  2  2–5  2    B  17 (14.3)  6 (18.8)  2  3  4    CRF14_BG  53 (44.5)  15 (46.9)  7  2–3  4    CRF35_AD  18 (15.1)  2 (6.2)  1  2  2    Subtotal  97 (81.5)  30 (93.8)  12  2–5  12  Non-PWID LTNs    A  7 (5.9)  …  …  …  …    B/D recombinant  1 (0.8)  …  …  …  …    CRF56_cpx  1 (0.8)  …  …  …  …    Subtotal  9 (7.5)  …  …  …  …  URFs including LTNs    13 (11.0)  2 (6.2)  1  2  2    Total  119 (100)  32 (100)  13  2–5  14  Transmission Subnetwork  HIV-1 Subtype/CRF  No. of Sequences (%)  No. of Clustered Sequences (%)  No. of Clusters  Range of Clusters  Individuals Within Clusters With Social Ties  PWID LTNs    A  9 (7.6)  7 (21.9)  2  2–5  2    B  17 (14.3)  6 (18.8)  2  3  4    CRF14_BG  53 (44.5)  15 (46.9)  7  2–3  4    CRF35_AD  18 (15.1)  2 (6.2)  1  2  2    Subtotal  97 (81.5)  30 (93.8)  12  2–5  12  Non-PWID LTNs    A  7 (5.9)  …  …  …  …    B/D recombinant  1 (0.8)  …  …  …  …    CRF56_cpx  1 (0.8)  …  …  …  …    Subtotal  9 (7.5)  …  …  …  …  URFs including LTNs    13 (11.0)  2 (6.2)  1  2  2    Total  119 (100)  32 (100)  13  2–5  14  Abbreviations: CRF, circulating recombinant form; HIV-1, human immunodeficiency virus type 1; LTN, local transmission network; PWID, people who inject drugs; URF, unique recombinant form. View Large Phylogenetic (Transmission) Clusters Phylogenetic analysis of the total number of sequences (n = 158) using all samples per participant (n = 118) showed that 2 different sequences sampled from the same participant at different time points clustered in 2 distinct LTNs (subtype A and CRF14_BG), suggesting superinfection. Subsequent phylogenetic analysis of the total number of sequences, using only 1 sample per participant for 117 participants and 2 samples for the 1 participant with sequences from 2 distinct LTNs (n = 119), revealed the existence of 13 highly supported phylogenetic clusters (transmission clusters) containing 2–5 individuals each. In total, these clusters consisted of 32 sequences (27.0%) (Figure 1). For the participant with sequences belonging to distinct LTNs, we analyzed both sequences. Hypothesis Testing We explored the potential association between transmission clusters, as estimated by phylogenetic analysis, and social network–based ties. We found that 7 of 13 phylogenetic clusters included 14 individuals (of 32 individuals found within clusters [43.8%]) who also had a first-degree social tie with at least 1 member of their phylogenetic cluster (Figure 2A). Further analysis revealed that phylogenetic clusters included 17 individuals (of 32 [53.1%]) who also had first-degree (n = 14) or second-degree (n = 3) social ties with at least 1 member of their phylogenetic cluster (Figure 2B). Then, we tested if the estimated proportion of individuals within clusters who had a first-degree social tie with at least 1 member of their phylogenetic cluster (43.8%) differed significantly from the distribution of the expected proportions under the null hypothesis of random mixing (panmixis hypothesis) (Figure 3). Our analysis revealed strong evidence for an observed proportion that was significantly higher than what was expected by chance (Figure 3). Figure 2. View largeDownload slide Social networks of individuals (n = 32) found within phylogenetic (transmission) clusters visualized by Social Network Visualizer (version 1.9): A, First-degree social ties. B, First- and second-degree social ties. Nodes are marked in different colors according to individual’s phylogenetic (transmission) cluster. Dotted circles in red represent the number of ties a node has to other nodes in its network. The nodes with the higher number of social ties are placed closer to the center of the circle. The degree of centrality measure quantifies the number of ties for node to others and it is often considered a measure of social tie activity. The number of ties is proportional to centrality. Diamond shape was used for individuals who had no social ties with other members of their phylogenetic (transmission) cluster. Figure 2. View largeDownload slide Social networks of individuals (n = 32) found within phylogenetic (transmission) clusters visualized by Social Network Visualizer (version 1.9): A, First-degree social ties. B, First- and second-degree social ties. Nodes are marked in different colors according to individual’s phylogenetic (transmission) cluster. Dotted circles in red represent the number of ties a node has to other