Introduction

Gastrointestinal neuroendocrine tumors (NET) arise from diffuse neuroendocrine system and account for only 0.5% of all malignancies and 2% of gastrointestinal malignancies. Gastric NETs (GNETs) are the most frequent of all digestive NETs, representing up to 23% of the cases ENT#91;¹, ²ENT#93;. The incidence and prevalence have been increasing, probably due to the wide spread use of endoscopic and radiological examinations and increased recognition by pathologists. Thus, GNETs are now often found by chance during endoscopy that was done for another reason.

In 2019, the World Health Organization (WHO) categorized these neoplasms as grade 1-3NET,neuro-endocrine carcinoma (NEC), and mixed neuroendocrine-nonneuroendocrine neoplasm based on histological classification, Ki-67 index, and mitotic activity ENT#91;³ENT#93;. Formal TNM staging classifications have been introduced for gastrointestinal NETs by the American Joint Committee on Cancer (AJCC), and the European Neuroendocrine Tumor Society (ENETS). In addition to grade and TNM, GNETs should be further subclassified into 4 subtypes according to clinicopathological characteristics since this classification has management implications. GNETs arising in the context of hypergastrinemia are classified into type I if associated with corpus atrophic gastritis/autoimmune gastritis and type II if associated with non-compensatory hypergastrinemia (produced by a gastrinoma) and Zollinger-Ellison syndrome. Types III and IV are considered sporadic, not being associated with hypergastrinemia nor any back-ground pathology. They are distinguished by histological grade, with type IV being grade 3 (mitotic index >20% and Ki-67% >20%) ENT#91;4ENT#93;. This classification is useful in predicting prognosis as well as in therapeutic decision-making.

Despite the increasing incidence of these types of lesions, it is not yet certain which factors are predictors of worse prognosis and whether these should alter the therapeutic strategy. Moreover, with the current use of advanced endoscopic techniques, these now play a role in the treatment of these patients, particularly in patients with type I GNET. However, evidence regarding the different therapeutic strategies in these patients is limited and optimal management is still not well established. The strategy adopted depends on the preference and experience of the endoscopist, while others propose a size threshold for endoscopic resection.

The aim of the study was to analyze the clinical and pathological characteristics of GNETs, identify prognostic factors associated with overall survival and metastasis, specifically in type I NETs, and evaluate the impact of serial measurements of chromogranin A (CgA) in these patients. We also assessed the clinical outcomes of patients with type I GNET treated with different strategies (polypectomy/surveillance, mucosectomy ENT#91;EMRENT#93;, endoscopic submucosal dissection ENT#91;ESDENT#93;, surgery, and somatostatin analogues).

Materials and Methods

Results

Patients

A total of 137 patients with GNET were identified in the Pathological database, of which 4 patients submitted to surgery due to carcinoma with the presence of GNET on the surgical specimen and 1 patient with less than 1 year of follow-up were excluded. We included 132 patients with GNET (type I, 113 patients; type II, 1 patient; type III, 14 patients; type IV, 2 patients; not classifiable, 2 patients), with 61% being female (n = 80) and a mean age at the diagnosis of 65.5 years (±12.2) (Table 1). Two patients were not classified since no data were available in these patients on serum gastrin, anti-parietal cell, or intrinsic factor antibodies or histology of the gastric mucosa.

At diagnosis, 52% of patients had symptoms (36 patients had dyspeptic symptoms, 18 abdominal pain, 14 symptoms related to gastroesophageal reflux and 1 diarrhea), 18% anemia, and 15% vitamin B12 deficiency. The median gastrin value was 689 pg/mL (IQR 184-1,141) and CgA 186 ng/mL (100-288). One-third of the patients (33/99) had used PPIs for more than 2 years. Antiparietal cell antibody and intrinsic factor assays were performed in 109 patients and were positive in 75 and 4 patients, respectively.

Table 1 Characteristics of NET patients 

Most lesions were solitary (79%, n = 104), with a median size of 5 mm (IQR 3-10). Thirty-five patients had a lesion larger than 10 mm. In the evaluation of gastric mucosa, 68.5% and 83.0% had atrophy of the antrum and atrophy of the body, respectively. IM of the antrum and body was present in 32.1% and 69.3%, respectively. Hp was present in 29 patients (25%).

Prognostic Factors

During the follow-up period (median 66 months), 3 patients died due to metastatic disease (1 patient with type III GNET and 2 patients with type IV GNET) and there were no deaths due to other causes. Overall, disease-specific survival was 100% for type I and type II gastric NETs, 92.9% for type III, and 0% for type IV (mean follow-up 69 months).

