ABSTRACT
Background
Endoscopic resection (ER) is increasingly utilized for early gastric cancer (EGC) due to its minimally invasive profile and comparable survival to gastrectomy (GR). However, concerns regarding oncologic efficacy and institutional volume impact persist. This study evaluates how ER volume quartiles impact outcomes and guideline adherence in EGC management.
Methods
A retrospective analysis of the NCDB from 2010 to 2021 for patients with cT1aN0M0 gastric adenocarcinoma was performed comparing ER and GR. Primary outcomes were oncologic outcomes, rates of guideline adherence, and survival stratified by institutional ER volume tertiles (high, mid, low). Multivariable logistic, linear regression, and Cox proportional hazards models were used.
Results
A total of 1461 patients were analyzed; 937 (61.4%) undergoing ER and 524 (35.7%) GR. 57.5% of ER occurred in the latter half of the study time period, and increasing proportional ER utilization in EGC was observed over time. When stratified by ER volume tertiles, margin positivity (14.6% high volume vs. 13.7% mid vs. 15.5% low, p = 0.85), LVI rates (6.3% high vs. 9.5% mid vs. 9.7% low, p = 0.32), guideline discordant care (22.6% high vs. 22.3% mid vs. 24.3% low, p = 0.82), and survival (vs. low volume: high HR 1.04, p = 0.87; mid HR 1.27, p = 0.39) did not differ by institutional ER volume. High‐volume centers preferentially used ER (OR 1.76, p < 0.01) compared to low and mid‐volume centers. Guideline‐discordant ER correlated with earlier years (OR 0.69, p = 0.03) but not institutional volume (vs. low volume: high OR 0.94, p = 0.76; mid OR 0.90, p = 0.61).
Conclusion
ER for EGC is rapidly expanding across US institutions at all volume levels. ER achieves comparable oncological outcomes and survival independent of ER volume when guideline‐concordant criteria are met, suggesting centralization may be unnecessary for EGC. The guideline discordant care rate of ~24% across all institutions highlights the need for adherence to ER criteria for EGC.
Keywords: early gastric cancer, endoscopic resection, gastrectomy, institutional volume, NCCN guidelines
1. Introduction
Endoscopic resection (ER) has emerged as a less invasive alternative to gastrectomy (GR) for early gastric cancer (EGC), offering comparable overall survival (OS) and oncologic efficacy with shorter hospital stays and fewer complications [1, 2, 3, 4, 5, 6]. ER has been documented to be curative in ~80% of patients, achieving similar R0 resection rates to GR, while avoiding higher rates of adverse events and re‐intervention and poorer quality of life scores compared to patients undergoing GR [7]. Thusly, ER has seen a boom in recent years and is slowly becoming the intervention of choice for select EGC [2]. Furthermore, ER's rise in popularity and prevalence in Eastern cohorts, such as Japan, is more apparent with significantly increasing proportion of ERs in EGC [8].
However, concerns persist regarding metachronous lesions, occurring in ~8% of endoscopically resected EGC cases, histologically incomplete resections, and the inability to evaluate lymph nodes as comprehensively as GR, which may contribute to incomplete prognostication, higher risk of disease recurrence, and ultimately poorer disease‐free survival [4, 5, 9, 10, 11, 12]. Nevertheless, current guidelines from the National Comprehensive Cancer Network (NCCN), Japanese Gastric Cancer (JGC) consensus, American Society for Gastrointestinal Endoscopy (ASGE), American Gastroenterological Association all support ER for EGC meeting criteria including lesion size ≤ 2 cm, well to moderately differentiation, no penetration beyond superficial submucosa, and without lymphovascular invasion (LVI) or positive margins [13, 14, 15, 16, 17]. Despite these standards, real‐world data reveal significant rates of guideline‐discordant ER use, particularly in Western populations [18].
Several Eastern centers in Japan and Korea have had great success with ER for EGC with high R0 resection rates and few complications, in part attributed to centralized expertise and standardized techniques [19, 20]. Conversely, some reports show equivalent survival and resection quality across low‐ and high‐volume centers, suggesting ER appropriateness depends more on adherence to guidelines than volume alone [21, 22]. Centralization efforts in Korea [23] highlight regional attempts to optimize outcomes, but whether similar strategies are effective in Western populations remains under‐investigated.
Most research into ER application for EGC is dominated by Japan contributing ~37.8% of all published manuscripts on the topic, followed by South Korea at ~29%, China at ~21.5%, and the United States at only ~6.2% [24], thus exists relatively limited data on Western practice. Studies on how institutional ER volume impacts perioperative outcomes, survival, or guideline adherence in US populations are lacking. This study aims to address these gaps by evaluating (1) national usage of ER over time, (2) the impact of institutional ER volume on short‐ and long‐term outcomes for EGC, and (3) predictors of ER utilization versus GR in EGC.
2. Methods
2.1. Database and Cohort Selection
We performed a retrospective analysis using the National Cancer Database (NCDB), which is a national database comprised of hospital data from more than 1500 Commission on Cancer (CoC) accredited facilities representing more than 72% of newly diagnosed cancers in the United States. The study period was from 2010 to 2021. The data available to CoC‐accredited facilities lack personal identifiers and were exempt from the Roger Williams Medical Center institutional review board review. The Strengthening the Reporting of Observational Studies in Epidemiology guidelines were followed in writing this manuscript.
