Clinical characteristics and long‐term prognosis of type 1 gastric neuroendocrine tumors in a large Japanese national cohort

Abstract

Objectives

Optimal management of type 1 gastric neuroendocrine tumors (T1‐GNETs) remains unknown, with few reports on their long‐term prognosis. This study investigated the clinical characteristics and long‐term prognosis of T1‐GNETs.

Methods

We reviewed the medical records of patients diagnosed with T1‐GNET during 1991–2019 at 40 institutions in Japan.

Results

Among 172 patients, endoscopic resection (ER), endoscopic surveillance, and surgery were performed in 84, 61, and 27, respectively, including 27, 77, and 2 patients with pT1a‐M, pT1b‐SM, and pT2 tumors, respectively. The median tumor diameter was 5 (range 0.8–55) mm. Four (2.9%) patients had lymph node metastasis (LNM); none had liver metastasis. LNM rates were significantly higher in tumors with lymphovascular invasion (LVI) (15.8%; 3/19) than in those without (1.1%; 1/92) (P = 0.016). For tumors <10 mm, LVI and LNM rates were 18.4% (14/76) and 2.2% (2/90), respectively, which were not significantly different from those of tumors 10–20 mm (LVI 13.3%; 2/15, P = 0.211; and LNM 0%; 0/17, P = 1.0). However, these rates were significantly lower than those of tumors >20 mm (LVI 60%; 3/5, P = 0.021; and LNM 40%; 2/5, P = 0.039). No tumor recurrence or cause‐specific death occurred during the median follow‐up of 10.1 (1–25) years. The 10‐year overall survival rate was 97%.

Conclusions

Type 1 gastric neuroendocrine tumors showed indolent nature and favorable long‐term prognoses. LVI could be useful in indicating the need for additional treatments. ER for risk prediction of LNM should be considered for tumors <10 mm and may be feasible for tumors 10–20 mm.

Trial registration

The study protocol was registered in the University Hospital Medical Information Network (UMIN) under the identifier UMIN000029927.

INTRODUCTION

Gastric neuroendocrine tumors (GNETs) are classified into three subtypes with different clinical manifestations and prognoses. ¹ , ² Type 2 GNETs are associated with hypergastrinemia due to multiple endocrine neoplasia‐I/Zollinger–Ellison syndrome, and type 3 GNETs have no known etiological factor. Type 1 GNETs (T1‐GNETs) are the most frequent subtype and occur in the setting of hypergastrinemia due to chronic gastritis, partly atrophic gastritis caused by Helicobacter pylori infection, ¹ , ² , ³ and mainly autoimmune gastritis (AIG). AIG has been considered a rare disease in Japan, ⁴ with a prevalence of 0.49%. ⁵ According to studies from Japan, 2–11% of AIG cases are complicated by GNET. ⁶ , ⁷ Therefore, due to the rarity of AIG, little is known about the clinicopathological characteristics of T1‐GNETs and the prognosis of patients with T1‐GNETs in Japan.

Although several previous reports have described a favorable long‐term prognosis, ⁸ , ⁹ , ¹⁰ the number of reports is limited, and are mainly from Western countries. Our group previously reported that the clinical outcomes of patients with T1‐GNETs in Japan were favorable regardless of differences in management. In that study, Sato et al. ¹¹ reported that 41 of 82 patients underwent endoscopic therapy, with no case of local recurrence or metastasis. However, the study was limited by a small sample size and a relatively short follow‐up period. Thus, the long‐term prognosis of T1‐GNET in Japan has not been fully investigated.

Several treatments are currently available for T1‐GNETs, including endoscopic surveillance, endoscopic resection (ER), and surgery. ¹² , ¹³ , ¹⁴ Recently, clinical guidelines for gastroenteropancreatic neuroendocrine neoplasms were revised in Japan. ¹⁵ In the previous guidelines, tumors sized 10–20 mm could be managed with ER or gastrectomy with lymph node (LN) dissection. ¹⁶ In the new guidelines, gastrectomy with LN dissection is recommended for tumors ≥10 mm. ¹⁵ However, there is debate regarding the treatment for T1‐GNETs sized 10–20 mm. ¹⁵ , ¹⁷ , ¹⁸ Thus, the management approach of T1‐GNET in Japan has not been established. Furthermore, it is not standardized worldwide; statements on the application of ER or additional treatments after ER differ between the guidelines of the European Neuroendocrine Tumor Society (ENETS) and the National Comprehensive Cancer Network (NCCN). ¹⁹ , ²⁰

