Improved Prediction of Response to Antiretroviral Combination Therapy using the Genetic Barrier to Drug Resistance

By André Altmann, Niko Beerenwinkel, Tobias Sing, I. Savenkov, M. Doumer, R. Kaiser, Soo-Yon Rhee, W. J. Fessel, R. Shafer, Thomas Lengauer. 2006

63 citations recorded by the source corpus.

Background The outcome of antiretroviral combination therapy depends on many factors involving host, virus, and drugs. We investigate prediction of treatment response from the applied drug combination and the genetic constellation of the virus population at baseline. The virus's evolutionary potential for escaping from drug pressure is explored as an additional predictor. Methods We compare different encodings of the viral genotype and antiretroviral regimen including phenotypic and evolutionary information, namely predicted phenotypic drug resistance, activity of the regimen estimated from sequence space search, the genetic barrier to drug resistance, and the genetic progression score. These features were evaluated in the context of different statistical learning procedures applied to the binary classification task of predicting virological response. Classifier performance was evaluated using cross-validation and receiver operating characteristic curves on 6,337 observed treatment change episodes from the Stanford HIV Drug Resistance Database and a large US clinic-based patient population. Results We find that the choice of appropriate features affects predictive performance more profoundly than the choice of the statistical learning method. Application of the genetic barrier to drug resistance, which combines phenotypic and evolutionary information, outperformed the genetic progression score, which uses exclusively evolutionary knowledge. The benefit of phenotypic information in predicting virological response was confirmed by using predicted fold changes in drug susceptibility. Moreover, genetic barrier and predicted phenotypic drug resistance were found to be the best encodings across all datasets and statistical learning methods examined. Availability THEO (THErapy Optimizer), a prototypical implementation of the best performing approach, is freely available for research purposes at http://www.geno2pheno.org.

Subfields: Genetics, Infectious Diseases, Virology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Genetics

1. Finale Doshi-Velez
2. Anne-Mieke Vandamme
3. Thomas Lengauer

In Infectious Diseases

1. Maurizio Zazzi
2. Jasmina Bogojeska
3. Vukosi Marivate

In Virology

1. Maurizio Zazzi
2. Jasmina Bogojeska
3. Finale Doshi-Velez
4. Sonali Parbhoo
5. Vukosi Marivate

Citations in the corpus

• Combining Kernel and Model Based Learning for HIV Therapy Selection (2017, methodology): “[2] use tree-based learning to predict a patient’s virological response using evolutionary information, while others [3, 4] predict virological response based on the patient’s treatment history.”
• Antiretroviral Therapy Optimisation without Genotype Resistance Testing: A Perspective on Treatment History Based Models (2010, background): “Accordingly, there is a plethora of GRT-based decision systems either for single drug susceptibility scoring or combined ART (cART) optimisation, from rule-based [23,24,25] to machine learning approaches [26,27,28,29,30,31,32].”
• Multi-task learning for HIV therapy screening (2008, background): “Altmann et al. (2007) approach the problem by including various phenotypic information and an estimate of future evolutionary de­velopment of the virus in the learning process.”
• Investigation of expert rule bases, logistic regression, and non-linear machine learning techniques for predicting response to antiretroviral treatment (2008, background): “More recently, machine learning approaches have been investigated with different technologies, as artificial neural networks modelling [14], fuzzy-rule-based systems [15,16], probabilistic models focusing on viral evolutionary pathways through mutagenetic trees (MT) [7,17] and in vivo viral fitness ”
• Selecting anti-HIV therapies based on a variety of genomic and clinical factors (2008, methodology): “Each therapy with a baseline and a follow-up VL is labeled a success if there is a drop of 2 log in VL or a drop below detection level (500 copies per milliliter); otherwise, the therapy is labeled as a failure.”