Deep learning‐based assessment of body composition and liver tumour burden for survival modelling in advanced colorectal cancer

Abstract

Background Personalized therapy planning remains a significant challenge in advanced colorectal cancer care, despite extensive research on prognostic and predictive markers. A strong correlation of sarcopenia or overall body composition and survival has been described. Here, we explore whether automated assessment of body composition and liver metastases from standard of care CT images can add to clinical parameters in personalized survival risk prognostication. Methods We retrospectively analysed clinical imaging data from 85 patients (50.6% female, mean age 58.9 SD 12.2 years) with colorectal cancer and synchronous liver metastases. Pretrained deep learning models were used to assess body composition and liver metastasis geometry from abdominal CT images before the initiation of systemic treatment. Abdominal muscle-to-bone ratio (MBR) was calculated by dividing abdominal muscle volume by abdominal bone volume. MBR was compared with body mass index (BMI), abdominal muscle volume, and abdominal muscle volume divided by height squared. Differences in overall survival based on body composition and liver metastasis parameters were compared using Kaplan-Meier survival curves. Results were correlated with clinical and biomarker data to develop a machine learning model for survival risk prognostication. ResultsThe MBR, unlike abdominal muscle volume or BMI, was significantly associated with overall survival (HR 0.39, 95% CI: 0.19-0.80, P = 0.009). The MBR (P = 0.022), liver metastasis surface area (P = 0.01) and primary tumour sidedness (P = 0.007) were independently associated with overall survival in multivariate analysis. Body composition parameters did not correlate with KRAS mutational status or primary tumour sidedness. A prediction model based on MBR, liver metastasis surface area and primary tumour sidedness achieved a concordance index of 0.69. Conclusions Automated segmentation enables to extract prognostic parameters from routine imaging data for personalized survival modelling in advanced colorectal cancer patients.

Decoding pan-cancer treatment outcomes using multimodal real-world data and explainable artificial intelligence

“…In 5,395 patients, body composition was automatically assessed from abdominal CT images taken before treatment initiation. 21,22 In total, we included 350 parameters in our analysis, consisting of different modalities and both patient- and tumor-specific parameters (see methods for a detailed description of the parameters).…”

“…In the medical domain, xAI has previously been applied to validate the model performance or assess feature importance across cohorts. 17,22,38 Few studies have made use of patient-wise xAI explanations. 17 Here, we built on xAI to contextualize complex multimodal patient data and systematically reveal the underlying mechanisms driving a patient’s disease progression.…”

Investigation of the Trajectory of Muscle and Body Mass as a Prognostic Factor in Patients With Colorectal Cancer: Longitudinal Cohort Study

“…Third, the trajectory of BMI is a critical independent prognostic factor, where it is a parameter containing longitudinal information of multiple compositions. Adiposity-related [ 17 ] and muscle-related indexes such as SMVI, muscle-to-bone ratio, or skeletal muscle radiodensity [ 16 , 18 ] are also essential prognostic factors, yet frequent assessments are limited since these parameters are products of occasional CT imaging studies that depend on cancer treatment progress, patient’s medical condition, health insurance, etc. Thus, synergizing CT-driven information and BMI data will promote the future clinical application.…”

“…Recently, the medical application of artificial intelligence (AI) using machine and deep learning algorithms has increased, enabling the resolution of medical problems with ease of use, robustness, and precision [15]. Moreover, studies have demonstrated that automated CT imaging evaluations using deep learning models are feasible for extracting complicated body composition parameters, and the prognostic effects of these measurements on mortality in CRC have been confirmed [16][17][18]. Although previous automated body profile evaluations have partially fulfilled the gap for clinical application, the evaluation of adiposity and muscularity is limited to one time point, and thus, not considered alternatively during various treatment or interventions.…”

Drug-eluting beads transcatheter arterial chemoembolization combined with systemic therapy versus systemic therapy alone as first-line treatment for unresectable colorectal liver metastases

“…Liver metastasis is a significant determinant of the prognosis for colorectal cancer (CRC) patients, often resulting in organ failure and a high mortality rate ( 1 , 2 ). Over 80% of patients diagnosed with colorectal liver metastases (CRLM) are not suitable candidates for surgical removal ( 3 ). Typically, patients with unresectable CRLM administered systemic FOLFOX (folinic acid, 5-fluorouracil, oxaliplatin) or FOLFIRI (folinic acid, 5-fluorouracil, and irinotecan) treatment, either with or without bevacizumab or cetuximab as first-line therapy ( 4 , 5 ).…”