Neuroendocrine cells produce various regulatory substances, including biogenic amines, peptides, and growth factors. These cells are predominantly found in the pancreas, lungs, and gastrointestinal tract. While most of these cells display slow growth and are well-differentiated with a benign form, aggressive and poorly differentiated malignant tumors also exist.¹ The 2019 World Health Organization classification categorizes neuroendocrine tumors into well-differentiated neuroendocrine tumors, graded as G1, G2, and G3. It also encompasses poorly differentiated neuroendocrine carcinomas such as small-cell neuroendocrine carcinoma and large-cell neuroendocrine carcinoma, along with mixed neuroendocrine-non-neuroendocrine neoplasms.²

Gastric neuroendocrine tumors (GNETs) are rare diseases arising from gastric mucosa enterochromaffin-like cells. They are typically categorized into three main types. Type 1 GNETs are the most frequently diagnosed, accounting for 70% to 80% of all GNETs. They are strongly associated with atrophic gastritis and are often located in the corpus or the fundus. Type 1 tumors are typically multiple, small (<1 cm), mainly confined to the mucosa and submucosa layers, and well differentiated (World Health Organization grades 1 and 2). They exhibit higher serum gastrin and intragastric pH levels.^1,3 Type II GNETs may be associated with gastrinoma, multiple endocrine neoplasia-1, and Zollinger-Ellison syndrome. Accounting for 5% to 10% of all GNETs, they are often located in the body, fundus, and antrum. Type II tumors are multiple and small (1 to 2 cm). Unlike type I, they exhibit low intragastric pH levels due to hyperacidity.^1,3 Type III tumors represent the second most common type, accounting for 15% to 20% of all GNETs. They behave aggressively, with metastasis rates >50% and a poorer prognosis. Type III tumors are often located in the antrum or corpus, can invade any depth of the gastric wall, and typically present as single large (>2 cm) masses. They exhibit a male preponderance, and unlike type I or II, both serum gastrin and intragastric pH levels are normal.³

Although GNETs are rare, there has been an increase in endoscopic surveillance and heightened awareness among clinicians in recent years, resulting in a reported rise in their incidence.⁴ According to a study utilizing Surveillance, Epidemiology, and End Results data, the incidence of GNETs increased from 0.31 per 1,000,000 patients in 1975 to 4.85 in 2014.⁵ Another study, based on data from European and U.S. cancer registries and PubMed, reported the prevalence per 10,000 population as a median of 0.32 for Europe, 0.17 for the United States, and 0.05 for Japan.⁶ A multicenter study in Korea between 2000 and 2009 noted an increasing incidence of GNETs over the last decade, with the stomach showing the second highest incidence (14.6%). There has been a continuous increase in the incidence of GNETs, with the incidence in 2009 being nine times that reported in 2000.⁷ A growing number of GNETs are incidentally detected during physical examination using esophagogastroduodenoscopy, even in the absence of symptoms.

However, owing to their low incidence rate, there is limited dedicated research on GNETs and their clinical course, resulting in gaps in the understanding of their progression and optimal management. Therefore, understanding the clinical characteristics and long-term outcomes of GNETs is crucial. This study aimed to explore the clinical characteristics, recurrence rates, and long-term outcomes of GNETs.

1. Study population

Patients who underwent management for GNETs at Seoul National University Hospital between January 1, 2013, and October 31, 2023, were assessed. Among these patients, those who were either diagnosed with GNETs through both endoscopic and histological examinations or received specific treatment for GNETs were included. To ensure data quality, patients lacking histological information or those lost during follow-up, specifically those with a follow-up duration of less than 2 years, were excluded from the study.

2. Study design

This retrospective study was conducted at Seoul National University Hospital. The Institutional Review Board of Seoul National University Hospital approved the study protocol (IRB number: E-2401-126-1505). Since this is a retrospective study, obtaining informed consent from patients has been waived. Baseline characteristics including age, underlying disease, and associated symptoms were evaluated. Treatment modalities such as observation, endoscopic resection, surgical resection, and chemotherapy were systematically examined. Histologic findings, size, depth of invasion, surgical/endoscopic resection margin, lymphovascular invasion, grade, mitotic count, Ki-67 index, and serologic parameters such as gastrin levels and pepsinogen I and II ratios were evaluated. The follow-up period, occurrence of recurrence, and endoscopic findings, such as atrophic gastritis and intestinal metaplasia, were reviewed. Types were categorized based on atrophic gastritis and gastrin levels; specifically, cases with atrophic gastritis and elevated gastrin levels were classified as type I, whereas patients with atrophic gastritis and normal gastrin levels were classified as type III.

