Determining the Direction of Human Immunodeficiency Virus Transmission: Benefits and Potential Harms of Taking Phylogenetic Analysis One Step Further

phylogenetic analysis, HIV, direction of transmission, medical ethics

(See the Major Article by Zhang et al on pages 30–7.)

Phylogenetic studies of human immunodeficiency virus (HIV) transmission dynamics are an important tool in addressing the HIV epidemic [1]. Phylogenetic techniques contributed to our understanding of the origins of the HIV epidemic [2, 3], transmission dynamics within localized epidemics [4–7], and the efficacy of treatment as prevention [8, 9]. Molecular cluster analysis is one of the 4 pillars of the US End the HIV Epidemic Effort, used by the US Centers for Disease Control and Prevention (CDC) to prioritize public health resource allocation [10]. In this issue of Clinical Infectious Diseases, Zhang and colleagues take phylogenetic analysis one step further to determine not just relatedness of HIV transmission but direction of transmission—literally, who infected whom. Their findings represent an advance in our ability to understand transmission dynamics and could be used to direct more effective outreach and prevention interventions for populations in need. However, they also raise important ethical concerns. We, as an HIV care provider and an advocate living with HIV, have direct experience with phylogenetic analysis when applied to a population. San Antonio was home to the first large molecular “cluster” of rapid HIV transmission identified by the CDC and Texas Department of State Health Services in 2016 [11]. Our knowledge of the community impact of those findings informs our response to the individual and societal implications of the current research.

Prior efforts to use phylogenetics to determine direction of transmission met with less success. Previously, investigators from the HIV Prevention Trials Network (HPTN) Study 052 used next-generation sequencing of the env gene from partner pairs for whom direction of transmission was already established. They correctly predicted direction of transmission in 67%–74% of dyads and incorrectly identified the direction of transmission in 13%–21% [12]. A second study determined direction of transmission among newly 70 diagnosed men who have sex with men in 4 of 8 known clusters (50%) using Sanger and ultra-deep sequencing of the pol region [13, 14]. The Phylogenetics and Networks for Generalized HIV Epidemics in Africa consortium Consortium conducted ultra-deep sequencing of 3758 HIV samples from the Rakai District in Uganda. Of 36 pairs where serial HIV testing established direction of transmission, phylogenetic analysis assigned the incorrect direction of transmission to 4 cases and could not ascertain direction of transmission for 5 pairs [15].

In this context, the study by Zhang and colleagues is highly successful. Using data from HPTN 052, they generated whole genome next-generation sequences from 105 paired samples from 32 couples with known direction of transmission. They used phyloscanner to infer linkage and direction of transmission of overlapping windows covering the entire HIV genome. Among the 105 sample pairs, 98 (93.3%) of them were correctly assigned direction of transmission by this method. Four pairs had unknown status, and 3 were linked but direction could not be inferred. Overall, this methodology inferred the correct direction of transmission for 31 of 32 couples and did not assign incorrect direction of transmission to any. Accuracy decreased when less than the whole genome sequences were used, with >10% of sample pairs assigned incorrect transmission direction when a short env fragment was used. The accuracy of this technique is higher than any reported previously.

These results may be difficult to replicate outside of a trial like HPTN 052, a relatively small sample of known sexual dyads that allowed for a gold standard against which to assess the accuracy of phylogenetic results. Applying this methodology to molecular surveillance datasets is currently impossible, as those surveillance strategies construct phylogenies based on single, incomplete sequences of the pol gene derived from HIV genotype testing [16]. All transmissions within HPTN 052 were likely sexual transmission in cisgender male–female partnerships, and the accuracy of these techniques applied to other types of partnerships or injection drug use transmission is unknown. However, previously published studies imply that researchers would like to extend phylogenetic direction of transmission assessment to population-based or public health surveillance cohorts [13, 15].

This leads to an important thought experiment: What would happen if we could use phylogenetic methods to determine direction of transmission in HIV public health surveillance? The scientific community has addressed the ethics of phylogenetic studies previously [17–20], but much of this discussion assumes that inferring direction of transmission from phylogenetic data is impossible. If we accept that these methodologies are evolving rapidly and we are moving closer to determining direction of transmission, the dialogue changes.

