CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

557 POINT-OF-CARE HIV-1 GENOTYPIC RESISTANCE TESTING USING MELT- CURVE ANALYSIS Dana S. Clutter 1 , Gelareh Mazarei 2 , Justen Manasa 3 , Sara Bolouki 2 , Ruma Sinha 2 , Peter Kuimelis 2 , Robert Kuimelis 2 , Benjamin Pinsky 1 , Gary Schoolnik 2 , Arjang Hassibi 2 , Robert Shafer 1 1 Stanford University, Stanford, CA, USA, 2 InSilixa, Inc, Sunnyvale, CA, USA, 3 African Institute of Biomedical Science and Technology, Harare, Zimbabwe Background: An inexpensive point-of-care (POC) genotypic resistance test (GRT) would enable HIV-1 providers to make informed decisions for ARV-naïve patients starting therapy and for patients on therapy with virological failure. The main challenge in developing a POC GRT is the variability at and surrounding each drug-resistance mutation (DRM) position. We performed in silico and in vitro experiments to determine whether melt-curve analysis could distinguish between 6 different codon 103 variants despite surrounding sequence variability. In vitro experiments were performed using a developer platform for the HYDRA-1K CMOS biochip (InSilixa, Inc, Sunnyvale, CA). Methods: We performed in silico thermodynamic modeling of 43,000 global RT sequences to design a probe set comprising 18 34-bp probes (6 codons x 3 flanking regions) predicted to discriminate the 6 most common codon 103 variants: AAA (K), AAG (K), AAC (N), AAT (N), AGA (R), AGC (S). We then printed each probe in the set onto silicon slides and performed 81 experiments using diverse fluorescently-labeled 80-bp targets. In each experiment, we added a single target or a mixture of AAA and AAC targets to a slide, which we then heated over 20 minutes while a fluorescent microscope captured serial images. The Tm (temperature at which 50% of the target was hybridized) was calculated for each probe and weighted by the initial signal intensity to reflect the degree of hybridization. Results: In 54 experiments, we tested 18 targets (6 codons x 3 flanking regions) in triplicate. In 27 experiments, we tested mixtures of AAC/AAA targets with 3 different flanking regions at ratios of 50/50, 25/75, and 10/90 in triplicate. In the 54 experiments using single targets, a mean weighted ∆Tm cutoff of 8.7°C was 100% sensitive and specific for correctly identifying single targets. A mean weighted ∆Tm cutoff of 13.1 °C was 97% sensitive and 100% specific for identifying targets present at 25% or 50%. A mean weighted ∆Tm cutoff of 16.3 °C was 90% sensitive and 89% specific for the detection of variants at 10% frequency. Conclusion: Despite genetic variability surrounding the DRM position, multiple codons at a DRM position can be correctly identified in isolation or as part of a mixture when the frequency of both species is ≥25% using a solid-phase melt curve analysis platform. Future work will include additional DRM positions, minority variant detection, and transfer to the HYDRA-1K CMOS biochip.