Patients were also classified into four groups according to their pathological diagnosis grades: grade 1, grade 2, grade 3, or poorly differentiated. Baseline characteristics for each grade were examined, and the study explored the differences in GNETs characteristics between endoscopic and surgical treatments. Additionally, a comprehensive analysis of the recurrence rate and associated factors was performed. Finally, the analysis covered the total overall survival, total recurrence-free survival, and overall survival for each grade, stage, and type.

3. Statistical analysis

The primary outcome was the overall survival of patients treated for GNETs, which was visualized using Kaplan-Meier survival plots. Baseline features, recurrences, and related factors were analyzed using statistical methods such as the chi-square test, Fischer exact test, or analysis of variance based on the nature of the variables. A significance level less than 0.05 for the p-value was considered statistically significant. All statistical analyses were conducted utilizing R studio version 4.2.1 (Posit PBC, Boston, MA, USA).

1. Study population

A total of 69 patients diagnosed with and managed for GNETs were included in this study. The mean±standard deviation age was 55±13.2 years, the proportion of men was 55.1% (38/69), and the mean tumor size was 9±8.1 mm (range, 0.4 to 25 mm). Most patients had follow-up durations ranging from 2 to 5 years, with a mean follow-up time of 47±28.4 months (range, 18 to 137 months). There were 49 (71.0%), 17 (24.6%), one (1.4%), and two (2.9%) patients categorized as grade 1, grade 2, grade 3, and poorly differentiated, respectively. The distribution comprised 69.6%, 1.4%, and 29.0% classified as types I, II, and III, respectively. Baseline characteristics, including sex, age, underlying disease, and follow-up period, showed no statistically significant differences among the grades, except for the treatment option. Sizes tended to be larger in higher grades, and the distribution of treatments showed that grade 1 patients received more endoscopic treatments, while grade 2 and above patients tended to undergo surgery or chemotherapy (p=0.015) (Table 1). The presence of Helicobacter pylori, endoscopic evaluation for atrophic gastritis and intestinal metaplasia; and serological assessments of gastrin, pepsinogen I/II ratio, and chromogranin were conducted. No statistically significant differences were found among the grades (Table 2).

Table 1 . Baseline Characteristics.

Characteristic	Grade 1 (n=49)	Grade 2 (n=17)	Grade 3 (n=1)	Poorly Diff (n=2)	p-value
Sex					0.562
Male	27 (55.1)	9 (52.9)	0	2 (100)
Female	22 (44.9)	8 (47.1)	1 (100)	0
Age, yr	59.0±14.1	54.0±10.7	42	61.5±9.2	0.769
Size, mm					0.037
<10	35 (71.4)	10 (58.8)	0	0
10–20	8 (16.3)	7 (41.2)	1 (100)	1 (50.0)
>20	4 (8.2)	0	0	1 (50.0)
Unknown	2 (4.1)	0	0	0
Mean size	8.1±7.1	9.5±5.1	20	21.0±7.1
Single/multiple					0.244
Single	38 (77.6)	13 (76.5)	0	1 (50.0)
Multiple	11 (22.4)	4 (23.5)	1 (100)	1 (50.0)
Type					1.000
I	33 (67.3)	12 (70.6)	1 (100)	2 (100)
II	1 (2.0)	0	0	0
III	15 (30.6)	5 (29.4)	0	0
Underlying disease
Cancer	8 (16.3)	2 (11.8 )	1 (100)	0	0.236
DM	11 (22.4)	5 (29.4)	0	2 (100)	0.104
HTN	11 (22.4)	5 (29.4)	0	2 (100)	0.104
Thyroid disease/MEN1	5 (10.2)	0	1 (100)	0	0.060
CVD	3 (6.1)	1 (5.9)	0	0	1.000
Hematologic disease	2 (4.1)	0	0	0	1.000
Treatment					0.015
Observation	2 (4.1)	1 (5.9)	0	0
Endoscopic	35 (71.4)	8 (47.1)	0	0
Surgical	7 (14.3)	3 (17.6)	0	1 (50.0)
Chemotherapy	1 (2.0)	0	1 (100)	0
Combination	4 (8.2)	5 (29.4)	0	1 (50.0)

Data are presented as number (%) or mean±SD..

