Optimal policy tree to assist in adjuvant therapy decision-making after resection of colorectal liver metastases

Artificial intelligence in colorectal cancer liver metastases: From classification to precision medicine

Colorectal cancer liver metastasis (CRLM) remains the leading cause of mortality among colorectal cancer (CRC) patients, with more than half eventually developing hepatic metastases. Achieving long-term survival in CRLM necessitates early detection, robust stratification, and precision treatment tailored to individual classifications. These processes encompass critical aspects such as tumor staging, predictive modeling of therapeutic responses, and risk stratification for survival outcomes. The rapid evolution of artificial intelligence (AI) has ushered in unprecedented opportunities to address these challenges, offering transformative potential for clinical oncology. This review summarizes the current methodologies for CRLM grading and classification, alongside a detailed discussion of the machine learning models commonly used in oncology and AI-driven applications. It also highlights recent advances in using AI to refine CRLM subtyping and precision medicine approaches, underscoring the indispensable role of interdisciplinary collaboration between clinical oncology and the computational sciences in driving innovation and improving patient outcomes in metastatic colorectal cancer.

Recurrence Timing and Risk Following Curative Resection of Colorectal Liver Metastases: Insights From a Hazard Function Analysis

INTRODUCTION

There is no consensus on the optimal surveillance interval for patients undergoing resection of colorectal liver metastases (CRLM). We sought to assess the timing and intensity of recurrence following curative-intent resection of CRLM utilizing a recurrence-free survival (RFS) hazard function analysis.

METHODS

Patients with CRLM who underwent curative-intent resection were identified from a multi-institutional database. The RFS hazard function was used to plot hazard rates and identify the peak of recurrence over time.

RESULTS

Among 1804 patients, the median RFS was 19.9 months. In the analytic cohort, the RFS hazard curve peaked at 5.9 months (peak hazard rate: 0.054) and gradually declined, indicative of early recurrence. In subgroup analyses, patients with high and medium tumor burden scores (TBS) had RFS hazard peaks at 4.9 months (peak hazard rate: 0.060) and 5.8 months (peak hazard rate: 0.054), respectively. In contrast, patients with low TBS had a later peak at 7.5 months, with the lowest peak hazard rate of 0.047.

CONCLUSIONS

The recurrence peak for CRLM patients occurred approximately 6 months postsurgery, highlighting the need for intensified early postoperative surveillance. Patients with high TBS experienced earlier recurrence, underscoring the importance of close monitoring, particularly during the first 6 months after surgery.

Artificial intelligence-powered clinical decision making within gastrointestinal surgery: A systematic review

BACKGROUND

Clinical decision-making in gastrointestinal surgery is complex due to the unpredictability of tumoral behavior and postoperative complications. Artificial intelligence (AI) could aid in clinical decision-making by predicting these surgical outcomes. The current status of AI-based clinical decision-making within gastrointestinal surgery is unknown in recent literature. This review aims to provide an overview of AI models used for clinical decision-making within gastrointestinal surgery.

METHODS

A systematic literature search was performed in databases PubMed, EMBASE, Cochrane, and Web of Science. To be eligible for inclusion, studies needed to use AI models for clinical decision-making involving patients undergoing gastrointestinal surgery. Studies reporting on reviews, children, and study abstracts were excluded. The Probast risk of bias tool was used to evaluate the methodological quality of AI methods.

RESULTS

Out of 1073 studies, 10 articles were eligible for inclusion. AI models have been used to make clinical decisions between surgical procedures, selection of chemotherapy, selection of postoperative follow up programs, and implementation of a temporary ileostomy. Most studies have used a Random Forest or Gradient Boosting model with AUCs up to 0.97. All studies involved a retrospective study design, in which external validation was performed in one study.

CONCLUSIONS

This review shows that AI models have the potentiality to select the most optimal treatments for patients undergoing gastrointestinal surgery. Clinical benefits could be gained if AI models were used for clinical decision-making. However, prospective studies and randomized controlled trials will reveal the definitive role of AI models in clinical decision-making.

Preoperative identification of early extrahepatic recurrence after hepatectomy for colorectal liver metastases: A machine learning approach

BACKGROUND

Machine learning (ML) may provide novel insights into data patterns and improve model prediction accuracy. The current study sought to develop and validate an ML model to predict early extra-hepatic recurrence (EEHR) among patients undergoing resection of colorectal liver metastasis (CRLM).

