Gene-Mutation-Based Algorithm for Prediction of Treatment Response in Colorectal Cancer Patients

Abstract

Purpose: Despite the high mortality of metastatic colorectal cancer (mCRC), no new biomarker tools are available for predicting treatment response. We developed gene-mutation-based algorithms as a biomarker classifier to predict treatment response with better precision than the current predictive factors. Methods: Random forest machine learning (ML) was applied to identify the candidate algorithms using the MSK Cohort (n = 471) as a training set and validated in the TCGA Cohort (n = 221). Logistic regression, progression-free survival (PFS), and univariate/multivariate Cox proportional hazard analyses were performed and the performance of the candidate algorithms was compared with the established risk parameters. Results: A novel 7-Gene Algorithm based on mutation profiles of seven KRAS-associated genes was identified. The algorithm was able to distinguish non-progressed (responder) vs. progressed (non-responder) patients with AUC of 0.97 and had predictive power for PFS with a hazard ratio (HR) of 16.9 (p < 0.001) in the MSK cohort. The predictive power of this algorithm for PFS was more pronounced in mCRC (HR = 16.9, p < 0.001, n = 388). Similarly, in the TCGA validation cohort, the algorithm had AUC of 0.98 and a significant predictive power for PFS (p < 0.001). Conclusion: The novel 7-Gene Algorithm can be further developed as a biomarker model for prediction of treatment response in mCRC patients to improve personalized therapies.

Artificial Intelligence Predictive Models of Response to Cytotoxic Chemotherapy Alone or Combined to Targeted Therapy for Metastatic Colorectal Cancer Patients: A Systematic Review and Meta-Analysis

“…Other articles were collected by checking reference lists of identified studies, reviews and systematic reviews. The final database consisted of 26 studies with 50,257 patients eligible for inclusion in the database [ 74 , 75 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 86 , 87 , 88 , 89 , 90 , 91 , 92 , 93 , 94 , 95 , 96 , 97 , 98 , 99 , 100 , 101 , 102 ]. Table 2 and Table 3 show the 26 original AI studies that were included in the systematic review.…”

“…When data quality was assessed in the 26 AI study included in the systematic review and the meta-analysis [ 74 , 75 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 86 , 87 , 88 , 89 , 90 , 91 , 92 , 93 , 94 , 95 , 96 , 97 , 98 , 99 , 100 , 101 , 102 ], we identified a common gap that consisted in the lack of an external validation cohort, that was missing in more than the half of the investigations ( Table 4 and Supplementary Table S1 ). Outliers, missing values and C.I.…”

“…In the following paragraphs, we provide a perspective of the progress in computer science by reviewing the roles that learning techniques have played in cancer research with predictive algorithms of response to chemotherapy alone or combined with targeted therapy as well as for predicting individual toxicity for mCRC patients. As shown in Table 2 and Table 3 , 26 studies were conducted to develop and validate different predictive algorithms for mCRC patients [ 74 , 75 , 79 , 80 , 81 , 82 , 83 , 84 , 85 , 86 , 87 , 88 , 89 , 90 , 91 , 92 , 93 , 94 , 95 , 96 , 97 , 98 , 99 , 100 , 101 , 102 ]. In the AI investigations, cohorts were generally divided into a training and a validation set to train and validate the developed models to avoid overfitting, with the exception of three studies [ 79 , 92 , 93 ], that did not split the data.…”

“…Johnson et al [ 100 ] have investigated by a ML approach with gene-mutation data the response to standard chemotherapy with or without target therapy in mCRC patients from the MSK cohort [ 124 ]. In this study, mutation profiles fed as inputs into a RF model, produced a final model containing a classifier based on mutation profiles of KRAS , BRAF , ERBB2 , MAP2K1 , TSC2 , TP53 , and APC .…”

“…An improved knowledge of the mechanisms involved in colorectal carcinogenesis and tumor progression has allowed the development of biological agents with targeted actions, as inhibitors of angiogenesis, EGFR-targeted therapy, and immunotherapy [ 5 ]. As shown in Table 3 , ML models [ 74 , 87 , 88 , 89 , 90 , 91 , 94 , 97 , 99 , 100 , 102 ] were mainly utilized to analyze the response to targeted therapy rather than DL networks [ 85 , 86 , 98 ]. Findings show that the prognostic and predictive values of radiomics and molecular biomarkers in predicting the benefit of targeted therapy in CRC metastases have been extensively analyzed, with seven radiomic studies demonstrating the utility of diagnostic imaging in predicting clinical outcomes [ 74 , 85 , 86 , 87 , 88 , 91 , 98 ] and with seven molecular biomarkers based investigations that developed successful predictive molecular classifiers [ 89 , 90 , 94 , 97 , 99 , 100 , 102 ].…”