Male gender (p = 0.030), lesion size >10 mm (p = 0.004), type III/IV versus I/II (p < 0.001), Ki-67 ≥3%versus <3% (p = 0.025), intermediate/high grade versus low grade (p = 0.025), WHO Classification G2 or more versus G1 (p = 0.007), invasion beyond the submucosa (p < 0.001), and presence of metastases (p < 0.001) were identified as risk factors for mortality (Table 2). Age, anemia, vitamin B12 deficit, chronic use of PPIs, and presence of gastric premalignant conditions were not related to mortality. Multivariate analysis did not identify any independent factors for mortality.

Table 2 Risk factors for mortality in patients with GNETs 

Seven patients presented metastatic disease, 4 at diagnosis and 3 during follow-up (5.3%, 3 patients with type I GNET, 2 patients type III, and 2 with type IV), diagnosed with Gallium-68-Dota-NOC-PET. Four patients presented distant metastases and 3 lymphatic metastases. The most common distant metastatic site was the liver (n = 4), followed by bone (n = 2) and peritoneum (n = 1).

Lesion size >10 mm (p = 0.001), Ki-67 > 20% (p < 0.001), high grade (p < 0.001), NET G3/NEC (p = 0.003), invasion beyond submucosa (p < 0.001), and lower gastrin value (p = 0.003) were associated with metastasis on univariate analysis. Multivariable analysis revealed that Ki-67 >20% (p = 0.016) was an independent risk factor for metastasis (Table 3). In our cohort, patients with type I and III GNETs, <10 mm, and confined to the mucosa did not develop metastasis or died due to GNET.

Table 3 Risk factors for metastasis in patients with GNETs 

Evaluation of Chromogranin A

Serial measurements of CgA were performed in 73 patients. Overall, CgA showed a sensitivity for detection of recurrence (recurrence of disease requiring a change in therapeutic strategy) of 20% and a specificity of 79%. In type I GNET patients, CgA demonstrated a sensitivity of 8% and a specificity of 71%.

Management of Patients with Type I GNET Overall, 113 patients with type I GNET were included, 62% were women, with a mean age of 60 years. At diagnosis, 18% of patients had anemia and 16% vitamin B12 deficiency.

In the evaluation of gastric mucosa, 70% and 87% had atrophy of the antrum and atrophy of the body, respectively. IM of the antrum and body was present in 32% and 71%, respectively. Hp was present in 25 patients (24.8%). Antiparietal cell and intrinsic factor antibodies were present in 76% and 4% of patients with autoimmune gastritis, respectively.

At diagnosis, the surveillance/polypectomy strategy was adopted in 77 patients. Twenty-four patients under-went EMR, 6 underwent ESD, 3 underwent surgery (2 for evidence of lymph node metastases), and 3 started somatostatin analogues.

Patients undergoing EMR and ESD had no statistically significant differences in maximum lesion diameter (16 vs. 13 mm, p =0.515),Ki-67index (p = 0.745), and grade (p = 0.855). The rate of complete histological resection (78 vs. 60%, p = 0.255) and the presence of negative horizontal and vertical margins (89 vs. 100%, p = 0.597; 77 vs. 60%, p =0.470, respectively) were similar in both groups. The presence of positive horizontal or vertical margins had no impact on the development of metastases during the follow-up (p = 0.917 and p = 0.790, respectively) (Table 4).

Table 4 Outcomes of EMR and ESD in type I GNET patients 

Lesions removed by ESD had more frequently lymphovascular invasion, without significant difference (40 vs. 6%, p = 0.120). All patients had at least one surveillance endoscopy. The development of metachronous lesions and local recurrence was similar in EMR and ESD groups (38 vs. 40%, p = 0.914 and 22 vs. 20%, p = 0.932, respectively). In the ESD group, 1 patient developed lymph node metastasis (p = 0.029) during surveillance - a patient with a type I GNET, with 12 mm (Ki-67 3-20%, G2) with vertical and lympho-vascular invasion on the endoscopic dissection specimen.

Three patients were submitted to surgery. One of the patients had multiple lesions and underwent antrectomy in 2010 to reduce hypergastrinemia. The other 2 patients underwent subtotal gastrectomy since at diagnosis they presented lesions 16 and 26 mm in size with lymph node metastasis on preoperative CT. Three patients started somatostatin analogues since they had multiple gastric lesions that were considered not amenable to surveillance/endoscopic resection.

