Table 3.

Studies combining phylogenetic and social network analysis in the context of an HIV outbreak investigation (N = 6)

Authors	Title	Pub. yeara	Study design	Locationb	Time period	Data source/project	Population characteristics	Social/sexual network data	Molecular cluster analysis	Data integration	Resultsc
Dennis et al.	Integration of Contact Tracing and Phylogenetics in an Investigation of Acute HIV Infection	2018	Outbreak investigation	NC	2013–2014	NC Department of Health and Human Services (NC-DHHS), North Carolina Screening and Tracing for Active HIV-1 Transmission (NC-STAT) program	68 AHI/RHI index cases: 85% male, median age of 25, 66% Black. 210 reported contacts: 189 first degree, 21 degree. All sexual (93%) or social (7%) contacts, no IDU contacts	In response to new foci of AHI cases in early 2014 detected by NC-STAT, all AHI or RHI cases in NC in 2014 were investigated. DIS interviewed all index cases and followed-up with all high-risk sexual contacts. Contact networks were constructed using contact tracing data for all index cases, their reported contacts, and second-degree contacts if linked first-degree contact had a new HIV diagnosis or an HIV diagnosis in 2013 or 2014	NC-DHHS requested HIV-1 pol sequences for all index cases and HIV-infected contacts. Partial pol sequences (n = 1672) from UNC Center for AIDS Research HIV Clinical Cohort were included as background references. Maximum-likelihood phylogenies were constructed in RAxML with general time reversible model of nucleotide substitution. Transmission clusters were defined as clades with bootstrap support >98% that had one or more index case sequences with<0.015 substitutions/site pairwise genetic distance from another case	Independent phylogenetic and contacttracing networks were constructed. These networks were compared by the study	Statewide contact tracing network had 58 independent components, with mean degree of 1.7 and overall network density of 1.2%. Mean index node degrees by contacts’ HIV statuses were 0.8 for HIV-infected contacts, 1.0 for HIV-negative contacts, and 1.3 for contacts with unknown HIV status. HIV-1 pol sequences were available for 56% of index cases and 48% of HIV-infected contacts. 13 identified transmission clusters involved at least two index cases or an index and a contact. 31% included a previously diagnosed contact. 4 clusters revealed links not apparent from contact tracing. The largest component of the contact network included members from distinct phylogenetic clusters
Hayman et al.	Phylogenetic Analysis of Multiple Heterosexual Transmission Events Involving Subtype B of HIV Type 1	2001	Outbreak investigation	Doncaster, United Kingdom	1996–1999	Not identified	13 individuals who tested HIV+ in the town of Doncaster between 1996 and 1999. 31% male, 69% female	Initial epidemiological follow-up of an HIV index case identified in 1996 found a cluster of previously undiagnosed cases. Further contact tracing demonstrated the cluster of 13 cases were linked by both sexual and venue-based histories	Frozen blood samples from the 13 cases were amplified by PCR and the same C1 to C4 region of the HXB2 peptide (930 bp) within the HIV genome was sequenced for phylogenetic analysis. Phylogeny was assessed using two methods, both maximum likelihood and neighbor joining, and both indicated the same phylogeny. Sequences were also compared to those available in GenBank	Phylogenetic relationships and known social contact links (sexual or IDU) were compared. Phylogenetic analysis was primarily used to assess the ability to discern relationships that were unclear from contact tracing due to multiple interlinked contacts among the 13 cases	Sequencing indicated three separate HIV strains among the 13 sequences. Ten of the sequences could be linked to the index case but could also be delineated into two separate clusters. One group was more closely genetically related to itself and the index case sequence (this group included the index case and five other cases), while the other consisted of six cases that were less related to the index and to one another. Contact tracing postulated sexual, social, or an indirect IDU contact between the index case and 10 other cases. The set of 11 cases connected by contact tracing was the same set connected by phylogenetic analysis apart from one possibly distantly related sequence
Monterosso et al.	Identifying and investigating a Rapidly Growing HIV Transmission Cluster in Texas	2017	Outbreak investigation	TX	2015–2016	Texas Department of State Health Services	27 newly diagnosed HIV+ individuals constituting the initial outbreak group and 112 HIV+ sex/IDU partners and social contacts. Of the 76 partners with available records, 100% men, 89% MSM, 87% Hispanic, 78% aged 13–29	Partner services were used to elicit index cases’ sex and IDU partners, the partners of identified partners, and social network contacts, along with information on demographics, risk behaviors, and partner meeting sites	Phylogenetic analysis conducted on HIV sequences collected through the National HIV Surveillance System identified a rapidly growing molecular cluster in Texas	Partner services data was integrated with molecular data to identify sex/IDU partners and social contacts without molecular sequences who could have composed part of the rapidly growing cluster	An initial cluster of 27 cases was identified via phylogenetic analysis. Twelve (44%) of the 27 cases were linked via social or sexual partnerships. From the original 27, an additional 112 cases were identified via partner services information
Nett et al.	Two Clusters of HIV-1 Infection, Rural Idaho, USA, 2008	2010	Outbreak investigation	ID	2008	Idaho Department of Health and Welfare	15 newly diagnosed HIV+ individuals comprising two outbreak clusters in southeastern Idaho. Median age of cluster members 24 and 26, respectively. Overall, 73% MSM, 27% suspected IDU	Participant demographics, drug use, and sexual activity was elicited through epidemiological investigation. This information was used to construct sexual and drug-use networks	HIV pol sequences for 10 participants were obtained from commercial laboratories. An additional sequence from the same region but not linked via contact tracing was also included. Control sequences were provided by 2 Idaho HIV clinics. Sequences were grouped into phylogenetic clusters using multiple sequence alignment, neighbor joining, and maximum likelihood tree analysis	Contact tracing and molecular data were combined to determine possible sources of HIV transmission for both outbreak clusters	The 11 sequences clustered into two phylogenetic groupings. The first grouping contained 4 sequences from outbreak cluster A and the sequence that was not linked to either outbreak cluster. The second grouping contained 5 sequences from outbreak cluster B. The two groupings had an average pol genetic distance of 4.8%, which was similar to that of the control group (5.1%), suggesting no linkage between the two clusters
Samoff et al.	HIV Outbreak Control with Effective Access to Care and Harm Reduction in North Carolina, 2017–2018	2020	Outbreak investigation	Western NC	2017–2018	North Carolina Department of Public Health	7 HIV+ individuals diagnosed with HIV from 2017 to early 2018 in western NC and their primary, secondary, and tertiary IDU contacts. 100% IDU, 62% male, 96% White, median age 36	Disease intervention specialists conducted contact tracing interviews with the 7 individuals part of the initial HIV outbreak as well as their listed sexual and IDU contacts (primary contacts). DIS then performed contact tracing on their partners (secondary contacts) as well as the partners of secondary contacts (tertiary contacts). Network diagrams were constructed using partnership information	New HIV nucleotide sequences were reported to reference laboratories monthly. Molecular clusters were constructed using a TN-93 pairwise genetic distance of 1.5% or less. BEAST version 2.4.8 was used to build phylogenetic trees for outbreak-associated clusters using the GTR+ gamma model, Bayesian skyline coalescent prior, and a relaxed log-normal molecular clock	The number of sexual and IDU contacts newly diagnosed with HIV was compared with the number of new diagnoses in genetic clusters. Markers of linkage to care (ex: pertinent laboratory and appointment dates) were used to determine success in decreasing HIV exposure within the transmission networks	DIS successfully contacted 6 out of 7 individuals part of the initial outbreak as well as 96 IDU contacts linked to the initial outbreak members. Of those 96, 14(15%) were HIV+. Five outbreak group members and 1 newly diagnosed HIV+ individual had sequences that grouped with a cluster of 14 individuals diagnosed between 2011 and 2018, 6 of whom were IDUs. As of March 2019 (12 months follow-up), 5 of the initial outbreak members remained virally suppressed, as did both individuals newly diagnosed during the investigation
Sizemore et al.	Using an Established Outbreak Response Plan and Molecular Epidemiology Methods in an HIV Transmission Cluster Investigation, Tennessee, January–June 2017	2020	Outbreak investigation	TN	2017	Tennessee Department of Health	31 individuals in eastern TN diagnosed with HIV after January 1, 2017, and their 107 named sex/drug-use partners. For index cases: 45% MSM, 42% recently incarcerated, 31% IDU	Index cases completed interviews and questionnaires detailing their sex/drug-use partnerships as well as high risk sexual and drug-use history. Social network analysis was performed to visualize the linkages between index cases and contacts	HIV pol sequences were amplified via PCR. COMET was used to determine Tamura-Nei pairwise distances and subtyping. A genetic distance threshold of under 1.5% was used to establish molecular clusters	Social and molecular data were both utilized to describe HIV transmission among study participants	Partner services interviews with 31 index cases resulted in 107 unique sex and drug-use partners. HIV testing of partners resulted in 2 new HIV+ diagnoses. 14 HIV-1 pol sequences were obtained from 7 of the 8 IDUs as well as 7 other participants. Molecular analysis revealed 3 clusters, 1 of which included 3 of the IDUs

