Comparative Clinical Analysis of Gastroenteropancreatic Neuroendocrine Carcinomas with Liver Metastasis and Primary Hepatic Neuroendocrine Carcinomas

Purpose The objective of this study was to analyze the clinical features and prognosis of gastroenteropancreatic (GEP) neuroendocrine carcinomas (NECs) with liver metastasis and primary hepatic neuroendocrine carcinomas (PHNECs), as these rare hepatic neuroendocrine carcinomas have not been exhaustively studied. Methods The clinical data of 47 patients with hepatic NECs were retrospectively reviewed and categorized to analyze features and prognosis. Results The 47 studied cases comprised 13 cases of primary hepatic NECs (primary group) and 34 cases of metastatic hepatic NECs (metastatic group). Male patients were slightly dominant in both groups, while no age predilection was present. PHNECs were mostly single nodules located in the right lobe of the liver. Metastatic hepatic NECs originated mostly from the pancreas and stomach without distinction of the lobes of the liver. Univariate analysis showed that the treatment protocol (radical operation or others) was correlated with the overall survival (OS; p < 0.05) in the primary group, while treatment protocol and cytokeratin 7 (CK7) were associated with OS (p < 0.05) in the metastatic group. Cox proportional hazard regression showed that radical operation was an independent prognostic factor (p < 0.05) for OS in the metastatic group. Conclusions No significant differences in the clinicopathological features between PHNECs and metastatic hepatic GEP NECs were found, but radical operation was significantly correlated with OS for both carcinomas. Radical operation is the first choice for patients who are eligible for operation.


Introduction
Neuroendocrine tumors (NETs), also known as amine precursor uptake decarboxylation (APUD) tumors, are an uncommon type of cancer originating from disseminated neuroendocrine cells. According to the World Health Organization (WHO), gastroenteropancreatic NETs are categorized into three grades G1 to G3 based on the mitotic rate and the Ki67 index (G1: <2 mitoses/10 high power field (HPF) and Ki67 index <3%; G2: 2-20 mitoses/10 HPF or Ki67 index 3-20%; and G3: >20 mitoses/10 HPF or Ki67 index >20%). NETs of the G1/G2 grade were regarded as well as differentiated. High-grade (G3) neoplasms have been regarded as synonymous with poorly differentiated NECs [1]. Outcome and treatment of NETs and NECs are strikingly different. Extrapulmonary NECs are most often found in the gastrointestinal (GI) tract [2]. The liver is the common site for metastasis, yet it is an uncommon site for the origin of carcinomas [3]. Little is known about PHNECs, and the diagnosis of PHNECs is a problem worthy of discussion. The differential diagnosis between PHNECs and metastatic hepatic GEP NECs is very important for the diagnosis of PHNECs.
Due to the rarity and similarity of PHNECs and metastatic hepatic GEP NECs, their clinical features and treatment outcomes are not well understood. In this study, we retrospectively reviewed experiences with these two carcinomas for contributing to the overall understanding and improved distinction of PHNECs and metastatic hepatic GEP NECs. Among all reviewed patients, 47 patients exhibited complete pathological and follow-up data and were thus eligible for histopathological and prognostic analysis. Ethical approval was requested and obtained from the Medical Ethics Committee of Tongji Medical College (Wuhan, China). Written informed consent was obtained from all participants.
A number of transcription factors involved in the development of neuroendocrine cells during fetal life can serve as specific histological markers to identify PHNEC. CDX2 is a good marker of midgut origin, TTF1 is expressed in a subset of lung carcinoids, and PDX1 seems to be a good marker of pancreatic origin as well as ISL1. However, these markers can only be used to estimate extrahepatic primary lesions approximately, since they can also be detected in other organizations. Therefore, after the histopathological and immunohistochemical examination, it is still necessary to combine with the comprehensive clinical examination to confirm the diagnosis. The diagnosis of all patients with PHNECs is considered after limiting the possibility of a metastatic focus from an unknown primary NEC.
indexes on the prognosis was performed by the Kaplan-Meier survival curve and the log-rank test. We used Cox proportional hazard models to assess the significance of the treatment protocol in the multivariate analysis. Values of p < 0 05 were considered statistically significant. In the primary group, 10 (76.92%) cases had a single nodule, and only 3 (23.08%) cases had multiple nodules. These carcinomas were located in the left lobe of the liver in 3 (23.08%) cases, in the right lobe of the liver in 9 (69.23%) cases, and in both lobes in 1 (7.69%) case. The mean diameter of the carcinoma in the liver was 7.95 ± 3.79 cm with a range of 4-16 cm. The primary carcinoma sites of the metastatic group were mostly the pancreas and stomach. The mean diameter of carcinoma was 4.26 ± 3.05 cm with a range of 1-15 cm. In this group, 25 (73.53%) cases had a single nodule, and 9 (26.47%) cases had multiple nodules in the liver. These nodules were located in the right lobe of the liver in 8 (23.53%) cases, in the left lobe of the liver in 10 (29.41%) cases, and in both lobes in 16 (47.06%) cases (Tables 1-3).