Diff, differentiation; DM, diabetes mellitus; HTN, hypertension; MEN, multiple endocrine neoplasia; CVD, cardiovascular disease..

Table 2 . Baseline Characteristics of Helicobacter pylori, Atrophic Gastritis, Intestinal Metaplasia, and Serological Examinations.

Characteristic	Grade 1 (n=49)	Grade 2 (n=17)	Grade 3 (n=1)	Poorly Diff (n=2)	p-value
H. pylori					0.118
Absent	19 (38.8)	3 (17.6)	1 (100)	0
Present	18 (36.7)	7 (41.2)	0	0
Unknown	12 (24.5)	7 (41.2)	0	2 (100)
Atrophic gastritis					1.000
Absent	8 (16.3)	3 (17.6)	0	0
Present	40 (81.6)	14 (82.4)	1 (100)	2 (100)
Unknown	1 (2.0)	0	0	0
Intestinal metaplasia					0.703
Absent	21 (42.9)	7 (41.2)	1 (100)	0
Present	27 (55.1)	10 (58.8)	0	2 (100)
Unknown	1 (2.0)	0	0	0
Gastrin (pg/mL)					0.934
13–115	15 (30.6)	5 (29.4)	0	0
>115	12 (24.5)	5 (29.4)	0	0
Unknown	22 (44.9)	7 (41.2)	1 (100)	2 (100)
Median (IQR)	58.5 (26.4–424.0)	119.5 (21.0–806.0)	-	-
Pepsinogen I/II ratio					0.778
Negative	5 (10.2)	2 (11.8)	0	0
Weakly positive	0	0	0	0
Positive	0	1 (5.9)	0	0
Strongly positive	6 (12.2)	2 (11.8)	0	0
Unknown	38 (77.6)	12 (70.6)	1 (100)	2 (100)
Mean±SD	3.2±2.9	2.4±2.2	-	-
Chromogranin					0.224
0–108	43 (87.8)	12 (70.6)	1 (100)	1 (50.0)
>108	4 (8.2)	4 (23.5)	0	1 (50.0)
Unknown	2 (4.1)	1 (5.9)	0	0
Median (IQR)	62.5 (27.0–205.0)	41.3 (39.0–90.0)	-	21.0 (21.0–21.0)

Data are presented as number (%) unless otherwise indicated..

Diff, differentiation; IQR, interquartile range..

2. Comparison between endoscopic and surgical treatments

Patients treated endoscopically, including endoscopic mucosal resection and endoscopic submucosal dissection, and those treated surgically, including wedge resection and gastrectomy, were analyzed individually. A total of 45 cases underwent endoscopic treatment, while 14 underwent surgical treatment. Among the various factors, both endoscopically measured tumor size (p=0.000) and pathologically measured tumor size (p=0.001) as well as depth (p=0.011) and stage (p=0.005) showed statistically significant differences. There was no significant difference between the two groups in terms of recurrence (Supplementary Table 1).

3. Recurrence

Among the 69 patients, five individuals encountered recurrence, three of them experienced metachronous recurrence, and two developed distant metastasis, primarily in the liver. All cases of metachronous recurrence were classified as type I, whereas those with distant metastases were included as types I or III. The time to recurrence spanned 1 to 9.8 years (Table 3). Two resection margin positive cases were identified, one with a vertical margin and the other with a lateral margin. The former was treated surgically, whereas the latter was treated endoscopically using argon plasma coagulation (APC). In exploring the potential factors associated with recurrence rate, two variables emerged as statistically significant. Margin positivity (p=0.017) and invasion beyond the submucosal layer (p=0.007) correlated with a higher recurrence rate (Table 4). However, no correlation was found between disease recurrence and grade, type, or treatment method. Factors, such as tumor size, serum gastrin level, and Ki-67 index, which were previously hypothesized to have potential correlations, were not significant in this study. Further analysis of the presence of intestinal metaplasia, atrophic gastritis, and H. pylori infection did not reveal any statistically significant correlation with recurrence (Table 5). The association between H. pylori infection and recurrence within type I GNETs was analyzed. Among the 48 ^type I GNET patients investigated for H. pylori infection, three out of 16 non-infected patients relapsed, and one out of 14 infected patients relapsed. There was no statistical significances (p=0.1).