The NCDB was queried for patients diagnosed with early‐stage gastric adenocarcinoma in our study period. Gastric adenocarcinoma was identified using relevant International Classification of Diseases for Oncology, 3rd Edition histology codes (8140, 8144, 8210, 8211, 8260, 8480, 8490). Inclusion criteria were patients aged 18 years or older, clinical stage T1aN0M0 gastric adenocarcinoma, and treatment with either ER (rx_summ_surg_prim_site codes 20‐27, local tumor excision) or GR (rx_summ_surg_prim_site codes 30‐80, GR). The rx_summ_surg_prim_site procedure code lists the most definitive procedure the patient underwent for the disease process; therefore, all ER patients defined by codes 20‐27 did not have subsequent GR. Clinical stage was determined through AJCC seventh and eighth edition T‐Tumor, N‐Lymph Nodes, M‐Metastasis staging for appropriate years. Exclusion criteria included patients with missing data on treatment modality, survival information, and/or those undergoing non‐resectional surgery (i.e., local tumor destruction). The primary outcomes included short‐term perioperative and long‐term survival outcomes. Short‐term perioperative outcomes included length of stay, unplanned 30‐day readmission, and 30‐ and 90‐day mortality. Long‐term outcomes included OS. Other secondary outcomes of interest were time from diagnosis to ER, and distance traveled from patient's residence to institution to undergo treatment. ER volume for EGC, as defined by cT1aN0M0 gastric adenocarcinoma where primary tumor size was ≤ 2 cm, per institution was totaled and subsequently divided by active years of ER usage to calculate annual institutional ER volume, Institutions were subsequently divided into volume tertiles based on patient volume: low volume centers (LVC), middle volume centers (MVC), and high volume centers (HVC). Institutions with no ER intervention for EGC were excluded from this analysis. A CONSORT diagram can be seen in Figure 1.
Figure 1.
CONSORT Diagram.
Key variables for descriptive and multivariable analyses were treatment type (ER, GR), LVI, sociodemographic factors (age, sex, race, insurance status), tumor characteristics (size, grade/differentiation, tumor site), Charleson Deyo Comorbidity Index (CCI), year of diagnosis, and facility type. Race was divided into White, Black, Asian, American Indian/Other, and Hispanic subgroups. The year of diagnosis was divided into two 6‐year periods: early (2010–2015) and late (2016–2021). Guideline discordant care (GD) was defined as those undergoing ER whose tumors had either poor differentiation or LVI or positive margins, which are factors associated with lymph node upstaging/metastasis and thus beyond the scope of current guideline recommendations by various societies (NCCN, JGC).
2.2. Statistical Analyses
Continuous variables were described as means and standard deviation (SD); they were analyzed with unpaired t‐test. Categorical variables were presented as totals and frequencies; they were analyzed with chi‐squared test. Kaplan–Meier curves and log‐rank tests were utilized to describe survival outcomes. Cox proportional hazards models were used to assess the impact of various patient/tumor factors on OS. Multivariable logistic regression models were used to calculate odds ratios for categorical variables including LVI, margin positivity, tumor differentiation, and GD care. Multivariable linear regression models were used to calculate coefficients for continuous variables including time from diagnosis to surgery and distance from the patient's residence to the hospital. All tests were two‐sided with a p < 0.05 as the threshold for significance. Statistical analyses were performed using Stata 14.0 (StataCorp, College Station, TX, USA).
3. Results
3.1. Patient and Tumor Characteristics
A total of 1461 patients with cT1aN0M0 EGC met inclusion criteria and were analyzed; 937 (61.4%) undergoing ER and 524 (35.7%) GR (Table 1). Patients undergoing ER were older (mean age 69.5 vs. 63.5 years, p < 0.01), similarly comorbid (p = 0.26), more frequently male (72.6% vs. 60.3%, p < 0.01), and were treated in more recent years (2016–2021: 57.5% vs. 44.7%, p < 0.01) (Table 1). Patients undergoing ER had tumors that were more often well/moderately differentiated (82.7% vs. 60.4%, p < 0.01) and located in the cardia (74.9% vs. 43.5%, < 0.01). LVI was equivalent between patients undergoing ER versus GR. ER contributed to increased margin positivity compared to GR (14.6% vs. 2.3%, p < 0.01). Patients undergoing ER were more likely insured through Medicare (62.6% vs. 53.6%, p < 0.01), whereas those undergoing GR were more likely to have private insurance (38.5% vs. 28.7%, p < 0.01) (Table 1). Over the time period, ER was proportionally increasingly used over GR for EGC (Figure 2).
Table 1.
Patient/tumor characteristics by approach among patients with stage cT1aN0M0 EGC from 2010 to 2021 at ER facilities.
| Patient characteristics | All patients (n = 1461) | ER (n = 937, 61.4%) | GR (n = 524, 35.7%) | p‐value* |
|---|---|---|---|---|
| Age, mean, years (SD) | 67.2 (12.2) | 69.5 (11) | 63.5 (13.1) | < 0.01 |
| Sex (Male %) | 68.2 | 72.6 | 60.3 | < 0.01 |
| Year of diagnosis % | < 0.01 | |||
| Early first half (2010–2015) | 47.1 | 42.5 | 55.3 | |
| Late secnd half (2016–2021) | 52.9 | 57.5 | 44.7 | |
| Race % | < 0.01 | |||
| NH White | 77.7 | 81.5 | 70.9 | |
| NH Black | 8.3 | 7.0 | 10.6 | |
| Asian | 7.4 | 6.3 | 9.4 | |
| American Indian/other | 0.7 | 0.8 | 0.6 | |
| Hispanic | 5.9 | 4.5 | 8.5 | |
| Tumor size, median mm (IQR) | 9 (4–13) | 8 (4–12) | 10 (4–15) | 0.25 |
| Tumor differentiation % | < 0.01 | |||
| Well | 26.8 | 30.8 | 21.7 | |
| Moderate | 46.1 | 51.9 | 38.7 | |
| Poor | 26.5 | 16.8 | 38.7 | |
| Undifferentiated | 0.7 | 0.5 | 0.9 | |
| Site % | < 0.01 | |||
| Cardia | 63.7 | 74.9 | 43.5 | |
| Fundus | 2.1 | 1.6 | 3.1 | |
| Body | 7.1 | 6.1 | 9.0 | |
| Antrum | 13.1 | 9.0 | 20.6 | |
| Pylorus | 1.4 | 1.1 | 2.1 | |
| Other | 12.5 | 7.4 | 21.7 | |
| Lymphovascular invasion % | 8.9 | 8.3 | 9.7 | 0.43 |
| Margin positive % | 9.7 | 14.6 | 2.3 | < 0.01 |
| Chemotherapy receipt % | 4.9 | 3.0 | 8.3 | < 0.01 |
| Charleson‐Deyo score % | 0.26 | |||
| 0 | 67.4 | 67.3 | 67.4 | |
| 1 | 20.4 | 19.3 | 22.3 | |
| 2 | 7.4 | 8.1 | 6.1 | |
| ≥ 3 | 4.9 | 5.2 | 4.2 | |
| Insurance Status % | < 0.01 | |||
| Uninsured | 1.0 | 1.0 | 1.2 | |
| Private | 32.2 | 28.7 | 38.5 | |
| Medicaid | 5.3 | 5.2 | 5.6 | |
| Medicare | 59.4 | 62.6 | 53.6 | |
| Other Government | 2.1 | 2.6 | 1.2 | |
| Facility type % | < 0.01 | |||
| Community | 1.3 | 1.6 | 0.6 | |
| Comprehensive community | 17.1 | 18.5 | 14.4 | |
| Integrated cancer network | 60.3 | 57.1 | 66.4 | |
| Academic/research | 21.3 | 22.7 | 18.6 |
Abbreviations: ER, endoscopic resection; GR, gastric resection; IQR, interquartile range; NH, Non‐Hispanic; SD, standard deviation.
