CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

291 OPTIMIZING HIV-1 VIRUS TEST PANELS FOR EFFECTIVE DETECTION OF NEUTRALIZATION BREADTH Claus Kadelka 1 , Peter Rusert 1 , Roger Kouyos 2 , Huldrych F. Günthard 2 , Alexandra Trkola 1 1 University of Zurich, Zurich, Switzerland, 2 University Hospital Zurich, Zurich, Switzerland Background: Potent broadly neutralizing antibodies (bnAbs) are a key focus of HIV-1 vaccine and therapy development but are only elicited at low frequency in natural infection. The identification of bnAb responses requires virus screening panels that effectively capture neutralization breadth. Varying procedures and thresholds to define breadth are applied in the literature calling for standardization efforts to enable unbiased assessment of vaccine responses. The Swiss 4.5K Screen, a systematic survey of bnAb activity in 4,484 HIV-1 infected individuals, identified 239 bnAb inducers (Rusert, Kouyos Nat Med 2016), which provided an ideal set of patients to investigate the composition of optimized virus panels for the detection of HIV-1 neutralization breadth. Methods: The Swiss 4.5K Screen was based on a multi-clade panel of 8 HIV-1 strains. 729 plasma samples, which showed >80% inhibition of at least one of these viruses, were screened against 15 additional viruses thus yielding neutralization information for a total of 23 viruses. For 162 plasma samples with the highest predicted bnAb activity (based on the 23-virus panel), we obtained ID50 values against a multi-clade 40-virus panel. The median ID50 served as a proxy for true neutralization breadth. Spearman correlations, Fisher’s exact test and the area under the ROC curve were used to evaluate virus screening panels of different size and composition. Results: A larger screening panel generally led to more accurate neutralization breadth predictions (see Figure). However, the gain in predictive strength decreased with the number of included viruses. A comparison of various transformation functions for the % inhibition values revealed max(value – 10%, 0) as best, yielding significantly better predictions than the frequently used 20-50-80 rule. As expected, virus panels of diverse subtypes performed better than single-clade panels. Intriguingly, inclusion of certain viruses like TRO clone11 led to consistently better average predictions than others, independent of panel size. General neutralization sensitivity of the 23 viruses had no impact on our findings. Conclusion: Collectively, our systematic survey of virus screening panels for the detection of HIV-1 neutralization breadth provides important information on the optimal design of test panels. This data provides a basis for a standardization of neutralization breadth assessment and prediction methods, which will be particularly important in the evaluation of forthcoming vaccine efficacy trials.

mechanisms. Moreover, we observed potent enhancement for ADCC activity of anti-V3/CD4i antibodies against HIV-1 infected cells in the presence of YIR-821. Conclusion: These data taken together suggest that YIR-821 has enhancing activity to ADCC with an excellent property of neutralization-enhancing activities for broad spectrum of HIV-1 in vitro. Combinational use of YIR-821 and anti-V3/CD4i antibodies represents a promising candidate for prevention or treatment of HIV-1 infection. 290 A 2-DIMENSIONAL APPROACH TO ANALYZING BROADLY NEUTRALIZING ANTIBODY COMBINATIONS Nicholas E. Webb , Colin M. Sevaried, Nicole Tobin, Grace M. Aldrovandi University of California Los Angeles, Los Angeles, CA, USA Background: Broadly neutralizing antibodies (bnAbs) show great promise for HIV treatment and prophylaxis. Immunotherapies involving combinations of bnAbs (BNACs) will be necessary to prevent transmission and the emergence of resistant variants. Current methods for evaluating BNACs use a constant dose ratio, which may not represent an ideal combination. These approaches focus on identifying synergy, with experimental and analytical methods that depend on combinatorial models. The models impose specific assumptions about the interactive mechanics of the system under study and their applicability to BNACs has never been formally determined. For example, the most commonly used model, Loewe Additivity, is explicitly invalid for assessing BNACs whose components exhibit different curvatures, or slopes, which are characteristic features of bnAb activity. Methods: We employed a two-dimensional matrix experimental strategy, where neutralization curves of one antibody are generated in the presence of a constant concentration of another antibody, for two-member permutations of PG9, VRC01, 10E8 and PG16. The neutralizing activity of the combination is then mapped as a two-dimensional surface, which can be analyzed empirically, without imposing any mechanistic assumptions. Because the primary target of neutralizing antibodies, the HIV-1 envelope (Env), is extremely variable, we characterized these BNACs across a panel of clade C, mother-to-child- transmitted Env isolates. Results: Combinatorial activity was assessed from both a dose and effect- based perspective, reflecting the two major combinatorial models, Loewe Additivity and Bliss Independence, respectively (see figure). Inclusion of a PG9/PG16 BNAC provided mechanistic insight, as this BNAC represents an ideal Loewe combination. Unlike other approaches, this method also measures combinatorial activity across a range of dose ratios, allowing us to assess clinical potential across a wider range of parameters. Conclusion: The matrix/surface approach is significantly more informative for evaluating BNACs, both clinically and mechanistically, than more common, fixed-ratio methods.

Poster Abstracts

CROI 2018 102