Human Immunodeficiency Virus Type 1 Phylodynamics to Detect and Characterize Active Transmission Clusters in North Carolina

Abstract

Background: Human immunodeficiency virus type 1 (HIV-1) phylodynamics can be used to monitor epidemic trends and help target prevention through identification and characterization of transmission clusters.

Methods: We analyzed HIV-1 pol sequences sampled in North Carolina from 1997 to 2014. Putative clusters were identified using maximum-likelihood trees and dated using Bayesian Markov Chain Monte Carlo inference. Active clusters were defined as clusters including internal nodes from 2009 to 2014. Effective reproductive numbers (Re) were estimated using birth-death models for large clusters that expanded ≥2-fold from 2009 to 2014.

Results: Of 14 921 persons, 7508 (50%) sequences were identified in 2264 clusters. Only 288 (13%) clusters were active from 2009 to 2014; 37 were large (10-36 members). Compared to smaller clusters, large clusters were increasingly populated by men and younger persons; however, nearly 60% included ≥1 women. Clusters with ≥3 members demonstrated assortative mixing by sex, age, and sample region. Of 15 large clusters with ≥2-fold expansion, nearly all had Re approximately 1 by 2014.

Conclusions: Phylodynamics revealed transmission cluster expansion in this densely sampled region and allowed estimates of Re to monitor active clusters, showing the propensity for steady, onward propagation. Associations with clustering and cluster characteristics vary by cluster size. Harnessing sequence-derived epidemiologic parameters within routine surveillance could allow refined monitoring of local subepidemics.

Keywords: HIV-1; molecular epidemiology; phylogeny; southeastern United States; transmission.

Figure 1.

Distribution of large clusters (≥10 members; n = 74 clusters) among 2264 subtype B clusters identified among 14 921 persons with HIV-1 pol sequences sampled, 1997–2014. Each cluster is a horizontal line on the y-axis. The x-axis indicates the number of samples (only the first per person) by sampling year. The cross (+) indicates the time of the most recent common ancestor (tMRCA) of the cluster, as estimated in BEAST version 1.8. Color gradient indicates the percentage of samples in each cluster per year from women.

Figure 2.

Distribution of clusters with high proportional detection rate (PDR) (≥2-fold increase in cluster size) between 2009 and 2014 and size ≥10 members by 2014 (n = 15 clusters). The y-axis indicates cluster identification number. A, Number of samples per year. Color gradient indicates PDR. The effective reproductive number (R_e) and 95% credibility intervals estimated in the most recent time span (~2011–2014) is indicated for each cluster. B, Number of samples per year, by North Carolina (NC) region of sampling and sex.

Figure 3.

Effective reproductive number (R_e) and section of maximum clade credibility tree among selected clusters with size ≥10 members by 2014 and high proportional detection rate from 2009 to 2014. Clades with posterior probability ≥0.90 are labeled. A, Cluster 1076 with increasing R_e and majority samples from the Raleigh region. B, Cluster 618, with R_e trending downward and sampling from multiple North Carolina regions.