• Remi Matsuyama, Keisuke Maeda, Yosuke Yamanaka, Yuria Ishida, Tomoyuki Nonogaki, Ryoko Kato, Akio Shimizu, Junko Ueshima, Yoshiaki Kazaoka, Tomio Hayashi, Kunihiro Ito, Akifumi Furuhashi, Takayuki Ono, and Naoharu Mori.
• Department of Oral and Maxillofacial Surgery, Graduate School of Medicine, Aichi Medical University, Nagakute, Japan.
• Nutrition. 2022 Jan 1; 93: 111475.

ObjectivesPeople with cancer have a high risk of cachexia and sarcopenia, which are associated with worse clinical outcomes. We evaluated the prediction accuracy of the Matsuyama et al. and Ishida et al. formulas using computed tomography (CT) slices from the twelfth thoracic vertebra (Th12) level in people with cancer.MethodsThis retrospective study included patients with advanced cancer who underwent thoracic and abdominal CT scans (n = 173). The cross-sectional area (CSA) on CT images was measured at the levels of Th12 and the third lumbar vertebra (L3). The Matsuyama et al. formula used the Th12 CSA, whereas the Ishida et al. formula used only the Th12 CSA of the spinal erectors; thus, the measurements were performed separately. The correlation between predicted and actual L3 CSA was assessed using r and the intraclass correlation coefficient. A prediction-accuracy analysis of the predicted values was also performed.ResultsThe mean participant age was 66.2 ± 12.8 y; 50.3% of participants were women and 49.7% were men. Strong correlations were observed between the predicted and measured L3 values calculated from the two prediction formulas. The prediction-accuracy analysis using previously reported cutoff values showed that the Ishida et al. method had high sensitivity and the Matsuyama et al. method had high specificity for low skeletal muscle index determined by the predicted and measured L3 skeletal muscle index.ConclusionsBoth the Matsuyama et al. and Ishida et al. formulas had good reliability on CT slices at the Th12 level in people with advanced cancer, indicating that these formulas can be applied in clinical practice.Copyright © 2021 Elsevier Inc. All rights reserved.

18 keywords...

hide…