Table 3 . Detailed Clinicopathologic Information at the Time of Initial Treatment for Patients with Recurrence.

Variable	Patient 1	Patient 2	Patient 3	Patient 4	Patient 5
Age, yr	51	65	64	61	34
Sex	F	F	M	F	F
Treatment	Surgery	ESD → surgery	ESD	ESD	ESD → APC
Single/multiple	Single	Single	Multiple	Multiple	Singe
Grade	G1	G1	G1	G2	G1
Site	HB	Antrum	MB, LB	Fundus, HB	HB
Site	GC	LC	GC	LC, AW	GC
Size, mm	23	8	1, 4	2.5, 1	15
Depth	Ss	Sm3	Sm1	M, Sm1	Sm1
Resection margin	Free	Vertical margin (+)	Free	Free	Lateral margin (+)
LVI	Present	Absent	Absent	Absent	Absent
Type	I	III	I	I	I
Stage	IIIB	I	I	I	I
Follow-up, yr	11.4	10	6	4	10.3
Gastrin, pg/mL	Unknown	97	401	>1,000	>1,000
Ki-67 index, %	<3	<3	3–20	3–20	<3
Recurrence type	Distant (liver)	Distant (liver)	Local	Local	Local
Time to recurrence, yr	6.7	8	1	1	9.8

ESD, endoscopic submucosal dissection; APC, argon plasma coagulation; HB, high body; MB, middle body; LB, low body; GC, greater curvature; LC, lesser curvature; AW, anterior wall; Ss, subserosa; Sm, submucosa; M, mucosa; LVI, lymphovascular invasion..

Table 4 . Analysis of Factors Associated with Recurrence.

	Total	Recurrence (%)	p-value
Grade			0.866
G1	47	4 (8.51)
G2	17	1 (5.88)
G3	0	0
Poorly differentiation	2	0
Initial treatment			0.933
Observation	1	0
Biopsy	7	0
APC	1	0
EMR/ESD	34	3 (8.82)
Surgery	23	2 (8.70)
Size			0.840
<1 cm	38	3 (7.89)
1–2 cm	17	1 (5.88)
>2 cm	7	1 (14.26)
Resection margin			0.017
Free	57	3 (5.26)
Positive	5	2 (40.00)
LVI			0.548
Absent	42	4 (9.52)
Present	12	1 (8.33)
Depth of invasion			0.007
Mucosa	16	0
Submucosa	42	4 (9.52)
Muscularis propria	2	0
Subserosa	1	1 (100)
Type			0.558
I	45	4 (8.89)
II	0	0
III	21	1 (4.76)
Stage			0.719
I	52	4 (9.16)
IIA	6	0
IIB	2	0
IIIA	0	0
IIIB	6	1 (16.67)
Gastrin			0.223
13–115 pg/mL	21	1 (4.76)
>115 pg/mL	16	3 (18.75)
Unknown	29	1 (3.45)
Ki-67 index			0.565
<3%	33	3 (9.09)
3%–20%	16	2 (12.50)
>20%	2	0

Data are presented as number (%). The number of valid data points differs across variables, leading to varying totals for each variable..

APC, argon plasma coagulation; EMR, endoscopic mucosal resection; ESD, endoscopic submucosal dissection; LVI, lymphovascular invasion..

Table 5 . Analysis Based on Intestinal Metaplasia, Atrophic Gastritis, and Helicobacter pylori Infection.

	Absent	Present	p-value
No. of intestinal metaplasia	29	39	0.307
No recurrence	25 (86.2)	36 (92.3)
Recurrence	2 (6.9)	3 (7.7)
Unknown	2 (6.9)	0
No. of atrophic gastritis	11	57	0.706
No recurrence	11 (100.0)	50 (87.7)
Recurrence	0	5 (8.8)
Unknown	0	2 (3.5)
No. of H. pylori infection	23	25	0.415
No recurrence	21 (91.3)	21 (84.0)
Recurrence	0	3 (12.0)
Unknown	2 (8.7)	1 (4.0)

Data are presented as number (%)..

4. Overall survival

Among the 69 patients, six died. Five died due to grade 1 tumors and one from grade 3 tumors. The overall 5-year survival rate was 93.8%, and the 5-year recurrence-free survival rate was 97.1%. Analysis based on different grades revealed statistically significant differences (p<0.001). Specifically, a significant difference was found between grade 1 and grade 3 tumors (p=0.002). However, no statistically significant differences were detected between the different stages (p=0.154) or types (p=0.920) (Fig. 1).