Therefore, this large multicenter study was led by the science committee of the Japan Gastroenterological Endoscopy Society (JGES). It aimed to confirm the long‐term prognosis of patients with T1‐GNETs in Japan, over a 10‐year median follow‐up period, and to investigate the clinical features of T1‐GNETs to help determine the optimal management for patients.

METHODS

Patients

Between 1991 and 2019, 190 patients with histologically confirmed T1‐GNETs were recruited from 40 institutions in the JGES Research Group. Of these, 18 patients were excluded due to missing data or age eligibility and 172 were included in the analysis.

Inclusion and exclusion criteria

The inclusion criteria were patients with a GNET and at least one of the following: (i) positive anti‐intrinsic factor antibody (APC‐Ab); (ii) presence of fundic gland atrophy and hyperplasia of enterochromaffin‐like cells; and (iii) evidence of hypergastrinemia (>450 pg/mL). The exclusion criteria were as follows: (i) presence of Zollinger–Ellison syndrome; and (ii) patients younger than 20 or older than 79 years of age. These criteria were the same as we previously reported. ¹¹

Assessment

Clinical, laboratory, and endoscopic data were collected at the time of initial diagnosis and during follow‐up for GNETs at each institution and were retrospectively analyzed based on institutional reports without being re‐evaluated. They included data on demographics, tumor size, depth of invasion, lymphovascular invasion (LVI), subclassification according to the World Health Organization (WHO) grading system, ² treatment approaches, presence of APC‐Ab, serum gastrin level, and the long‐term prognosis of tumors: recurrence‐free survival, overall survival, and disease‐specific survival (DSS). The use of immunohistochemistry (IHC) staining for evaluating LVI or grading varied among the institutions. Abdominal computed tomography (CT) or magnetic resonance imaging was performed for tumor staging. Tumor size in the nonresected cases was measured endoscopically in reference to the size of the biopsy forceps.

Follow‐up

This study was based on medical records. Therefore, a unified follow‐up protocol and treatment plan were not instituted. Consequently, the course of treatment was decided according to the expertise at each institution. During the follow‐up period, the patients underwent esophagogastroduodenoscopy every 6–12 months following the initial diagnosis. Physical and hematological/biochemical evaluation and abdominal CT were performed every year if possible. The absence of lymph node metastasis (LNM) was defined histopathologically and radiologically during the follow‐up period in cases treated by surgical resection with LN dissection. A follow‐up duration of at least 5 years was required in cases treated by ER, endoscopic surveillance, or antrectomy to confirm the absence of LNM.

Ethics

This retrospective multicenter study was designed according to the Helsinki Declaration of the World Medical Association and approved by the ethics committee of each participating institution.

Statistical analyses

Qualitative data are reported as numbers with frequencies, and data of continuous variables are expressed as medians with ranges. We applied Fisher's exact test or the Mann–Whitney U‐test for data comparison between the two groups; Fisher's exact test or Kruskal–Wallis test followed by Bonferroni correction for data comparison between the three groups. A P‐value ≤0.05 was considered statistically significant. The survival rates were calculated using the Kaplan–Meier method. All statistical analyses were performed with EZR (Saitama Medical Center, Jichi Medical University, Shimotsuke, Japan), which is a graphical user interface for R (version 4.1.2; The R Foundation for Statistical Computing, Vienna, Austria). ²¹

RESULTS

Clinical features and treatments of T1‐GNETs

Table 1 summarizes the clinical features and treatments of the 172 patients. The male–female ratio was nearly even at 89:83. The median age and follow‐up period were 63 years and 10.1 years (1–25 years), respectively. Multiple lesions and single lesions were each found in 84 cases, and four cases had missing data. The most common lesion site was the corpus (139 lesions, 88.5%), and 107 (84.3%) cases were positive for APC‐Ab. Among the data of evaluated endoscopic findings, central depression was observed in 39 cases (26.2%). The most common color of tumors was yellow (82 cases, 64.6%). Twelve cases (7%) had additional lesions in the stomach, including nine adenocarcinomas, two adenomas, and one large hyperplastic polyp, all of which were treated with endoscopic therapy. ER was performed for 84 cases. Among the 27 cases undergoing surgery, antrectomy was the most common procedure (10 cases), followed by total gastrectomy (8 cases). Three cases received additional surgery after ER due to LVI. Surveillance was conducted for 61 cases. LNM was found in four cases (2.9%).

