We conducted a retrospective study at Changhai Hospital affiliated with Navy/Second Medical University. The Institutional Review Board approved the study. Patients who underwent surgical resection from January 2013 to June 2019 for pathologically identified SPN were included in our study. The following inclusion criteria were applied: (1) patients pathologically diagnosed with SPN; (2) patients whose full electronic medical records and imaging records could be obtained; and (3) patients whose follow-up data could be obtained. Exclusion criteria included (1) specimens obtained from reresections; (2) concomitant other neoplasms on final pathology (e.g., neuroendocrine tumor, cholangiocarcinoma); and (3) patients with unavailable pathological and follow-up data. The selection procedure of the study participants is presented in Figure 1 .

The decision on surgical treatment was made by multidisciplinary hepatopancreatobiliary teams. If the lesion was diagnosed as a pancreatic cystic lesion by imaging modality, the surgical indications followed the International Consensus Guideline (18). If diagnosed with invasive cancer, the surgery indications would follow the European Society for Medical Oncology guidelines (19). The choices of surgical procedures depended on the location, degree, extent of diseases and experiences of the surgeon.

The operations for SPN were performed by experienced surgeons in Changhai hospital. Before surgery, routine preoperative examinations were performed to exclude surgical contraindications. These routine examinations included electrocardiogram, pulmonary function, coagulation function, liver function, renal function, electrolyte and hemoglobin.

After surgery, amylase analysis from drainage fluid was performed to determine whether postoperative pancreatic fistula (POPF) existed. Routine blood examinations were performed to determine whether infection existed and whether antibiotics were used. Plain CT was performed to determine whether pancreatic fluid collection existed and to detect the causes of infection. If any clinically significant complications occurred, further treatments were needed.

The parameters included in our study were composed of five parts: baseline characteristics, imaging data, pathological outcomes, short-term complications and long-term follow-up data.

The baseline characteristics included age, sex, initial symptoms, hospital stay days and surgical methods. The surgical methods included pancreatoduodenectomy (PD), total pancreatectomy (TP), distal pancreatectomy (DP), central pancreatectomy (CP) and enucleation.

The imaging data included tumor size, location, imaging diagnoses by radiologists and proportion of solid components. The tumor might be located in head/body/tail/multiple sites of the pancreas. If multiple tumors occurred, only the size of the largest tumor was measured. The proportion of solid components was evaluated by T2 sequences obtained by contrast-enhanced MRI. If the proportion of the solid component was inconsistent on different layers, only the largest proportion was measured. The imaging diagnosis was divided into 2 parts: diagnostic accuracy (the diagnostic conclusions were SPNs) and surgical indication accuracy [diagnostic conclusions were malignancy, SPNs, neuroendocrine tumor (NET)].

The pathological outcomes included margin status, peripheral tissue invasion status and immunohistochemical outcomes. Positive margin status was defined as a tumor component ≤5 mm from the incisal margin. Peripheral tissue invasion status consisted of perineural invasion, vascular invasion, cancerization of ducts, lymphatic metastasis, common bile duct invasion, peripancreatic fat invasion, spleen invasion and duodenum invasion. The immunohistochemical outcomes included the Ki-67 index, beta-catenin, lymphoid enhancer-binding factor (LEF1), cyclin D1 and vimentin, which were measured in all included cases. The Ki-67 index of the tumor was divided into 3 grades: < 3%, 3%-20% and >20% (20).

Short-term complications were adverse events that occurred within 30 days after surgery, including POPF, delayed gastric emptying, hemorrhage, abdominal infection and bile leakage. The grade of complications was based on the Clavien-Dindo score. Complications that scored Clavien Dindo grade III or greater were considered severe complications.

The long-term follow-up data included long-term complications (exocrine insufficiency, endocrine insufficiency, alimentary stricture), disease-free survival (DFS) rate and overall survival (OS) rate. Follow-up data were obtained from telephone interviews and/or outpatient interviews in this study. Endocrine insufficiency was defined as a fasting plasma glucose level > 7.0 mmol/L and/or the need for diet modification, oral medication, or insulin use to control plasma glucose levels. Exocrine insufficiency was defined as symptoms (steatorrhea or weight loss) resolving after pancreatic enzyme supplementation (21). Recurrence was defined as a local or a metastatic tumor confirmed by radiology or histology during postoperative follow-up.

The patients were divided into 2 groups according to their sex: male group and female group. The parameters were compared between the 2 groups. Quantitative parameters were expressed as the medians and range. Continuous data are reported as the mean ± standard deviation (SD) or as the median and range according to the distribution of the parameter. Categorical parameters were compared between the 2 groups using χ² or Fisher’s exact test. Kaplan–Meier survival curves were established to estimate and compare the RFS rate and OS rate between the 2 groups. A logistic regression model was established to identify the independent risk factors for tumor recurrence. Statistical analysis was performed using SPSS version 22.0 (SPSS Inc., Chicago, IL, United States), and survival curves were drawn using GraphPad Prism (version 7.00). All tests were two-sided, and a P value < 0.05 was considered statistically significant.

