After approval (2014-0580) from the Institutional Review Board, 182 patients with primary PanNETs who underwent surgical resection at our institution from 1995 to 2013 were selected from the files of the Department of Pathology. Medical records were reviewed to evaluate clinical data, such as age, sex, symptoms, and follow-up data. Pathologic information, including tumor size, extension, metastases to regional lymph nodes, distant metastases, and lymphovascular and perineural invasions were carefully reviewed. Hematoxylin and eosin–stained slides were independently reviewed by three pathologists (S.-M.H., J.Y.K., and E.-M.S.). All PanNET cases were confirmed by immunohistochemical staining using neuroendocrine markers, synaptophysin, chromogranin, and/or CD56. Immunohistochemical staining for synaptophysin and chromogranin was performed in 144 and 138 cases, respectively. All 144 cases (100%) were positive for synaptophysin, and 113 of 138 (81.9%) cases were positive for chromogranin. All PanNET cases were re-classified into grades 1, 2, or 3 based on mitotic counts (per 10 high-power fields) and the Ki-67 labeling index according to the scheme of the 2010 WHO classification [8]. Tumor extension was assessed based on the T classification of the 2010 AJCC/UICC cancer staging system.

Tissue microarrarys (TMAs) were constructed using three 2-mm-diameter tumor cores from donor blocks using a manual tissue microarryer (Uni TMA Co., Ltd., Seoul, Korea). The sections of TMAs were stained using an automatic immunohistochemistry staining device (Benchmark XT, Ventana Medical System, Tucson, AZ, USA). Briefly, 5-µm-thick formaldehydefixed paraffin-embedded tissue sections were transferred onto adhesive slides and dried at 62°C for 30 minutes. Standard heat epitope retrieval was performed for 30 minutes in ethylene diamine tetraacetic acid, pH 8.0, in the autostainer. The samples were then incubated with antibodies against KIT (1:400, Dako-Cytomation, Glostrup, Denmark) and CK19 (1:100, Cell Marque, Rocklin, CA, USA). The sections were subsequently incubated with biotinylated anti-mouse immunoglobulins, peroxidase-labeled streptavidin (LSAB kit, DakoCytomation), and 3,30-diaminobenzidine. Negative control samples were processed without the primary antibody. Slides were counterstained with Harris hematoxylin. Nuclear labeling of intra-tumoral mast cells was used as an internal positive control for KIT immunohistochemical staining. Normal pancreatic acinar cells, ductal epithelial cells, and islet cells were negative for KIT staining, while mast cells in the pancreatic parenchyma were positive (Fig. 1A). Membranous immunolabeling in normal pancreatic ductal epithelial cells and centroacinar cells was used as an internal positive control for CK19 staining. Normal pancreatic islet cells and acinar cells were negative for CK19 staining (Fig. 1B). Sections demonstrating >5% positivity for KIT- and CK19-labeled tumor cells were considered to be positive, as previously described [2,13].

SPSS ver. 18.0 (SPSS, Chicago, IL, USA) was used for the statistical analysis. The overall survival time was calculated from the date of diagnosis of PanNET to that of death from any cause. The overall survival rate was calculated with the Kaplan-Meier method, and the associations between overall survival rate and clinicopathologic factors were compared using the log-rank test. Correlations between KIT and CK19 expression and other prognostic factors were analyzed using chi-square and Fisher exact tests. Possible prognostic factors associated with survival probability were calculated by the Cox’s proportional hazard regression model. A p<.05 was considered statistically significant.

Of a total of 182 PanNET patients, 44.5% (81 cases) were men, 55.5% (101 cases) were women, and the mean age was 51.4±13.10 years. The mean tumor size was 3.06±2.31 cm. Tumors were classified as functional tumors if they were associated with the distinct clinical manifestations of hormonal alterations, including hypoglycemia, sweating, or syncope; 45 cases (24.7%) were consequently classified as functional PanNETs. There were 71 head (39.0%), 7 uncinate (3.8%), 29 body (15.9%), 67 tail (36.8%), 2 head to body (1.1%), and 6 body to tail (3.3%) tumors. Using the TNM staging system, PanNETs were classified as pT1 in 74 (40.7%), pT2 in 47 (25.8%), and pT3 in 61 (33.5%) cases. Because only surgically resected PanNETs were included in the study, pT4 tumors were not included. Lymphovascular and perineural invasion was noted in 48 (26.4%) and 27 (14.8%) cases, respectively. Evaluation of the lymph nodes was possible in 99 cases (54.4%), and lymph node metastasis was present in 29 cases (15.9%). Distant metastasis was observed in 36 cases (19.8%), and liver was the most common site of these distant metastases (32 cases, 88.9%). Adequate follow-up of survival information was available in 181 patients, with a median follow-up period of 30 months (range, 1 to 142 months), and 153 patients (84.1%) were alive and 28 patients (15.3%) had died; 9 of these patients died of an unrelated cause.

KIT expression was observed in 16 of 182 PanNETs (8.8%) (Fig. 1D) and CK19 expression was noted in 97 cases (53.3%) (Fig. 1F). All 16 PanNET cases that showed KIT expression were also positive for CK19. The associations between KIT or CK19 expression and clinicopathological factors are summarized in Table 1. KIT expression was more frequently observed in larger PanNETs (p=.030), those with extrapancreatic extension (p<.001), advanced pT classification (p<.001), distant metastasis (p=.001), higher WHO grade (p=.001), and lymphovascular invasion (p=.014). There was no statistically significant correlation between KIT expression and sex, age, functional status, lymph node metastasis, expression of synaptophysin and chromogranin, or perineural invasion.