Table 1.

Clinical features and treatments of type I gastric neuroendocrine tumors

Characteristics	(n = 172)
Sex
Male	89 (51.7%)
Female	83 (48.3%)
Age (years)
Median (range), years	63 (24–74)
Follow‐up period
Median (range), years	10.1 (1–25)
APC‐Ab
Positive	107 (84.3%)
Negative	20 (15.7%)
Missing	45
IFA‐Ab
Positive	29 (35.8%)
Negative	52 (64.2%)
Missing	91
ECL cell hyperplasia
Positive	89 (77.4%)
Negative	26 (22.6%)
Missing	57
Gastrin
Median (range), pg/mL	2962 (168–9100)
Size of tumor
Median (range), mm	5 (0.8–55)
Location
Angle	4 (2.6%)
Corpus	139 (88.5%)
Fornix‐corpus	12 (7.6%)
Fornix	2 (1.3%)
Missing	15
Occurrence
Multiple	84 (50%)
Solitary	84 (50%)
Missing	4
Central depression of tumor
Positive	39 (26.2%)
Negative	110 (73.8%)
Missing	23
Color of the tumor
Yellow	82 (64.6%)
Red	38 (29.9%)
Pale	5 (3.9%)
Same in background mucosa	2 (1.6%)
Missing	45
Other complications in the stomach
Positive (early GC/adenoma/large HPP)	12 (7.0%) (9/2/1)
Negative	160 (93.0%)
Treatment method
ER	84 (48.9%)
Surgery † (TG/DG/AR/PG/PR/LECS)	27 (15.7%) (8/4/10/1/2/2)
Endoscopic surveillance	61 (35.4%)
Lymph node metastasis
Positive	4 (2.9%)
Negative ‡	135 (97.1%)
Not evaluated	33

^† Including three additional cases with surgery after endoscopic resection (ER).

^‡ Negativity for lymph node metastasis was evaluated in patients with follow‐up period over 5 years.

APC‐Ab, anti‐parietal cell antibody; AR, antrectomy; DG, distal gastrectomy; ECL, enterochromaffin‐like; GC, gastric cancer; HPP, hyperplastic polyp; IFA‐Ab, intrinsic factor antibody; LECS, laparoscopy and endoscopy cooperative surgery; PG, proximal gastrectomy; PR, partial resection; TG, total gastrectomy.

Treatment and outcome of T1‐GNETs

Table 2 shows the pathological features and outcomes of resected lesions. In the 111 resected lesions, the depth of invasion reached the mucosa, submucosa, and muscularis propria in 27 (25.5%), 77 (72.6%), and two (1.9%) lesions, respectively. Nineteen cases (17.1%) had LVI, and 21 cases (35%) were subclassified as G2. No recurrence was observed in any of the patients with T1‐GNET.

Table 2.

Pathological features and outcome of resected type 1 gastric neuroendocrine tumors (GNETs)

Characteristics	(n = 111)
Depth of invasion
M	27 (25.5%)
SM	77 (72.6%)
MP	2 (1.9%)
Missing	5
Lymphovascular invasion
ly−, v−	92 (82.9%)
ly−, v+	9 (8.1%)
ly+, v−	5 (4.5%)
ly+, v+	5 (4.5%)
WHO classification
G1	39 (65%)
G2	21 (35%)
Missing	51
Recurrence of GNET
Positive	0 (0%)
Negative	111 (100%)

M, mucosa; MP, muscularis propria; SM, submucosa; WHO, World Health Organization.

