Poster presentation at:
14th European AIDS Conference
16th October 2013 - 19th October 2013
Revell AD1, Wang D1, Streinu-Cercel A2, Ene L3, Dragovic GJ4, Hamers R5, Morrow C6, Wood R6, Tempelman H7, AMJ Wensing8, Reiss P5, 9, van Sighem A9, Pozniak A10, Montaner JS11, Lane HC12, Larder BA1 on behalf of the RDI study group
1 The HIV Resistance Response Database Initiative (RDI), London UK
2 National Institute of Infectious Diseases Prof.Dr. Matei BalAY and Carol Davila University of Medicine and Pharmacy, Bucharest, Romania
3 "Dr. Victor Babes" Hospital for Infectious and Tropical Diseases, Bucharest, Romania
4 Department of Pharmacology, Clinical Pharmacology and Toxicology, School of Medicine University of Belgrade, Belgrade, Serbia
5 Dept of Global Health, Academic Medical Centre of the University of Amsterdam, Amsterdam Institute for Global Health and Development, Amsterdam, The Netherlands
6 Desmond Tutu HIV Centre, Cape Town, South Africa
7 Ndlovu Care Group, Elandsdoorn, South Africa
8 University Medical Centre, Utrecht, The Netherland
9 Stichting HIV Monitoring, Amsterdam, The Netherlands
10 Chelsea and Westminster Hospital, London, UK
11 BC Centre for Excellence in HIV/AIDS, Vancouver, Canada
12 National Institute of Allergy and Infectious Diseases, Bethesda, MD, USA
Optimisation of antiretroviral therapy in resource-limited settings (RLS) can be challenging without access to the newest drugs, or genotyping to help tailor therapy against resistance. Computational models have been developed to predict virological response to therapy but most require a genotype, have been developed using western data and are less accurate for patients from RLS. Here we describe the development of models that do not require a genotype, using substantial data from RLS.
Random forest (RF) models were trained to predict the probability of virological response (viral load <50 copies/ml) using baseline viral load, CD4 count, treatment history, new drug regimen and time to follow-up from 22,567 treatment change episodes (TCEs) following failure, including 1,090 from southern Africa and 328 from Eastern Europe. The models were assessed during cross-validation, with an independent test set (n=1,000 including 582 from Europe, 23 from Eastern Europe and 100 from southern Africa). The area under the ROC curve (AUC) was the main outcome. The predictive accuracy of the models was compared to genotyping using three rules-based interpretation systems to produce genotypic sensitivity scores (GSS).
The models achieved a mean AUC of 0.82 during cross validation, 0.80 with the 1,000 test set, 0.81 for the European, 0.82 for the Eastern European and 0.78 for the southern African test cases. Of the 1,000 test TCEs, 346 had genotypes and the AUC for the GSS ranged between 0.56 and 0.57 - significantly inferior to the accuracy of the models (p<0.0001).
These models, trained with the largest dataset so far, were able to predict response to HIV therapy significantly more accurately than genotyping with rules-based interpretation. They were accurate for cases from different RLS. The models are freely available online and have the potential to predict and avoid treatment failure by identifying potentially effective, alternative regimens.
New predictive models accurately predict HIV treatment response despite missing data
Predicting HIV therapy response despite missing data - new models published
2021 update to HIV-TRePS: a highly flexible and accurate system for the prediction of treatment response from incomplete baseline information in different healthcare settings
New models that predict treatment outcome without the need for a baseline CD4 count or genotype added to HIV-TRePS
The most accurate models to date for predicting response to HIV therapy are released for clinical use
RDI expands the range of models available through HIV-TRePS to include their most accurate models to date
AI models predict absolute virological response to HIV therapy over time