Sarcopenia diagnosis in patients with colorectal cancer: a review of computed tomography-based assessments and emerging ways to enhance practicality.

Traditionally, cancer treatment has focused on the stages of the disease; however, recent studies have highlighted the importance of considering the overall health status of patients in the prognosis of cancer. Loss of skeletal muscle, known as sarcopenia, has been found to significantly affect outcomes in many different types of cancers, including colorectal cancer. In this review, we discuss the guidelines for diagnosing sarcopenia, with a specific focus on CT-based assessments. Many groups worldwide, including those in Europe and Asia, have introduced their own diagnostic guidelines for sarcopenia. Seemingly similar yet subtle discrepancies, particularly in the cutoff values used, limit the use of these guidelines in the general population, warranting a more universal guideline. Although CT-based measurements, such as skeletal muscle index and radiodensity, have shown promise in predicting outcomes, the lack of standardized values in these measurements hinders their universal adoption. To overcome these limitations, innovative approaches are being developed to assess changes in muscle mass trajectories and introduce new indices, such as skeletal and appendicular muscle gauges. Additionally, machine learning models have shown superior performance in predicting sarcopenic status, providing an alternative to CT-based diagnosis, particularly after surgery. CT has tremendous benefits and a significant role in visually as well as quantitatively retrieving information on patient body composition. In order to compensate for the limitation of standard cutoff value, 3-dimensional analysis of the CT, artificial intelligence-based body composition analysis, as well as machine learning algorithms for data interpretation and analysis have been proposed and are being utilized. In conclusion, despite the varying definitions of sarcopenia, CT-based measurements coupled with machine-learning models are promising for evaluating patients with cancer. Standardization efforts can improve diagnostic accuracy, reduce the reliance on CT examinations, and make sarcopenia assessments more accessible in clinical settings.