Maarten van Egmond

69 The association of muscle strength with muscle mass 4 METHODS Study design and participants For this cross-sectional study, all patients with an indication for curative-intent esophagectomy at the Gastro Intestinal Oncologic Center Amsterdam (GIOCA) of the Academic Medical Center (AMC) in Amsterdam were eligible for this study between March 2012 and October 2015. Only patients, who agreed to have their muscle strength measured preoperatively were referred to the physiotherapy department and included in this study. We did not apply cut-off scores for muscle strength to be included in the study. We excluded patients for assessment of muscle strength if severe cognitive, functional or neurological impairments would make reliable assessment outcomes impossible. We measured all patients 3 months preoperatively, before neoadjuvant chemoradiation. The Medical Ethics Committee of the AMC waived the need for informed consent, because the measurements in this study were part of standard care. Measurements Clinical characteristics We prospectively recorded clinical characteristics and the presence of conventional risk factors: age, gender, height, weight, body mass index (BMI), the presence of diabetes type II, cardiovascular- and pulmonary diseases and pulmonary function. Body composition All included patients underwent computed tomography scans (CT-scans) before neoadjuvant chemoradiation. We used CT-scans meeting the following criteria to assess for the muscle surface areas: 1. The vertebral spine was entirely visible at the level of the third lumbar vertebra (L3); 2. The whole cross-section of the body was displayed on the image at the level of L3; 3. The CT-scan was displayed in the portal venous phase; 4. The quality of the CT-scan was high enough to be able to distinguish different tissues. We assessed cross-sectional muscle surface areas (cm 2 ) at the level of L3, because tissue areas in this region are significantly related to whole-body muscle mass (Figure 1). 22,23 We selected plain images and we obtained measurements of the psoas, paraspinal, transverse abdominal, internal and external oblique and rectus abdominis muscles using computer software SliceOmatic 5.0 (Tomovision, Montreal, Canada). We used Hounsfield unit thresholds of −29 to 150 to differentiate muscle from other tissues. 24 We computed cross-sectional areas (cm 2 ) for the muscle by summing tissue pixels and multiplying by the pixel surface area. We corrected cross-sectional muscle areas for height to calculate the L3 muscle index expressed in cm 2 /m 2 .