The individualized genetic barrier predicts treatment response in a large cohort of HIV-1 infected patients

Abstract

The success of combination antiretroviral therapy is limited by the evolutionary escape dynamics of HIV-1. We used Isotonic Conjunctive Bayesian Networks (I-CBNs), a class of probabilistic graphical models, to describe this process. We employed partial order constraints among viral resistance mutations, which give rise to a limited set of mutational pathways, and we modeled phenotypic drug resistance as monotonically increasing along any escape pathway. Using this model, the individualized genetic barrier (IGB) to each drug is derived as the probability of the virus not acquiring additional mutations that confer resistance. Drug-specific IGBs were combined to obtain the IGB to an entire regimen, which quantifies the virus' genetic potential for developing drug resistance under combination therapy. The IGB was tested as a predictor of therapeutic outcome using between 2,185 and 2,631 treatment change episodes of subtype B infected patients from the Swiss HIV Cohort Study Database, a large observational cohort. Using logistic regression, significant univariate predictors included most of the 18 drugs and single-drug IGBs, the IGB to the entire regimen, the expert rules-based genotypic susceptibility score (GSS), several individual mutations, and the peak viral load before treatment change. In the multivariate analysis, the only genotype-derived variables that remained significantly associated with virological success were GSS and, with 10-fold stronger association, IGB to regimen. When predicting suppression of viral load below 400 cps/ml, IGB outperformed GSS and also improved GSS-containing predictors significantly, but the difference was not significant for suppression below 50 cps/ml. Thus, the IGB to regimen is a novel data-derived predictor of treatment outcome that has potential to improve the interpretation of genotypic drug resistance tests.

Figure 1. Schematic illustration of I-CBN model and individualized genetic barrier (IGB).

(A) A partially ordered set of three mutations, , , and , is considered with the two relations and , resulting in two possible escape pathways of the virus, namely or . (B) The partial order constraints give rise to the genotype lattice consisting of genotypes 000, 001, 100, 101, and 111 indicated with bold arrows, where genotypes are encoded as binary strings such that 000 is the wild type (no mutations), 100 is defined by mutation and identified with , 101 with , etc. The genotype lattice is shown inside the embedding hypercube . For each antiretroviral drug, genotypes are labeled as either susceptible (green) or resistant (red). (C) Genotype lattice isolated from the embedding hypercube. The IGB is the probability of the virus not reaching a resistant state.

Figure 2. Data flow.

Matched pairs of viral genotype and drug resistance phenotype from the Stanford HIV Drug Resistance Database (top right) were used to learn I-CBN models for all drugs separately. The drug-specific individualized genetic barriers (IGBs) are derived from these models. The IGB to regimen is computed for each genotype-therapy pair in the Swiss HIV Cohort Database and its predictive power is assessed in prediction models that also account for classical demographic, clinical, and genetic covariates.

Figure 3. Multivariate analysis of predictors of response to antiretroviral combination therapy in the SHCS database.

Associations have been tested using a logistic regression model and odds ratios of therapeutic success, defined as viral load reduction below 50 cps/ml (A) and 400 cps/ml (B), are reported together with their 95% confidence intervals on a logarithmic scale. Benjamini-Hochberg-corrected p-values are represented as black () and grey () symbols. Only predictors with a p-value smaller than 0.01 are included.

Figure 4. ROC curves quantifying the performance of elastic net regularized logistic regression models in predicting treatment outcome, defined as a reduction of viral load below 50 cps/ml (A) and 400 cps/ml (B).

The areas under the ROC curves (AUC values) are reported in Table S5 and Table S6. Prediction models are encoded by the sets of predictors used, where C refers to the demographic and clinical variables, D refers to drugs, and M to mutations. For example, the model IGB+CDM includes as predictors IGB to regimen, clinical and demographic predictors, applied drugs, and mutations. The models with all predictors perform significantly better than all other models.