Management of Patients with Type III/IV Gastric NETs Overall, 14 patients were treated for type III GNET in our center: 5 underwent polypectomy at diagnostic endoscopy (lesions <10 mm), 5 underwent EMR, 1 under-went ESD, and 3 underwent surgery. In the group of patients submitted to polypectomy, there was no local recurrence or evidence of metastases during the follow-up. The complete resection rate after EMR was 40% and 1 patient developed lymph node metastasis (20% of the patients submitted to EMR). In 1 patient, ESD was curative and without local or distant recurrence during a follow-up of 10.6 years. Also, surgery was curative for the 3 patients, but 1 patient developed liver metastasis and died during follow-up. Of the 2 patients with type IV GNET, 1 underwent surgery and chemotherapy after the diagnosis of metastatic disease and the other started chemotherapy at diagnosis.

Discussion

The incidence of GNETs has increased over the past few decades and type I GNETs represent 70-80% ENT#91;⁵ENT#93;. In our study, type I GNETs accounted for 86% of the patients, type III, 11%, type IV, 2%, and only 1 patient had type II. The lack of a standard staging system for GNETs substantially hinders the prediction of the risk of recurrence and prognosis of patients suffering from GNETs. The AJCC, ENETS, and WHO staging classifications provide a better stratification of gastric NETs and they are routinely employed ENT#91;³, ⁶ENT#93;. Our retrospective cohort with a long follow-up confirms that type I GNETs have an excellent prognosis, and that endoscopic resection is effective and safe for the treatment of type I and small type III gastric NETs (selected lesions in the latter). We also identified risk factors for metastasis and mortality that should dictate a more intensive surveillance in these cases. Our findings of low sensitivity and specificity of sequential CgA for tumor recurrence show that it is not useful neither as diagnostic nor as tumor biomarker for surveillance of these patients.

Most GNETs are diagnosed incidentally during screening endoscopy and classifying them is crucial in deciding the proper treatment as they can be either indolent or aggressive. In our study, as expected, we found type of GNET as a prognostic factor, with no deaths in the group of patients with type I GNET, even though 3 patients presented lymph node metastasis. Type III and IV GNETs have a more malignant biological behavior with a metastasis rate in our study of 8.7 and 100%, which was different to that reported in the literature (40%), probably due to the low number of patients ENT#91;⁷, ⁸ENT#93;.

In our cohort, we found male sex as a mortality predictor, which is in line with previous studies ENT#91;⁹, ¹⁰ENT#93;. The protective effect of female gender observed in various studies is still unclear.

All patients with metastases or who died during the follow-up period had lesions larger than 10 mm in size. In a Korean study, a tumor size was significantly associated with survival and a tumor size of ≥2 cm was a risk factor for metastasis ENT#91;¹¹ENT#93;. We also demonstrated that patients with Ki-67 >20%, an intermediate/high grade (vs. low), and invasion beyond submucosa had a poor prognosis, with a high risk of metastasis and mortality. According to Nadler et al. ENT#91;¹²ENT#93;, Ki-67 is a very reliable marker for NETs when used to stage tumors with ENETS and WHO staging methods. GNETs with poor differentiation (G3) are more aggressive tumors and have poorer prognosis than well-differentiated (G1-G2) ENT#91;¹³ENT#93;. Tian et al.ENT#91;¹⁴ENT#93; found a 5-year survival of patients with well-differentiated tumors of 80.4%, which was nearly 4 times the overall survival of patients with G3 tumors. In our study, there was only 1 patient with NEC and two with G3 lesions, and the mortality was 100 and 50%, respectively. In multivariate analysis, we did not identify any independent factor for mortality probably due to the low number of deaths (n = 3). For metastasis, Ki-67 was identified as an independent risk factor. According to our results, patients with prognostic factors such as lesions >10 mm, Ki-67 >20%, intermediate/high grade, and invasion beyond the submucosa should be monitored with PET-DOTANOC initially every 6 months and after 2 years annually.

According to ENETS guidelines, published in 2016, GNETs should be treated according to their classification and size. Type I and II small tumors should be monitored or treated with minimal endoscopic or laparoscopic surgery and type III tumors require radical gastrectomy.

However, some studies showed that endoscopic resection is possible for larger lesions (>10 mm) and even for smaller type III GNETs ENT#91;¹⁵-¹⁷ENT#93;.