CK19 expression was more commonly observed in functioning PanNETs (p<.001), larger tumors (p=.006), extrapancreatic extension (p<.001), advanced pT classification (p<.001), lymph node metastasis (p=.009), distant metastasis (p=.004), higher WHO grade (p=.002), and lymphovascular (p=.012) and perineural (p=.019) invasion. There was no statistically significant correlation between CK19 expression and sex, age, or expression of synaptophysin and chromogranin. The overall survival rate of PanNET patients with KIT expression was significantly lower (5-year survival rate, 62%) than that of patients without KIT expression (77%, p=.011) (Fig. 2A). However, the overall survival rate of PanNET patients with CK19 expression (5-year survival rate, 74%) was not significantly different from that of patients lacking CK19 expression (79%, p=.242) (Fig. 2B).

The relationships between the survival of PanNET patients and other clinicopathologic factors are summarized in Table 2. Of these clinicopathologic factors, increased age (p<.001), extrapancreatic extension (p=.003), increased pT classification (p=.009), lymph node metastasis (p=.026), distant metastasis (p=.003), and higher WHO grade (p<.001) were correlated with worse survival. In contrast, overall survival was not correlated with sex, functional status, or lymphovascular and perineural invasion.

Multivariate analysis was performed to assess which factors remained independent predictors of survival after adjusting for factors that were found to be significant by univariate analysis. Only WHO grade remained as an independent prognostic factor (p=.001) (Table 3). Conversely, KIT expression, pT classification, and lymph node and distant metastases were not found to be independent prognostic factors in the present model.

KIT is a cytoplasmic membrane-bound receptor tyrosine kinase and a stem cell marker. Activation of KIT signaling has been found to mediate cell survival, migration, and proliferation [14]. Signaling from KIT is crucial for normal hematopoiesis, pigmentation, fertility, and gut movement. Deregulated KIT kinase activity has been observed in a number of pathological conditions, including small cell lung cancer, breast cancer, colon cancer, melanoma, gastrointestinal stromal tumors, and allergy [14]. KIT is involved in β-cell development and survival and in regulation of glucose metabolism of the endocrine pancreas via receptor phosphorylation [15-19]. Normally, KIT expression in the endocrine pancreas is present during the embryonic and early fetal periods; however, it is not expressed in the adult pancreas [15,19]. Thus, KIT expression in the endocrine pancreas is considered to be a stem cell feature. Several pancreatic neoplasms with KIT expression have been reported, including ductal adenocarcinomas, intraductal papillary mucinous neoplasms, mucinous cystic neoplasms, serous cyst adenomas, and neuroendocrine tumors, although the proportion of pancreatic neoplasms expressing KIT was small [20,21].

Several previous studies have reported KIT expression in PanNETs, with 22% to 46% of PanNETs expressing KIT [2,13,22]. In the present study, we observed KIT expression in 8.8% (16/182) of PanNET cases. This is a significantly lower percentage than reported in previous studies and might be related to ethnicity, as all previous studies were performed with Caucasians while all the cases included in the present study were Koreans. KIT expression is generally associated with worse prognosis in cancers of other organs, including colorectal, stomach, bile duct, and endometrial cancers, and malignant melanomas of the vulva [23-27]. In contrast, KIT expression in patients with neuroblastomas and hepatocellular carcinomas is an indicator of good survival [28,29]. There have been controversies with respect to the prognostic significance of KIT expression in PanNETs [2,13]. Zhang et al. [13] reported that KIT expression was an independent worse prognostic factor for PanNETs, while Han et al. [2] observed no significant survival differences based on KIT expression. PanNETs with KIT expression resulted in significantly worse survival than those without KIT expression in the present study; a finding that concurs with those of the previous study of Zhang et al. [13]. A plausible explanation for the worse prognosis in PanNET cases with KIT expression is that these PanNETs may possess stem cell features. The mechanisms of KIT expression in PanNETs are unknown. One previous study demonstrated no KIT mutations in exons 9 and 11 in any of the 21 examined PanNETs expressing KIT [13]. Plausible explanations for KIT overexpression without mutations in exons 9 and 11 include KIT mutations that occur in other exons.

CK19 is one of a 20-member cytokeratin family that encompasses the intermediate filaments of epithelial cells. In the early embryonic stage, endocrine cells in the pancreas strongly express CK19. This expression is gradually lost such that most islet cells no longer express CK19 by 41 gestational weeks, whereas all ductal and acinar cells maintain CK19 labeling [30,31].

Previous studies have demonstrated CK19 expression in 49% to 70% of PanNETs [2,9,12,13]. In the present study, CK19 expression was found in 53.3% (97/182 cases) of PanNETs, which is concordant with the results of previous studies. In previous studies, CK19-expressing PanNETs were associated with aggressive pathologic features, including increased tumor size, mitoses, lymphovascular invasion, and necrosis [9-12]. In accordance with the results of these previous studies, we also observed that PanNETs expressing CK19 were associated with larger tumor size, higher WHO grade, higher TNM stage, and lymphovascular and perineural invasions. Several previous studies have demonstrated that PanNET patients with CK19 expression had decreased survival rates [9-12]. However, in contrast to the results of previous studies, we did not observe any prognostic significance of CK19 expression in PanNET patients, although CK19 expression was related with aggressive clinical behavior in PanNET patients.

In summary, we observed that 1) subsets of PanNETs express KIT or CK19, 2) KIT and CK19 expression was associated with aggressive behavior of PanNETs, and 3) KIT expression was correlated with decreased survival of PanNET patients but was not an independent prognostic factor.