An update to the HIV-TRePS system: The development and evaluation of new global and local computational models to predict HIV treatment outcomes, with or without a genotype

on behalf of the RDI Data and Study Group

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Objectives: Optimizing antiretroviral drug combination on an individual basis in resource-limited settings is challenging because of the limited availability of drugs and genotypic resistance testing. Here, we describe our latest computational models to predict treatment responses, with or without a genotype, and compare the potential utility of global and local models as a treatment tool for South Africa. Methods: Global random forest models were trained to predict the probability of virological response to therapy following virological failure using 29 574 treatment change episodes (TCEs) without a genotype, 3179 of which were from South Africa and were used to develop local models. In addition, 15 130 TCEs including genotypes were used to develop another set of models. The 'no-genotype' models were tested with an independent global test set (n = 1700) plus a subset from South Africa (n = 222). The genotype models were tested with 750 independent cases. Results: The global no-genotype models achieved area under the receiver-operating characteristic curve (AUC) values of 0.82 and 0.79 with the global and South African tests sets, respectively, and the South African models achieved AUCs of 0.70 and 0.79. The genotype models achieved an AUC of 0.84. The global no-genotype models identified more alternative, locally available regimens that were predicted to be effective for cases that failed their new regimen in the South African clinics than the local models. Both sets of models were significantly more accurate predictors of outcomes than genotyping with rules-based interpretation. Conclusions: These latest global models predict treatment responses accurately even without a genotype, out-performed the local South African models and have the potential to help optimize therapy, particularly in resource-limited settings.

Original languageEnglish
Article numberdkw217
Pages (from-to)2928-2937
Number of pages10
JournalJournal of Antimicrobial Chemotherapy
Volume71
Issue number10
DOIs
Publication statusPublished - 1 Oct 2016

Cite this

@article{45f96d178b3c4d12a01f020a0496b030,
title = "An update to the HIV-TRePS system: The development and evaluation of new global and local computational models to predict HIV treatment outcomes, with or without a genotype",
abstract = "Objectives: Optimizing antiretroviral drug combination on an individual basis in resource-limited settings is challenging because of the limited availability of drugs and genotypic resistance testing. Here, we describe our latest computational models to predict treatment responses, with or without a genotype, and compare the potential utility of global and local models as a treatment tool for South Africa. Methods: Global random forest models were trained to predict the probability of virological response to therapy following virological failure using 29 574 treatment change episodes (TCEs) without a genotype, 3179 of which were from South Africa and were used to develop local models. In addition, 15 130 TCEs including genotypes were used to develop another set of models. The 'no-genotype' models were tested with an independent global test set (n = 1700) plus a subset from South Africa (n = 222). The genotype models were tested with 750 independent cases. Results: The global no-genotype models achieved area under the receiver-operating characteristic curve (AUC) values of 0.82 and 0.79 with the global and South African tests sets, respectively, and the South African models achieved AUCs of 0.70 and 0.79. The genotype models achieved an AUC of 0.84. The global no-genotype models identified more alternative, locally available regimens that were predicted to be effective for cases that failed their new regimen in the South African clinics than the local models. Both sets of models were significantly more accurate predictors of outcomes than genotyping with rules-based interpretation. Conclusions: These latest global models predict treatment responses accurately even without a genotype, out-performed the local South African models and have the potential to help optimize therapy, particularly in resource-limited settings.",
author = "Revell, {Andrew D.} and Dechao Wang and Robin Wood and Carl Morrow and Hugo Tempelman and Raph Hamers and Peter Reiss and {van Sighem}, {Ard I.} and Mark Nelson and Montaner, {Julio S.G.} and Lane, {H. Clifford} and Larder, {Brendan A.} and Richard Harrigan and {de Wit}, {Tobias Rinke} and Raph Hamers and Kim Sigaloff and Brian Agan and Vincent Marconi and Scott Wegner and Wataru Sugiura and Maurizio Zazzi and Rolf Kaiser and Eugen Schuelter and Adrian Streinu-Cercel and Gerardo Alvarez-Uria and Jose Gatell and Elisa Lazzari and Brian Gazzard and Anton Pozniak and Sundhiya Mandalia and Daniel Webster and Colette Smith and Lidia Ruiz and Bonaventura Clotet and Schlomo Staszewski and Carlo Torti and Cliff Lane and Julie Metcalf and Perez-Elias, {Maria Jesus} and Stefano Vella and Gabrielle Dettorre and Andrew Carr and Richard Norris and Karl Hesse and Emanuel Vlahakis and Roos Barth and Chris Hoffmann and Luminita Ene and Gordana Dragovic and Ricardo Diaz and {on behalf of the RDI Data and Study Group}",
year = "2016",
month = "10",
day = "1",
doi = "10.1093/jac/dkw217",
language = "English",
volume = "71",
pages = "2928--2937",
journal = "Journal of Antimicrobial Chemotherapy",
issn = "0305-7453",
publisher = "Oxford University Press",
number = "10",

}

An update to the HIV-TRePS system : The development and evaluation of new global and local computational models to predict HIV treatment outcomes, with or without a genotype. / on behalf of the RDI Data and Study Group.