*p < 0.05 deemed significant.
Figure 2.
Endoscopic resection (ER) versus gastrectomy (GR) cases for early‐stage gastric cancer (EGC) over time.
3.2. ER Volume Tertiles and Outcomes
Patients undergoing ER were stratified by annual ER volume tertiles in Table 2, with LVCs performing 0.18–0.8 ER for EGC annually, MVCs performing 0.83–1.13 cases annually, and HVCs performing 1.17–3.33 ER for EGC annually. When stratified by annual ER volume tertiles, guideline‐discordant care did not differ across groups (24.3% LVC vs. 22.3% MVC vs. 22.6% HVC, p = 0.82) (Table 2). High‐volume centers treated younger patients (mean age 68.2 in HVC vs 70.7 in MVC & 69.7 in LVC, p = 0.04), and more non‐Hispanic White patients (85% HVC vs. 78.4% MVC & 80.9% LVC, p < 0.01). Patients being treated at high‐volume centers had primary EGC more frequently located in the cardia (79.5% HVC vs. 71% MVC & 73.8% LVC, p = 0.03) while patients with antral tumors were more likely to receive ER at lower volume centers (11.7% LVC vs. 9.5% MVC vs. 6% HVC, p = 0.03). High‐volume facilities were more likely to be integrated cancer networks or academic institutions (92.9% HVC vs. 80.8% LVC & 64.4% MVC, p < 0.01) (Table 2). Over our study period, all volume tertiles had an increasing proportion of EGC cases treated with ER over GR (Figure 3).
Table 2.
ER patient/tumor characteristics by ER volume tertiles among patients with stage cT1aN0M0 EGC from 2010 to 2021.
| Patient characteristics | Low volume | Mid volume | High volume | p‐value* |
|---|---|---|---|---|
| Annual volume | 0.0–0.8 | 0.8–1.1 | 1.2–3.3 | |
| Guideline discordant* | 24.3 | 22.3 | 22.6 | 0.82 |
| Age, mean, years (SD) | 69.7 (11.1) | 70.7 (10.4) | 68.2 (11.3) | 0.04 |
| Sex (male %) | 73.8 | 71.6 | 72.3 | 0.83 |
| Year of diagnosis % | ||||
| Early first half (2010–2015) | 46.9 | 35.8 | 44.3 | 0.02 |
| Late second half (2016–2021) | 53.1 | 64.2 | 55.7 | |
| Race % | < 0.01 | |||
| NH White | 80.9 | 78.4 | 85.0 | |
| NH Black | 7.3 | 8.6 | 5.2 | |
| Asian | 4.6 | 6.5 | 7.7 | |
| American Indian/other | 2.3 | 0.0 | 0.0 | |
| Hispanic | 5 | 6.5 | 2.2 | |
| Tumor size, median mm (IQR) | 10 (5–13) | 8 (4–14) | 8 (4–11) | 0.12 |
| Tumor differentiation % | 0.47 | |||
| Well | 34.5 | 27.1 | 30.7 | |
| Moderate | 46.9 | 54.8 | 53.7 | |
| Poor | 18.1 | 17.0 | 15.6 | |
| Undifferentiated | 0.6 | 1.1 | 0.0 | |
| Site | 0.03 | |||
| Cardia | 73.8 | 71.0 | 79.5 | |
| Fundus | 1.9 | 1.4 | 1.5 | |
| Body | 6.5 | 6.1 | 5.7 | |
| Antrum | 11.7 | 9.5 | 6.0 | |
| Pylorus | 0.7 | 0.7 | 1.8 | |
| Other | 5.5 | 11.5 | 5.4 | |
| Lymphovascular invasion % | 9.7 | 9.5 | 6.3 | 0.32 |
| Margin positive % | 15.5 | 13.7 | 14.6 | 0.85 |
| Chemo receipt % | 3.3 | 4.5 | 1.5 | 0.10 |
| Charleson‐Deyo score % | 0.20 | |||
| 0 | 67.0 | 66.9 | 68.1 | |
| 1 | 16.5 | 19.3 | 22.0 | |
| 2 | 10.7 | 7.8 | 6.0 | |
| ≥ 3 | 5.8 | 6.1 | 3.9 | |
| Insurance status % | ||||
| Uninsured | 0.3 | 2.4 | 0.3 | 0.02 |
| Private | 31.3 | 23.1 | 31.3 | |
| Medicaid | 3.9 | 7.1 | 4.6 | |
| Medicare | 62.5 | 64.6 | 60.7 | |
| Other government | 2.0 | 2.7 | 3.1 | |
| Facility type % | < 0.01 | |||
| Community | 0.7 | 4.4 | 0.0 | |
| Comprehensive community | 18.6 | 31.2 | 7.1 | |
| Integrated cancer network | 57.0 | 38.3 | 74.2 | |
| Academic/research | 23.8 | 26.1 | 18.7 |
Abbreviations: ER, endoscopic resection; GR, gastric resection; IQR, interquartile range; NH, Non‐Hispanic; SD, standard deviation.
*p < 0.05 deemed significant; guideline discordant patients are those receiving ER with a high risk of lymph node metastasis predicted by either having positive margins, poor differentiation, or positive LVI.
Figure 3.
