Patients
A total of 136 patients who underwent PD between 2006 and 2018 at the Department of Surgery, Onomichi General Hospital, were enrolled in this study. Patients who underwent palliative surgery were excluded from the study. Clinical and pathological data and preoperative CT findings, of the enrolled patients, were collected. Patients with pancreatic ductal adenocarcinoma, papillary adenocarcinoma, extrahepatic cholangiocarcinoma, intraductal papillary-mucinous carcinoma, and chronic pancreatitis were considered eligible. The study was approved by the local institutional review board (OJH-201642), and written informed consent was obtained from all patients.
Preoperative CT analysis of body composition
Eight weeks prior to surgery, all patients underwent a preoperative evaluation using non-enhanced and enhanced multi-detector CT (Discovery CT 750 HD, GE Healthcare, Milwaukee, WI, USA; and Sensation 16, Siemens, Forchheim, Germany, respectively), with a section thickness of 2.5–5 mm. The CT slices at the third lumbar vertebra (L3) level were analyzed using Advantage Workshop 4.5 (GE Healthcare) and Virtual Place Fujin (AZE Ltd., Tokyo, Japan). The body composition assessment included calculations of the VATA, subcutaneous adipose tissue area (SATA), and skeletal muscle area. CT analysis was performed by trained investigators who were blinded to the patients’ characteristics and clinical outcomes (HY and MY). The tissue Hounsfield unit thresholds were as follows: −29 to 150 for the skeletal muscle area, −190 to −30 for the SATA, and −150 to −50 for the VATA. Skeletal muscle area was defined at the L3 level and included the psoas major and minor, paraspinal muscles (i.e., erector spinae and quadratus lumborum), and abdominal wall muscles (i.e., transversus abdominus, external and internal obliques, and rectus abdominis). The skeletal muscle area was normalized for height in meters squared (m2) to obtain the skeletal muscle index (SMI). The VATA/SMI ratio was defined as high if ≥1.4 and low if <1.4 (Figure 1). The cutoff value was estimated using receiver-operating characteristic (ROC) curve analysis.
Definition of sarcopenia and pancreatic fistula
Sarcopenia was defined as follows: in men, an SMI < 43 cm2/m2 with a BMI < 25 kg/m2 or an SMI < 53 cm2/m2 with a BMI of 25 kg/m2; and in women, a SMI of <41 cm2/m2.13 Pancreatic fistula was defined in accordance with the current ISGPF criteria,12 and only grade B and C fistulas were considered.
Surgical procedure and postoperative care
Surgery type was chosen according to tumor location; subtotal stomach-preserving PD was routinely performed. Lymph nodes near the anterior and posterior surfaces of the pancreatic head were resected en bloc (5, 6, 8a, 12b1-2, 12p, and 12c), and 14a-b nodes of the superior mesenteric artery were resected en bloc when technically possible. The pancreatic stump was sutured using two-layer sutures end to end with invagination of the jejunum. End-to-side hepaticojejunostomy was performed 5 cm distal to the pancreaticojejunostomy followed by end-to-side duodenojejunostomy. A round surgical drain was positioned near the pancreatic anastomosis, while a flat drain was positioned near the biliary anastomosis. Epidurals and non-steroid anti-inflammatory agents were routinely used during the procedure.
The patients with CR-POPF received total parenteral nutrition and octreotide (Sandostatin, Novartis, Rueil Malmaison, France), and a percutaneous drain was inserted under radiological guidance when suspicious infected fluid collection was detected on CT.
Statistical analysis
Continuous variables as median values were analyzed using the nonparametric Mann-Whitney U test. Categorical variables were compared using Fisher’s exact test. Logistic regression was used to model categorical outcomes. Variables that showed statistically significant associations in the univariate analysis were entered into a multivariate logistic regression model. Differences between the results of the comparative tests were considered significant at p values < 0.05. An FRS > 4 was defined as high based on the ROC curve analysis (Table 1). ROC curve analysis was applied to determine the cutoff values for the variables identified as independent risk factors. All statistical analyses were performed using SPSS version 22 (IBM Corp., Armonk, NY, USA).