In: Journal of Antimicrobial Chemotherapy, Vol. 71, No. 10, dkw217, 01.10.2016, p. 2928-2937.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - An update to the HIV-TRePS system

T2 - The development and evaluation of new global and local computational models to predict HIV treatment outcomes, with or without a genotype

AU - Revell, Andrew D.

AU - Wang, Dechao

AU - Wood, Robin

AU - Morrow, Carl

AU - Tempelman, Hugo

AU - Hamers, Raph

AU - Reiss, Peter

AU - van Sighem, Ard I.

AU - Nelson, Mark

AU - Montaner, Julio S.G.

AU - Lane, H. Clifford

AU - Larder, Brendan A.

AU - Harrigan, Richard

AU - de Wit, Tobias Rinke

AU - Hamers, Raph

AU - Sigaloff, Kim

AU - Agan, Brian

AU - Marconi, Vincent

AU - Wegner, Scott

AU - Sugiura, Wataru

AU - Zazzi, Maurizio

AU - Kaiser, Rolf

AU - Schuelter, Eugen

AU - Streinu-Cercel, Adrian

AU - Alvarez-Uria, Gerardo

AU - Gatell, Jose

AU - Lazzari, Elisa

AU - Gazzard, Brian

AU - Pozniak, Anton

AU - Mandalia, Sundhiya

AU - Webster, Daniel

AU - Smith, Colette

AU - Ruiz, Lidia

AU - Clotet, Bonaventura

AU - Staszewski, Schlomo

AU - Torti, Carlo

AU - Lane, Cliff

AU - Metcalf, Julie

AU - Perez-Elias, Maria Jesus

AU - Vella, Stefano

AU - Dettorre, Gabrielle

AU - Carr, Andrew

AU - Norris, Richard

AU - Hesse, Karl

AU - Vlahakis, Emanuel

AU - Barth, Roos

AU - Hoffmann, Chris

AU - Ene, Luminita

AU - Dragovic, Gordana

AU - Diaz, Ricardo

AU - on behalf of the RDI Data and Study Group

PY - 2016/10/1

Y1 - 2016/10/1

N2 - Objectives: Optimizing antiretroviral drug combination on an individual basis in resource-limited settings is challenging because of the limited availability of drugs and genotypic resistance testing. Here, we describe our latest computational models to predict treatment responses, with or without a genotype, and compare the potential utility of global and local models as a treatment tool for South Africa. Methods: Global random forest models were trained to predict the probability of virological response to therapy following virological failure using 29 574 treatment change episodes (TCEs) without a genotype, 3179 of which were from South Africa and were used to develop local models. In addition, 15 130 TCEs including genotypes were used to develop another set of models. The 'no-genotype' models were tested with an independent global test set (n = 1700) plus a subset from South Africa (n = 222). The genotype models were tested with 750 independent cases. Results: The global no-genotype models achieved area under the receiver-operating characteristic curve (AUC) values of 0.82 and 0.79 with the global and South African tests sets, respectively, and the South African models achieved AUCs of 0.70 and 0.79. The genotype models achieved an AUC of 0.84. The global no-genotype models identified more alternative, locally available regimens that were predicted to be effective for cases that failed their new regimen in the South African clinics than the local models. Both sets of models were significantly more accurate predictors of outcomes than genotyping with rules-based interpretation. Conclusions: These latest global models predict treatment responses accurately even without a genotype, out-performed the local South African models and have the potential to help optimize therapy, particularly in resource-limited settings.

AB - Objectives: Optimizing antiretroviral drug combination on an individual basis in resource-limited settings is challenging because of the limited availability of drugs and genotypic resistance testing. Here, we describe our latest computational models to predict treatment responses, with or without a genotype, and compare the potential utility of global and local models as a treatment tool for South Africa. Methods: Global random forest models were trained to predict the probability of virological response to therapy following virological failure using 29 574 treatment change episodes (TCEs) without a genotype, 3179 of which were from South Africa and were used to develop local models. In addition, 15 130 TCEs including genotypes were used to develop another set of models. The 'no-genotype' models were tested with an independent global test set (n = 1700) plus a subset from South Africa (n = 222). The genotype models were tested with 750 independent cases. Results: The global no-genotype models achieved area under the receiver-operating characteristic curve (AUC) values of 0.82 and 0.79 with the global and South African tests sets, respectively, and the South African models achieved AUCs of 0.70 and 0.79. The genotype models achieved an AUC of 0.84. The global no-genotype models identified more alternative, locally available regimens that were predicted to be effective for cases that failed their new regimen in the South African clinics than the local models. Both sets of models were significantly more accurate predictors of outcomes than genotyping with rules-based interpretation. Conclusions: These latest global models predict treatment responses accurately even without a genotype, out-performed the local South African models and have the potential to help optimize therapy, particularly in resource-limited settings.

UR - http://www.scopus.com/inward/record.url?scp=84994791060&partnerID=8YFLogxK

U2 - 10.1093/jac/dkw217

DO - 10.1093/jac/dkw217

M3 - Article

VL - 71

SP - 2928

EP - 2937

JO - Journal of Antimicrobial Chemotherapy

JF - Journal of Antimicrobial Chemotherapy

SN - 0305-7453

IS - 10

M1 - dkw217

ER -