Proportional endoscopic resection (ER) usage for early‐stage gastric cancer (EGC) by ER volume tertile.
3.3. Predictors of Receiving ER Versus GR
The odds of patients receiving ER versus GR are summarized in Table 3. On multivariate analysis, older age (OR 1.06, 95% CI 1.04–1.08), recent diagnosis time frame (OR 1.63, 95% CI 1.17–2.28), margin positivity (OR 6.55, 95% CI 3.14–13.68), and treatment at high‐volume centers (OR 1.76, 95% CI 1.17–2.64) were associated with ER usage. Conversely, poor differentiation (OR 0.37, 95% CI 0.23–0.60), antral location (OR 0.28, 95% CI 0.17–0.46), and increasing tumor size (OR 0.96, 95% CI 0.94–0.99) were associated with GR (Table 3).
Table 3.
Logistic regression of ER versus GR (odds of getting ER).
| Factors | Multivariable | |
|---|---|---|
| OR | 95% CI | |
| Age | 1.06 | 1.04–1.08 |
| Sex | ||
| Male | Ref. | — |
| Female | 0.97 | 0.67–1.40 |
| Year of diagnosis | ||
| Early first half (2010–2015) | Ref. | — |
| Late second half (2016–2021) | 1.63 | 1.17–2.28 |
| Race | ||
| NH White | Ref. | — |
| NH Black | 1.04 | 0.57–1.92 |
| Asian | 0.83 | 0.44–1.59 |
| American Indian/others | 4.70 | 0.40–55.24 |
| Hispanic | 0.63 | 0.31–1.29 |
| Tumor size | 0.96 | 0.94–0.99 |
| Tumor differentiation | ||
| Well | Ref. | — |
| Moderate | 1.01 | 0.68–1.50 |
| Poor | 0.37 | 0.23–0.60 |
| Undifferentiated | 1.29 | 0.21–7.88 |
| Site | ||
| Cardia | Ref. | — |
| Fundus | 0.34 | 0.10–1.12 |
| Body | 0.45 | 0.23–0.89 |
| Antrum | 0.28 | 0.17–0.46 |
| Pylorus | 0.29 | 0.10–0.86 |
| Other | 0.22 | 0.12–0.40 |
| Lymphovascular invasion | 0.66 | 0.37–1.17 |
| Margin | 6.55 | 3.14–13.68 |
| Charleson‐Deyo score | ||
| 0 | Ref. | — |
| 1 | 0.82 | 0.55–1.22 |
| 2 | 1.17 | 0.60–2.29 |
| ≥ 3 | 0.71 | 0.31–1.59 |
| Insurance status | ||
| Private | Ref. | — |
| Medicaid | 1.53 | 0.70–3.35 |
| Medicare | 0.85 | 0.55–1.32 |
| Other government | 3.21 | 0.70–14.76 |
| Uninsured | 1.89 | 0.41–8.73 |
| Volume | ||
| Low | Ref. | — |
| Mid | 0.72 | 0.49–1.08 |
| High | 1.76 | 1.17–2.64 |
Abbreviations: CI, confidence interval; GR, gastric resection; HR, hazard ratio; NH, Non‐Hispanic.
*p < 0.05 deemed significant.
3.4. Patient Access and ER Quality Indicator Variables by ER Volume
Descriptive and multivariable regression analyses of ER patient outcomes by volume tertiles are described in Table 4a, 4b. Margin positivity (vs. LVC: MVC OR 0.65, 95% CI 0.30–1.43; HVC OR 1.16, 95% CI 0.60–2.27), LVI (vs. LVC: MVC OR 1.26, 95% CI 0.49–3.26; HVC OR 0.74, 95% CI 0.30–1.85), and poor differentiation (vs. LVC: MVC OR 0.86, 95% CI 0.41–1.78; HVC OR 0.63, 95% CI 0.32–1.24) did not differ among institutional ER volume categories. Annual ER volume did not impact time from diagnosis to ER (vs. LVC: MVC coef −0.94, 95% CI −11.49 to 9.61; HVC coef −1.98, 95% CI −11.66 to 7.69). Patient's distance traveled to hospital was associated with hospital's ER volume for EGC, with HVCs having significantly longer travel distance compared to LVCs (coef 37.51, 95% CI 14.67–60.35).
Table 4a.
ER quality variables and patient access by ER volume tertile.
| Outcome | Low volume | Mid volume | High volume | p‐value |
|---|---|---|---|---|
| LVI % | 9.7 | 9.5 | 6.3 | 0.32 |
| Margin positivity % | 15.5 | 13.7 | 14.6 | 0.85 |
| Poor differentiation % | 18.6 | 18.1 | 15.6 | 0.69 |
| Time from diagnosis to ER, median days (IQR) | 27 (6–54) | 23 (0–55) | 30 (0–52) | 0.92 |
| Distance from patient's residence to hospital, median miles (IQR) | 21.2 (7.3–46) | 9.4 (4.5–23.5) | 25.1 (11–80.1) | < 0.01 |
Abbreviations: ER, endoscopic resection; IQR, interquartile range; LVI, lymphovascular invasion.
*p < 0.05 deemed significant.
Table 4b.
Regression comparison of ER quality variables and patient access by ER volume tertile versus the lowest volume centers.
| Outcome | Mid volume | High volume |
|---|---|---|
| LVI | 1.26 (0.48, 3.26) | 0.74 (0.30, 1.85) |
| Margin positivity | 0.65 (0.30, 1.43) | 1.16 (0.60, 2.27) |
| Poor differentiation | 0.86 (0.41, 1.78) | 0.63 (0.32, 1.24) |
| Time from diagnosis to ER | −0.94 (−11.49, 9.61) | −1.98 (−11.66, 7.69) |
| Distance from the patient's residence to the hospital | 0.04 (−24.06, 24.14) | 37.51 (14.67, 60.35) |
Note: Multivariable analysis including: age, sex, race, tumor size, tumor differentiation, LVI, margin, Charleson Deyo score, insurance, and year of diagnosis. Expressed as Odds ratio/OR in logistic regression for categorical variables (LVI, margin, poor differentiation) and expressed as a coefficient in linear regression for continuous variables (time from diagnosis, distance) with 95% confidence intervals.
