Determining how network-level factors influence individual risk of human immunodeficiency virus (HIV) acquisition is vital in preventing disease transmission. Debate remains as to how risk of HIV acquisition is affected given one’s network composition of those recently infected with HIV or those with long-term HIV infection, but who are not virally suppressed. Additionally, phylogenetic analysis can be utilized to build molecular networks and identify clusters of HIV transmission by determining similarities among HIV genetic sequences from persons infected with HIV. In this dissertation work social and sexual network data were combined with HIV molecular network data: 1) to determine characteristics associated with both membership in an HIV molecular cluster and the number of clustered sequences within these clusters; 2) to examine whether new HIV seroconversions occurring among young Black men who have sex with men (YBMSM) are proximal to either recently or long-term HIV infected individuals; and 3) to examine potential overlap and to ascertain the benefits of combining these types of analyses. A cohort of YBMSM (N = 618) was generated through respondent driven sampling with survey data collected across three waves from 2013 through 2016. Dried blood spots were obtained at each wave and were assessed for HIV seropositivity and/or acute HIV infection using 4th generation HIV immunoassay, HIV-1/-2 Ab differentiation, and HIV RNA testing. Chicago Department of Public Health (CDPH) HIV surveillance data was utilized to determine date of HIV diagnosis in combination with survey data. Pairwise genetic distances of HIV-1 pol sequences were used to identify potential molecular ties among HIV-infected persons whose sequences were ≤2.0% genetically distant. Putative molecular clusters were defined as ≥1 connection to another individual. We then determined demographic and risk attributes associated with both membership in an HIV molecular cluster and the number of ties to other persons within the cluster. RDS-weighted Cox and logistic regressions were utilized to examine proximity of HIV transmission events to other recent or long-term HIV infected individuals in the risk network. Finally, confidant, sexual, and Facebook© network data were matched between named partners across all three waves utilizing a computer algorithm with all matches confirmed manually by two separate analysts. An absence of tie overlap was observed between the molecular network and first, second and third degree confidant network as well as the molecular network and first, second and third degree sexual network. We did, however, find that of the 15 individuals with both molecular and Facebook© data available, 3 (25.0%) of the 12 network ties overlapped between the two networks. We also found a consistent 45-50% overlap of network ties between the social, sexual, and Facebook © networks. Across all waves, 266 (43.0%) participants were identified as HIV-positive with 139 identified as having an undectecable viral load. Of the remaining HIV-positive individuals, we successfully sequenced the pol region of the viral genome for 42 (30.2%) individuals. We obtained a further 44 viral sequences from CDPH HIV surveillance data for a total of 86 (61.9%) available sequences., Thirty-five (40.7%) of these sequences were tied to ≥1 other sequence with a total of 55 ties between all individuals. Through multivariable analyses, we determined that those who identified as straight and those who reported symptoms of depression were significantly less likely to be members of a molecular cluster. Additionally, as the number of confidants in a participant’s social network increased, the odds of membership in a molecular cluster decreased significantly. We also found that those who identified as straight, had stable housing, and reported less frequent marijuana use had significantly fewer ties to other individuals’ sequences in molecular clusters. Within the cohort, 343 (55.5%) participants were identified as HIV seronegative at baseline. Of these, 33 (9.6%) seroconverted during the study period. We found that the odds of seroconversion increased significantly with each additional recently HIV infected individual in one’s network (AOR = 12.96; 95% CI: 5.69-29.50). The number of long-term infected individuals in one’s network did not significantly alter odds of seroconversion. We also found that for each additional member of one’s network who used PrEP, the odds of seroconversion decreased significantly, adjusting for overall network size as well as the number of HIV-negative individuals in one’s network (AOR = 0.44; 95% CI: 0.20-0.96). This work presented in this dissertation demonstrates the potential for combining molecular, social, and other individual and network attributes in a sociomolecular approach to target HIV control efforts to persons with potentially higher transmission risk. This work also suggests some unappreciated specific predictors of transmission risk among YBSM in Chicago for future study. An increase in the number of recently HIV infected individuals in one’s network was associated with an increased rate of seroconversion and the odds of seroconversion are significantly reduced by an increase in the number of network members who use PrEP. Limited overlap was also observed between reported social network ties and molecular ties in this young MSM population, suggesting that transmissions may have occurred in networks that existed more than six months ago (limited by the recall period) before the observed networks. Increasing the social network cluster size by two and three degrees only marginally improved overlap between reported social and molecular network ties. Virtual network data, such as Facebook, may be particularly useful in developing a more complete picture of one’s risk environment and identify most at risk community members not observed in phylogenetic networks.




Downloads Statistics

Download Full History