A data mining approach for identifying pathway-gene biomarkers for predicting clinical outcome: A case study of erlotinib and sorafenib

Drug-Drug Interaction between Metformin and Sorafenib Alters Antitumor Effect in Hepatocellular Carcinoma Cells

Hepatocellular carcinoma (HCC) is the most common primary liver malignancy and is one of the leading causes of cancer-related deaths worldwide. The multitarget inhibitor sorafenib is a first-line treatment of patients with advanced unresectable HCC. Recent clinical studies have evidenced that patients treated with sorafenib together with the antidiabetic drug metformin have a survival disadvantage compared with patients receiving sorafenib only. Here, we examined whether a clinically relevant dose of metformin (50 mg/kg per day) could influence the antitumoral effects of sorafenib (15 mg/kg per day) in a subcutaneous xenograft model of human HCC growth using two different sequences of administration, i.e., concomitant versus sequential dosing regimens. We observed that the administration of metformin 6 hours prior to sorafenib was significantly less effective in inhibiting tumor growth (15.4% tumor growth inhibition) than concomitant administration of the two drugs (59.5% tumor growth inhibition). In vitro experiments confirmed that pretreatment of different human HCC cell lines with metformin reduced the effects of sorafenib on cell viability, proliferation, and signaling. Transcriptomic analysis confirmed significant differences between xenografted tumors obtained under the concomitant and the sequential dosing regimens. Taken together, these observations call into question the benefit of parallel use of metformin and sorafenib in patients with advanced HCC and diabetes, as the interaction between the two drugs could ultimately compromise patient survival. SIGNIFICANCE STATEMENT: When drugs are administered sequentially, metformin alters the antitumor effect of sorafenib, the reference treatment for advanced hepatocellular carcinoma, in a preclinical murine xenograft model of liver cancer progression as well as in hepatic cancer cell lines. Defective activation of the AMP-activated protein kinase pathway as well as major transcriptomic changes are associated with the loss of the antitumor effect. These results echo recent clinical work reporting a poorer prognosis for patients with liver cancer who were cotreated with metformin and sorafenib.

A deep learning architecture for metabolic pathway prediction

MOTIVATION

Understanding the mechanisms and structural mappings between molecules and pathway classes are critical for design of reaction predictors for synthesizing new molecules. This article studies the problem of prediction of classes of metabolic pathways (series of chemical reactions occurring within a cell) in which a given biochemical compound participates. We apply a hybrid machine learning approach consisting of graph convolutional networks used to extract molecular shape features as input to a random forest classifier. In contrast to previously applied machine learning methods for this problem, our framework automatically extracts relevant shape features directly from input SMILES representations, which are atom-bond specifications of chemical structures composing the molecules.

RESULTS

Our method is capable of correctly predicting the respective metabolic pathway class of 95.16% of tested compounds, whereas competing methods only achieve an accuracy of 84.92% or less. Furthermore, our framework extends to the task of classification of compounds having mixed membership in multiple pathway classes. Our prediction accuracy for this multi-label task is 97.61%. We analyze the relative importance of various global physicochemical features to the pathway class prediction problem and show that simple linear/logistic regression models can predict the values of these global features from the shape features extracted using our framework.

AVAILABILITY AND IMPLEMENTATION

https://github.com/baranwa2/MetabolicPathwayPrediction.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.