AI-driven CT assessment of erector spinae myosteatosis: muscle quality and its association with strength and frailty in older surgical patients.
Authors
Affiliations (5)
Affiliations (5)
- Department of Orthopedics and Trauma Surgery, University Medical Center Freiburg, Freiburg, Germany.
- Department of Diagnostic and Interventional Radiology, Faculty of Medicine, Medical Center - University of Freiburg, University of Freiburg, Freiburg, Germany.
- Department of Neurology and Neuroscience, Faculty of Medicine, Medical Center, University of Freiburg, Freiburg, Germany.
- Division of Medical Physics, Department of Diagnostic and Interventional Radiology, Faculty of Medicine, University Medical Center Freiburg, University of Freiburg, Freiburg, Germany.
- Department of Orthopedic Surgery and Traumatology, Odense University Hospital, Odense, Denmark.
Abstract
Frailty and sarcopenia are overlapping syndromes that increase adverse outcomes in older adults. Computed tomography (CT) is a reference standard for body composition analysis, but manual segmentation is time-consuming. This exploratory study used a fully automated deep learning model (DLM) to investigate associations between CT-derived muscle quality and quantity and clinical frailty. We performed a retrospective, cross-sectional secondary analysis of 53 geriatric surgical patients (mean age 77.4 ± 4.4 years). A DLM segmented skeletal muscle (SM), subcutaneous and visceral adipose tissue at the third lumbar vertebral level. Intermuscular adipose tissue (IMAT) was quantified exclusively within the erector spinae muscle and is therefore reported as a surrogate marker of trunk muscle quality rather than whole-muscle myosteatosis. Frailty was assessed with the Clinical Frailty Scale (CFS), handgrip strength and the Timed Up and Go (TUG) test. Because functional testing was incomplete, effective sample sizes varied (<i>n</i> = 21-53) and are reported for every model. In the total cohort, a higher IMAT Index was associated with lower handgrip strength (<i>ρ</i> = -0.434, <i>p</i> = 0.004, <i>n</i> = 43) and remained independent after adjustment for age, sex, BMI and ASA status (B = -1.808, 95% CI -2.994 to -0.622, <i>p</i> = 0.004, <i>n</i> = 42). Its association with the CFS was borderline univariately (<i>ρ</i> = 0.273, <i>p</i> = 0.051, <i>n</i> = 52) and absent after adjustment (<i>p</i> = 0.184, <i>n</i> = 51). Neither index was independently associated with TUG performance (<i>p</i> = 0.134 and <i>p</i> = 0.071, <i>n</i> = 25). In an exploratory sex-stratified analysis, the IMAT Index correlated with the CFS in women (<i>ρ</i> = 0.727, <i>p</i> < 0.001, <i>n</i> = 19) but not in men. Because the female subgroup was small (<i>n</i> = 20), no sex-stratified regression or interaction test was performed. In this small, retrospective, cross-sectional cohort, an erector-spinae-based IMAT Index was associated with handgrip strength but not independently with global clinical frailty. Automated opportunistic CT analysis may therefore identify specific muscle weakness rather than global frailty. The observed sex differences are hypothesis-generating and require confirmation in larger, adequately powered longitudinal studies.