nodes in its network. The nodes with the higher number of social ties are placed closer to the center of the circle. The degree of centrality measure quantifies the number of ties for node to others and it is often considered a measure of social tie activity. The number of ties is proportional to centrality. Diamond shape was used for individuals who had no social ties with other members of their phylogenetic (transmission) cluster. Figure 3. View largeDownload slide Proportions of individuals within phylogenetic (transmission) clusters with social ties. The red dot corresponds to the estimated proportion of individuals within transmission clusters who had first-degree social ties with at least 1 other member of their cluster (43.8%). The distribution of the proportion was estimated after 100 reshuffling rounds at the tips of the maximum likelihood tree (panmixis hypothesis). Figure 3. View largeDownload slide Proportions of individuals within phylogenetic (transmission) clusters with social ties. The red dot corresponds to the estimated proportion of individuals within transmission clusters who had first-degree social ties with at least 1 other member of their cluster (43.8%). The distribution of the proportion was estimated after 100 reshuffling rounds at the tips of the maximum likelihood tree (panmixis hypothesis). DISCUSSION Social network–based extensions of traditional contact tracing methods are promising public health interventions to prevent HIV transmissions [7, 30]. TRIP aimed to reduce HIV transmission by identifying recently HIV-infected individuals and tracing their social networks where transmissions occur at high rates [6]. Early identification and treatment of recently infected individuals are expected to have benefits at both the individual and the public health level [31, 32]. An important question, however, is the extent to which phylogenetically based clusters are related to social network–based ties. This relationship can affect the efficiency and value of social network–based interventions to reduce HIV incidence. We found, using a combination of molecular and social network–based tracing methods, that a high proportion of individuals (43.8%) in phylogenetic clusters also have first-degree social ties with at least 1 member of their cluster. This means that almost half of the individuals with viral linkage are also socially linked, suggesting that during an HIV outbreak among PWID, transmissions occurred preferentially within socially linked individuals and that social network–based methods helped us locate them. To our knowledge, this is one of the few studies to combine molecular and social network–based data. This is of public health importance, especially for HIV prevention in highly vulnerable, marginalized populations (PWID, transgender people, undocumented migrants, racial/ethnic groups, homeless people, and, in some countries, gay and bisexual men). In a previous study, Wertheim et al estimated the proportions of named partners infected with genetically closely related viruses and vice-versa (ie, the proportion of individuals for whom viruses were closely related) and they were linked by naming their partners in New York [12]. They found that 53% of individuals with genetically similar viruses were also linked based on their reports about the partners they have. Their estimates were close to ours (43.8% and 53.1% for first- and second-degree of social ties, respectively). Wertheim et al estimated transmission clusters based on pairwise comparisons that fell below a genetic threshold [12]. In the current study, we selected all sequences within well-defined phylogenetic clusters to belong to transmission clusters. The approach that Wertheim et al [12] used is probably better suited to identify linked pairs, especially among sexual partners; however, for studying an HIV outbreak situation, where genetically identical viruses have spread rapidly among PWID, choosing the appropriate genetic threshold is challenging [33]. In the absence of previously well-established threshold for genetic distances, we defined transmission clusters based on bootstrap/Shimodaira-Hasegawa/posterior support that probably corresponds to PWID infected from a common source rather than those with direct transmission links. The high similarity of the genetic sequences analyzed in the current study might be also due to the rather high