Figure 1. Survival plots. (A) Total OS and total RFS. (B) OS associated with each grade. (C) OS associated with each stage. (D) OS associated with each type. OS, overall survival; RFS, recurrence-free survival.

In this retrospective single-center study, we examined the clinical features, recurrence rates, and treatment outcomes of GNETs. Kim et al.⁸ reported a significant increase in the mean tumor size with higher tumor grades (p=0.025) in a single-center retrospective study conducted in Korea. They also found that the proportion of patients who underwent surgical treatment was significantly higher in the GNET grades 2-3 group than in the GNET grade 1 group (10.9% for grade 1 vs 38.3% for grades 2-3, p<0.001). Our results are consistent with these findings. In our baseline characteristic analysis, we observed a significant gradual increase in mean GNET size with higher grades (p=0.037). The distribution of treatment indicated that grade 1 patients received more endoscopic treatment, while grade 2 and above patients tended to undergo surgery or chemotherapy (p=0.013).

Upon investigating the potential factors associated with the recurrence rate, two statistically significant variables emerged. Resection margin positivity and invasion deeper than the submucosal layer were correlated with a higher recurrence rate. This observation aligns with the findings of Noh et al.,⁹ who reported that incomplete endoscopic resection was associated with GNETs recurrence, consistent with the results of our recurrence analysis. This finding highlights the importance of early detection and diagnosis, which can lead to early treatment before the lesion advances locally.

Our results did not indicate any association between disease recurrence and grade, type, size, or treatment method. Specifically, tumor size and grade are crucial factors in treatment decisions, as international guidelines recommend more aggressive treatments for larger tumors and higher grades. For type I tumors, guidelines recommend endoscopic surveillance or resection for tumors <10 mm and surgery for tumors 10 to 20 mm and >20 mm. Grade 3 tumors typically require surgery. Patients with type III tumors require surgery for high-grade tumors. For grade 1/2 tumors, surgery is recommended for tumors >20 mm in size, while endoscopic treatment is recommended for tumors <10 mm in size and low grade. Wedge resection or gastrectomy is often suggested for sizes between 10 and 20 mm, with endoscopic resection for smaller tumors (<15 mm) with grade 2 histology if the surgical risk is high. National Comprehensive Cancer Network guidelines favor partial or total gastrectomy regardless of size for grade 1/2 tumors.^10,11 As suggested by these guidelines, tumor size and grade are critical factors in determining disease prognosis, with some scholars proposing that lesions measuring 5 mm or larger are significantly associated with recurrence.¹² However, in our study, neither tumor size nor grade showed a significant association with recurrence. Some studies have proposed no significant association between tumor size and recurrence, which aligns with our findings.^13-15 The inconsistent results regarding the relationship between tumor size and recurrence make it controversial and underscore the need for further research.

Serum gastrin level and Ki-67 index are other controversial factors that have been suggested to be potentially correlated with the recurrence rate. Daskalakis et al.¹⁴ reported an association between high serum gastrin levels and recurrence, while another study by Sheikh-Ahmad et al.¹³ suggested that high gastrin levels and a lower Ki-67 index were associated with recurrence. In our study, high gastrin levels and Ki-67 index were not significantly associated with recurrence. This discrepancy highlights the uncertainty regarding their role in disease prognosis and emphasizes the need for additional research.

In our study, most patients with GNETs were grade 1 and type I, with a recurrence rate of 7.25%. The recurrence rate in our study was lower than that reported in previous studies. Tsolakis et al.,¹⁶ in a systematic review and meta-analysis from 2019, pooled results eight studies reported local recurrence rate of 17.8% (n=72/422). In another recent retrospective study conducted by Chen et al.,¹⁷ it was found that 41.2% of patients experienced their first recurrence. There could be several reasons why our study reported a lower recurrence rate compared to other studies. In Korea, the National Cancer Screening Program recommends that individuals aged 40 and older undergo endoscopy or upper gastrointestinal series every 2 years. The high accessibility of endoscopic examinations in Korea may have contributed to the detection of tumors at earlier stages, before the disease became locally advanced. This systematic national screening program could potentially result in a better prognosis regarding recurrence.

