ABSTRACT
Predicting clinical response to anticancer drugs remains a major challenge in cancer treatment. Emerging reports indicate that the tumour microenvironment and heterogeneity can limit the predictive power of current biomarker-guided strategies for chemotherapy. Here we report the engineering of personalized tumour ecosystems that contextually conserve the tumour heterogeneity, and phenocopy the tumour microenvironment using tumour explants maintained in defined tumour grade-matched matrix support and autologous patient serum. The functional response of tumour ecosystems, engineered from 109 patients, to anticancer drugs, together with the corresponding clinical outcomes, is used to train a machine learning algorithm; the learned model is then applied to predict the clinical response in an independent validation group of 55 patients, where we achieve 100% sensitivity in predictions while keeping specificity in a desired high range. The tumour ecosystem and algorithm, together termed the CANScript technology, can emerge as a powerful platform for enabling personalized medicine.
Subject(s)
Algorithms , Antineoplastic Agents/pharmacology , Extracellular Matrix Proteins/metabolism , Precision Medicine/methods , Tissue Engineering/methods , Tumor Microenvironment/drug effects , Analysis of Variance , Chromatography, Liquid , DNA Mutational Analysis , Gene Expression Profiling , Humans , Machine Learning , Microscopy, Electron, Scanning , Predictive Value of Tests , Tandem Mass SpectrometryABSTRACT
The PI3K/AKT/mTOR pathway is an important signaling axis that is perturbed in majority of cancers. Biomarkers such as pS6RP, GLUT1, and tumor FDG uptake are being evaluated in patient stratification for mTOR pathway inhibitors. In the absence of a clear understanding of the underlying mechanisms in tumor signaling, the biomarker strategy for patient stratification is of limited use. Here, we show that no discernible correlation exists between FDG uptake and the corresponding Ki67, GLUT1, pS6RP expression in tumor biopsies from patients with head and neck cancer. Correlation between GLUT1 and pS6RP levels in tumors was observed but elevated pS6RP was noticed even in the absence of concomitant AKT activation, suggesting that other downstream molecules of PI3K/AKT and/or other pathways upstream of mTOR are active in these tumors. Using an ex vivo platform, we identified putative responders to rapamycin, an mTOR inhibitor in these tumors. However, rapamycin did not induce antitumor effect in the majority of tumors with activated mTOR, potentially attributable to the observation that rapamycin induces feedback activation of AKT. Accordingly, treatment of these tumors with an AKT inhibitor and rapamycin uniformly resulted in abrogation of mTOR inhibition-induced AKT activation in all tumors but failed to induce antitumor response in a subset. Phosphoproteomic profiling of tumors resistant to dual AKT/mTOR inhibitors revealed differential activation of multiple pathways involved in proliferation and survival. Collectively, our results suggest that, in addition to biomarker-based segregation, functional assessment of a patient's tumor before treatment with mTOR/AKT inhibitors may be useful for patient stratification.