Abbreviations: ER, endoscopic resection; LVI, lymphovascular invasion.
3.5. Survival and Guideline Discordance
In multivariable Cox regression, institutional ER volume for EGC did not independently predict survival (vs. LVC: MVC HR 1.32, 95% CI 0.80–2.20; HVC HR 1.02, 95% CI 0.64–1.65) (Table 5). Guideline‐discordant care, which comprised ~23% of the ER cohort, was not associated with institutional volume (vs. LVC: MVC OR 0.89, 95% CI 0.60–1.32; HVC OR 0.98, 95% CI 0.66–1.43) (Table 6). However, GD care was associated with earlier years of the study period (OR 0.71, 95% CI 0.52–0.99).
Table 5.
Multivariable cox regression for survival in ER patients.
| Factors | Multivariable | |
|---|---|---|
| HR | 95% CI | |
| Age | 1.06 | 1.04–1.09 |
| Sex | ||
| Male | Ref. | — |
| Female | 0.82 | 0.52–1.31 |
| Year of diagnosis | ||
| Early first half (2010–2015) | Ref. | — |
| Late second half (2016–2021) | 0.69 | 0.43–1.10 |
| Race | ||
| NH White | Ref. | — |
| NH Black | 1.18 | 0.48–2.91 |
| Asian | 0.65 | 0.20–2.07 |
| American Indian/others | 29.26 | 5.50–155.66 |
| Hispanic | 0.16 | 0.02–1.08 |
| Tumor size | 1.02 | 0.99–1.06 |
| Tumor differentiation | ||
| Well | Ref. | — |
| Moderate | 1.00 | 0.61–1.65 |
| Poor | 1.06 | 0.57–2.00 |
| Undifferentiated | — | — |
| Site | ||
| Cardia | Ref. | — |
| Fundus | 2.83 | 0.90–8.89 |
| Body | 0.96 | 0.39–2.34 |
| Antrum | 1.34 | 0.71–2.52 |
| Pylorus | 0.61 | 0.08–4.77 |
| Other | 0.87 | 0.32–2.37 |
| Lymphovascular invasion | 1.44 | 0.77–2.70 |
| Margin | 1.41 | 0.87–2.30 |
| Charleson‐Deyo score | ||
| 0 | Ref. | — |
| 1 | 0.61 | 0.37–1.02 |
| 2 | 1.16 | 0.56–2.37 |
| ≥ 3 | 4.36 | 1.94–9.78 |
| Insurance status | ||
| Private | Ref. | — |
| Medicaid | 2.08 | 0.63–6.91 |
| Medicare | 1.93 | 0.99–3.76 |
| Other government | 2.04 | 0.58–7.21 |
| Uninsured | 17.82 | 2.39–132.68 |
| Volume | ||
| Low | Ref. | — |
| Mid | 1.32 | 0.80–2.20 |
| High | 1.02 | 0.64–1.65 |
Abbreviations: CI, confidence interval; GR, gastric resection; HR, hazard ratio; NH, Non‐Hispanic.
*p < 0.05 deemed significant.
Table 6.
Multivariable logistic regression for predictors of guideline discordant care.
| Factors | Multivariable | |
|---|---|---|
| OR | 95% CI | |
| Age | 1.02 | 1.01–1.04 |
| Sex | ||
| Male | Ref. | — |
| Female | 1.08 | 0.74–1.56 |
| Year of diagnosis | ||
| Early first half (2010–2015) | Ref. | — |
| Late second half (2016–2021) | 0.71 | 0.52–0.99 |
| Race | ||
| NH White | Ref. | — |
| NH Black | 0.98 | 0.49–1.96 |
| Asian | 0.92 | 0.42–2.01 |
| American Indian/others | 1 | — |
| Hispanic | 0.64 | 0.25–1.65 |
| Tumor size | 1.05 | 1.02–1.08 |
| Site | ||
| Cardia | Ref. | — |
| Fundus | 1.40 | 0.43–4.53 |
| Body | 0.70 | 0.34–1.44 |
| Antrum | 0.80 | 0.43–1.49 |
| Pylorus | 0.40 | 0.05–3.30 |
| Other | 0.59 | 0.27–1.30 |
| Charleson‐Deyo score | ||
| 0 | Ref. | — |
| 1 | 1.15 | 0.76–1.73 |
| 2 | 1.45 | 0.82–2.54 |
| ≥ 3 | 1.81 | 0.94–3.49 |
| Insurance status | ||
| Private | Ref. | — |
| Medicaid | 1.18 | 0.51–2.74 |
| Medicare | 1.01 | 0.65–1.56 |
| Other government | 1.56 | 0.58–4.15 |
| Uninsured | 1.51 | 0.28–8.05 |
| Volume | ||
| Low | Ref. | — |
| Mid | 0.89 | 0.60–1.32 |
| High | 0.98 | 0.66–1.43 |
Abbreviations: OR, odds ratio; NH, Non‐Hispanic.
*p < 0.05 deemed significant.
4. Discussion
Our study demonstrates that ER is increasingly being used to treat EGC and that institutional volume of ER for EGC influences procedural utilization patterns but does not independently impact survival or guideline‐concordant care. While HVCs were more likely to utilize ER over GR and treated favorable patients that were younger with cardia‐predominant tumors, margin positivity, LVI, and survival outcomes remained comparable across volume tertiles, possibly indicating that centralization of ER for EGC may not be necessary. Guideline‐discordant ER care was associated with earlier study years, suggesting improvements in patient selection and quality of ER over time across all institutions. These findings highlight ER as a broadly viable strategy for EGC, with outcomes driven by adherence to risk stratification rather than the institution's procedural volume.