METHODS

Patients with CRLM who underwent curative-intent resection between 2000 and 2020 were identified from an international multi-institutional database. An eXtreme gradient boosting (XGBoost) model was developed to estimate the risk of EEHR, defined as extrahepatic recurrence within 12 months after hepatectomy, using clinicopathological factors. The relative importance of factors was determined using Shapley additive explanations (SHAP) values.

RESULTS

Among 1410 patients undergoing curative-intent resection, 131 (9.3%) patients experienced EEHR. Median OS among patients with and without EEHR was 35.4 months (interquartile range [IQR] 29.9-46.7) versus 120.5 months (IQR 97.2-134.0), respectively (p < 0.001). The ML predictive model had c-index values of 0.77 (95% CI, 0.72-0.81) and 0.77 (95% CI, 0.73-0.80) in the entire dataset and the validation data set with bootstrapping resamples, respectively. The SHAP algorithm demonstrated that T and N primary tumor categories, as well as tumor burden score were the three most important predictors of EEHR. An easy-to-use risk calculator for EEHR was developed and made available online at: https://junkawashima.shinyapps.io/EEHR/.

CONCLUSIONS

An easy-to-use online calculator was developed using ML to help clinicians predict the chance of EEHR after curative-intent resection for CRLM. This tool may help clinicians in decision-making related to treatment strategies for patients with CRLM.

Preoperative and postoperative predictive models of early recurrence for colorectal liver metastases following chemotherapy and curative-intent one-stage hepatectomy

INTRODUCTION

Accurate prediction of patients at risk for early recurrence (ER) among patients with colorectal liver metastases (CRLM) following preoperative chemotherapy and hepatectomy remains limited.

METHODS

Patients with CRLM who received chemotherapy prior to undergoing curative-intent resection between 2000 and 2020 were identified from an international multi-institutional database. Multivariable Cox regression analysis was used to assess clinicopathological factors associated with ER, and an online calculator was developed and validated.

RESULTS

Among 768 patients undergoing preoperative chemotherapy and curative-intent resection, 128 (16.7 %) patients had ER. Multivariable Cox analysis demonstrated that Eastern Cooperative Oncology Group Performance status ≥1 (HR 2.09, 95%CI 1.46-2.98), rectal cancer (HR 1.95, 95%CI 1.35-2.83), lymph node metastases (HR 2.39, 95%CI 1.60-3.56), mutated Kirsten rat sarcoma oncogene status (HR 1.95, 95%CI 1.25-3.02), increase in tumor burden score during chemotherapy (HR 1.51, 95%CI 1.03-2.24), and bilateral metastases (HR 1.94, 95%CI 1.35-2.79) were independent predictors of ER in the preoperative setting. In the postoperative model, in addition to the aforementioned factors, tumor regression grade was associated with higher hazards of ER (HR 1.91, 95%CI 1.32-2.75), while receipt of adjuvant chemotherapy was associated with lower likelihood of ER (HR 0.44, 95%CI 0.30-0.63). The discriminative accuracy of the preoperative (training: c-index: 0.77, 95%CI 0.72-0.81; internal validation: c-index: 0.79, 95%CI 0.75-0.82) and postoperative (training: c-index: 0.79, 95%CI 0.75-0.83; internal validation: c-index: 0.81, 95%CI 0.77-0.84) models was favorable (https://junkawashima.shinyapps.io/CRLMfollwingchemotherapy/).

CONCLUSIONS

Patient-, tumor- and treatment-related characteristics in the preoperative and postoperative setting were utilized to develop an online, easy-to-use risk calculator for ER following resection of CRLM.

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

Colorectal liver metastasis (CRLM) is a disease entity that warrants special attention due to its high frequency and potential curability. Identification of "high-risk" patients is increasingly popular for risk stratification and personalization of the management pathway. Traditional regression-based methods have been used to derive prediction models for these patients, and lately, focus has shifted to artificial intelligence-based models, with employment of variable supervised and unsupervised techniques. Multiple endpoints, like overall survival (OS), disease-free survival (DFS) and development or recurrence of postoperative complications have all been used as outcomes in these studies. This review provides an extensive overview of available clinical prediction models focusing on the prognosis of CRLM and highlights the different predictor types incorporated in each model. An overview of the modelling strategies and the outcomes chosen is provided. Specific patient and treatment characteristics included in the models are discussed in detail. Model development and validation methods are presented and critically appraised, and model performance is assessed within a proposed framework.