The principle of beneficence, outlined in the 1987 Belmont Report [21], guides researchers to ensure that risks to participants are reasonable in relation to potential benefits. In the specific case of direction of transmission, researchers believe this knowledge will enable targeting of prevention measures to specific populations [1, 12, 15]. These theoretical benefits are an extension of the benefits from molecular surveillance. In San Antonio, state and local health departments worked to find individuals in the molecular cluster and relink them to care [11]. The data also served to galvanize the community, with political leaders signing on to the Fast Track Cities Initiative and the formation of a community-wide collaboration with leadership from people living with HIV.

What about risks to participants or, extending beneficence beyond the research context of the Belmont Report, to the community? Zhang and colleagues describe the need to avoid releasing data on HIV transmission direction that could be linked to individuals, and they acknowledge the risk of incarceration. Their proposed solution is anonymization of samples and sequence data. However, multiple legal precedents exist for the use of phylogenetic analysis in criminal prosecutions [22–24]. HIV-specific criminal laws or sentence enhancements exist in 34 states in the United States, and a disproportionate burden of prosecutions falls on communities of color [25]. Some states require sharing of public health information with law enforcement [26]. Globally, at least 68 countries have laws that criminalize HIV, and 33 others have applied other criminal law provisions to such cases [27]. Though direction of transmission cannot be determined definitively from phylogenetics because there is always the possibility of an intermediary, unknown partner [17], the use of these techniques in criminal prosecutions and some law enforcement’s access to molecular surveillance data lead to a concerning permeability in protection of individuals.

A second risk associated with phylogenetic determination of direction of transmission, even if exclusively confined to research and public health, is its potential to increase stigma and discrimination of specific populations. In San Antonio, health officials avoided referencing demographics of the molecular cluster in their health advisory. Despite this, headlines like “Cluster of HIV Cases Involves Hispanic Men in San Antonio” may have unintentionally increased stigma in the community [28]. The addition of direction of transmission to these data—for example, the ability to determine whether older Latinx men were transmitting to younger persons—could inform prevention interventions yet, simultaneously, increase stigma.

Considering the risks of stigma, discrimination, and possible prosecution, we believe calling for protections to preserve anonymity is an insufficient ethical response. The creation of true protections in any context in which direction of transmission is being determined is a formidable task, as demonstrated by the work of organizations such as the Georgia Coalition to End HIV Criminalization and the HIV Modernization Movement Indiana, which work for legislation that aligns with scientific knowledge. Robust protections require changes in laws in the approximately 101 countries and 34 US states where HIV criminalization exists [26, 27]. If direction of transmission data increase stigma and discrimination, this undermines efforts to promote HIV testing and linkage to care, particularly among key populations [29].

In this circumstance, we have ethical complexity overlaid on top of scientific complexity. Research methodologies evolve quickly, and it is difficult to tell how soon it will be before we can assess direction of transmission with relative accuracy in datasets beyond trials like HPTN 052. It would be simple to say that the benefit of more effective interventions using direction of transmission data outweigh any potential harm to individuals or communities, because we have an urgent need to respond to the expanding global HIV epidemic and should use all tools at our disposal to save lives. Conversely, it would be easy to say that the discrimination and stigma direction of transmission data could propagate in already marginalized communities and that the possibility of these data being used in the criminal justice system is dangerous and the work should be stopped. The answer lies somewhere in between. We must find solutions that allow for the development of effective tools to combat the epidemic while offering robust protections and incorporating the voices of the most impacted individuals and communities.

Note

Potential conflicts of interest. B. T. has received personal fees from the International Antiviral Society–USA and grants from the Health Resources and Services Administration, the University of Texas at Austin, the Texas Department of State Health Services, and the University of Texas Health Science Center San Antonio Clinical and Translational Science Award, outside the submitted work. H. S. has received personal fees from Gilead Sciences, outside the submitted work. Both 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.