The rising adoption of ER for EGC in our findings, particularly in recent years, aligns with US and global trends. In western populations, some have reported an increase in proportional utilization of ER from ~15% to ~39% of cases from 1998 to 2014 [2]. Similarly, eastern populations that have already had strong ER implementation have had a continued increase in the proportion of ER usage for EGC in Japan from 47% to 57% from 2014 to 2021 [2, 8]. Newland et al. showed this trend to be significant across CoC‐designated institutions on a yearly level, with a ~77% annual increase in ER cases for EGC from 2010 to 2017 using the NCDB [18]. Our data shows a similar increase, with ER usage of only 42.5% in EGC cases in the earlier years of the study compared to 57.5% in later years. Regardless of center volume, both low‐ and high‐volume centers paralleled this trend in our data, with ER proportional use increasing from ~44% in 2012 to ~83% in 2018 in LVCs and similarly from ~48% in 2011 to ~88% in 2018 in HVCs. Notably, only HVCs independently predicted utilization of ER over GR, suggesting that institutional experience and volume, and by proxy academic presence, integrated tumor boards, and increased availability of specialists and resources, plays a role in the selection of ER treatment modality versus GR.
As ER utilization for EGC increases globally, centralization of ER cases is happening in the eastern cohorts. EGC treatment experience in Korea, wherein high and very HVC perform ~250 and ~540 gastric ER cases annually, and ~45% of ER centers are concentrated in the Seoul capital area [23], highlight centralization efforts. Some have advocated for centralization, demonstrating increased institutional volume of ER for EGC in Japan contributed to shorter LOS and lower complication rates, particularly for lesions of the cardia and fundus [25]. Others in Korea have also similarly reported higher rates of ER‐related adverse events in LVCs [26]. Our data does not allow for deep investigation into complication rates at LVCs, but we were able to demonstrate equivalent LOS and unplanned perioperative readmission rates among volume tertiles in the US Nonetheless, aligning with our finding, some have shown no difference in outcomes such as bleeding, complete en‐bloc resection rate, DSS, or RFS when low and HVC were compared [21, 22]. This may be due to the smaller size and invasiveness of cT1a gastric cancer resulting in a simpler endoscopic procedure achievable by interventionalist of varying levels of expertise, and that non‐EGC ER experience may contribute to flattening of the learning curve when applying ER to EGC specifically. However, comparatively, while the United States and European populations have had good success with ER, some report not always achieving the R0 and curative resection rates seen in the East [27, 28]. This may in part be due to higher volume and longer experience in Eastern proceduralists. While our data suggest that ER clinical efficacy and outcomes depend more on appropriate patient selection than procedural centralization, more investigation is warranted to assure benchmark outcomes in Western populations at LVCs. However, as ER for EGC becomes more prevalent and higher volume proceduralists and institutions separate themselves more from lower volume centers, we may see improvements and oncologic quality differences at HVCs. We surmise this may be most evident in tumors of the gastric cardia, which, while presumably more technically challenging, are likely more rewarding as it may obviate the need for a more morbid total GR. Less technically challenging ER of distal or antral EGC should be able to be safely done at LVCs, thus there may potentially be legitimacy for the implementation of selective centralization of ER for difficult, high‐risk EGC.
ER for EGC can be curative in the majority of patients while avoiding GR‐related morbidity. Some have even achieved improved long‐term survival rates with low recurrence with ER over GR, even in expanded criteria patients [29, 30]. However, with the growth in ER usage overall and in expanded criteria patients comes concerns regarding overuse or inappropriate use of ER [31, 32], which may lead to poorer outcomes if histologically incomplete resections, lack of lymph node sampling, or undertreatment of metachronous lesions are prevalent. Despite strict societal criteria for ER use in EGC [13, 14, 15, 16, 17], some have found nearly 85% of patients treated with ER were deemed inappropriate according to JGC guidelines [18]. While our data is less pessimistic with GD care seen in ~24% of ER patients, the fact remains that overzealous implementation of ER could have long‐term consequences. We observed that treatment at an HVC independently predicted receiving ER in multivariable analysis, possibly suggesting bias towards ER utilization. However, despite potential biases and preferences for ER versus GR across volume tertiles, this did not translate into institutional volume‐dependent association with GD care, meaning ER is selected appropriately across volume tertiles. Additionally, the rate of GD care is decreasing over time, suggesting increasing ER usage is being appropriately used and the desire to use ER is not clouding the decision‐making process. With appropriate patient selection, a focus on R0 resection, screening of lymph nodes in high‐risk individuals, and close post‐procedural surveillance, ER appears to be an increasingly safe alternative to GR for EGC even in lower volume institutions.
Our study has limitations inherent to retrospective NCDB analyses, including potential for selection bias, unmeasured confounders, and lack of recurrence or disease‐specific survival data. Additionally, rare clinical staging inaccuracies may exist given various staging variables provided by the NCDB (e.g., tnm_clin_stage_group, analytic_stage_group, tnm_clin_t etc.). Also, the database does not capture detailed information about the decision‐making process for each patient in order to understand the choice of ER versus GR. Additionally, our original intent in creating volume tertiles based on hospital volume (i.e., institutions with < 5 cases annually, 5–10 cases annually, etc.) was unsuccessful due to the low volume of ER for EGC seen in the NCDB, with the highest volume center only having ~3.3 cases per year. Thus, we altered our approach to create volume tertiles based on patients, potentially limiting observed differences due to tertiles having more similar case volumes; future studies with either larger datasets or as ER becomes increasingly utilized for EGC may be able to capture higher annual volume institutions for more apt analysis. Despite these constraints, our findings reinforce ER as a growing, safe, effective option for EGC when guidelines are followed.
5. Conclusion
Our analysis demonstrates that ER for cT1aN0M0 EGC achieves comparable oncologic outcomes to GR across institutional volume tertiles. While HVCs preferentially utilize ER, margin positivity, survival, and guideline discordance were volume‐independent, challenging the necessity of regional centralization. Temporal improvements in guideline adherence highlight the importance of national guidelines, standardized training, and surveillance protocols, particularly as ER use expands. To optimize outcomes, clinicians should prioritize strict adherence to established criteria (e.g., tumor size, differentiation, LVI status) and ensure structured post‐ER surveillance to mitigate recurrence risks. Future research should focus on prospective evaluation of recurrence and complication patterns, disease‐specific survival, and quality‐of‐life metrics, which are underrepresented in the current database.
Funding
The authors received no specific funding for this work.
Conflicts of Interest
The authors declare no conflicts of interest.
Synopsis
Utilization of endoscopic resections is proportionally increasing over time across institutions in the United States. We explored the impact of institutional endoscopic resection volume on guideline‐discordant usage and survival outcomes. There were no differences in guideline‐discordant endoscopic resection usage and no overall survival advantages among higher volume centers compared to low volume centers.