From January 2013 to December 2018, 221 patients who underwent pancreatic surgery were included in our study. All tumors were confirmed as SPNs according to the final histology examination. Among them, 53 patients (25.1%) were males, and 168 patients (74.9%) were females. The mean age of the overall study cohort was 33.3 years ± 12.7. Male patients were older than female patients (39.1 vs 31.6 years, t=3.283, P=0.001).

The majority of them were incidentally found (158/221, 71.5%). In the patients who were symptomatic, abdominal pain was the most common symptom (40/63, 63.5%), followed by abdominal distension (12/63, 19.0%), nausea and vomiting (8/63, 12.7%) and jaundice (3/63, 4.8%). The proportion of initial symptoms was not significantly different between males and females (χ^{2 =} 5.37, P=0.251).

In male patients, PD was performed in 13 patients (24.5%), DP was performed in 25 patients (47.2%), CP was performed in 10 patients (18.9%) and enucleation was performed in 5 patients (9.5%). No TP was performed in male patients. In female patients, PD was performed in 48 patients (28.6%), DP was performed in 82 patients (48.8%), CP was performed in 20 patients (11.9%), enucleation was performed in 8 patients (4.8%) and TP was performed in 10 patients (6.0%). The proportion of surgical methods was also not significantly different between males and females (χ^{2 =} 4.28, P=0.37).

The mean hospital stay was 12.53 days ± 6.87, with no difference between the 2 groups (14.24 vs 12.1 days, t=2.131, P=0.08).

The baseline characteristics of the study cohort are shown in Table 1 .

The tumor size for the overall study cohort was 47.98 ± 27.75 mm, with a significantly larger tumor size in female patients (50.38 vs 39.65 mm, P=0.038). In the male group, 23/5/4/21 tumors were located in multiple head/body/tail sites of the pancreas, while 67/19/19/63 tumors were located in multiple head/body/tail sites of the pancreas. The tumor location was not different between the 2 groups (χ^{2 =} 0.482, P=0.92).

All patients underwent preoperative imaging evaluations. The imaging diagnoses were correct, which means that the diagnosis was SPN, which was observed in 146 patients (66.9%). In the male group, 28 patients (54.0%) had correct imaging diagnoses. In the female group, 118 patients (70.5%) had correct imaging diagnoses. SPNs in male patients tended to be misdiagnosed with malignant tumors more often than SPNs in female patients (37.7% vs 10.7%, χ^{2 =} 20.662, P<0.0001). The diagnostic accuracy was significantly higher in the female group (χ^{2 =} 5.446, P=0.02). SPNs in male patients tended to be misdiagnosed with malignant tumors more often than SPNs in female patients (37.7% vs 10.7%, χ^{2 =} 20.662, P<0.0001).

Moreover, 49 male patients (92.5%) were diagnosed with having surgical indications (28 SPNs, 20 malignant tumors, 1 NET). Meanwhile, 147 male patients (87.5%) were diagnosed with having surgical indications (118 SPNs, 18 malignant tumors, 11 NETs). The diagnostic accuracy of having surgical indication was not significantly different between the 2 groups (χ^{2 =} 0.415, P=0.519). However, SPNs in male patients tended to be misdiagnosed with malignant tumors more often than SPNs in female patients (37.7% vs 10.7%, χ^{2 =} 20.662, P<0.0001).

Completely cystic tumors were observed in 0 and 32 (19.0%) male and female patients, respectively. Completely solid tumors were observed in 25 (47.2%) and 40 (23.8%) male and female patients, respectively. Tumors that had a solid component <50% were observed in 5 (9.4%) and 41 (24.4%) males and females, respectively. Tumors that had a solid component >50% were observed in 23 (43.4%) and 55 (32.7%) males and females, respectively. The mean solid component was 66.8% ± 24.5 for male patients and 24.7% ± 20.5 for female patients, which was significantly higher in male patients (t=23.67, P<0.0001). The imaging data of the study cohort are shown in Table 1 .

According to the pathological specimens in male patients, a positive margin status was observed in 3 patients (5.7%), and peripheral tissue invasion was observed in 4 patients (7.5%)

For immunohistochemical staining, beta catenin was expressed in 47 cases (89.2%), cyclin D1 was expressed in 50 cases (94.6%), LEF1 was expressed in 33 cases (62.3%), and vimentin was expressed in 44 cases (83.0%). Grade I Ki67 was identified in 47 tumors (88.7%), Grade II Ki67 was identified in 4 tumors (7.5%) and Grade III Ki67 was identified in 2 tumors (3.8%).