Table 3 shows the summary of patients with LNM. All the tumors with LNM were submucosal (SM) in the depth of invasion and were classified as G1 or G2 in WHO grading. Three of them had LVI, including one tumor <10 mm (4.9 mm). The remaining tumor was 6 mm in size without LVI. Figure 1 shows the representative case of LNM, which was case 1 in Table 3.

Table 3.

Summary of patients with lymph node metastasis (LNM)

Case	Age, years	Sex	Serum gastrin	Location	Size, mm	Central depression	Occurrence	Depth of invasion	WHO classification	LVI	Station of LNM
1	46	F	8000	Corpus	4.9	Negative	Multiple	SM	G1	Positive (ly−, v+)	#6
2	68	F	3000	Corpus	6	Negative	Multiple	SM	G1	Negative	#4d
3	49	F	5167	Corpus	25	Positive	Solitary	SM	G1 or G2 †	Positive (ly+, v+)	#1, #3
4	78	M	N/A	Corpus	30	Positive	Solitary	SM	G1 or G2 †	Positive (ly+, v−)	N/A

^† Whether G1 or G2 could not be diagnosed by only the samples of hematoxylin and eosin staining.

F, female; LVI, lymphovascular invasion; M, male; N/A, not available; SM, submucosa; WHO, World Health Organization.

Figure 1.

Endoscopic and histopathological findings of a representative case of lymph node metastasis. (A, B) Endoscopic findings of the background mucosa revealed proximal‐predominant gastric mucosal atrophy. In comparison to the antrum (B), the degree of mucosal atrophy was severe in the fundus (A) based on visible submucosal vessels. (C, D) Endoscopic findings of type 1 gastric neuroendocrine tumors revealed a subepithelial‐like elevated lesion with dilated vessels on the anterior wall of the lower gastric body (C, white‐light image; D, indigo carmine dye‐spraying image). (E, F) Hematoxylin and eosin (H&E) staining of the specimen resected by endoscopic submucosal dissection showed a neuroendocrine tumor (E, low power magnification, scale bar = 500 μm; F, middle power magnification, scale bar = 250 μm). (G) The specimen showed strong immunohistochemical staining for chromogranin A (middle power magnification, scale bar = 100 μm). (H) Yellow arrowheads indicate vascular invasion (Victoria blue‐H&E staining, middle power magnification, scale bar = 100 μm). (I, J) H&E staining of the surgically resected specimen showed metastasis of a neuroendocrine tumor in the lymph node (I, low power magnification, scale bar = 500 μm; J, middle power magnification, scale bar = 100 μm).

Table 4 shows the association between clinicopathological characteristics and LNM in patients with a follow‐up period over 5 years. Univariate analysis showed a significantly higher LNM rate in tumors with LVI (15.8%) than in those without LVI (1.1%) (P = 0.016). The median diameter of the tumor in the patients with LNM tended to be larger than that in their counterparts (15.5 mm vs. 5 mm, P = 0.09).

Table 4.

Association between clinicopathological characteristics and lymph node metastases (LNM) in patients with a follow‐up period over 5 years

Characteristics	LNM positive (n = 4)	LNM negative (n = 135)	Rate of LNM (%)	P‐value
Tumor size (mm)
Median (range)	15.5 (4.9–30)	5 (1–15)	–	0.090
Gastrin
Median (range), pg/mL	5167 (3000–8000)	2920 (168–9100)	–	0.106
Occurrence
Multiple	2	71	2.7	0.900
Solitary	2	64	3.0
Central depression of tumor
Positive	2	28	6.7	0.220
Negative	2	100	2.0
Missing	–	7
Lymphovascular invasion
Positive	3	16	15.8	0.016
Negative	1	91	1.1
Missing	–	28

Clinicopathological features of T1‐GNETs according to tumor size

Table 5 shows the clinicopathological outcomes of T1‐GNETs stratified by tumor size. Of all the lesions, 132 tumors of different size ranges were evaluated. Patients with tumor >20 mm were provided with significantly different (P = 0.045) treatments (surgery 66.7%, 4/6; ER 33%, 2/6; endoscopic surveillance 0%, 0/6) compared to those with tumor <10 mm (surgery 15.1%, 16/106; ER 59.4%, 63/106; endoscopic surveillance 25.5%, 27/106). For tumors <10 mm, LVI and LNM rates were 18.4% (14/76) and 2.2% (2/90), respectively, which were not significantly different from those of tumors sized 10–20 mm (LVI 13.3%, 2/15, P = 0.211; and LNM 0%, 0/17, P = 1.0). Tumors <10 mm had significantly lower LVI and LNM rates than tumors >20 mm (LVI 60%, 3/5, P = 0.021; and LNM 40%, 2/5, P = 0.039). There were no significant differences in other clinicopathological characteristics among the three groups.