Data Availability Statement
The data that support the findings of this study are openly available in the American College of Surgeons National Cancer Database at https://www.facs.org/quality‐programs/cancer‐programs/national‐cancer‐database/.
References
- 1. Najmeh S., Cools‐Lartigue J., Mueller C., and Ferri L. E., “Comparing Laparoscopic to Endoscopic Resections for Early Gastric Cancer in a High Volume North American Center,” Journal of Gastrointestinal Surgery 20, no. 9 (2016): 1547–1553, 10.1007/s11605-016-3176-1. [DOI] [PubMed] [Google Scholar]
- 2. Pourmousavi M. K., Wang R., Kerdsirichairat T., et al., “Comparable Cancer‐Specific Mortality of Patients With Early Gastric Cancer Treated With Endoscopic Therapy vs Surgical Resection,” Clinical Gastroenterology and Hepatology 18, no. 12 (2020): 2824–2832, 10.1016/j.cgh.2020.04.085. [DOI] [PubMed] [Google Scholar]
- 3. Shin D. W., Hwang H. Y., and Jeon S. W., “Comparison of Endoscopic Submucosal Dissection and Surgery for Differentiated Type Early Gastric Cancer Within the Expanded Criteria,” Clinical Endoscopy 50, no. 2 (2017): 170–178, 10.5946/ce.2016.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Pyo J. H., Lee H., Min B. H., et al., “Long‐Term Outcome of Endoscopic Resection vs. Surgery for Early Gastric Cancer: A Non‐Inferiority‐Matched Cohort Study,” American Journal of Gastroenterology 111, no. 2 (2016): 240–249, 10.1038/ajg.2015.427. [DOI] [PubMed] [Google Scholar]
- 5. An L., Gaowa S., Cheng H., and Hou M., “Long‐Term Outcomes Comparison of Endoscopic Resection With Gastrectomy for Treatment of Early Gastric Cancer: A Systematic Review and Meta‐Analysis,” Frontiers in Oncology 9 (2019): 725, 10.3389/fonc.2019.00725. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Dao Q., Chen K., Zhu L., et al., “Comparison of the Clinical and Prognosis Risk Factors Between Endoscopic Resection and Radical Gastrectomy for Early‐Stage Gastric Cancer,” World Journal of Surgical Oncology 21, no. 1 (2023): 147, 10.1186/s12957-023-03018-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Libânio D., Braga V., Ferraz S., et al., “Prospective Comparative Study of Endoscopic Submucosal Dissection and Gastrectomy for Early Neoplastic Lesions Including Patients' Perspectives,” Endoscopy 51, no. 1 (2019): 30–39, 10.1055/a-0628-6601. [DOI] [PubMed] [Google Scholar]
- 8. Sako A., Yada T., Fujiya K., et al., “Clinical Epidemiology of the Endoscopic, Laparoscopic, and Surgical Resection of Malignant Gastric Tumors in Japan, 2014–2021: A Retrospective Study Using Open Data From a National Claims Database,” Gastric Cancer 28, no. 1 (2025): 1–11, 10.1007/s10120-024-01553-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Kamarajah S. K., Markar S. R., Phillips A. W., Salti G. I., and Dahdaleh F. S., “Local Endoscopic Resection Is Inferior to Gastrectomy for Early Clinical Stage T1a and T1b Gastric Adenocarcinoma: A Propensity‐Matched Study,” Annals of Surgical Oncology 28, no. 6 (2021): 2992–2998, 10.1245/s10434-020-09485-4. [DOI] [PubMed] [Google Scholar]
- 10. Nakajima T., Oda I., Gotoda T., et al., “Metachronous Gastric Cancers After Endoscopic Resection: How Effective Is Annual Endoscopic Surveillance?,” Gastric Cancer 9, no. 2 (2006): 93–98, 10.1007/s10120-006-0372-9. [DOI] [PubMed] [Google Scholar]
- 11. Meng F. S., Zhang Z. H., Wang Y. M., Lu L., Zhu J. Z., and Ji F., “Comparison of Endoscopic Resection and Gastrectomy for the Treatment of Early Gastric Cancer: A Meta‐Analysis,” Surgical Endoscopy 30, no. 9 (2016): 3673–3683, 10.1007/s00464-015-4681-0. [DOI] [PubMed] [Google Scholar]
- 12. Bestetti A. M., de Moura D. T. H., Proença I. M., et al., “Endoscopic Resection Versus Surgery in the Treatment of Early Gastric Cancer: A Systematic Review and Meta‐Analysis,” Frontiers in Oncology 12 (2022): 939244, 10.3389/fonc.2022.939244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Wang A. Y., Hwang J. H., Bhatt A., and Draganov P. V., “AGA Clinical Practice Update on Surveillance After Pathologically Curative Endoscopic Submucosal Dissection of Early Gastrointestinal Neoplasia in the United States: Commentary,” Gastroenterology 161, no. 6 (2021): 2030–2040.e1, 10.1053/j.gastro.2021.08.058. [DOI] [PubMed] [Google Scholar]
- 14. Draganov P. V., Wang A. Y., Othman M. O., and Fukami N., “AGA Institute Clinical Practice Update: Endoscopic Submucosal Dissection in the United States,” Clinical Gastroenterology and Hepatology 17, no. 1 (2019): 16–25, 10.1016/j.cgh.2018.07.041. [DOI] [PubMed] [Google Scholar]
- 15. Ajani J. A., D'Amico T. A., Bentrem D. J., et al., “Gastric Cancer, Version 2.2025, NCCN Clinical Practice Guidelines in Oncology,” Journal of the National Comprehensive Cancer Network 23, no. 5 (2025): 169–191, 10.6004/jnccn.2025.0022. [DOI] [PubMed] [Google Scholar]
- 16. Japanese Gastric Cancer Association , Japanese Gastric Cancer Treatment Guidelines 2010 (ver 3),” Gastric Cancer 14, no. 2 (2011): 113–123, 10.1007/s10120-011-0042-4. [DOI] [PubMed] [Google Scholar]