Type I GNETs are known to be associated with chronic autoimmune atrophic gastritis.^18-20 H. pylori infection is one of the most important factors in the development of atrophic gastritis and gastric cancer, and its prevalence is higher in East Asia than in the United States than in Western countries.^21,22 In our study, out of 69 individuals, 21 were infected with H. pylori, with 14 individuals with type I and seven with type III infected. Considering its involvement in the development of atrophic gastritis, which leads to hypergastrinemia, it is natural to assume that H. pylori is associated with type I GNETs. However, our analysis revealed no significant differences in H. pylori infection between these types (Supplementary Table 2). However, in many cases, H. pylori infection was not tested, leading to incomplete information for drawing definitive conclusions from the findings. Further research on the relationship between H. pylori infection and GNETs as well as their prognosis could provide valuable insights into the disease.

In the survival analysis, we observed a total of six deaths among 69 patients, with a 5-year survival rate of 93.8 and a 5-year recurrence-free survival of 97.1%. This result again demonstrates that the disease is indolent and has a good prognosis. Among these six patients, only one had a grade 3 tumor, while the other five patients were categorized as having grade 1 tumors. Type III was present in only one patient, whereas the other five patients had type I. Our results revealed a significant difference in the survival rate between grade 1 and grade 3 tumors (p=0.002). However, it is logical to associate a worse prognosis with a higher grade. Further investigation of each patient’s death revealed that only one patient died of GNET, while the others died of old age or other comorbidities. The one patient who died of GNET-related causes was classified as grade 3 and type I, with a Ki-67 index of 20.2%. The patient had liver metastases at the time of initial presentation at our hospital. Therefore, if the cause of death had been further analyzed and considered, the results could have been different. This also underscores the fact that locally advanced disease may lead to worse prognosis.

This study has some limitations. First, as a retrospective study, our research had inherent limitations, including a lack of available information, insufficient data, and the absence of randomization. Also, as a single-center study, the generalizability of our findings is limited, necessitating larger, multicenter studies to validate our conclusions and explore additional factors influencing recurrence. Despite the low incidence rate of GNETs, we believe our study includes a significant number of patients with detailed clinical and pathologic data. However, the small number of patients with recurrence may limit our ability to draw solid conclusions about associated factors. Our analysis found a significant correlation between margin positivity and recurrence, prompting further investigation of patients with positive margins after endoscopic submucosal dissection. Among five patients with positive margins, four had positive vertical margins, and one had a positive lateral margin. One patient exhibited lymphovascular invasion, but none showed additional lymphatic or venous invasion post-endoscopic submucosal dissection. Subsequent treatments included additional surgery for three patients, APC for one patient with a residual tumor, and follow-up for another patient without further procedures. One patient who underwent APC experienced local recurrence after 9.8 years and subsequently had a total gastrectomy, while another developed distant liver metastasis after 8 years and was treated with octreotide and lutetium oxodotreotide. These findings suggest that patients with higher grade or type III tumors underwent more aggressive interventions, while those with lower-grade, type I tumors were managed with less invasive approaches such as APC or short-term follow-up. However, no clear trend was found between patient characteristics and the risk of tumor recurrence (Supplementary Table 3). Furthermore, we were unable to establish a definite diagnosis of autoimmune gastritis owing to a lack of sufficient evidence, such as anti-parietal cell antibodies, anti-intrinsic factor antibodies, and vitamin B₁₂ levels.

In conclusion, our study identified margin positivity and deep invasion after resection as significant factors associated with higher recurrence rates, highlighting the importance of closer monitoring during short-term follow-up. Furthermore, we advocate for long-term, large-scale, multicenter prospective studies to provide further evaluation and validation of our findings.

S.J.C. is an editorial board member of the journal but was not involved in the peer reviewer selection, evaluation, or decision process of this article. No other potential conflicts of interest relevant to this article were reported.

Study concept and design: H.C. Data acquisition: J.Y.K., Q.Z. Data analysis and interpretation: J.Y.K., Q.Z. Drafting of the manuscript; critical revision of the manuscript for important intellectual content: H.C., J.Y.K., Q.Z. Statistical analysis: J.Y.K., Q.Z. Administrative, technical, or material support; study supervision: S.J.C., S.G.K., H.C. Approval of final manuscript: all authors.

Supplementary materials can be accessed at https://doi.org/10.5009/gnl240272.