A machine learning model for colorectal liver metastasis post-hepatectomy prognostications

Abstract

Background: Currently, surgical resection is the mainstay for colorectal liver metastases (CRLM) management and the only potentially curative treatment modality. Prognostication tools can support patient selection for surgical resection to maximize therapeutic benefit. This study aimed to develop a survival prediction model using machine learning based on a multicenter patient sample in Hong Kong.Methods: Patients who underwent hepatectomy for CRLM between 1 January 2009 and 31 December 2018 in four hospitals in Hong Kong were included in the study. Survival analysis was performed using Cox proportional hazards (CPH). A stepwise selection on Cox multivariable models with LASSO regression was applied to a multiply-imputed dataset to build a prediction model. The model was validated in the validation set, and its performance was compared with that of Fong Clinical Risk Score (CRS) using concordance index.Results: A total of 572 patients were included with a median follow-up of 3.6 years. The full models for overall survival (OS) and recurrence-free survival (RFS) consist of the same 8 established and novel variables, namely colorectal cancer nodal stage, CRLM neoadjuvant treatment, Charlson Comorbidity Score, prehepatectomy bilirubin and CEA levels, CRLM largest tumor diameter, extrahepatic metastasis detected on PET-scan as well as KRAS status. Our CRLM Machine-learning Algorithm Prognostication model (CMAP) demonstrated better ability to predict OS (C-index =0.651), compared with the Fong CRS for 1-year (C-index =0.571) and 5-year OS (C-index =0.574). It also achieved a C-index of 0.651 for RFS. Conclusions:We present a promising machine learning algorithm to individualize prognostications for patients following resection of CRLM with good discriminative ability.

Clinical Prediction Models for Prognosis of Colorectal Liver Metastases: A Comprehensive Review of Regression-Based and Machine Learning Models

“…The most commonly used are patient age (14 studies), gender (5 studies), American Society of Anesthesiologists (ASA) score and comorbidities (each in 3 studies). Comorbidities were defined using the Charlson comorbidity score in two of three studies [23,83]. Body mass index (BMI), marital status and no prior liver surgery were also reported as variables (each in one study).…”

Immune Escape and Metabolic Reprogramming in Colon Cancer: Insights from Endocytosis-Related Genes

“…In this study, we utilized the ''glmnet'' R package to analyze the expression values of 759 prognostic genes as features. We then condensed these prognostic genes into more important feature genes employing the root mean square error as the cost function [29].…”

Necessity of integrating statistical inference and clinical practice in developing clinical models: bias introduced by neoadjuvant chemotherapy

“…We read the article by Dr. Lam et al ( 1 ). This article is a multicenter study involving Cox multivariable models with the least absolute shrinkage and selection operator (LASSO) to construct a survival prediction model for colorectal liver metastasis (CRLM) patients after hepatectomy.…”