In our cohort, we had 113 patients with type I GNET. The strategy adopted depended on the characteristics of the lesions (morphology and size), but also on endoscopist experience and preferences. In a recent study by Chin et al. ENT#91;¹⁸ENT#93;, the rate of disease progression during surveillance was 29.5%, suggesting that simple surveillance may not be safe in these patients. However, Esposito et al. ENT#91;¹⁵ENT#93; demonstrated that a noninterventional approach might be justified for small lesions (<5 mm), since none of the patients underwent disease progression after a median follow-up of almost 4 years. Also, the authors agree that endoscopic resection remains the preferred approach for larger tumors, owing the risk of disease progression, although low, as suggested by European guidelines ENT#91;¹⁹ENT#93;. There are a few studies with a limited number of patients comparing endoscopic resection by EMR and ESD for type I GNETs. A retrospective study (n = 87) has compared both techniques for <10 mm lesions and ESD showed a trend to a better pathologically complete resection rate (95% vs. 83%, p = 0.17) and a trend to a higher adverse event rate (perforation 2.6% vs. 0%, delayed bleeding 5.1% vs. 4.2%), but with no statistically significant differences ENT#91;²⁰ENT#93;. Sato et al. ENT#91;¹⁶ENT#93; analyzed 13 patients and showed a better complete resection with ESD versus conventional EMR. In our study, the characteristics of the lesions submitted to ESD and EMR were comparable and the rate of complete histological resection was similar in both groups (78 vs. 60%), but lympho-vascular invasion was more frequent in the ESD group (40 vs. 6%). Besides, the recurrence rate was also similar in both groups. Of note, the presence of positive vertical or horizontal margins after EMR or ESD had no impact on prognosis. There is no evidence of clear superiority of one technique over another, and the choice of technique should take into account not only size but also morphology and lifting, and the technique with a high likelihood of en bloc/R0 resection should be chosen. For type III GNETs, ENETs recommend surgery and endoscopic resection is still not debated ENT#91;⁶ENT#93;. However, there is now evidence to support endoscopic resection. Kwon et al. ENT#91;¹⁷ENT#93; analyzed 50 patients with type III GNETs treated with endoscopic resection (EMR in 41 and ESD in 9 cases). Pathological incomplete resection was observed in 10 cases (no differences between the two techniques), and during the follow-up period (43 months), there was no evidence of tumor recurrence). In our cohort, polypectomy was performed in 5 patients at diagnosis because these lesions were small (5-10 mm) and were not suspected to be NETs. Endoscopic resection was performed in 6 patients (5 EMR and 1 ESD). A complete resection rate after EMR was 40% and 100% after ESD. In the authors opinion, in selected patients with lesions <15-20 mm, Ki-67 <3% (consider up to 3-20%), without evidence of lymph node metastasis, endoscopic resection should be attempted (with ESD being the preferred technique for lesions >10 mm). CgA level has been shown to be elevated in various diseases, including benign and malignant diseases and associated with acid-suppressive medications ENT#91;²¹, ²²ENT#93;. The specificity of CgA for the diagnosis or prognosis of NETs is compromised by the non oncological and non-NET oncological situations, but is the most common circulating biomarker used for the follow-up of NETs. Tsai et al. ENT#91;²³ENT#93; suggested that baseline CgA level is associated with overall survival of gastroenteropancreatic NET (GEP-NET) patients (HR = 13.52, p = 0.045). A 40%or greater increase of change of CgA level may predict tumor progression or recurrence during treatment or surveillance of gastroenteropancreatic NETs. In our study, CgA showed a low sensitivity for detection of recurrence (20%) but a specificity of 79%. The sensitivity was even lower in type I GNET patients (8%). Routine determination of serum CgA in all patients with GNET does not seem to be beneficial. However, it may be helpful in patients with distant disease to assess response to therapy.

This study has some potential limitations. The retrospective nature of our cohort presented limitations mainly in the selection of patients for each therapeutic group and in the collection of patients’ data. Also, this unicenter design can limit the generalizability of findings. Furthermore, an important limitation in our work was that, although a significant number of patients were included, due to the reduced rate of recurrence and mortality of most GNETs, only 3 patients died and 7 patients had metastasis making it difficult to identify prognostic factors.

In conclusion, type I GNET is the most frequent in our series. Identification of risk factors for the presence of metastases and for mortality in these groups of patients can help in individualizing the therapeutic strategy and possibly reducing the burdensome surveillance for patients and hospital endoscopy resources as, even after endoscopic resection, lifelong endoscopic surveillance is required due to the multifocal nature of this tumor. Regardless of disease stage, the overall survival of our patients with type I GNET is excellent. ESD and EMR achieved similar short- and long-term results in this group of patients.