[Paper Review] Predicting in vivo escape dynamics of HIV-1 from a broadly neutralizing antibody

Matthijs Meijers, Kanika Vanshylla+3 more|arXiv (Cornell University)|Aug 6, 2020

HIV Research and Treatment49 references27 citations

TL;DR

This study develops a biophysically grounded fitness model that predicts in vivo HIV-1 escape dynamics from broadly neutralizing antibodies (bnAbs), using clinical trial data from 11 individuals treated with the bnAb 10-1074. The model identifies antibody dosage and viral load as key drivers of viral fitness, revealing a dosage-dependent fitness ranking of sensitive and resistant strains due to an evolutionary tradeoff between resistance and replication cost, which successfully predicts strain frequency turnover during treatment.

ABSTRACT

Broadly neutralizing antibodies are promising candidates for treatment and prevention of HIV-1 infections. Such antibodies can temporarily suppress viral load in infected individuals; however, the virus often rebounds by escape mutants that have evolved resistance. In this paper, we map an in vivo fitness landscape of HIV-1 interacting with broadly neutralizing antibodies, using data from a recent clinical trial. We identify two fitness factors, antibody dosage and viral load, that determine viral reproduction rates reproducibly across different hosts. The model successfully predicts the escape dynamics of HIV-1 in the course of an antibody treatment, including a characteristic frequency turnover between sensitive and resistant strains. This turnover is governed by a dosage-dependent fitness ranking, resulting from an evolutionary tradeoff between antibody resistance and its collateral cost in drug-free growth. Our analysis suggests resistance-cost tradeoff curves as a measure of antibody performance in the presence of resistance evolution.

Motivation & Objective

• To understand the in vivo dynamics of HIV-1 escape from broadly neutralizing antibodies (bnAbs) in human hosts.
• To identify the key host and viral factors governing viral fitness during bnAb treatment.
• To develop a predictive fitness model that captures strain-specific replication rates and resistance evolution.
• To quantify the evolutionary tradeoff between bnAb resistance and viral fitness cost in the absence of antibodies.
• To evaluate the model’s predictive power across different host environments using cross-validation.

Proposed method

• Uses time-resolved in vivo data from a clinical trial (Caskey et al.) including viral load, bnAb concentration, and single-virion genome sequences from 11 HIV-1-infected individuals.
• Develops a biophysical fitness model where viral replication rate depends on antibody dosage (A) and effective viral load (Nu), with strain-specific parameters: baseline growth rate (b0_i), resistance (Ki), and death rate (d).
• Applies a saturation model for ecological constraints: replication rate is reduced by antibody neutralization (1/(1 + A/Ki)) and by viral load saturation (exp(−CNu)).
• Infers fitness parameters (b0_i, Ki, d, C) via Bayesian inference from longitudinal data, using pseudocounts for zero-frequency observations.
• Validates predictions using a leave-one-out cross-validation scheme: trains on data from 10 individuals and predicts trajectories for the excluded individual.
• Tests prediction accuracy by comparing predicted frequency changes (with sampling noise) to observed changes between sampling time points.

Experimental results

Research questions

• RQ1 How do antibody dosage and viral load jointly influence the in vivo fitness of HIV-1 strains during bnAb treatment?
• RQ2 What is the nature and magnitude of the fitness cost associated with bnAb resistance mutations in HIV-1?
• RQ3 Can a universal fitness model predict strain-specific escape dynamics across different human hosts?
• RQ4 How does the fitness ranking of sensitive and resistant strains change with bnAb dosage, and what drives the observed frequency turnover?
• RQ5 To what extent can the model predict viral escape evolution using only baseline host-specific data?

Key findings

• The fitness model successfully predicts viral escape dynamics, including the characteristic frequency turnover between sensitive and resistant strains during bnAb treatment.
• The model identifies two universal fitness factors—antibody dosage and viral load—that determine viral reproduction rates across different hosts.
• A resistance-cost tradeoff is quantified: strains with higher resistance to bnAb 10-1074 exhibit reduced replication capacity in the absence of antibodies.
• The saturation model (with exponential viral load constraint) fits the data significantly better than the linear ecological model, with a Δ(BIC) = 24 in favor of saturation.
• The predicted frequency changes between sampling points closely match the observed changes, validating the model’s predictive power under realistic sampling noise.
• Fitness parameters (b0_i, Ki, d, C) inferred from the full dataset are consistent with those from leave-one-out training sets, supporting approximate universality across hosts.