Abstract Body

The use of broadly neutralizing antibodies (bNAbs) in HIV-1 cure-related clinical trials has greatly increased in recent years. Even though bNAbs are broadly neutralizing, not all HIV-1 variants are sensitive to a given bNAb due to tremendous viral diversity and development of escape mutations. Screening of participants for bNAb sensitivity before or after inclusion in a clinical trial is therefore crucial to obtain valid antibody efficacy data. Different assays are currently being used to screen bNAb sensitivity. Here we compare three different methods for bNAb sensitivity prediction of the two bNAbs: 10-1074 and 3BNC117.

The three methods tested in this study were the phenotypic ‘PhenoSense® HIV Monoclonal Antibody Assay’ developed by Monogram Biosciences, and two genotypic prediction algorithms; the ‘HIV screening analysis’ developed at Rockefeller University and the ‘bNAb-ReP’ developed by VRC/NIH. The PhenoSense assay is a pseudovirus neutralization assay. The prediction pipelines were fed with intact HIV-1 envelope sequences from single genome amplification of HIV-1 envelope performed on pre-ART plasma RNA. The threshold for sensitivity in the PhenoSense assay was set to IC90<1.5 ?g/ml for 10-1074 and IC90<2.0 ?g/ml for 3BNC117. When using the genotypic prediction algorithms, a study participant was defined as harboring bNAb sensitive viruses if ?90% of all envelope sequences were sensitive. Fifty-nine clinical samples from ART-naïve participants in the eCLEAR trial (NCT03041012) were evaluated.

Concordant classification across all three methods were obtained for 79% and 52% of the participants for 10-1074 and 3BNC117, respectively. Comparing the HIV screening analysis and bNAb-ReP prediction algorithms to the PhenoSense assay, agreement for 10-1074 sensitivity was 87% and 83%, respectively, while agreement for 3BNC117 sensitivity was 60% and 75%, respectively. For 3BNC117, the HIV screening analysis was more likely to classify the envelope sequences as sensitive whereas, no pattern was observed for the bNAb-ReP pipeline.

We observed good agreement in bNAb sensitivity between the PhenoSense assay and both genotypic prediction algorithms for the V3-loop binding bNAb 10-1074. The agreement between phenotypic and genotypic sensitivity was lower for the CD4-binding site bNAb 3BNC117, which may pose greater challenges when screening participants for clinical studies and comparing results across different trials.