Predicting Cancer Drug Response In Vivo by Learning an Optimal Feature Selection of Tumour Molecular Profiles.

Linh C. Nguyen,Pedro J. Ballester,Alejandra Bruna,Stefan Naulaerts,Ghita Ghislat

doi:10.3390/biomedicines9101319

Abstract

(1) Background: Inter-tumour heterogeneity is one of cancer’s most fundamental features. Patient stratification based on drug response prediction is hence needed for effective anti-cancer therapy. However, single-gene markers of response are rare and/or may fail to achieve a significant impact in the clinic. Machine Learning (ML) is emerging as a particularly promising complementary approach to precision oncology. (2) Methods: Here we leverage comprehensive Patient-Derived Xenograft (PDX) pharmacogenomic data sets with dimensionality-reducing ML algorithms with this purpose. (3) Results: Combining multiple gene alterations via ML leads to better discrimination between sensitive and resistant PDXs in 19 of the 26 analysed cases. Highly predictive ML models employing concise gene lists were found for three cases: paclitaxel (breast cancer), binimetinib (breast cancer) and cetuximab (colorectal cancer). Interestingly, each of these multi-gene ML models identifies some treatment-responsive PDXs not harbouring the best actionable mutation for that case. Thus, ML multi-gene predictors generally have much fewer false negatives than the corresponding single-gene marker. (4) Conclusions: As PDXs often recapitulate clinical outcomes, these results suggest that many more patients could benefit from precision oncology if ML algorithms were also applied to existing clinical pharmacogenomics data, especially those algorithms generating classifiers combining data-selected gene alterations.

Highlights

(4) Conclusions: As Patient-Derived Xenograft (PDX) often recapitulate clinical outcomes, these results suggest that many more patients could benefit from precision oncology if Machine Learning (ML) algorithms were applied to existing clinical pharmacogenomics data, especially those algorithms generating classifiers combining data-selected gene alterations
We will determine whether this is the case in vivo. This would mean that many more patients could benefit from precision oncology if multi-gene ML methods are applied to existing clinical pharmacogenomics data
Note that this study always reports the median performances of each algorithm on held-out PDXs that were not used to train or select the model providing the prediction

Summary

Introduction

Single-Gene Markers, the Predominant Approach to Precision Oncology, Are Scarce. It is well-established that the efficacy of cancer drugs is strongly patient-dependent. The predominant approach to date has been to identify a specific somatic mutation to act as a single-gene biomarker discriminating between therapy responders and non-responders [2]. Such a predictive biomarker is commonly referred to as an actionable mutation (either a point mutation, deletion or amplification of a specific gene in the tumour sample). Despite being able to predict the response to some drugs [3,4], most patients cannot benefit from single-gene markers because these have not been

Methods

Results

Discussion

Conclusion