This was a retrospective, observational study using the medical records of Keio University Hospital, Tokyo, Japan. In the present study, all patients who underwent one-stage radical resection for thoracic esophageal cancer between March 2016 and July 2019 at the Department of Surgery were included. CT was routinely performed six months after esophagectomy for follow-up, and patients in whom CT images were not available or were not taken by December 31, 2019 were excluded. This study was approved by the institutional ethics review board (20190268). The outline of the study was published on the institution’s public website, and the participants were guaranteed the right to refuse participation.
Clinical treatment and perioperative rehabilitation
Participants underwent esophagectomy with or without neoadjuvant therapy. Detailed surgical procedures and indications for neoadjuvant therapy have been presented elsewhere [20-22]. All patients hospitalized for esophagectomy were referred to the Department of Rehabilitation and received perioperative rehabilitation. Physical therapy consisted of preoperative pulmonary rehabilitation, postoperative early mobilization, walking training, and strengthening and endurance exercises. At the time of discharge, physical therapists provided instructions on home exercise programs including stretching, strengthening, and aerobic exercises. The content of speech therapy is described below.
Dysphagia rehabilitation and initiation of enteral feeding and oral intake
Speech therapy consisted of preoperative clinical swallowing assessment and postoperative non-swallowing exercises started after extubation or tracheostomy. Enteral nutrition was started early after esophagectomy by a placed jejunostomy. Anastomotic leakage was assessed using CT and esophagography at postoperative day (POD) 7. Then, the patients without anastomotic leakage received swallowing assessment using a videofluoroscopic swallowing study (VFSS) and fiberoptic endoscopic evaluation of swallowing (FEES) at POD 8 or POD 12. For patients diagnosed with anastomotic leakage, oral intake was suspended until the findings improved. Vocal fold immobility and its severity were evaluated using FEES. VFSS consisted of 3 and 5 ml of thin and thickened liquids with iopamidol. Patients were seated upright or semi-reclined at a 60-degree angle and viewed in the lateral position. Aspiration was defined as an entry of bolus below the vocal folds, equivalent to penetration-aspiration scale (PAS) 6-8, and penetration was defined as entry of bolus into the airway above/on contact with the vocal folds (PAS 2-5) . A rehabilitation physician and a speech therapist, both with at least 10 years of experience in dysphagia rehabilitation, reviewed the findings and determined by consensus whether to start oral intake.
If it was determined that oral intake was not possible, non-swallowing training was continued. With the improvement of swallowing function through the training, VFSS was performed again, and the decision about whether oral intake was possible was made. When it was decided to start oral intake, an appropriate dietary modification was planned according to the swallowing function, with most patients starting from pureed and blenderized foods on the day after VFSS. Swallowing training was performed by a speech therapist using head rotation, chin-tuck, and supraglottic swallowing maneuvers. Signs of aspiration, body and head position, eating pace, and gastrointestinal and/or respiratory symptoms were also checked during and after meals. The consistency of food was changed according to swallowing function, from pureed, to soft, to regular food , and this modification required a longer time with more severe dysphagia. The jejunostomy was removed after discharge when sufficient nutrition and hydration was established by oral intake alone.
Nutrition education was provided by a dietitian at discharge. After discharge, medical interviews including weight changes, physical examinations, imaging, and laboratory studies were routinely performed. Additional treatments such as chemotherapy and esophageal dilation of benign anastomotic strictures were provided as needed.
Primary outcome measures
Skeletal muscle mass was measured using CT images taken six months after esophagectomy for regular follow-up purposes. Cross-sectional images at the level of the third lumbar vertebra were selected, and skeletal muscle was identified using Hounsfield unit (HU) thresholds between -29 and +150, according to previous studies [25, 26]. Image analysis was performed on a GE Healthcare AW Server version 2.0 (GE Healthcare). Skeletal muscle mass was reported as SMI (cm2/m2). One of the authors, blinded to participant characteristics and interventions, performed the measurements.
The diagnosis of sarcopenia was made using the cut-off criteria of SMI <52.4 cm2/m2 for male and SMI <38.5 cm2/m2 for female patients . Skeletal muscle quality was reported as mean muscle attenuation (HU) for the entire muscle area at the third lumbar vertebra .
Secondary outcome measures
The following data were obtained for univariate and multivariate analyses:
- Participant characteristics consisted of age, sex, past medical history, height, and weight. Body mass index (BMI) was assessed from the preoperative period to the time of discharge and six months after surgery. Past medical history consisted of cerebrovascular disease, neuropsychiatric disease, head and neck surgery, and pneumonia, which could result in dysphagia and malnutrition. %VC and FEV1% were measured as indicators of preoperative pulmonary function.
- Tumor characteristics consisted of pathological tumor stage and histology, categorized as squamous cell carcinoma or adenocarcinoma. The tumor stage was classified according to the tumor-node-metastasis (TNM) classification, 7th edition . Neoadjuvant therapy consisted of chemotherapy, radiation therapy, or a combination of the two.
- Surgical factors consisted of: surgical procedures, categorized into thoracotomy versus video-assisted thoracoscopic surgery (VATS) and laparotomy versus hand-assisted laparoscopic surgery (HALS), lymph node dissection (three-field lymphadenectomy), operation time (minutes), and blood loss (mL).
- Postoperative complications included reoperation, pneumonia confirmed by chest X-ray or CT, anastomotic leakage, chylothorax, intra-abdominal abscess, empyema, and sepsis. Oral intake-related complications were defined as aspiration pneumonia that required discontinuation of oral feeding.
- Postoperative muscle strength was evaluated using handgrip strength on POD 8, measured according to the protocol recommended by AWGS 2019 .
- Nutritional status was assessed using albumin and the prognostic nutritional index (PNI)  from the preoperative period to six months after surgery.
- The inflammatory reaction was assessed using C-reactive protein (CRP) from the preoperative period to six months after surgery.
- Postoperative swallowing status was assessed using the Food Intake Level Scale (FILS)  until the time of discharge. FILS is one of the assessment tools for dysphagia in sarcopenic patients . Based on the initial swallowing assessment using VFSS/FEES, patients with FILS ≤3 were classified as the non-oral intake group, whereas the patients with FILS ≥4 were classified as the oral intake group.
A list of assessments is shown in Supplementary Table 1.
For interval scales, normality was checked using a histogram and the Shapiro-Wilk test. Normally distributed variables are expressed as means ± standard deviation, whereas non-normally distributed variables (e. g., POD, blood loss, operative time) are expressed as medians (interquartile range, IQR). Nominal and ordinal scales are expressed as numbers/percentages of patients. For comparisons between two groups, Student’s t-test or the Mann-Whitney U test was used for interval scales, and the chi-squared test or Fisher’s exact test was used for nominal and ordinal scales, split-plot analysis of variance and multiple comparisons with Bonferroni adjustment were used for time-dependent variables.
Multivariate logistic regression analysis was performed to identify perioperative risk factors during hospitalization that could be related to sarcopenia six months after esophagectomy, with the results presented as odds ratios (95% confidence interval, CI). The basic model adjusted for age and tumor stage was used, and all variables with p<0.20 were included in the following multivariable model. Because of their clinical relevance, age and tumor stage were forcibly entered into the multivariable model. Multivariate analysis was performed by the backward elimination method (likelihood ratio, elimination criteria, p>0.10). Statistical analysis was performed with SPSS version 25.0 (IBM). Values with p<0.05 were considered significant.