This study is retrospective study. It was approved by the Human Ethics Review Board of the University of Miyazaki on 2018/11/21 (reference number O-0426), and it conforms to the provisions of the Declaration of Helsinki. Informed consent for data collection was obtained from the patients using the opt-out procedure. Patient data were retrieved from the departmental database. Between April 2003 and December 2011, 195 patients underwent PD for hepatobiliary pancreatic disease at the Division of Hepato-Biliary-Pancreatic Surgery, Department of Surgery, University of Miyazaki Faculty of Medicine in Japan. Plasma ghrelin was measured in 83 of these patients. Among them, 14 patients had benign disease (tumors and chronic inflammatory disease), and 69 had malignant tumors. Short-term outcomes were evaluated in all 83 patients, and long-term outcomes were evaluated in 69 patients after excluding those with benign disease. To assess patient condition, we used the following indices to respectively score combined immunonutrition and inflammation status: The prognostic nutritional index (PNI) and the high-sensitivity modified Glasgow Prognostic Score (HS-mGPS) (15–18). The following factors were examined: Age, sex, body mass index (BMI), diabetes mellitus, preoperative biliary drainage, nutritional score (PNI), inflammation-based prognostic score (HS-mGPS), serum tumor markers [carcinoembryonic antigen (CEA), carbohydrate antigen 19-9], tumor location, tumor type, tumor size, pathological status based on the international Union Against Cancer (UICC) tumor-node-metastasis (TNM) classification (8th edition) (19), surgical procedure (SPD, PPPD, SSPPD), portal vein resection, operation time, blood loss, severe complications, postoperative hospital stay, adjuvant chemotherapy, plasma ghrelin level and DG-to-AG ratio (D/A). We used the median for the basic values for the analysis and the cut-off values derived from receiver operating characteristic (ROC) analysis for the tumor markers, plasma ghrelin and D/A. The PNI was calculated as 10 × albumin (g/dl) + 0.005 × lymphocyte count (/µl). The HS-mGPS was determined based on C-reactive protein (CRP) and albumin levels. The scoring system was as follows: (1) Patients with a normal CRP value (<0.3 mg/dl) were allocated a score of 0, regardless of the albumin level; (2) patients with a CRP level ≥0.3 mg/dl combined with an albumin level ≥3.5 g/dl were allocated a score of 1; and (3) patients with a CRP ≥0.3 mg/dl and an albumin <3.5 g/dl were allocated a score of 2. Postoperative complications were classified using the Clavien-Dindo grading system (20), where grade III or higher represents morbidity.

Blood samples were collected in a tube containing aprotinin and ethylenediaminetetraacetic acid (Wako, Osaka, Japan) before breakfast after an overnight fast. Plasma total ghrelin (TG) concentrations were calculated as the AG concentration plus the DG concentration. Blood sampling points were preoperatively, 2 weeks, and 1, 6 and 12 months after surgery.

The same surgical team performed all operations and oversaw the perioperative management. The operative technique was essentially similar to that described previously (21). The organs resected during PD included the gallbladder, common hepatic duct, head of the pancreas, duodenum except for a 3–4 cm portion nearby the bulb and approximately 10 cm of the proximal jejunum. The pancreas was divided with a scalpel. The duodenum was transected approximately 3–4 cm distal to the pyloric ring for PPPD, and the stomach was transected approximately 2–3 cm proximal to the pyloric ring at the antrum in SSPPD. We performed pancreatojejunostomy by the duct-to-mucosa method with eight interrupted sutures of 5-0 PDS-II (polydioxanone; Johnson & Johnson Co.) and the modified Kakita method with 3-0 Prolene (polypropylene; Johnson & Johnson Co.) (22). An end-to-side hepaticojejunostomy was then performed 5–10 cm distal from the pancreatojejunostomy. Finally, end-to-side duodeno- or gastrojejunostomy was performed to place the stomach and duodenum into a straight vertical line as described for the retrocolic route or the antecolic route, which was decided by randomized controlled trial (23). A Braun anastomosis was then created. Closed drains were inserted behind the hepaticojejunostomy and at the upper side of the pancreatojejunostomy. All patients received prophylactic antibiotics for 3 days and an H₂ blocker or proton pump inhibitor to prevent stress peptic ulcer. None of the patients received octreotide.

In 2016, the Japan Adjuvant Study Group of Pancreatic Cancer (JASPAC) reported the results of a phase III study of S-1 vs. gemcitabine (GEM) for resected pancreatic cancer. S-1 significantly prolonged overall survival (OS) and relapse-free survival, and S-1 therapy is currently recommended (24). At the time of the present study, there was no evidence on adjuvant therapy for pancreatic cancer, and various methods such as S-1, GEM or S-1+GEM were selected. Evidence for adjuvant chemotherapy for biliary carcinoma and ampullary carcinoma was still lacking at the time GEM or GEM+cisplatin were chosen.

Clinical parameters are expressed as the number (%) or median value and range. Comparisons between groups were made using the χ² test for categorical variables and the Mann-Whitney U test for continuous variables. The difference between plasma ghrelin concentrations at different time-points was tested for significance using a Friedman test followed by a post hoc Bonferroni correction. Survival was analyzed using the Kaplan-Meier method, and the log-rank test was used to determine the differences between the two groups. The cutoff for definition of a subgroup was determined by a ROC analysis. A univariate analysis was performed to assess significant differences in clinicopathological characteristics. The independent risk factors affecting severe complications and prognostic factors were determined by multivariate analysis using a Cox proportional hazards model for variables that were significant following univariate analysis. The associated 95% confidence interval (CI) was subsequently determined. P<0.05 was considered statistically significant. Statistical analyses were performed using JMP 11^® (SAS Institute Inc., Cary, NC, USA).

