Methodological synthesis of Bayesian phylodynamics, HIV-TRACE, and GEE: HIV-1 transmission epidemiology in a racially/ethnically diverse Southern U.S. context

Abstract

This study introduces an innovative methodological approach to identify potential drivers of structuring HIV-1 transmission clustering patterns between different subpopulations in the culturally and racially/ethnically diverse context of Houston, TX, the largest city in the Southern United States. Using 6332 HIV-1 pol sequences from persons newly diagnosed with HIV during the period 2010-2018, we reconstructed HIV-1 transmission clusters, using the HIV-TRAnsmission Cluster Engine (HIV-TRACE); inferred demographic and risk parameters on HIV-1 transmission dynamics by jointly estimating viral transmission rates across racial/ethnic, age, and transmission risk groups; and modeled the degree of network connectivity by using generalized estimating equations (GEE). Our results indicate that Hispanics/Latinos are most vulnerable to the structure of transmission clusters and serve as a bridge population, acting as recipients of transmissions from Whites (3.0 state changes/year) and from Blacks (2.6 state changes/year) as well as sources of transmissions to Whites (1.8 state changes/year) and to Blacks (1.2 state changes/year). There were high rates of transmission and high network connectivity between younger and older Hispanics/Latinos as well as between younger and older Blacks. Prevention and intervention efforts are needed for transmission clusters that involve younger racial/ethnic minorities, in particular Hispanic/Latino youth, to reduce onward transmission of HIV in Houston.

Figure 1

Phylogenetic analytical design. The three-step process of estimating viral transmission between epidemiological characteristics that structure HIV transmission clusters within Houston/Harris County is summarized. In the first step, transmission clusters were identified. In the second step, a phylogenetic analysis was conducted to estimate the time of the most recent common ancestor for each cluster. In the third step, a joint estimation of transition rates of a single viral transition matrix for all clusters was conducted.

Figure 2

Visualization of HIV-1 transmission network in Houston/Harris County (2010–2018). MSM refers to cisgender men who report being MSM, cis-women refers to cisgender women who did not report injection drug use, trans-women refers to transgender women who did not report injection drug use, cis-men refers to cisgender men who did not report injection drug use or MSM, and PWID refers to anyone who reported injection drug use, including MSM. Hispanic/Latino refers to Hispanic/Latino of all races; Black/African American refers to non-Hispanic Black/African American; and White/Other refers to non-Hispanic Whites, non-Hispanic Asians, and non-Hispanic other racial categories, including American Indians, Alaskan Natives, and multi-race categories.

Figure 3

(A) BEAST phylogeny of all clusters; (B) chord diagram for viral transmission between racial/ethnic groups (Hispanics, Blacks, Whites/Others); (C) chord diagram for viral transmission between racial/ethnic groups plus age group (born in or before 1990, born after 1990). (A) Shows a maximum clade credibility phylogenetic tree with all clusters that contain five or more cluster members. Tips are labeled with the race/ethnicity associated with the taxa at that tip. In (B) and (C), chord thickness indicates the magnitude of the transition rate, and color indicates Bayes Factor support. Only supported transition rates from discrete trait analysis are displayed. YoB Year of Birth.