- 17. Forbes N., Elhanafi S. E., Al‐Haddad M. A., et al., “American Society for Gastrointestinal Endoscopy Guideline on Endoscopic Submucosal Dissection for the Management of Early Esophageal and Gastric Cancers: Summary and Recommendations,” Gastrointestinal Endoscopy 98, no. 3 (2023): 271–284, 10.1016/j.gie.2023.03.015. [DOI] [PubMed] [Google Scholar]
- 18. Newland J. J., Johnson A. M., Feng Z., et al., “Increasing Utilization of Endoscopic Local Excision for Early‐Stage Gastric Cancers in the United States,” Surgery 172, no. 1 (2022): 169–176, 10.1016/j.surg.2022.01.034. [DOI] [PubMed] [Google Scholar]
- 19. Suzuki H., Takizawa K., Hirasawa T., et al., “Short‐Term Outcomes of Multicenter Prospective Cohort Study of Gastric Endoscopic Resection: ‘Real‐World Evidence’ in Japan,” Digestive Endoscopy 31, no. 1 (2019): 30–39, 10.1111/den.13246. [DOI] [PubMed] [Google Scholar]
- 20. Chung I. K., Lee J. H., Lee S. H., et al., “Therapeutic Outcomes in 1000 Cases of Endoscopic Submucosal Dissection for Early Gastric Neoplasms: Korean ESD Study Group Multicenter Study,” Gastrointestinal Endoscopy 69, no. 7 (2009): 1228–1235, 10.1016/j.gie.2008.09.027. [DOI] [PubMed] [Google Scholar]
- 21. Dohi O., Hatta W., Gotoda T., et al., “Long‐Term Outcomes After Non‐Curative Endoscopic Submucosal Dissection for Early Gastric Cancer According to Hospital Volumes in Japan: A Multicenter Propensity‐Matched Analysis,” Surgical Endoscopy 33, no. 12 (2019): 4078–4088, 10.1007/s00464-019-06710-4. [DOI] [PubMed] [Google Scholar]
- 22. Hotta K., Oyama T., Akamatsu T., et al., “A Comparison of Outcomes of Endoscopic Submucosal Dissection (ESD) for Early Gastric Neoplasms Between High‐Volume and Low‐Volume Centers: Multi‐Center Retrospective Questionnaire Study Conducted by the Nagano ESD Study Group,” Internal Medicine 49, no. 4 (2010): 253–259, 10.2169/internalmedicine.49.2816. [DOI] [PubMed] [Google Scholar]
- 23. Park J. Y., Kim M. S., Kim B. J., and Kim J. G., “A 6‐Year Nationwide Population‐Based Study on the Current Status of Gastric Endoscopic Resection in Korea Using Administrative Data,” Scientific Reports 13, no. 1 (2023): 7203, 10.1038/s41598-023-34215-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Liu Y., Wen H., Wang Q., and Du S., “Research Trends in Endoscopic Applications in Early Gastric Cancer: A Bibliometric Analysis of Studies Published From 2012 to 2022,” Frontiers in Oncology 13 (2023): 1124498, 10.3389/fonc.2023.1124498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Murata A., Okamoto K., Muramatsu K., and Matsuda S., “Endoscopic Submucosal Dissection for Gastric Cancer: The Influence of Hospital Volume on Complications and Length of Stay,” Surgical Endoscopy 28, no. 4 (2014): 1298–1306, 10.1007/s00464-013-3326-4. [DOI] [PubMed] [Google Scholar]
- 26. Park J. Y., Kim M. S., Kim B. J., and Kim J. G., “Influence of Procedural Volume on the Outcome of Gastric Endoscopic Submucosal Dissection: A Nationwide Population‐Based Study Using Administrative Data,” Gastroenterology 166, no. 4 (2024): 680–689.e4, 10.1053/j.gastro.2023.12.010. [DOI] [PubMed] [Google Scholar]
- 27. Ngamruengphong S., Ferri L., Aihara H., et al., “Efficacy of Endoscopic Submucosal Dissection for Superficial Gastric Neoplasia in a Large Cohort in North America,” Clinical Gastroenterology and Hepatology 19, no. 8 (2021): 1611–1619, 10.1016/j.cgh.2020.06.023. [DOI] [PubMed] [Google Scholar]
- 28. Manta R., Galloro G., Pugliese F., et al., “Endoscopic Submucosal Dissection of Gastric Neoplastic Lesions: An Italian, Multicenter Study,” Journal of Clinical Medicine 9, no. 3 (2020): 737, 10.3390/jcm9030737. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Fukunaga S., Nagami Y., Shiba M., et al., “Long‐Term Prognosis of Expanded‐Indication Differentiated‐Type Early Gastric Cancer Treated With Endoscopic Submucosal Dissection or Surgery Using Propensity Score Analysis,” Gastrointestinal Endoscopy 85, no. 1 (2017): 143–152, 10.1016/j.gie.2016.06.049. [DOI] [PubMed] [Google Scholar]
- 30. Tanabe S., Ishido K., Matsumoto T., et al., “Long‐Term Outcomes of Endoscopic Submucosal Dissection for Early Gastric Cancer: A Multicenter Collaborative Study,” Gastric Cancer 20, no. Suppl 1 (2017): 45–52, 10.1007/s10120-016-0664-7. [DOI] [PubMed] [Google Scholar]
- 31. Choi S., Song J. H., Lee S., et al., “Lymphovascular Invasion: Traditional but Vital and Sensible Prognostic Factor in Early Gastric Cancer,” Annals of Surgical Oncology 28, no. 13 (2021): 8928–8935, 10.1245/s10434-021-10224-6. [DOI] [PubMed] [Google Scholar]
- 32. Al‐Haddad M. A., Elhanafi S. E., Forbes N., et al., “American Society for Gastrointestinal Endoscopy Guideline on Endoscopic Submucosal Dissection for the Management of Early Esophageal and Gastric Cancers: Methodology and Review of Evidence,” Gastrointestinal Endoscopy 98, no. 3 (2023): 285–305.e38, 10.1016/j.gie.2023.03.030. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are openly available in the American College of Surgeons National Cancer Database at https://www.facs.org/quality‐programs/cancer‐programs/national‐cancer‐database/.