Table 5.

Clinicopathological features of type 1 gastric neuroendocrine tumors (T1‐GNETs) according to tumor size

Tumor size	<10 mm	10–20 mm	>20 mm	P‐value
Clinical characteristics	(n = 106)	(n = 20)	(n = 6)
Age
Median (range), years	60 (35–85)	58 (39–78)	68 (49–78)	0.37900
Gastrin
Median (range), pg/mL	3000 (168–9100)	2720 (1100–8140)	2848 (2790–3608)	0.93500
Occurrence
Multiple	52 (49.1%)	9 (50%)	1 (16.7%)	0.36900
Solitary	54 (50.9%)	9 (50%)	5 (83.3%)
Missing	2	2	–
Central depression of tumor
Positive	18 (18.8%)	5 (31.3%)	2 (66.7%)	0.08300
Negative	78 (81.2%)	11 (68.7%)	1 (33.3%)
Missing	10	4	3
Treatment method
ER	63 (59.4%)	11 (55%)	2 (33.3%)	0.03580*
Surgery (TG/DG/AR/others † )	16 (15.1%) (5/2/5/4)	6 (30%) (1/0/3/2)	4 (66.7%) (1/1/2/0)
Endoscopic surveillance	27 (25.5%)	3 (15%)	0 (0%)
Lymph node metastasis
Positive	2 (2.2%)	0 (0%)	2 (40%)	0.01020**
Negative ‡	88 (97.8%)	17 (100%)	3 (60%)
Not evaluated	16	3	1

Pathological characteristics	(n = 79)	(n = 16)	(n = 6)
Depth of invasion
M	17 (22.1%)	4 (25.0%)	0 (0%)	0.20600
SM	59 (76.6%)	12 (75.0%)	5 (83.3%)
MP	1 (1.3%)	0 (0%)	1 (16.7%)
Missing	2	–	–
Lymphovascular invasion
ly−, v−	62 (81.6%)	13 (86.7%)	2 (40%)	0.00606***
ly−, v+	9 (11.8%)	0 (0%)	0 (0%)
ly+, v−	4 (5.3%)	0 (0%)	1 (20%)
ly+, v+	1 (1.3%)	2 (13.3%)	2 (40%)
Missing	3	1	1
WHO classification
G1	27 (71.1%)	4 (44.4%)	0 (0%)	0.23900
G2	11 (28.9%)	5 (55.6%)	0 (0%)
G3	0 (0%)	0 (0%)	0 (0%)
Missing	41	7	6

^*Bonferroni‐adjusted P‐values were 0.767 (<10 mm vs. 10–20 mm), 0.045 (<10 mm vs. >20 mm), and 0.985 (10–20 mm vs. >20 mm).

^**Bonferroni‐adjusted P‐values were 1.000 (<10 mm vs. 10–20 mm), 0.039 (<10 mm vs. >20 mm), and 0.130 (10–20 mm vs. >20 mm).

^***Bonferroni‐adjusted P‐values were 0.211 (<10 mm vs. 10–20 mm), 0.021 (<10 mm vs. >20 mm), and 0.218 (10–20 mm vs. >20 mm).

^† Including PG, PR, and LECS.

^‡ Negativity for lymph node metastasis was evaluated in patients with follow‐up period over 5 years.

AR, antrectomy; DG, distal gastrectomy; ER, endoscopic resection; M, mucosa; MP, muscularis propria; SM, submucosa; TG, total gastrectomy.