^aYear of publication

^bCity and/or state abbreviation

^cMSM Men who have sex with men, HIV human immunodeficiency virus, IDU injection drug user, CDC centers for disease control and prevention, BLAST basic local alignment search tool, BRAI bio-rad avidity incidence, IVDU intravenous drug user, STD sexually transmitted disease, TB tuberculosis, ECVC Encuesta Centroamericano de Vigilancia de VIH y Comportamiento en Poblaciones Vulnerables, FSW female sex worker, STI sexually transmitted infection, Hx history; NC-DHHS North Carolina Department of Health and Human Services, NS-STAT North Carolina Screening and Tracing for Active HIV-1 Transmission, AHI acute HIV infection, RHI recent HIV infection, DIS disease intervention specialists; iMAN integrated molecular & affiliation network analysis of HIV transmission, AIC Akaike’s Information Criterion, RDS respondent-driven sampling, DPH department of public health, TRACE secure HIV transmission cluster engine, PLWHA people living with HIV/AIDS, PCR polymerase chain reaction; PWID people who inject drugs; COMET context-based modeling for expeditious typing, GRT genotypic resistance testing, CDPH Chicago Department of Public Health, MUSCLE multiple sequence comparison by log-expectation, HCV hepatitis C virus, SD-PIRC San Diego Primary Infection Resource Consortium, CNICS center for AIDS research network of integrated clinical systems, RTPCR reverse transcription polymerase chain reaction, AIEDRP acute infection and early disease research program, FCP first choice program, ART antiretroviral therapy, GHWP good health for women project, UCSD University of California San Diego, SDC San Diego County, NYCDOH New York City Department of Health