conservation of the gene that was sequenced (partial pol). In another study related to the outbreak investigation among PWID in Indiana, Campbell et al inferred transmission clusters based on a combined analysis of epidemiological and molecular data [33]. For those reported high-risk contacts, 34% were found as potential transmission partners. This proportion is close to our estimate, although different approaches have been used. Based on these findings, it seems that HIV transmissions occur frequently among persons with measurable social ties, which highlights the importance of social network research for HIV prevention and epidemiology. Another group of studies was based on the respondent-driven sample (RDS) strategy, which limits the number of people recruited per individual and therefore has certain limitations for the description of social networks. The benefits of social network–based tracing approaches over RDSs are that they allow the recruitment of an unlimited number of individuals reporting social ties to an index case and that they collect data on, and allow analysis of, social and risk linkages of a given participant with people other than the ones who gave them a coupon or whom they gave a coupon to. In previous studies using molecular methods, RDS was performed mostly among men who have sex with men (MSM) and female sex workers. Lepej et al suggested that 5 of 12 (42%) MSM recruited with RDS belonged to a single phylogenetic cluster; however, they were not directly linked and 3 of them belonged to different RDS chains [34]. In a similar analysis in El Salvador, among 34 MSM who belonged to transmission clusters, 8 (24%) were recruited in the same RDS chain [35]. On the other hand, Fujimoto et al showed that viral transmissions do not occur between the socially connected young adult black MSM [36]. Additionally, phylogenetic studies have been used extensively to study characteristics of transmission networks between different transmission risk and racial groups and across diverse geographic locations and between subpopulations [9, 10, 37–50]. As of now there are no consensus criteria of phylogenetic confidence (eg, bootstrap or posterior probability support) for the definition of phylogenetic clusters [10, 51]. Evidence from a previous study suggests that the selection of a stricter definition of phylogenetic clusters (higher bootstrap values in combination with lower genetic distance thresholds) resulted in higher proportions of viral linkage from individuals with recent-phase transmissions [52]. We chose less strict phylogenetic criteria to avoid potential bias toward identifying recently infected individuals within clusters. We chose, however, to perform our analysis using different methods and 3 criteria for clustering (Shimodaira-Hasegawa, bootstrap, and posterior probability support). Phylogenetic analyses have certain limitations for identifying transmission clusters—for example, that clustering inferences depend on sampling coverage, which is the proportion of individuals for whom viral sequences are available [14, 15]. In our study, the sampling coverage was 79% (118 of 150 participants who were infected with HIV-1). Moreover, phylogenetic resolution depends on the number of nucleotides analyzed and also on the levels of genetic diversity within the alignment. Given these limitations, it has been suggested that phylogenetic trees represent a proxy of an actual HIV transmission pathway [15]. The design of TRIP and the recent nature of the HIV-1 outbreak among PWID in Athens, Greece, which started approximately between late 2010 and early 2011 [3], suggest that the high proportion of individuals with social ties within their phylogenetic clusters was probably due to the proximity in time of interviews and transmissions. Specifically, TRIP recruitment took place between June 2013 and July 2015, and among the 150 HIV-infected participants, 45 (30%) were recently infected. Social networks are complex and evolve over time as individuals modify their relationships or their risk behavior [15]. We show here that, under the conditions of our study, phylogenetically based transmission pathways were significantly associated with social networks traced relatively close to transmission dates. Specific conditions under which this association exists in different epidemics and populations remain to be elucidated. A limitation of this work is