HIV-1 Genetic and Contact Networks Among Men Who Have Sex With Men May Inform Public Health Efforts

Despite decreases in overall HIV-1 incidence in the United States, incidence has been stable or increasing among Black men who have sex with men (MSM), particularly among those aged 25 to 34 years.¹ National incidence trends also obscure significant localized trends, with estimated HIV-1 prevalence rates higher than 30% among MSM in some US counties and metropolitan areas.² Remarkable progress in HIV-1 prevention can be made through expansion of antiretroviral therapy for treatment and preexposure prophylaxis. However, reaching and engaging persons who would benefit the most are challenging and often hampered by societal factors that disproportionately affect US Black men, including high rates of incarceration, poverty, unemployment, and lack of access to affordable preventive medical care and health insurance. Societal factors coupled with smaller sexual networks, more same-race partnerships, and higher HIV-1 prevalence compared with other racial/ethnic groups contribute to the heightened HIV infection risk among Black MSM. Altogether, new avenues leveraging network-based interventions to disrupt ongoing HIV-1 transmission are needed.

UCONNECT

In this issue of AJPH, Morgan et al. (p. 1528) report on the combination of HIV-1 molecular, social, and sexual network data derived from uConnect, a longitudinal population-based cohort study of young, Black MSM residing in Chicago, Illinois. The investigators sought to examine potential overlap between the three networks to show how phylogenetic analyses can inform existing network recruitment approaches and vice versa. In total, data from 86 MSM with available HIV-1 pol sequences were evaluated with comparison of recruitment, social, and sexual ties. Despite this attempt to link networks, Morgan et al. found no direct sexual links between persons with closely phylogenetically related viruses, who could have been part of the same HIV-1 transmission chain. However, a few overlapping links were found among Facebook and molecular networks.

These results immediately raise the question of how such molecular data can be exploited if they do not overlap with the current sexual or social contact networks. Furthermore, how could these data be used in the context of the public health system and resource-limited field services? Although the authors provide a potential public health framework, other pitfalls remain. For example, HIV-1 molecular data used to guide public health responses carry concern for ethical and legal implications, which were not mentioned in this article.

Although unexpected on first glance, the finding of minimal or no direct overlap between the molecular and contact networks is not surprising given how the data were collected. Participants were recruited through respondent-driven sampling, which limits the number of recruits per individual. In uConnect, each participant could recruit up to six other persons who met the eligibility criteria. Furthermore, participants were asked to name sexual partners and confidants, but this was restricted to six recent sexual partners in the past six months and up to five confidants. Only these named persons who met eligibility criteria were enrolled in the respondent-driven sampling, and it appears that only these individuals (if an HIV-1 sequence was available) were included in the phylogenetic analysis, thus limiting its sample size. Given the dynamic nature of sexual and social networks, direct overlap with HIV-1 transmission networks could be difficult to recover in this limited sample.

HIV molecular networks (also termed genetic or transmission clusters) are historical events, which may be uncovered by improved sampling after diagnosis and may not necessarily indicate increased transmission rates; this is an important consideration when interpreting such clusters.³ Fewer than one third of HIV-infected recruits had a sequence available for analysis, and many were previously diagnosed. Some persons had sequences obtained through the public health surveillance system, indicating linkage to HIV clinical care where preantiretroviral drug resistance testing is routinely done. A lag between HIV diagnosis, entry to care and HIV sequencing, and cluster or network analysis could limit the usefulness of genetic information to inform near–real-time HIV transmission events.

CURRENT AT-RISK NETWORK

Social and sexual contact data, however, do provide important information about the current at-risk network, which cannot be derived from HIV molecular networks. Therefore, the combination of networks may uncover important geographic and demographic subsets at highest risk where interventions such as preexposure prophylaxis could be more intensely allocated. However, contact network–based strategies and field services must address the frequent use of Web-based platforms and “hook-up” apps by MSM to facilitate sexual encounters.⁴ In this study, more than 30% of the participants reported frequent use of hook-up apps.

Notably, the only overlap between social and molecular ties identified in this study was among a subset with Facebook contacts, another online platform used to facilitate encounters. Anonymous sexual partnerships or an unwillingness to provide information necessary to locate partners significantly limits successful contact tracing by field services—including the paradigm suggested by Morgan et al.—should a large number be unable or unwilling to provide partner contact information.

MOLECULAR NETWORK

The potential for HIV-1 sequence data to inform the local epidemiology extends beyond analyzing pairwise genetic distances from a small group. In the reported study, because the phylogenetic analysis was restricted to only MSM who participated in the respondent-driven sampling (thus restricted to Black race and young age), the full extent of how the molecular network overlapped within and outside the young, Black MSM community was not fully realized. More sophisticated phylodynamic methods, which take into account sequence sampling dates and population genetic modeling, can lead to improved estimates of source attribution and transmission rates among subpopulations.⁵

The study presented by Morgan et al. highlights the potential of network-guided approaches that could leverage social network interventions to reduce HIV incidence. In fact, the Centers for Disease Control and Prevention recently expanded the Molecular HIV Surveillance Program to describe and respond to HIV transmission clusters nationwide.⁶ However, ethical and legal consideration should be noted, including potential concerns about the invasion of privacy, incitement of stigma, and promotion of discrimination or HIV criminalization. Molecular network and phylogenetic analyses cannot confirm direct transmission between individuals or establish directionality, because unsampled persons may be involved in the transmission chain. Nonetheless, communication with public health departments and communities about the interpretation, use, and safeguarding of HIV genetic and surveillance data is essential.

Although investigators are excited about the potential of using HIV-1 genetic clustering to guide public health interventions, a systematic analysis to evaluate the effectiveness of such approaches has not yet been done. Furthermore, more work is needed to determine which clusters to prioritize, such as incorporating models to predict cluster growth or early detection of outbreaks. Integration of timely data, including sequences, RNA viral loads, and acute or recent infection, would capitalize further on the potential public health use of the molecular network information.

To be successful, field investigation responses likely will need to be multifactorial, including “enhanced” services allocated to contacts or network members, such as expanded HIV testing, expedited linkage to care and antiretroviral therapy for new HIV-positive patients, re-engagement and field services for out-of-care HIV-positive patients, and facilitated preexposure prophylaxis linkage for HIV-negative persons. Even though such activities could be resource intensive, they may ultimately be more effective in reducing HIV incidence because of intense resource allocation toward ongoing transmission. Such activities must be done in the context of sensitivity toward interpretation of molecular network data, confidentiality provided by legal jurisdiction, and trust in local public health authorities to avoid implicating persons as potential transmitters and prevent further stigmatization of already difficult-to-reach populations.