Prognosis

Surveillance status was evaluated in 163 cases. No tumor recurrence occurred during the median follow‐up period of 10.1 years (0–25 years). Nine patients died due to other diseases; the causes of death were hepatocellular carcinoma in two cases, pneumonia in one case, cirrhosis in one case, colorectal cancer in one case, rectal neuroendocrine tumor in one case, and other causes in three cases. There was no death caused by T1‐GNET, resulting in a DSS rate of 100%.

Figure 2 shows the survival curve of all patients by treatment method. The 10‐year overall survival rate of all patients was 97%. For patients managed with ER, surgery, and endoscopic surveillance, the 10‐year survival rate was 98%, 100%, and 95%, respectively.

Figure 2.

Long‐term survival rates of patients with type 1 gastric neuroendocrine tumors in all patients (ALL) and by treatment methods. Kaplan–Meier curves showed the overall survival curve in all patients and patients managed with endoscopic resection (ER), surgery, and endoscopic surveillance.

DISCUSSION

In 2014, our study group first conducted a retrospective multicenter study on patients with T1‐GNETs in Japan and reported the treatment methods and clinical outcomes. ¹¹ In that study of 82 cases recruited from 11 institutions initially diagnosed between 1991 and 2011, endoscopic surveillance was conducted for 25 cases, ER for 41 cases, and surgery for 16 cases. There were eight cases with LVI (pT1b‐SM 7 cases, pT2 1 case) but no cases of metastasis. There was no death from the primary disease during a median follow‐up period of 7 years. ¹¹ With 29 additional institutions and an 8‐year extension of the registration period, our group registered additional cases to obtain a sample of 172 cases with a median follow‐up period of 10.1 years (0–25 years). The present study included 84, 27, and 61 cases of ER, surgery, and surveillance, respectively. There were two cases of invasion of the muscularis propria, whose sizes were 5 and 45 mm.

Type 1 gastric neuroendocrine tumor has been reported to have low malignant potential and a good prognosis. ⁸ , ⁹ , ¹⁰ , ¹⁷ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ Thomas et al. ¹⁰ reported long‐term outcomes of 111 T1‐GNET cases from Greece, Sweden, and Israel, with a follow‐up period of 12–348 months (mean, 76 months). During follow‐up, there were no deaths from the primary disease, but there were deaths from other causes, including myocardial infarction and hepatocellular carcinoma. The survival curve showed good prognoses with a 10‐year survival rate of 98%. Another retrospective study by Borch et al. ⁹ on long‐term follow‐up of a total of 65 cases of type 1–4 GNETs, including 51 cases of type 1, reported that long‐term survival in T1‐GNETs did not differ from that of the general population. In a systematic review and meta‐analysis of T1‐GNETs by Tsolakis et al., including 13 studies on 11–152 cases with a median follow‐up ranging from 42 to 87 months, the 5‐year DSS rate of patients was 100% for both ER and surgery. ²⁶

In the present study, no cases of distant or local recurrence, or cause‐specific death were observed regardless of the management approach, with a 10‐year overall survival rate of 97%, which obtained similar results in the long‐term prognosis. ⁸ , ⁹ , ¹⁰ , ¹⁷ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ Considering the number of patients and follow‐up periods in the above‐mentioned meta‐analysis, ²⁶ we believe that the current study is of importance because it provides information regarding long‐term outcomes with a relatively large cohort.

Previous studies reported an LNM rate of 0–5.9% and a liver metastasis rate of 0–3.2%, ⁹ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ which are consistent with our results; however, the risk factors for metastasis remain unclear. Vanoli et al. ²⁷ reported that metastasis was more common in tumors with diameters ≥10 mm. Grozinsky et al. ³¹ showed that the characteristics of metastasis cases included tumor diameter ≥10 mm, high Ki‐67 levels, and significantly different gastrin levels compared with those of nonmetastasis cases; LVI was not included in the risk factors. Furthermore, it was reported that LVI was not a significant risk factor for metastasis in type 3 GNET, ³² although it has been conventionally considered a strong indicator of the risk of metastasis in other gastrointestinal cancers. ³³ , ³⁴ , ³⁵ The clinical significance of LVI in determining the need for additional surgery after ER for T1‐GNETs is unknown. The guidelines in ENETS, NCCN, and Japan do not require additional surgery for GNETs with LVI. ¹⁵ , ¹⁹ , ²⁰ The absence of a significant difference may be because there were fewer cases of metastasis than in other cancers. We showed a significantly higher proportion of LVI in patients with LNM in univariate analysis (P = 0.016). For tumors with LVI, 15.8% (3/19) of patients had LNM, in contrast to only 1.1% (1/106) of patients without LVI. Thus, we believe that LVI is a risk factor for LNM and is useful in indicating the need for additional treatments. Although the application of additional treatments after ER remains controversial, given surgery with LN dissection is recommended for patients with suspected LNM by images in the Japanese or NCCN guidelines, ¹⁵ , ²⁰ the presence of LVI harboring risk for LNM should be treated with additional surgery. However, further studies with larger sample sizes are needed to confirm whether LVI is a risk factor for LNM using multivariate analysis. An international collaborative study is considered essential to recruit enough cases for the analysis.