Patients and Clinical
The magnetic resonance imaging (MRI) scan of PHNECs showed slightly longer T1 and T2 signal masses and nodules with clear boundaries. Contrast-enhanced MRI revealed an irregular mixed appearance ( Figure 1). The MRI scan of metastatic hepatic NECs showed multiple round nodules in the primary lesions, which were similar to the signal and enhancement pattern in the liver. The contrast-enhanced CT scan of metastatic hepatic NECs showed the uneven mass of the soft tissue in the primary lesions, and the liver showed multiple sizes of nodules with abnormal enhancement ( Figure 2).

Clinical Prognosis Analysis.
In the primary group, the mean and median survival times were 12.9 and 9 months, respectively, while in the metastatic group, the mean and median survival times were 18.5 and 12.5 months, respectively. There were two recurrences in the metastatic group with disease-free survival (DFS) times of 23 and 34 months. Univariate analysis showed that the treatment protocol was correlated with the overall survival (OS; p < 0 05) in the primary group (Table 5, Figure 5). In the metastatic group, treatment protocol and CK7 were correlated with OS (p < 0 05; Table 6, Figure 6). Cox proportional hazard models demonstrated that radical operation was a good independent prognostic factor (p < 0 05) for OS (Table 7). We compared the survival of 34 cases of metastatic hepatic NECs from different primary lesions. We found differences in the overall prognosis between them (p < 0 05; Figure 7), which may be related to the other metastatic sites of the metastatic group besides the liver foci (Table 8).