the high genetic similarity of sequences from the study population (which may be due to the short period of time since the start of the outbreak). To overcome this limitation, we selected as transmission (phylogenetic) clusters only those highly supported by 3 different criteria for clustering. Moreover, we tested if the proportion of those within transmission clusters with social ties is higher than panmixis by simulating the null hypothesis across the optimized topology (tree). The proportion of individuals within transmission clusters and, consequently, the fraction of those with social ties might be higher if more genetically diverse genomic regions were used. Finally, our analysis benefits from the combination of detailed social network–based and molecular data to explore whether HIV transmission pathways follow social ties. This is important to understand the role of social networks in HIV transmission and consequently to identify critical parameters of the epidemic (eg, populations at higher risk to transmit). Molecular analyses can be used to trace past transmission events and, when combined with social data, can be valuable for public health interventions. Notes Financial support. The study was supported by the National Institute on Drug Abuse (award number DP1 DA034989) and the Hellenic Scientific Society for the Study of AIDS and Sexually Transmitted Diseases. Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed. Presented in part: 13th International Conference on Molecular Epidemiology and Evolutionary Genetics of Infectious Diseases, Antwerp, Belgium, May 2016. References 1. Paraskevis D, Nikolopoulos G, Tsiara Cet al.   HIV-1 outbreak among injecting drug users in Greece, 2011: a preliminary report. Euro Surveill  2011; 16. pii: 19962. Google Scholar CrossRef Search ADS PubMed  2. Pharris A, Wiessing L, Sfetcu Oet al.   Human immunodeficiency virus in injecting drug users in Europe following a reported increase of cases in Greece and Romania, 2011. Euro Surveill  2011; 16. pii: 20032. Google Scholar PubMed  3. Paraskevis D, Paraschiv S, Sypsa Vet al.   Enhanced HIV-1 surveillance using molecular epidemiology to study and monitor HIV-1 outbreaks among intravenous drug users (IDUs) in Athens and Bucharest. Infect Genet Evol  2015; 35: 109– 21. Google Scholar CrossRef Search ADS PubMed  4. Hatzakis A, Sypsa V, Paraskevis Det al.   Design and baseline findings of a large-scale rapid response to an HIV outbreak in people who inject drugs in Athens, Greece: the ARISTOTLE programme. Addiction  2015; 110: 1453– 67. Google Scholar CrossRef Search ADS PubMed  5. Sypsa V, Psichogiou M, Paraskevis Det al.   Rapid decline in HIV incidence among persons who inject drugs during a fast-track combination prevention program after an HIV outbreak in Athens. J Infect Dis  2017; 215: 1496– 505. Google Scholar PubMed  6. Friedman SR, Downing MJJr, Smyrnov Pet al.   Socially-integrated transdisciplinary HIV prevention. AIDS Behav  2014; 18: 1821– 34. Google Scholar CrossRef Search ADS PubMed  7. Nikolopoulos GK, Pavlitina E, Muth SQet al.   A network intervention that locates and intervenes with recently HIV-infected persons: the Transmission Reduction Intervention Project (TRIP). Sci Rep  2016; 6: 38100. Google Scholar CrossRef Search ADS PubMed  8. Friedman SR, Ompad DC, Maslow Cet al.   HIV prevalence, risk behaviors, and high-risk sexual and injection networks among young women injectors who have sex with women. Am J Public Health  2003; 93: 902– 6. Google Scholar CrossRef Search ADS PubMed  9. Paraskevis D, Nikolopoulos GK, Magiorkinis G, Hodges-Mameletzis I, Hatzakis A. The application of HIV molecular epidemiology to public health. Infect Genet Evol  2016; 46: 159– 168. Google Scholar CrossRef Search ADS PubMed  10. Grabowski MK, Redd AD. Molecular tools for studying HIV transmission in sexual networks. Curr Opin HIV AIDS  2014; 9: 126– 33. Google Scholar CrossRef Search ADS PubMed  11. Brenner BG, Wainberg MA. Future of phylogeny in HIV prevention. J Acquir Immune Defic Syndr  2013; 63( Suppl 2): S248– 54. Google Scholar CrossRef Search ADS PubMed  12. Wertheim JO, Kosakovsky Pond SL, Forgione LAet al.   Social and genetic networks of HIV-1 transmission in New York City. PLoS Pathog  2017; 13: e1006000. Google Scholar CrossRef Search ADS PubMed  13. Friedman SR, Curtis R, Neaigus A, Jose B, Des Jarlais DC. Social networks, drug injectors’ lives, and HIV/AIDS . New York: Kluwer/Plenum, 1999. 