Application of the SLAPNAP statistical learning tool to broadly neutralizing antibody HIV prevention research

Abstract

Combination monoclonal broadly neutralizing antibody (bnAb) regimens are in clinical development for HIV prevention, necessitating additional knowledge of bnAb neutralization potency/breadth against circulating viruses. Williamson et al. (2021) described a software tool, Super LeArner Prediction of NAb Panels (SLAPNAP), with application to any HIV bnAb regimen with sufficient neutralization data against a set of viruses in the Los Alamos National Laboratory's Compile, Neutralize, and Tally Nab Panels repository. SLAPNAP produces a proteomic antibody resistance (PAR) score for Env sequences based on predicted neutralization resistance and estimates variable importance of Env amino acid features. We apply SLAPNAP to compare HIV bnAb regimens undergoing clinical testing, finding improved power for downstream sieve analyses and increased precision for comparing neutralization potency/breadth of bnAb regimens due to the inclusion of PAR scores of Env sequences with much larger sample sizes available than for neutralization outcomes. SLAPNAP substantially improves bnAb regimen characterization, ranking, and down-selection.

Keywords: Immunological methods; Immunology; Mathematical biosciences; Virology.

Graphical abstract

Figure 1

Relative efficiency (ratio of the sample variances of the estimated mean outcome value for the Reference approach vs. SLAPNAP-augmented approach, taken over 1000 simulated replicates) versus estimated SLAPNAP-prediction performance for each bnAb regimen listed in Table S1 Top row: Prediction of IC₈₀. Bottom row: Prediction of binary IC₈₀ < 1 μg/mL. Columns denote the percentage increase in the number of viruses included in the SLAPNAP-augmented approach when adding viruses with data on Env sequence data only to viruses with data on both Env sequence and IC₈₀. The bnAb regimens are differentiated by color; each point represents the ratio of the Monte-Carlo variances taken over 1000 simulated replicates. See also Figures S3–S8.

Figure 2

Bounded relative efficiency (ratio of the squared widths of the 95% percentile bootstrap confidence interval for the mean neutralization outcome for the Reference approach vs. SLAPNAP-augmented approach bounded above by the maximum possible efficiency for the given combination of bnAb regimen and country, $\frac{\#\left\{sequencesavailableinbothLANLandCATNAP\right\}}{\#\left\{sequencesavailableinCATNAP\right\}}$) versus the proportion of additional numbers of Env sequences in LANL compared to CATNAP Top row: Prediction of IC₈₀. Values were imputed in SLAPNAP at two times the right censoring value reported in CATNAP, as in ref. Bottom row: Prediction of binary IC₈₀ < 1 μg/mL. Columns show the estimated prediction performance for each bnAb regimen listed in Table S1. The bnAb regimens are differentiated by color, while countries are differentiated by the plotting symbol (1–8). Only country/bnAb regimen combinations with at least 30 pseudoviruses in CATNAP were analyzed. Clades represented overall: 01_AE, 02_AG, 07_BC, A1, B, C, D, Other. Clades by country in the figure: China: 01_AE, 02_AG, 07_BC, B, C, Other. (majority: Other, 01_AE, B); Germany: 02_AG, B, C, Other. (majority: B); Tanzania: A1, C, D, Other. (majority: C, Other); United States: 01_AE, 02_AG, A1, B, C, D, Other. (majority: B); South Africa: A1, B, C, D, Other (majority: C).

Figure 3

Empirical power of sieve analyses under different sieve alternative hypotheses, ranging from a large sieve effect (left-column) to the null hypothesis of no sieve effect (right-column) (A “sieve effect” is differential prevention efficacy against S230 HIV-1 vs. against not S230 HIV-1.) Power of a Lunn and McNeil test for detecting a sieve effect is displayed for an unbiased site-scanning analysis over 414 amino acid (AA) positions in gp120 that passed a minimum variability filter in AMP, and for a priority hypothesis-driven analysis that restricted to the 15 AA positions in gp120 pre-identified as important by SLAPNAP.