CT Derived Measurement of Body Composition: Observations from a Comparative Analysis of Patients with Colorectal and Lung Cancer

Abstract

Background CT-derived measures of body composition have been shown to have prognostic value in patients with cancer. However, few studies have compared these observations across tumor types and stages of disease. The aim of the present study was to compare body composition measures between two types of cancers, i.e. colorectal cancer (CRC), which is less inflammatory and patients maintain body composition over a longitudinal study period, whereas lung cancer (LC) is proinflammatory and patients lose more fat and muscle mass using a standard methodology. Methods Clinicopathological characteristics, including those pertaining to nutritional risk/status and systemic inflammation in patients with colorectal cancer (CRC, n = 1047) and lung cancer (LC, n = 662), were compared. The CT image at L3 was used to assess body composition. Comparison of these cohorts was carried out using the chi-square test. Binary logistic regression analysis was performed to assess the impact of clinico-pathological variables on body composition, and scatter plots were used to examine the relationship between body mass index (BMI) and CT-derived measures of body composition. Results According to CT-derived body composition, high subcutaneous (SFI) and visceral fat index (VFI) were common (>70%) in both CRC and LC. Also, low skeletal muscle index (SMI) and density (SMD) were approximately 40–50% and 60–70% in both CRC and LC. Compared with CRC, patients with LC had a higher American Society of Anaesthesia (ASA) (P < 0.001), Malnutrition Universal Screening Tool (MUST) (P < 0.001), modified frailty index (mFI) (P < 0.001), modified Glasgow Prognostic Score (mGPS) (P < 0.001), and neutrophil lymphocyte ratio (NLR) (P < 0.001) scores. On binary logistic regression analysis, MUST, mFI, and NLR were predictors of subcutaneous adiposity (P < 0.05); type of cancer, MUST, and mFI were predictors of visceral obesity (P < 0.001); age, type of cancer, MUST, and mGPS were predictors of low SMI (P < 0.001); and age, type of cancer, mFI, and mGPS were predictors of low SMD (P < 0.05). There was a similar relationship between BMI and other measures of CT-derived body composition across two types of cancers. Conclusion Obesity and low skeletal muscle mass were common in both CRC and LC cohorts despite large differences in comorbidity, nutritional risk, systemic inflammation, and survival, even when normalized for TNM stage. These observations would support the hypothesis that, although prognostic, CT derived body composition analysis primarily reflects patient constitution rather than the effect of tumor stage in patients with cancer. The systemic inflammatory response, as evidenced by mGPS, can be considered as an important therapeutic target and loss of muscle mass in patients with advanced cancer is related to the systemic inflammatory response.