- Design and participants
This study was a retrospective cohort study. Subjects were over 70 years of age and had undergone curative esophagectomy for esophageal cancer at the National Cancer Center East Hospital between September 2015 and December 2018. The inclusion criteria were as follows: 1) completion of physical function and nutritional assessments; 2) computed tomography (CT) within 2 months prior to and 4 ± 2 months after esophagectomy. The exclusion criteria were as follows: 1) readmission, death, or recurrence by the day of postoperative CT; 2) distant metastasis at esophagectomy; 3) untreated or undertreated duplicate cancer; 4) missing data. Informed consent was obtained through an opt‐out consent process due to the retrospective nature of the study. This study was approved by the research ethics committee of the National Cancer Center (2019-075) in accordance with the Declaration of Helsinki.
- Perioperative rehabilitation
Perioperative rehabilitation was performed on all subjects. Preoperative rehabilitation is a home-based intervention, consisting of respiratory training with incentive spirometry; resistance training, such as squats and heel-lift exercises; and aerobic exercise with walking. Postoperative rehabilitation was performed from the first postoperative day at an intensive care unit up until discharge for 20–40 minutes per day, including early mobilization, respiratory training, resistance training, and aerobic exercise depending on the individual’s condition.
We obtained patient characteristics from medical records, and the variables assessed were as follows: age, sex, neoadjuvant chemotherapy (NAC), preoperative C-reactive protein (≥0.5 mg/dL vs. <0.5 mg/dL) , neutrophil–lymphocyte ratio (≥3.5 vs. <3.5) , pathological UICC-TNM classification 7th edition  (pT 3/1–2, pN positive/negative, and pStage III/I–II ), presence of postoperative complications, including pneumonia and anastomotic leak (≥I according to Japan Clinical Oncology Group postoperative complications criteria in line with the Clavien-Dindo classification ), and length of hospital stay (≥median vs. <median).
Physical function and nutrition
Preoperative and postoperative physical function, including isometric quadriceps muscle strength (QS) (IsoForce GT-330, OG GIKEN, Japan) , usual gait speed of 4 meters , preoperative and postoperative nutrition (prognostic nutrition index [PNI] ), and body mass index (BMI), were collected from medical records. For isometric QS, the side with a greater muscle strength was analyzed. Physical function and nutrition were measured within 3 months before esophagectomy and at the first visit after discharge (within the first month after discharge). In the variables of physical function and nutrition, the change ratio (%) after esophagectomy was calculated as follows: (postoperative value − preoperative value) ÷ preoperative value × 100%.
Ratio change in SMI
We used SMI , which was calculated from CT images at the level of L3, as an indicator of skeletal muscle mass. CT was performed twice within 3 months prior to and 4 ± 2 months after esophagectomy. Regarding preoperative CT images, CT images after NAC were used if the patient was treated with NAC. The cross-sectional area of the Hounsfield unit (−29 to 150) at the level of L3 on axial CT images was measured in the skeletal muscle area using SliceOmatic (Imagelabo, Canada) . SMI was calculated as follows: cross-sectional skeletal muscle area ÷ (height2) [4, 5, 8, 9, 22]. The ratio change of SMI was calculated as follows: [(post-SMI − pre-SMI) ÷ pre-SMI] × 100% [4, 8, 9].
Descriptive statistics are presented as number of people and mean ± standard deviation. The difference in SMI before, compared with after, esophagectomy was analyzed with a paired t-test. With a univariate analysis, associations between the ratio change in SMI and physical function, nutrition, and patient characteristics were analyzed with simple linear regression. Multiple regression was performed using the forced entry method. The dependent variable was the ratio change in SMI. Explanatory variables were potential predictors of a ratio change in SMI with significance in a univariate analysis. Confounding variables were age, sex, preoperative SMI , and pStage . A stratified analysis for SMI was performed using these predictors. Patients were divided into two groups with a median value as a cut-off point if the predictor was a continuous variable. Differences in preoperative and postoperative SMI between the two groups were compared with one-way analysis of variance adjusted using the Bonferroni method. Then, the association of predictors with other factors was analyzed using logistic regression. Statistical significance was considered as a two-tailed p value of <0.05. All analyses were performed with SPSS version 26 (IBM Corp., Japan) for Windows. Linear regression was drawn using R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria).