Endoscopic ultrasound in early gastric cancer: Diagnostic accuracy and Tis/T1a lesion over staging risks

Abstract

BACKGROUND

Accurate T-staging of early gastric cancer (EGC) is critical for selecting the most appropriate treatment strategy. Although the reported diagnostic performance of endoscopic ultrasound (EUS) in determining T stage varies among studies, it continues to serve as an important tool in the pretreatment assessment of EGC.

AIM

To identify clinicopathological factors that affect the diagnostic accuracy of EUS in EGC, with particular emphasis on mucosal lesions (Tis/T1a).

METHODS

We included EGC patients who underwent endoscopic resection or gastrectomy. The diagnostic accuracy of EUS was assessed by comparing its T-stage evaluation with the histopathological results of the resected specimens. Additionally, potential risk factors associated with the over staging of Tis/T1a lesions by EUS were investigated.

RESULTS

A total of 209 patients were included in this study. The overall diagnostic accuracy of EUS for T staging was 74.64%. However, the sensitivity of EUS in identifying Tis/T1a gastric cancers was relatively low at 41.25%. Risk factor analysis showed that younger age (P = 0.035) and the presence of ulceration (P < 0.001) were significantly associated with over staging of Tis/T1a lesions. In multivariate analysis, ulceration remained the only independent predictor of over staging, with an odds ratio of 15.25 (95% confidence interval: 3.23-71.98; P < 0.001).

CONCLUSION

The accuracy of EUS for staging Tis/T1a early gastric cancer is markedly reduced when gastric ulceration is present. Therefore, EUS-based staging should be interpreted cautiously in patients with ulcer-related lesions.

Key Words: Endoscopic ultrasound; Early gastric cancer; Tumor staging; Over staging; Diagnostic accuracy

Core Tip: This study demonstrates that gastric ulceration is a major independent risk factor contributing to the over staging of Tis/T1a early gastric cancer by endoscopic ultrasound (EUS). The presence of ulceration markedly reduces the sensitivity of EUS (41.25%) and underscores the need for careful interpretation of EUS findings in ulcer-associated lesions to avoid unnecessary or inappropriate overtreatment.

INTRODUCTION

Early gastric cancer (EGC) is defined as a carcinoma confined to the mucosa or submucosa, irrespective of lymph node status. This early stage offers a pivotal opportunity for curative treatment, making accurate assessment of tumor invasion depth (T staging) fundamental to therapeutic planning[1-3]. Therefore, distinguishing between T1a (mucosal) and T1b (submucosal) disease is especially crucial, as these categories correspond to two markedly different treatment strategies. T1a lesions are generally suitable for minimally invasive, organ-preserving endoscopic procedures such as endoscopic mucosal resection (EMR) or endoscopic submucosal dissection (ESD)[4]. Conversely, T1b tumors typically warrant radical gastrectomy with lymph node dissection due to their significantly increased risk of lymph node metastasis. Errors in staging can have serious implications: Over staging a T1a lesion may lead to unnecessary surgical intervention, whereas under staging a T1b tumor risks inadequate treatment and potential recurrence[1]. Hence, precise T staging delineates the critical threshold between organ-preserving therapy and more extensive surgical management.

Endoscopic ultrasound (EUS) is widely regarded as the preferred pre-treatment imaging modality for assessing the depth of invasion in EGC[3,5]. Its ability to provide high-resolution, layer-by-layer visualization of the gastric wall offers a distinct advantage over cross-sectional imaging techniques such as multidetector computed tomography (MDCT) or magnetic resonance imaging, which lack the spatial resolution necessary to reliably differentiate between mucosal and submucosal differentiation[6]. Despite these strengths, the reported diagnostic accuracy of EUS remains highly variable, ranging from 65% to 90% across published studies, underscoring substantial inter-study heterogeneity[3,7-10]. Most existing research has concentrated on improving the detection of submucosal invasion (T1b/submucosal invasion), an area in which EUS generally demonstrates strong performance[11-13]. In contrast, systematic evaluation of its accuracy for strictly mucosal (T1a/mucosal) lesions is comparatively limited and the available evidence remains inconsistent. This gap carries significant clinical implications, as both under staging of T1b tumors and over staging of T1a tumors can lead to inappropriate management either under-treatment of cancers with higher metastatic risk or unnecessary surgical intervention in patients who could otherwise be cured endoscopically. Since endoscopic resection techniques such as ESD or EMR are recommended for T1a lesions, whereas radical gastrectomy is typically required for T1b disease, over staging of T1a cancers by EUS may directly result in avoidable, invasive surgery. Consequently, identifying clinicopathological factors that predispose to T1a over staging is critical for improving staging accuracy, guiding more judicious clinical decision-making, and minimizing the risk of overtreatment.

Therefore, the present study seeks to assess the diagnostic performance of EUS in distinguishing Tis/T1a (mucosal) lesions from T1b (submucosal) EGCs, using histopathological evaluation of resected specimens as the definitive reference standard. In addition, we aim to identify clinicopathological factors that contribute to inaccuracies in EUS-based staging, with a particular focus on the propensity of EUS to over stage Tis/T1a tumors. However, unlike previous studies, which have predominantly centered on improving the detection of submucosal invasion, our investigation shifts the emphasis toward understanding why ostensibly superficial lesions are frequently misclassified as deeper tumors by EUS. Moreover, by examining variables such as patient demographics, tumor morphology, and the presence of surface changes, we highlight the notably confounding effect of gastric ulceration, a feature that may obscure normal wall layer structures and mimic deeper invasion on ultrasonographic imaging. Overall, through this targeted approach, our study aims to clarify the specific factors that compromise EUS accuracy in mucosal cancers and to inform more cautious interpretation of EUS findings in clinical decision-making.

MATERIALS AND METHODS

We conducted a retrospective analysis of patients with biopsy-confirmed gastric adenocarcinoma who underwent EUS-based staging at Sun Yat-sen University Cancer Center between January 2015 and December 2022. The study protocol was approved by the Institutional Review Board and adhered to the ethical standards of the 1964 Declaration of Helsinki and its subsequent revisions.

Patients

Eligible patients had histologically confirmed EGC, defined as tumor invasion confined to the mucosa or submucosa. All patients underwent curative-intent treatment, either through endoscopic resection or gastrectomy, with additional gastrectomy performed when endoscopic resection was deemed incomplete. Before treatment, each patient received a standardized staging evaluation that included both EUS and MDCT. Patients were excluded if final histopathology demonstrated invasion into the muscularis propria.

Procedure and endoscopic ultrasonography staging

Endoscopic ultrasonography examinations were performed by experienced endoscopists with at least three years of dedicated EUS experience. Radial ultrasound probes operating at frequencies of 5-12 MHz were used (Olympus EU-ME2, Tokyo, Japan; Fujifilm SU-9000, Tokyo, Japan). Patients were premedicated with dyclonine, dimeticone, and a protease preparation. EUS was conducted either in the awake state or under intravenous sedation. After aspiration of intragastric air, the stomach was filled with deaerated water to facilitate optimal acoustic coupling and visualization.

On EUS, the normal gastric wall appears as five alternating echogenic layers: The mucosa (hyper-/hypoechoic; layers 1-2), submucosa (hyperechoic; layer 3), muscularis propria (hypoechoic; layer 4), and subserosa/serosa (hyperechoic; layer 5)[14]. Tumor invasion was identified based on focal or diffuse wall thickening, disruption of the normal layered architecture, or irregularity of the echogenic pattern. T staging followed the American Joint Committee on Cancer 8^th edition criteria, with T1a defined as invasion limited to layers 1-2, T1b as involvement of the submucosa (layer 3), T2 as invasion of the muscularis propria (layer 4), T3 as penetration into the sub-serosal layer, T4a as serosal involvement, and T4b as invasion of adjacent structures.

Histopathology

Histopathological assessment of the resected specimens was carried out on serial sections cut at 2 mm intervals and stained with hematoxylin and eosin. An experienced pathologist reviewed all samples. Tumors showing well- or moderately differentiated tubular or papillary adenocarcinoma were classified as differentiated, whereas poorly differentiated tubular adenocarcinomas and signet-ring cell carcinomas were categorized as undifferentiated.

Statistical analysis

EUS T-staging performance was evaluated using histopathology as the reference standard. Continuous variables were summarized as medians with interquartile ranges (IQR), while categorical variables were reported as frequencies and percentages. Diagnostic metrics including sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy were calculated based on pathology-confirmed T stage. Comparisons between groups were conducted using χ² or Fisher’s exact tests for categorical variables and t-tests or Kruskal-Wallis tests for continuous variables, as appropriate. Risk factors associated with over staging of Tis/T1a lesions were examined using univariate and multivariate logistic regression analyses, with results expressed as adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs). Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., United States), and a P value < 0.05 was considered statistically significant.

RESULTS

Baseline characteristics

A total of 209 patients were included in the final analysis. The median age was 58.00 years (IQR: 50.00-66.00), and 56.94% of the cohort were male. The median tumor diameter measured 2.00 cm (IQR: 1.50-3.00). Ulceration was present in 155 patients (74.16%). Among the 80 patients with pathologically confirmed Tis/T1a lesions, 56 (70.00%) exhibited ulceration. Nineteen patients (23.80%) were younger than 50 years, and 52 (65.00%) had a body mass index (BMI) below 24 kg/m². According to final histopathological grading, 33.97% (71/209) of tumors were well or moderately differentiated, while 66.03% (138/209) were poorly differentiated. The demographic and clinicopathological characteristics of all patients are summarized in Table 1.

Table 1 The characteristics of 209 patients with early gastric cancer, n (%).

Characteristics
Age, year, median (IQR)	58.00 (50.00-66.00)
BMI, kg/m2, median (IQR)	22.01 (19.88-24.06)
Size, cm, median (IQR)	2.00 (1.50-3.00)
Male	119 (56.94)
Ulcer
Absent	54 (25.84)
Present	155 (74.16)
Location
Upper	26 (12.56)
Middle	65 (31.40)
Lower	116 (56.04)
Differentiation
Well/moderate	71 (33.97)
Poorly	138 (66.03)
Depth of invasion
Tis/T1a	80 (38.28)
T1b	129 (61.72)
Lymph node metastasis	51 (24.40)
Lymph vascular invasion	21 (10.05)
Perineural invasion	4 (1.91)

IQR: Interquartile range; BMI: Body mass index.

Accuracy rates of EUS for T staging of EGC

EUS demonstrated a moderate overall accuracy of 74.64% (156/209) for T staging of EGC when benchmarked against histopathological findings. Notably, its performance differed substantially between superficial and deeper lesions. The sensitivity for detecting Tis/T1a tumors was relatively low at 41.25% (33/80), reflecting a considerable tendency to miss mucosal-confined disease. In contrast, EUS showed high specificity (95.35%, 123/129) and a strong positive predictive value (84.62%, 33/39) for identifying Tis/T1a lesions. A detailed summary of diagnostic performance parameters is presented in Table 2.

Table 2 Diagnostic value of endoscopic ultrasonography for T staging early gastric cancer.

EUS diagnosis	Histology diagnosis		Total
	Tis/T1a	T1b
Tis/T1a	33	6	39
T1b or deeper	47	123	170
Total	80	129	209

Tis/T1a sensitivity = 33/(33 + 47) × 100% = 41.25%. Tis/T1a specificity = 123/(6 + 123) × 100% = 95.35%. Tis/T1a positive predictive value = 33/(33 + 6) × 100% = 84.62%. Tis/T1a negative predictive value = 123/(47 + 123) × 100% = 72.35%. Overall accuracy = (33 + 123)/(33 + 6 + 47 + 123) × 100% = 74.64%. EUS: Endoscopic ultrasound.

Clinical factors affecting the overestimation of Tis/T1a stage by endoscopic ultrasonography

In the subgroup of 80 patients with pathologically confirmed Tis/T1a EGC, substantial variation was observed in the accuracy of EUS staging. Only 33 lesions (41.25%) were correctly classified, whereas 47 lesions (58.75%) were over staged. Comparative analysis identified ulceration as the most prominent factor associated with over staging: 89.4% (42/47) of over staged lesions demonstrated ulceration, compared with 54.5% (18/33) of accurately staged lesions (P < 0.001). Younger age (< 50 years) was also significantly associated with over staging (40.43% vs 18.18%, P = 0.035). In contrast, no significant differences were observed between the two groups with respect to sex, BMI, tumor size, tumor location, or histological differentiation (all P > 0.05). These findings are summarized in Table 3.

Table 3 Factors associated with overestimation of Tis/T1a early gastric cancer by endoscopic ultrasonography, n (%).

	Accuracy (n = 33)	Overestimation (n = 47)	P value
Age, year
< 50	6 (18.18)	19 (40.43)	0.035
≥ 50	27 (81.82)	28 (59.57)
Gender
Male	13 (39.39)	24 (51.06)	0.365
Female	20 (60.61)	23 (48.94)
BMI, kg/m2
< 24	23 (69.70)	38 (80.85)	0.249
≥ 24	10 (30.30)	9 (19.15)
Size, cm
< 3	22 (66.67)	31 (65.96)	0.947
≥ 3	11 (33.33)	16 (34.04)
Ulcer finding
Absent	15 (45.45)	5 (10.64)	< 0.001
Present	18 (55.55)	42 (89.36)
Location
Upper	3 (9.09)	3 (6.38)	0.626
Middle	12 (36.36)	13 (27.66)
Lower	8 (24.24)	31 (65.96)
Differentiation
Well/moderate	14 (42.42)	21 (44.68)	1.000
Poorly	19 (57.58)	26 (55.31)

BMI: Body mass index.

Logistic regression analysis identified ulceration as the strongest independent predictor of EUS over staging in Tis/T1a EGC (Table 4). In the multivariate model, ulcerated lesions demonstrated a markedly elevated risk of over staging, with a 15.25-fold increase compared with non-ulcerated lesions (adjusted OR = 15.25; 95%CI: 3.23-71.98; P < 0.001). Younger age (< 50 years) was also significantly associated with over staging (OR = 6.24; 95%CI: 1.31-29.63; P = 0.021), and a lower BMI (< 24 kg/m²) increased the risk by 4.38-fold (OR = 4.38; 95%CI: 1.06-18.18; P = 0.042). In contrast, tumor size, location, sex, and histological differentiation were not significant predictors in the final multivariate model (all P > 0.05).

Table 4 Univariate and multivariate analysis for factors influencing overestimation of Tis/T1a early gastric cancer.

	Univariate analysis		Multivariate analysis
	OR (95%CI)	P value	OR (95%CI)	P value
Age, year
≥ 50	1.00		1.00
< 50	3.05 (1.06-8.81)	0.039	6.24 (1.31-29.63)	0.021
Gender
Male	1.00		1.00
Female	0.62 (0.25-1.54)	0.304	0.37 (0.11-1.17)	0.090
BMI, kg/m2
≥ 24	1.00		1.00
< 24	1.84 (0.65-5.19)	0.252	4.38 (1.06-18.18)	0.042
Size, cm
< 3	1.00		1.00
≥ 3	1.03 (0.40-2.65)	0.947	1.21 (0.35-4.19)	0.759
Ulcer
Absent	1.00		1.00
Present	7.00 (2.21-22.17)	< 0.001	15.25 (3.23-71.98)	< 0.001
Location
Upper	1.00		1.00
Middle	1.08 (0.18-6.44)	0.745	0.36 (0.03-4.63)	0.175
Lower	1.72 (0.31-9.45)	0.354	1.48 (0.16-13.38)	0.193
Differentiation
Well/moderate	1.00		1.00
Poorly	0.91 (0.37-2.24)	0.841	0.32 (0.08-1.24)	0.100

BMI: Body mass index; OR: Odds ratio; CI: Confidence interval.

DISCUSSION

Our study highlights several important findings regarding the performance of EUS in staging EGC. Although the overall accuracy of EUS was moderate (74.64%), its diagnostic capability differed substantially according to tumor depth. Sensitivity for detecting Tis/T1a lesions was notably low at 41.25%, contributing to a high over staging rate of 58.75% among pathologically confirmed mucosal cancers. Multivariate analysis further revealed that ulceration was the strongest independent predictor of over staging, with younger age (< 50 years) and lower BMI (< 24 kg/m²) also significantly associated with increased misclassification.

The overall diagnostic accuracy of EUS observed in our cohort aligns with the broad range reported in the literature (65%-92.1%)[5,9,10,15-19]. However, the sensitivity for detecting Tis/T1a lesions in our study was markedly lower than many previous reports. This discrepancy is likely related to the characteristics of our study population, particularly the high proportion of ulcerated lesions (74.16%). Ulceration is a well-documented factor that negatively affects EUS accuracy, and many earlier studies, especially those emphasizing the detection of submucosal invasion (T1b) may have included fewer ulcerated early cancers, resulting in higher reported overall accuracy[8,20,21]. Our findings highlight that global accuracy metrics can obscure substantial limitations in the assessment of mucosal cancers, especially in populations with a high prevalence of confounding features such as ulceration.

The pathophysiological basis for the identified risk factors is also biologically plausible. The pronounced impact of ulceration on over staging (adjusted OR = 15.25) likely stems from severe distortion of the normal gastric wall architecture. Active ulcers induce intense inflammation, edema, and fibrotic remodeling in the subepithelial layers, producing a hypoechoic thickened area that disrupts the typical layer pattern seen on EUS. This distortion makes it difficult to distinguish inflammatory fibrosis from true submucosal tumor invasion[22-24]. Furthermore, ulcer-related fibrosis and peritumoral inflammation frequently present as hypoechoic regions on EUS, further compromising image clarity and interpretive accuracy[8,25]. Younger age may serve as a surrogate marker for increased gastric wall tension or a more robust inflammatory and fibrotic response to mucosal injury, both of which could exaggerate apparent wall thickening on EUS and contribute to over staging. Similarly, a lower BMI may reduce peri-gastric adipose tissue, diminishing the hyperechoic acoustic interface that facilitates clear visualization of gastric wall layers. This loss of contrast can impair the ability to distinguish individual layers, thereby increasing the likelihood of misclassification.

Several limitations of this study merit consideration. First, the retrospective, single-center design may introduce selection bias and limit the broader applicability of the findings. Second, all EUS examinations were conducted by highly experienced endoscopists, which may not reflect performance outcomes in centers with less experienced operators and thus may reduce generalizability. Third, inter-observer variability in EUS interpretation was not evaluated, representing a potential source of diagnostic inconsistency. Future research should incorporate multicenter, prospective designs, include endoscopists with varying levels of expertise, and formally assess inter-observer agreement to strengthen external validity. Despite these limitations, the clinical implications of our results are both immediate and significant. Our study demonstrates that although EUS shows excellent specificity for confirming deeper submucosal invasion, its sensitivity for excluding invasion in Tis/T1a lesions remains markedly limited particularly in the presence of ulceration. This underscores the importance of cautious interpretation of EUS findings when evaluating superficially invasive, ulcerated EGC.

CONCLUSION

The diagnostic reliability of EUS for assessing Tis/T1a EGC is markedly reduced in the presence of gastric ulceration. As a result, EUS impressions suggestive of submucosal (T1b) invasion should be interpreted with particular caution when evaluating lesions that exhibit high-risk features, most notably ulceration, younger patient age, and lower BMI. In these contexts, sole reliance on EUS is inappropriate and may lead to overtreatment. To improve staging accuracy, clinical decision-making must incorporate a more comprehensive diagnostic approach. Magnifying endoscopy with narrow-band imaging provides enhanced visualization of surface pit patterns and microvascular architecture, enabling more precise differentiation between mucosal and submucosal involvement. Such optical evaluation is especially valuable when EUS findings are equivocal or potentially confounded by ulcer-related structural distortion. Thus, when a lesion appears endoscopically resectable but uncertainty regarding invasion depth persists, diagnostic endoscopic resection remains the ultimate allocation, but also minimizes the risk of unnecessary radical gastrectomy in patients who may be adequately treated with organ-preserving endoscopic techniques.