PD was performed in 83 patients (49 men, 34 women) with a median age of 70 (range 36–85) years: 63 underwent PPPD, 11 SSPPD and 9 SPD for diagnoses that included pancreatic disease (n=47), bile duct disease (n=24) and ampullary tumor (n=12). The clinicopathological characteristics of the SPD group (n=9) and Non-SPD group (n=74) are summarized in Table I. Of the 83 patients, 69 had a malignant tumor (pancreatic carcinoma: 36, bile duct carcinoma: 22, ampullary carcinoma: 11 patients). Thirty-three patients (47.8%) had diabetes mellitus, and 44 (63.8%) required biliary drainage for preoperative jaundice and cholangitis. The progression of cancer by TNM classification tended toward advanced cases (Stage I: 9, II: 25, III: 35 patients). The clinicopathological characteristics of the pancreatic carcinoma group (PCG, n=36) and the non-pancreatic carcinoma group (Non-PCG, n=33) are summarized in Table II. The PCG included many patients with deeper invasive lesions (P=0.0156) and portal vein resection (38.9% vs. 0.0%, P<0.0001) and adjuvant chemotherapy (86.1% vs. 63.6%, P=0.0304). In the Non-PCG, more patients required preoperative biliary drainage than in the PCG (78.8% vs. 50.0%, P=0.0130).

Fig. 1 shows the chronological change of plasma TG and D/A after PD. The TG in patients who underwent SPD was significantly lower than that in those who underwent PPPD or SSPPD (Fig. 1A). In SPD patients, TG had declined significantly at 2 weeks after surgery (mean, preoperative 32.8 vs. 2 weeks postoperatively 15.9 fmol/ml, P=0.0388). There were no significant differences at other times points. Table I shows the significant difference between the SPD group and the Non-SPD group in AG and TG, but shows no difference in the D/A. The perioperative D/A declined significantly at 6 months after surgery in patients with SPD (mean; preoperative, 10.6 vs. 6 months postoperatively, 4.5; P=0.0234), but there were no significant differences at other time points (Fig. 1B). In all 83 patients, the median preoperative values of AG, DG, TG and D/A were 6.49, 40.86, 47.33 fmol/ml and 4.92. The cutoff values for morbidity of AG, DG, TG and D/A calculated by ROC analysis were 10.6, 41.6, 47.3 fmol/ml and 3.1. The median PNI was 43.7, and 55 patients (66.3%) had a HS-mGPS score of 0. The plasma ghrelin levels of the high PNI group (PNI ≥40) were significantly lower than those of the low PNI group (PNI <40) and showed a negative association. The D/A of the high PNI group (PNI ≥40) tended to be higher than low PNI group, but there was no significant difference (Fig. 2A). The TG level in the patients with a HS-mGPS score of 1 was significantly lower than that of those with a score of 2 (score 1: 26.9 vs. score 2: 78.8 fmol/ml, P=0.0451). The D/A in the patients with a HS-mGPS score of 0 tended to be higher than other score group, but there was no significant difference (Fig. 2B).

In the 83 patients, the median operation time and blood loss were 607 min and 1520 ml, respectively. The morbidity rate was 31.3%, and the severe complications were pancreatic fistula (14.5%), intra-abdominal abscess (15.7%), intra-abdominal bleeding (6.0%) and liver abscess (1.2%). As shown in Table I, the Non-PCG had more severe complications than the PCG (54.6% vs. 5.6%, P<0.0001).

Clinical factors affecting severe complications were evaluated by univariate and multivariate analyses (Table III). A univariate analysis showed that sex [male, hazard ratio (HR): 3.22, P=0.0291], location of the disease (not pancreas, HR: 8.54, P<0.0001) and D/A (<3.1, HR: 6.23, P=0.0018) were significant variables influencing unfavorable morbidity. Among them, location of the disease (not pancreas, HR: 9.64, P=0.0005) and D/A (<3.1, HR: 5.34, P=0.0292) remained as independent risk factors for severe complications in a subsequent multivariate analysis.

Long-term results were examined in the 69 patients with malignant tumor. The rate of 5-year OS in all patients was 41.5%. The OS rates of the PCG and Non-PCG were 40.7 and 42.4%, respectively, and the survival rate between the two groups was not significantly different (P=0.8773; Fig. 3A). The survival rate between the SPD group and PPPD/SSPPD group was also not significantly different (44.4% vs. 41.0%, P=0.8199; Fig. 3B).

Factors associated with OS were evaluated by univariate and multivariate analyses (Table IV). A univariate analysis showed that diabetes mellitus (Yes, P=0.0001), CEA (≥4.6 ng/ml, P=0.0301), pathological depth of invasion (T4, P=0.0043), performed operation (PPPD, P=0.0313), blood loss (≥1565 g, P=0.0103), postoperative hospital stay (≥36 days, P=0.0144) and D/A (<3.1, P=0.0012) were significant variables influencing unfavorable OS. Among them, diabetes mellitus (Yes, HR: 3.47, P=0.0014), postoperative hospital stay (≥36 days, HR: 2.28, P=0.0371) and D/A (<3.1, HR: 4.73, P=0.0010) remained as independent risk factors for OS in a subsequent multivariate analysis. The rates of OS of the low D/A group (D/A <3.1) and high D/A group (D/A ≥3.1) were 15.4 and 47.6%, respectively (P=0.0134). In the patients of the PCG and Non-PCG, the rates of OS between the low D/A group and high D/A group were 0.0% vs. 50.8% (P=0.0040) and 11.1% vs. 50.0% (P=0.0247), respectively, and were significantly different (Fig. 4A and B).