14. Vasylyeva TI, Friedman SR, Paraskevis D, Magiorkinis G. Integrating molecular epidemiology and social network analysis to study infectious diseases: towards a socio-molecular era for public health. Infect Genet Evol  2016; 46: 248– 55. Google Scholar CrossRef Search ADS PubMed  15. Delva W, Leventhal GE, Helleringer S. Connecting the dots: network data and models in HIV epidemiology. AIDS  2016; 30: 2009– 20. Google Scholar CrossRef Search ADS PubMed  16. Friedman SR, Neaigus A, Jose Bet al.   Sociometric risk networks and risk for HIV infection. Am J Public Health  1997; 87: 1289– 96. Google Scholar CrossRef Search ADS PubMed  17. Woodhouse DE, Rothenberg RB, Potterat JJet al.   Mapping a social network of heterosexuals at high risk for HIV infection. AIDS  1994; 8: 1331– 6. Google Scholar CrossRef Search ADS PubMed  18. Centers for Disease Control and Prevention. Recommendations for partner services programs for HIV infection, syphilis, gonorrhea, and chlamydial infection. MMWR Recomm Rep  2008; 57: 1– 83; quiz CE1-4. 19. Paraskevis D, Nikolopoulos G, Fotiou Aet al.   Economic recession and emergence of an HIV-1 outbreak among drug injectors in Athens metropolitan area: a longitudinal study. PLoS One  2013; 8: e78941. Google Scholar CrossRef Search ADS PubMed  20. Martin DP, Lemey P, Lott M, Moulton V, Posada D, Lefeuvre P. RDP3: a flexible and fast computer program for analyzing recombination. Bioinformatics  2010; 26: 2462– 3. Google Scholar CrossRef Search ADS PubMed  21. Lole KS, Bollinger RC, Paranjape RSet al.   Full-length human immunodeficiency virus type 1 genomes from subtype C-infected seroconverters in India, with evidence of intersubtype recombination. J Virol  1999; 73: 152– 60. Google Scholar PubMed  22. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol Biol Evol  2011; 28: 2731– 9. Google Scholar CrossRef Search ADS PubMed  23. Drummond AJ, Suchard MA, Xie D, Rambaut A. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Mol Biol Evol  2012; 29: 1969– 73. Google Scholar CrossRef Search ADS PubMed  24. Stamatakis A. RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics  2014; 30: 1312– 3. Google Scholar CrossRef Search ADS PubMed  25. Price MN, Dehal PS, Arkin AP. FastTree 2–approximately maximum-likelihood trees for large alignments. PLoS One  2010; 5: e9490. Google Scholar CrossRef Search ADS PubMed  26. Wilgenbusch JC, Swofford D. Inferring evolutionary trees with PAUP. Curr Protoc Bioinformatics  2003; chapter 6, unit 6.4. Google Scholar PubMed  27. Ronquist F, Teslenko M, van der Mark Pet al.   MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol  2012; 61: 539– 42. Google Scholar CrossRef Search ADS PubMed  28. Maddison WP, Maddison DR. Mesquite: a modular system for evolutionary analysis. Version 3.10. 2018. Available at http://mesquiteproject.org. 29. Kostaki E, Magiorkinis G, Psichogiou Met al.   Detailed molecular surveillance of the HIV-1 outbreak among people who inject drugs (PWID) in Athens during a period of four years. Curr HIV Res  2017; 15: 396– 404. Google Scholar PubMed  30. Green N, Hoenigl M, Chaillon Aet al.   Partner services in adults with acute and early HIV infection. AIDS  2017; 31: 287– 93. Google Scholar CrossRef Search ADS PubMed  31. Cohen MS, Chen YQ, McCauley Met al.   HPTN 052 Study Team. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med  2011; 365: 493– 505. Google Scholar CrossRef Search ADS PubMed  32. Lundgren JD, Babiker AG, Gordin Fet al.   Initiation of antiretroviral therapy in early asymptomatic HIV infection. N Engl J Med  2015; 373: 795– 807. Google Scholar CrossRef Search ADS PubMed  33. Campbell EM, Jia H, Shankar Aet al.   Detailed transmission network analysis of a large opiate-driven outbreak of HIV infection in the United States. J Infect Dis  2017; 216: 1053– 62. Google Scholar CrossRef Search ADS PubMed  34. Lepej SZ, Vrakela IB, Poljak M, Bozicevic I, Begovac J. Phylogenetic analysis of HIV sequences obtained in a respondent-driven sampling study of men who have sex with men. AIDS Res Hum Retroviruses  2009; 25: 1335– 8. Google Scholar CrossRef Search ADS PubMed  35. Dennis AM, Murillo W, de Maria Hernandez Fet al.   Social network-based recruitment successfully reveals HIV-1 transmission networks among high-risk individuals in El Salvador. J Acquir Immune Defic Syndr  2013; 63: 135– 41. Google Scholar CrossRef Search ADS PubMed  36. Fujimoto K, Coghill LM, Weier CAet al.   