CROI 2020 Abstract eBook

Abstract eBook

Poster Abstracts

Background: The identification of recent (incident) HIV infections among people with newly diagnosed infections is critical to HIV prevention. We developed a Multiplexed Primer ID-Next Gen Sequencing (MPID-NGS) approach to identify recent infection by measuring the intra-host viral diversity over multiple regions of the HIV genome. Here we summarize the field implementations of this approach to identify recent infection, and include surveillance of drug resistance mutations (DRMs) in new diagnoses from the Public Health Laboratory in the state of North Carolina in 2018. Methods: The MPID-NGS libraries were constructed covering the coding regions for protease (PR), a portion of reverse transcriptase (RT), integrase (IN), and the V1 to V3 region of the env gene from the HIV positive serums. The MiSeq platform was used for sequencing. The TCS-DR pipeline was used for bioinformatics analysis and to identify DRMs. Recent infection was defined as within 9-month of infection, and the RT and V1/V3 regions were used to define recency. Results: A total of 547 HIV+ samples from diagnostic testing in 2018 were subjected to sequencing; of these 294 were considered new diagnosis and ART naïve as the sample used for testing was less than 30 days from diagnosis date reported by Health Dept. The sequencing success rate for the newly diagnoses was 91.2%. Overall recent infection was identified in 94 subjects (35%). Multivariate regression shows that people between 18 to 24 were more likely to be diagnosed at recent infection (OR=3.34, p=0.018) and those with unknown risk factors were less likely to be diagnosed at recent infection (OR=0.34, p=0.047). We observed that RT regions had more SDRMs than PR and IN, and RT K103N was the most common mutation overall. RT mutations M184V, R65R and major IN DRMs were rarely seen (Table 1). We used the RT sequence to explore close transmission clusters and we found a total of 28 clusters (size 2 to 4) using a similarity cut-off at 1%. Conclusion: MPID-NGS combines recency identification and DRM screening for new HIV diagnosis in near real-time. Young individuals had highest recent infection rate while those with unknown risk factors had the lowest. The overall DRM rate was high but clinically important mutations were low. Rapid identification of transmission clusters containing recently infected individuals facilitates targeted prevention efforts.

whether shifts in testing methods and sequencing technology have implications for PH surveillance of DR and transmission clusters. Methods: We identified ~115,000 RNA-S, ~11,000 DNA-NGS, and ~5,000 RNA-NGS sequences reported to New York during 2010-2019 from a single commercial lab. Inferred DR was compared for 1,350 persons with two or more sequence types. For cluster analyses, pairwise genetic distances were calculated between sequences for the same person with collection dates within 1 year (n=7,771 comparisons from 2,823 individuals) using Secure HIV-TRACE default settings and a 2% genetic distance threshold, stratified by sequence type. Results: Overall, DR was 37%more likely to be inferred from DNA-NGS sequences than RNA-based sequences from the same individual. Time between tests was not a significant factor, and individual drug classes showed similar results. For clustering, over 25% of DNA-NGS were rejected by Secure HIV-TRACE due to high levels of ambiguities compared to RNA-NGS (11%) and RNA-S (8%). Based on pairwise distances for sequences from the same individual, RNA-NGS and especially DNA-NGS sequences, clustered less frequently than RNA-S sequences and at a higher distance threshold if they did cluster. Mean number of years since diagnosis was high and varied by sequence types but did not explain the results (Table 1). Conclusion: We found significant differences between consensus DNA-NGS and RNA-NGS sequences compared to RNA-S sequences for cluster inference and between DNA-NGS and RNA-based sequences for DR. Hence, reporting of sequence type for PH surveillance is critical for ensuring appropriate inclusion of sequences for accurate HIV DR and transmission cluster analyses. Monitoring changes in sequencing technology is critical for assessing impact on PH and clinical decisions. 933 PERVASIVE AND NONRANDOM RECOMBINATION IN NEAR FULL-LENGTH HIV GENOMES FROM UGANDA Heather E. Grant 1 , Emma B. Hodcroft 2 , Deogratius Ssemwanga 3 , John Kitayimbwa 4 , Paul Kellam 5 , Tulio De Oliveira 6 , Rebecca N. Nsubuga 7 , Samantha Lycett 1 , David L. Robertson 8 , Oliver Ratmann 5 , Christophe Fraser 9 , Deenan Pillay 10 , Pontiano Kaleebu 3 , Andrew Leigh Brown 1 , for the PANGEA-HIV Consortium 1 University of Edinburgh, Edinburgh, UK, 2 University of Basel, Basel, Switzerland, 3 Uganda Virus Research Institute, Entebbe, Uganda, 4 Makerere University, Kampala, Uganda, 5 Imperial College London, London, UK, 6 University of KwaZulu- Natal, Durban, South Africa, 7 Medical Research Council, Entebbe, Uganda, 8 University of Glasgow, Glasgow, UK, 9 University of Oxford, Oxford, UK, 10 University College London, London, UK Background: Recombination is an important feature of HIV evolution, occurring both within and between the major branches of diversity (subtypes). The Ugandan epidemic is primarily composed of two subtypes, A1 and D, that have been co-circulating for 50 years, frequently recombining in dually infected patients. We have investigated the frequency of recombinants in this population (both inter- and intra-subtype), and the location of breakpoints along the genome. Methods: As part of the PANGEA-HIV consortium project 1472 consensus genome sequences over 5kb were obtained from 1857 samples collected by the MRC/UVRI & LSHTM Research Unit in Uganda, 465 (31.6%) of which were near- full length (NFL) genomes (>8kb). The subtyping tool SCUEAL was used with a reference dataset of 218 full length subtype and circulating recombinant form genomes to identify recombination events both between and within subtypes. Genomic distribution of inter-subtype breakpoints was characterised using K-means clustering and generalized linear modeling. Results: 233 of the 465 (50.1%) NFL genomes contained only one subtype; 143 A (30.8%), 82 D (17.6%) and 8 C (1.7%), while 232 (49.9%) contained more than one subtype (including A1/D (n=164), A1/C (n=13), C/D (n=9), A1/C/D

Poster Abstracts

932 IMPLICATIONS OF NEXT-GENERATION SEQUENCING FOR DRUG RESISTANCE AND CLUSTER DETECTION Randall V. Collura 1 , Wendy Patterson 1 , Carol-Ann Swain 1 , Eva Pradhan 1 , WilliamM. Switzer 2 , Alexandra M. Oster 2 , Bridget J. Anderson 1 1 New York State Department of Health, Albany, NY, USA, 2 CDC, Atlanta, GA, USA Background: HIV-1 polymerase (pol) sequences from routine HIV drug resistance (DR) testing are used to monitor DR and identify molecular transmission clusters as part of public health (PH) surveillance. Proviral DNA DR testing using next-generation sequencing (DNA-NGS) has been used clinically since 2015 to provide DR information in the setting of viral suppression. Since DNA-NGS covers the same part of the HIV genome and DNA-NGS consensus sequences mimic traditional RNA-Sanger (RNA-S) sequences, they have likely been reported to PH as RNA-S sequences. Some clinical labs are also using (and others are considering) NGS for RNA-based DR testing (RNA-NGS). We evaluated

CROI 2020 350