Can Psoas muscle density predict the development of metastasis in non-metastatic adrenocortical carcinomas?: A CT-based AI-assisted automated segmentation analysis study

Abstract

Aim: Our study aimed to investigate whether artificial intelligence-based body composition analysis can predict metastasis development during follow-up in patients with non-metastatic adrenocortical carcinoma (ACC) at the time of diagnosis. Methods: Forty-five patients with non-metastatic ACC were included at the time of diagnosis. From the patients' non-contrast computed tomography (CT) scans, visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), psoas area, psoas density, total muscle area, and total muscle density were automatically measured from sections taken at the level of the inferior endplate of the L3 vertebra. Patients were followed for developing liver, lung, and lymph node metastases. The relationship between body composition and liver and lymph node metastasis development was investigated. Propensity score matching (PSM) was performed for patients with metastases. Results: Forty-five patients, 27 of whom were female, with non-metastatic ACC at the time of diagnosis, were included in the study. The mean age of the patients was 53±17.4 years. Significant differences were found between the groups that developed liver metastases and those that did not, and between the groups that developed lymph node metastases and those that did not, in terms of correct Psoas HU, left Psoas HU, PMD, Wall Muscle HU, and age (p