According to the revised Japanese guidelines, gastrectomy with LN dissection is recommended for T1‐GNETs ≥10 mm. For tumors <10 mm without any signs of metastasis or muscularis propria invasion on images, if the number of lesions is small, ER or endoscopic surveillance is recommended, which is consistent with the ENETS consensus guidelines. ¹⁹ In the present study, 18.4% (14/76) of resected tumors <10 mm had LVI, which is similar to the LVI rate in pT1 gastric cancer. ³⁵ , ³⁷ Considering this result and the significantly higher presence of LVI in patients with LNM, ER to evaluate LVI for T1‐GNETs <10 mm may be a better choice rather than endoscopic surveillance.

In the present study, treatment policies were based on previous reports from each institution, which showed a positive effect on prognoses. Although gastrectomy with LN dissection is recommended for T1‐GNETs ≥10 mm in the revised Japanese guidelines, ¹⁵ the treatment for T1‐GNETs 10–20 mm is still debatable. ¹⁵ , ¹⁷ , ¹⁸ In the present study, for those tumors, ER without additional surgery was conducted in 55% (11/20), initial surgery in 30% (6/20), and endoscopic surveillance in 15% (3/20) of cases, leading to no case of LNM. Furthermore, LVI and LNM rates of tumors 10–20 mm were not significantly different compared to those of tumors <10 mm. These results indicated that tumors 10–20 mm can be preceded by ER. It should be noted that more tumors 10–20 mm tended to be subclassified as G2 than tumors <10 mm (55.6%, 5/9 vs. 28.9%, 11/39). The presence of G2 does not indicate the requirement of additional surgery for T1‐GNETs; however, it is the indicator of requiring additional treatments for other gastrointestinal neuroendocrine tumors such as duodenal and colorectal neuroendocrine tumors based on their LNM rate. ¹⁵ , ³⁷ , ³⁸ Further studies are needed to determine whether additional surgery is required for T1‐GNET classified as G2.

This study has several limitations. It is a retrospective, observational study, and the management of T1‐GNETs varied at each institution. Second, the sample size was small and lacked sufficient statistical power despite using a relatively large cohort. Third, there were some missing data. Although the number of missing data for clinical characteristics was relatively small, that for pathological factors was larger. Fourth, histopathological factors were not re‐evaluated. In particular, standardized criteria for the application of IHC are important because the diagnosis of LVI varies between hematoxylin and eosin staining and IHC staining. Further exploration with larger sample sizes and a centralized pathology review is needed to fully understand the relationship between pathological features and LNM in T1‐GNETs. Moreover, the present study provides some implications, however, questions regarding whether the patient needs surgical treatment or not could not be answered. Therefore, further research is required to ascertain the optimal treatment for T1‐GNETs.

CONCLUSION

We reported the characteristics and long‐term prognosis of T1‐GNETs in a Japanese national cohort. We showed the indolent nature and favorable long‐term prognoses of T1‐GNETs with adequate treatments. Our results indicate that LVI is useful in indicating the need for additional treatments. ER for risk prediction of LNM should be considered for tumors <10 mm, and may also be feasible for tumors 10–20 mm.

CONFLICT OF INTEREST

Author T.G. is a Deputy Editor‐in‐Chief of Digestive Endoscopy. The other authors declare no conflict of interest for this article.

FUNDING INFORMATION

None.