558 PROTEASE, GAG AND GP41 MUTATIONS ASSOCIATED WITH VIROLOGICAL RESPONSE TO PI REGIMEN Marine Perrier 1 , Louise Castain 2 , Eve Todesco 2 , Roland Landman 1 , Benoit Visseaux 1 , Yazdan Yazdanpanah 1 , Leslie Regad 3 , Veronique Joly 1 , Vincent Calvez 2 , Anne-Geneviève Marcelin 2 , Diane Descamps 1 , Charlotte Charpentier 1 1 Bichat–Claude Bernard Hospital, Paris, France, 2 Pitié-Salpêtrière Hospital, Paris, France, 3 University Paris Diderot, Paris, France Background: Protease (PR) resistance associated mutations are rarely observed in case of virological failure (VF) of a first-line PI-based regimen. The aim of this study was to assess the impact of baseline determinants in PR, but also in gag and gp41 regions, on virological response of a first-line PI-based regimen. Methods: In an observational cohort we enrolled all ARV-naïve patients initiating a first cART including 2 NRTI associated with DRV/r (n=131) or with ATV/r (n=23) between January 2012 and March 2015, including 36 experiencing VF. Baseline ultra-deep sequencing of PR, gag and gp41 regions was performed using Illumina® technology. Supervised data mining analysis were performed to identify mutations associated with virological response. Statistical analyses were performed with Fisher’s exact test and Bonferroni correction was applied. Structural analyses were performed to assess impact of mutations on PR conformation. Results: Among 154 patients enrolled, HIV-1 sequences were successfully obtained in 127, 138 and 134 samples for PR, gag and gp41, respectively. 31% of samples were subtype B, 38% CRF02_AG and 31% others “non-B” subtypes. Overall in PR, 2 mutations were identified as associated with VF: T4A, S37T (p=0.02; p=0.05, respectively). Among CRF02_AG sequences, mutations I72M and E21D were associated with VF (p=0.03 for both). Structural modeling analysis showed that all these mutations induced some conformational changes of some PR side-chain residues located near mutated residues. In gag, mutations associated with VF were G62D, N315H and Y441S (p=0.05, p=0.07 and p=0.0003, respectively). All these mutations are localized outside gag cleavage site (G62D: matrix, N315H: capsid and Y441S: p1). In gp41, I270T mutation was associated with VF (p=0.003). This mutation is located in the cytoplasmic tail, a region whose impact on PI resistance has been recently showed. In addition, mutation I4L, localized in fusion peptide, was associated with virological success (p=0.004). Conclusion: In this study, based on patients initiating a first-line DRV or ATV/r- based regimen, we identified baseline mutations associated with virological response as well inside as outside PR, in gag and gp41 regions. Further in vitro studies are needed to better characterize the impact of these newmutations on PI phenotypic susceptibility. 559 HIV-1 GAG REDUCES PI-SUSCEPTIBILITY IN THE ABSENCE OF PROTEASE RESISTANCE MUTATIONS Oscar Blanch-Lombarte , José Ramón Santos, Ruth Penya, Alba Ruiz, Esther Jimenez-Moyano, Roger Paredes, Bonaventura Clotet, Julia G Prado IrsiCaixa Institute for AIDS Research, Badalona, Spain Background: The HIV-1 genetic determinants of virological failure (VF) to Protease inhibitors in the absence of protease mutations are poorly understood. However, their identification would be crucial in improving the actual genotypic tools to define VF to PIs in B and particularly, non-B clade HIV-1 subtypes. Methods: We performed a retrospective review of 520 HIV-1 infected patients who started monotherapy with Protease inhibitors (PIs) LPV/r or DRV/r in our clinic. Eleven experienced VF and were further analysed. We amplified the HIV-1 Gag-Protease from 9 subjects at the time of VF. Signature mutations in the Gag-Protease region were identified by comparison with 2000 naïve B clade sequences from Los Alamos database using VESPA (www.hiv.lanl.gov). Also, we evaluated replicative capacity and drug susceptibility to LPV and DRV in Gag- Protease recombinant virus generated from plasma samples at VF. Results: All study patients experienced VF to PIs in the absence of Protease resistance mutations. We observed a high frequency of mutations previously described outside Gag cleavage sites at positions R76K (55%), I389T (44%), E12K (33%), V370A (33%) and T81A (11%). VESPA analyses provided a signature pattern of mutations in Gag including residues K95R, E203D, V215M, R286K (p<0.01). Replicative capacity experiments demonstrated a preservation of viral fitness by direct comparison with the NL43 laboratory strain. Most virus did not show significant changes in drug susceptibility. However, a Gag- Protease recombinant virus from one patient harboring the K95R, R286K Gag mutations without Protease mutations revealed 8- to 10-fold reduction in

Poster Abstracts

CROI 2018 205