Automated three-dimensional body composition analysis identifies visceral adipose tissue radiodensity as a predictor of mortality and recurrence in colorectal cancer

Area of Science:

• Medical Imaging
• Oncology
• Artificial Intelligence

Background:

• Artificial intelligence (AI) enables advanced three-dimensional (3D) volumetric body composition (BC) analysis from computed tomography (CT) scans.
• Traditional BC analysis often relies on single L3 vertebral slices, limiting comprehensive assessment.
• This study investigates the utility of 3D BC analysis in predicting colorectal cancer (CRC) outcomes.

Purpose of the Study:

• To identify relationships between 3D body composition parameters and postoperative outcomes in colorectal cancer (CRC) patients.
• To explore the predictive value of sarcopenia, myosteatosis, visceral obesity, and adipose tissue radiodensity for CRC survival and recurrence.
• To assess the association between elevated visceral adipose tissue radiodensity and tumor characteristics.

Main Methods:

• Quantitative analysis of abdominopelvic skeletal muscle (SM), subcutaneous, and visceral adipose tissue volume and radiodensity from preoperative CT scans of 750 CRC patients.
• Size-adjusted indices for BC parameters were calculated using abdominopelvic length and height.
• Multivariate Cox regression models were employed to evaluate overall survival and disease-free survival, with sex-specific cut-offs for BC parameters.

Main Results:

• Sarcopenia, myosteatosis, visceral obesity, subcutaneous adipose tissue depletion, and elevated visceral adipose tissue radiodensity were significant predictors of poorer overall survival.
• Sarcopenia and elevated visceral adipose tissue radiodensity predicted disease recurrence.
• Elevated visceral adipose tissue radiodensity was associated with advanced T4 tumor stage and lymphovascular invasion.

Conclusions:

• Three-dimensional body composition analysis, including sarcopenia and myosteatosis, effectively predicts CRC survival outcomes, consistent with L3 slice studies.
• Elevated abdominopelvic visceral adipose tissue radiodensity emerges as a potentially novel predictor of both CRC mortality and recurrence.
• This novel predictor is linked to a more aggressive tumor phenotype, suggesting its clinical significance in CRC management.