A proposed molecular mechanism for pathogenesis of severe RNA-viral pulmonary infections
URL with Digital Object Identifier
Background. Cancer chemotherapy responses have been related to multiple pharmacogenetic biomarkers, often for the same drug. This study utilizes machine learning to derive multi-gene expression signatures that predict individual patient responses to specific tyrosine kinase inhibitors (TKIs), including erlotinib, gefitinib, sorafenib, sunitinib, lapatinib and imatinib. Methods. Support vector machines predicting sensitivity versus resistance to TKIs were trained using a novel systems biology-based approach. This began with expression of genes previously implicated in specific TKI responses, and then expanded to evaluate genes whose products were related through biochemical pathways and interactions. Results. Optimal pathway-extended (PE) support vector machines predicted responses in patients at accuracies of 65% (imatinib), 71% (lapatinib and gefitinib), 78% (sunitinib), 83% (erlotinib) and 88% (sorafenib). These highest performing PE models demonstrated improved balance predicting both sensitive and resistant patient categories, with most PE genes comprising these models found to have previous roles in cancer etiology. Ensemble machine learning-based averaging of multiple PE high performance models derived for an individual TKI increased accuracy to >80% for all TKIs, except lapatinib. Conclusions. Through incorporation of novel cancer biomarkers, machine learning-based PE expression signatures display strong efficacy predicting both sensitive and resistant patient responses to chemotherapy.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.