Short communication: Lack of support for socially connected HIV-1 transmission among young adult black men who have sex with men. AIDS Res Hum Retroviruses  2017; 33: 935– 40. Google Scholar CrossRef Search ADS PubMed  37. Chan PA, Hogan JW, Huang Aet al.   Phylogenetic investigation of a statewide HIV-1 epidemic reveals ongoing and active transmission networks among men who have sex with men. J Acquir Immune Defic Syndr  2015; 70: 428– 35. Google Scholar CrossRef Search ADS PubMed  38. Poon AF, Joy JB, Woods CKet al.   The impact of clinical, demographic and risk factors on rates of HIV transmission: a population-based phylogenetic analysis in British Columbia, Canada. J Infect Dis  2015; 211: 926– 35. Google Scholar CrossRef Search ADS PubMed  39. Kharsany AB, Buthelezi TJ, Frohlich JAet al.   HIV infection in high school students in rural South Africa: role of transmissions among students. AIDS Res Hum Retroviruses  2014; 30: 956– 65. Google Scholar CrossRef Search ADS PubMed  40. Avila D, Keiser O, Egger Met al.   Swiss HIV Cohort Study. Social meets molecular: combining phylogenetic and latent class analyses to understand HIV-1 transmission in Switzerland. Am J Epidemiol  2014; 179: 1514– 25. Google Scholar CrossRef Search ADS PubMed  41. Middelkoop K, Rademeyer C, Brown BBet al.   Epidemiology of HIV-1 subtypes among men who have sex with men in Cape Town, South Africa. J Acquir Immune Defic Syndr  2014; 65: 473– 80. Google Scholar CrossRef Search ADS PubMed  42. Robertson AM, Garfein RS, Wagner KDet al.   Proyecto El Cuete IV and STAHR II. Evaluating the impact of Mexico’s drug policy reforms on people who inject drugs in Tijuana, B.C., Mexico, and San Diego, CA, United States: a binational mixed methods research agenda. Harm Reduct J  2014; 11: 4. Google Scholar CrossRef Search ADS PubMed  43. Volz EM, Frost SD. Inferring the source of transmission with phylogenetic data. PLoS Comput Biol  2013; 9: e1003397. Google Scholar CrossRef Search ADS PubMed  44. Volz EM, Ionides E, Romero-Severson EO, Brandt MG, Mokotoff E, Koopman JS. HIV-1 transmission during early infection in men who have sex with men: a phylodynamic analysis. PLoS Med  2013; 10: e1001568; discussion e1001568. Google Scholar CrossRef Search ADS PubMed  45. Lin H, He N, Zhou Set al.   Behavioral and molecular tracing of risky sexual contacts in a sample of Chinese HIV-infected men who have sex with men. Am J Epidemiol  2013; 177: 343– 50. Google Scholar CrossRef Search ADS PubMed  46. Levy I, Mor Z, Anis Eet al.   Men who have sex with men, risk behavior, and HIV infection: integrative analysis of clinical, epidemiological, and laboratory databases. Clin Infect Dis  2011; 52: 1363– 70. Google Scholar CrossRef Search ADS PubMed  47. Lewis F, Hughes GJ, Rambaut A, Pozniak A, Leigh Brown AJ. Episodic sexual transmission of HIV revealed by molecular phylodynamics. PLoS Med  2008; 5: e50. Google Scholar CrossRef Search ADS PubMed  48. Brenner BG, Roger M, Moisi DDet al.  ; Montreal PHI Cohort and HIV Prevention Study Groups. Transmission networks of drug resistance acquired in primary/early stage HIV infection. AIDS  2008; 22: 2509– 15. Google Scholar CrossRef Search ADS PubMed  49. Vasylyeva TI, Liulchuk M, Friedman SRet al.   Molecular epidemiology reveals the role of war in the spread of HIV in Ukraine. Proc Natl Acad Sci U S A  2018; 115: 1051– 6. Google Scholar CrossRef Search ADS PubMed  50. Paraskevis D, Kostaki E, Nikolopoulos GKet al.   Molecular tracing of the geographical origin of human immunodeficiency virus type 1 infection and patterns of epidemic spread among migrants who inject drugs in Athens. Clin Infect Dis  2017; 65: 2078– 84. Google Scholar CrossRef Search ADS PubMed  51. Hassan AS, Pybus OG, Sanders EJ, Albert J, Esbjörnsson J. Defining HIV-1 transmission clusters based on sequence data. AIDS  2017; 31: 1211– 22. Google Scholar CrossRef Search ADS PubMed  52. Marzel A, Shilaih M, Yang WLet al.  ; Swiss HIV Cohort Study. HIV-1 transmission during recent infection and during treatment interruptions as major drivers of new infections in the Swiss HIV Cohort Study. Clin Infect Dis  2016; 62: 115– 22. Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

The Journal of Infectious DiseasesOxford University Press

Published: Apr 24, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off