In female patients, positive margin status was observed in 14 patients (8.3%), and peripheral tissue invasion was observed in 11 patients (6.5%). For immunohistochemical staining, beta catenin was expressed in 165 cases (98.2%), cyclin D1 was expressed in 164 cases (97.6%), LEF1 was expressed in 105 cases (62.5%), and vimentin was expressed in 108 cases (64.3%). Grade I Ki67 was identified in 141 tumors (83.9%), Grade II Ki67 was identified in 25 tumors (14.9%) and Grade III Ki67 was identified in 2 tumors (1.2%).

Comparing the pathological outcomes between the 2 groups. Positive margin status and peripheral tissue invasion were not different between the 2 groups (χ^{2 =} 0.405 and 0.564, P=0.524 and 0.489). The expression of Cyclin D1 and LEF1 was not significantly different between the 2 groups (χ2 = 0.463, 0.059, P=0.541, 0.808). The Ki67 grade was not different between the 2 groups (χ2 = 1.263, P=0.532). However, the expression of beta catenin was significantly lower in male patients (χ2 = 9.377, P=0.002), and the expression of vimentin was significantly higher in male patients (χ2 = 6.584, P=0.01). The pathological outcomes of the study cohort are shown in Table 2 .

In the male group, perioperative complications occurred in 10 patients (18.9%). POPF grade II or above developed in 2 patients (2/10, 20.0%), delayed gastric emptying developed in 2 patients (2/10, 20.0%), abdominal infection developed in 4 patients (4/10, 40.0%), bleeding developed in 1 patient (1/10, 10.0%), and 1 patient (1/10, 10.0%) developed both delayed gastric emptying and abdominal infection. Three complications (3/10, 30%) scored Clavien Dindo grade III or above and were considered severe complications.

In the female group, perioperative complications occurred in 30 patients (17.9%). Five patients (5/30, 16.7%) developed severe POPF, 6 patients (6/30, 20.0%) developed delayed gastric emptying, 5 patients developed abdominal infection (5/30, 16.7%), 9 patients developed bleeding (9/30, 3%), and 5 patients (5/30, 16.7%) developed severe POPF, abdominal infection and bleeding. Ten complications (10/30, 33.3%) scored Clavien Dindo grade III or above and were considered severe complications.

The overall perioperative complication rate and severe complication rate were comparable between the groups (χ2 = 0.028 and 0.00, P=0.868 and 1.0). The short-term complications of the study cohort are presented in Table 3 .

In the male group, alimentary strictures were observed in 1 patient (1.9%). Three patients (5.7%) experienced pancreatic endocrine insufficiency, and 11 patients (20.8%) experienced exocrine insufficiency. Two patients (3.8%) experienced both pancreatic exocrine insufficiency and endocrine insufficiency. In the female group, alimentary strictures were observed in 1 patient (0.6%). Five patients (3.0%) experienced pancreatic endocrine insufficiency, and 31 patients (18.5%) experienced exocrine insufficiency. Ten patients (6.0%) experienced both pancreatic exocrine insufficiency and endocrine insufficiency. The incidence rates of long-term complications were not different between the 2 groups (χ2 = 0.002, P=0.963).

The final follow-up date was September 30, 2021. The median follow-up period was 54 months (27-105 months). Only 2 patients died due to perioperative complications (both in the female group), and the 3-, 5-, and 8-year overall survival (OS) rates were estimated to be 99.1%, 99.1%, and 99.1%, respectively. In total, 10 patients (4.5%) developed recurrence in the overall study cohort, with 7 patients (5 local recurrence and 2 liver metastasis) in the female group and 3 patients (2 local recurrence and 1 liver metastasis) in the male group. The median time to recurrence was 50 months (range 6–76 months). The 3-, 5-, and 8-year RFS rates were estimated at 98.6%, 96.2%, and 95.3% for the study cohort, respectively. The 3-, 5-, and 8-year RFS rates for the male group were 96.2%, 94.3%, and 94.3%, respectively. The 3-, 5-, and 8-year RFS rates for the female group were 99.4%, 97.0%, and 95.8%, respectively. Kaplan–Meier analysis and the log-rank test showed that the recurrence rate was not significantly different between the groups (χ2 = 1.286, P=0.206) ( Figure 2 ). The long-term follow-up data of the study cohort are presented in Table 3 .

The parameters, including age, symptoms, surgical method, tumor size, tumor location, margin status, peripheral tissue invasion and immunohistochemical staining, were included in univariate analysis ( Table 4 ). The results demonstrated that older age (42.4 vs 32.9 years, P=0.05), large tumor size (67.9 vs 47.1 mm, P=0.04) and a high KI 67 index were significantly associated with tumor recurrence, and the three parameters were included in multivariate logistic regression analysis. Moreover, sex was also included in multivariate analysis to adjust the baseline features.

Based on multivariate analysis, older age [hazard ratio (HR)= 1.094, 95% confidence interval (CI): 1.005-1.190, P=0.039] and KI 67 index grade III (HR=12.029, 95% CI: 2.399-60.311, P=0.002) were independent risk factors for tumor recurrence ( Table 5 ).