CROI 2016 Abstract eBook

Abstract Listing

Poster Abstracts

223 Potential Misclassification in Standard HIV Phylogenetic Clusters Association Studies Benoit Visseaux 1 ; Oliver Ratmann 2 ; Christopher Monit 3 ; David Dunn 4 ; Kholoud Porter 5 ; Stephane Hue 6 1 INSERM UMR 1137, Paris, France; 2 Imperial Coll London, London, UK; 3 Univ Coll London, London, UK; 4 Med Rsr Council Clinical Trials Unit at Univ Coll London, London, UK; 5 Med Rsr Council, London, UK; 6 London Sch of Hygiene & Trop Med, London, UK

Background: The phylogenetic identification of transmission chains from partial HIV pol sequence data is a cornerstone of HIV molecular epidemiology. The standard method proceeds by identifying clusters in the viral phylogeny under a combination of strict criteria (e.g. minimal intra-cluster genetic distance, high branch support). Individuals with sequences in the same cluster are then declared phylogenetically linked. However, HIV sequences from infected individuals are typically sampled some time after infection, which could lead to substantial bias under these criteria. To quantify potential bias in the phylogenetic identification of HIV linkage, we mined data from two British HIV observational cohorts for epidemiologically linked sequences, and evaluated what proportion are excluded from cluster association studies under standard protocols. Methods: Routine drug genotyping HIV sequence data from the UK HIV Drug Resistance Database and UK Register cohorts were mined for longitudinally sampled intra-host sequences (n=16,842 pairs). These were considered linked. Maximum likelihood phylogenies were reconstructed with FastTree v.2.1.7 and clustered pairs of sequences identified using a range of criteria. Potential bias was defined as the proportion of linked sequences that are not recognised as such under any given set of viral phylogenetic clustering criteria. Results: Figure 1 quantifies the extent of potential bias under standard protocols in the UK cohorts. With 95% support and 2% genetic distance thresholds, the proportion of linked pairs that are excluded from further analyses was 69%. With 90% bootstrap and 5% genetic distance thresholds, the proportion of linked sequences that are excluded from further analyses remained substantial at 52%. Larger genetic distances are expected between linked sequences that had more time to diverge. Therefore, authentic transmitters that are excluded from viral phylogenetic cluster association studies are more likely to be late presenters, untreated for a long period, or treated and successfully virally suppressed for long periods. Conclusions: Standard viral phylogenetic protocols may exclude a substantial proportion of authentic transmitters from further epidemiological analysis in cluster association studies.

224

Partner Services in Acute and Early HIV-Infected Adults Nella Green; Christy Anderson; Sergei Kosakovsky Pond; Martin Hoenigl; Davey M. Smith; Susan Little Univ of California San Diego, San Diego, CA, USA Background: We hypothesized that provision of Partner Services (PS) to acute and early HIV infected (AEH) index subjects in San Diego would prove effective in identifying genetically linked HIV-infected sex partners not previously aware of their HIV positive status. Methods: We analyzed data generated by the San Diego Primary Infection Resource Consortium from 1996 to 2014. Persons with AEH infection were offered PS and enrollment to PIRC. Sex partners of AEH index subjects were offered HIV screening and HIV drug resistance testing if infected. Partnerships were categorized as genetically linked if viral protease and partial reverse transcriptase sequences isolated from the partners were at most 1.5% distant under the TN93 model. Number needed to interview (NNTI), i.e. number of AEH index subjects required to interview to find one partner or new HIV diagnosis was calculated. Mixed-effects logistic regression and two-tailed χ 2 analyses were used to test whether partners who enrolled within 30 days of the index subject enrollment were more likely to be 1) newly HIV diagnosed and 2) genetically linked to the index subject. Results: PS provided to 96 AEH index subjects led to recruitment of 105 sex partners (total of 114 partnerships); 55 (52.4%) sex partners were HIV-infected, of whom 35 (63.6%) were newly HIV-diagnosed. Thirteen (37.1%) newly diagnosed sex partners were themselves AEH infected. Table 1 illustrates PS yield and efficiency for each diagnostic outcome. Demographics and risk behaviors between HIV uninfected and infected sex partners were similar. HIV-concordant sex partners (N=66) enrolled within 30 days of the index partner (47/66 [71%]) were 5.7 times more likely to be newly diagnosed with HIV infection (p < 0.01), and 3.25 times more likely (p = 0.04) to be genetically linked to each other (41/66 [62%]), compared to those enrolled after 30 days. There was no difference in the rates of genetic linkage to index subjects between chronic and AEH sex partners (p = 0.166). Conclusions: PS delivered immediately following an AEH diagnosis is an effective strategy to identify HIV infected and previously unaware persons, including sex partners with AEH. It is likely that PS delivered within the first month of AEH diagnosis benefits from a more accurate recall of recent sexual contacts, with 62% of sex partners reprrepresenting putative transmission links as confirmed by genetic linkage. PS delivered to chains of sex partners identified with AEH may be a promising strategy for finding HIV unawares.

Poster Abstracts

87

CROI 2016