Discussion
GEP NECs with liver metastasis and PHNECs are rare malignancies. Diagnosis of PHNECs is considered challenging in view of the common initial presentation of GEP NECs as metastatic liver lesion. Hepatic neuroendocrine cell may originate from intrahepatic bile duct epithelial cells, heterotopic pancreatic cells, or adrenal tissue [4,5]. PHNECs can secrete a variety of polypeptides and biogenic amines, including 5-HT, pancreatic polypeptides, gastrin, prostaglandin, and calcitonin. However, only about 5% of patients with the carcinoid syndrome have obvious biological effects. These effects are manifested as skin flushing, asthma, and diarrhea and result from the direct secretion of tumor products, degraded by liver enzymes, into the portal vein circulation, the release of neuroendocrine substances, or the presence of functional defects [6][7][8]. Clinically, symptoms of epigastric discomfort, loss of appetite, fatigue, and weight loss are often present when the tumor grows to a larger level.
No obvious carcinoid syndrome-related symptoms were found in the primary group, whereas the metastatic group was associated with the typical carcinoid syndrome. However, there were many reasons for the carcinoid syndrome in the patients, especially for the metastatic group, because it may involve the corresponding symptoms caused by the metastasis of other parts except the liver or the symptoms of other diseases in the patients. PHNECs are difficult to diagnose before operation. AFP, CEA, CA19-9, and other tumor markers have no specific diagnostic value in both groups. In this study, all 13 patients in the primary group had normal serum CEA, 1 patient (7.69%) had elevated serum AFP, 2 patients (15.38%) had elevated CA125, and 4 patients (30.77%) had elevated CA19-9. In the metastatic group, the serum AFP levels were normal in all 34 patients, 4 cases (11.76%) had elevated CA125, 9 cases (26.47%) had elevated CEA, and 10 cases (29.41%) had elevated CA19-9. Preoperative diagnosis of PHNECs can only be achieved by the exclusion of extrahepatic primary lesions using imaging. It has been reported that no particular CT/MR imaging feature is specific for PHNECs [9,10] and the results of PHNEC imaging are often mixed with those of other liver tumors, such as primary hepatocellular carcinoma and primary intrahepatic cholangiocarcinoma [11][12][13]. Other detection techniques include somatostatin receptor scintigraphy and positron emission tomographic (PET) scanning. For metastatic hepatic GEP NECs, gastroscopy, colonoscopy, endoscopic ultrasound of the pancreas, video capsule endoscopy, and balloon enteroscopy are important examination methods to evaluate for a primary source [14,15]. Immunohistochemistry has an important value to the diagnosis of NETs. CgA and Syn are generally accepted as highly sensitive immunohistochemical markers for the diagnosis of NETs [16]. It has been reported that Syn is usually positive in NECs, while CgA may be negative [17]. In our study, the positive rate of Syn in both groups was larger than 90%, while the positive rate of CgA was between 60 and 80%. It has been reported that the elevated levels of CgA correlate significantly with carcinoid heart disease, treatment of proton pump inhibitors, chronic atrophic gastritis, and impaired renal function [18,19]. CK7 is a member of the large CK family and is classified as basal type II [20]. It has been found in most epithelial cells and transitional epithelial cells. Previous studies have shown that CK7 is closely related to tumor prognosis [21]. In our study, the Kaplan-Meier survival curve showed significant differences in CK7 and the prognosis of the metastatic group (p < 0 05). The median and mean survival times for a positive   expression of CK7 were 15 and 21 months, respectively, and 2 and 8.57 months, respectively, for a negative expression of CK7. These results indicated that positive expression of CK7 is positively correlated with the prognosis of metastatic hepatic GEP NECs, suggesting that CK7 may inhibit tumor growth and that negative or low CK7 expression may predict a poor prognosis of patients, and more attention should be paid to this subset of patients. However, the exact mechanism needs to be studied further.
We found that men were slightly dominant and middleaged in both groups. However, previous reports also suggested that PHNECs are more common among middleaged women [22][23][24][25]. This discrepancy may be explained by the small number of cases in our study or an increase in the incidence of PHNECs in men. In the primary group, a single nodule was located in the right lobe of the liver, while mostly multiple nodules were located in both lobes of the liver in the metastatic group, which is in accordance with previous reports [26,27]. Surgical resection is still the first choice for the treatment of PHNECs. Zhang et al. reported that the 5-year survival rate and mean survival time in 58 cases of resected PHNECs were 80% and 148 months, respectively [28], in contrast to 33% and 54 months, respectively, for unresectable neuroendocrine tumors. In our study, four patients with resectable tumors were alive 20.75 months after treatment (range, 5-30 months), while nine patients with carcinomas that could not be surgically removed survived only for 9.44 months (range, 2-22 months), and the 5-year survival rate was 69.23% in the primary group. In the metastatic group, 17 patients with resectable carcinomas were alive 26.47 months after treatment (range, 1-53 months), while other patients with carcinomas that could not be surgically removed survived for only 10.47 months (range, 1-26 months), and the 5-year survival rate was 26.47%. The Kaplan-Meier survival curve also showed that radical surgery was an effective prognostic factor in the two groups. However, the clinical progression of the two groups may vary according to the studied cases. In addition, neuroendocrine tumors are blood-rich tumors and sensitive to ischemia. Therefore, transcatheter arterial chemoembolization (TACE) is also effective in patients who cannot undergo surgery [29,30]. Local treatment also includes radiofrequency ablation, and chemotherapy is available for patients with distant metastasis. No other protocols for nonradical surgery were summarized and compared to our values in this study. The

Conclusions
Liver NEC is an extremely rare tumor, and no specific clinical features of the disease are reported. Primary hepatic neuroendocrine carcinoma should be considered when no hepatitis or cirrhosis has been diagnosed, AFP is not high, imaging findings suggest solid occupying lesions, liquefaction, and  Figure 6: Prognostic values of cytokeratin 7 and radical operation in gastroenteropancreatic neuroendocrine carcinomas with liver metastasis. (a) The survival curve shows that the total survival rate of the cytokeratin 7-positive group is higher than that of the cytokeratin 7-negative group (p < 0 05). (b) The survival curve shows that the total survival rate of the radical operation group is higher than that of the other group (p < 0 05).  The survival curve shows that the total survival rate of patients with the gallbladder as the primary tumor site is higher than that of the stomach and pancreas, apart from the rectum and cecum (p < 0 05).

Data Availability
We cannot share the data privately because we have not got the publishing license of the data original owner.