Comprehensive Computed Tomography Evaluation of Gastric Cancer: 2026 Consensus Recommendation From the Korean Society of Abdominal Radiology

Abstract

Gastric cancer continues to be a significant global health challenge, particularly in East Asia, necessitating precise imaging for optimal management. Although multidetector computed tomography (MDCT) has become the primary tool for staging and response evaluation, clinical practice is hampered by a lack of standardized protocols and interpretation criteria. To address these needs, the Korean Society of Abdominal Radiology established a consensus-based guideline for gastric cancer CT. Using a modified Delphi process with a panel of experts, 10 key statements were developed and validated with a high level of agreement (≥80%). These recommendations emphasize the necessity of adequate gastric distention and a dedicated arterial phase to identify critical vascular variations, such as aberrant left hepatic arteries, for surgical planning. The guideline provides refined criteria for T, N, and M staging, including the use of multiplanar reformations and three-dimensional visualization to improve diagnostic accuracy. Furthermore, it introduces dedicated criteria for re-staging after chemotherapy, highlighting the differentiation of viable tumor from treatment-induced fibrosis. Finally, a risk-stratified approach for postoperative surveillance is proposed, suggesting that surveillance may be de-escalated in very-low-risk groups, such as pT1aN0 disease. These recommendations are intended to standardize CT practice and improve clinical decision-making in the management of gastric cancer.

INTRODUCTION

Gastric cancer remains a major global health burden, ranking as the fifth most commonly diagnosed malignancy and the fifth leading cause of cancer-related mortality worldwide [1]. In East Asian countries such as Korea and Japan, nationwide screening programs have led to a substantial increase in the detection of early gastric cancer (EGC), with more than half of surgically treated cases diagnosed at an early stage [2,3]. While this epidemiologic shift has improved patient prognosis, it has also introduced new challenges in imaging-based diagnosis, staging, and treatment planning.

Computed tomography (CT) has traditionally been regarded as a staging modality for gastric cancer, especially for advanced gastric cancer (AGC) [4,5,6]. However, recent advances in multidetector CT (MDCT) including isotropic volumetric acquisition and three-dimensional (3D) post-processing techniques have substantially expanded the role of CT across the entire disease spectrum of gastric cancer [7,8,9,10,11,12]. CT is now increasingly relied upon not only for TNM staging, but also for precise tumor localization, assessment of vascular anatomy relevant to surgical planning, evaluation of treatment response to chemotherapy, re-staging after neoadjuvant or palliative treatment, and surveillance following curative resection [6].

Despite this expanding clinical role, there is a lack of standardized and comprehensive imaging guidelines that define optimal CT acquisition protocols, interpretation criteria, and reporting elements specific to gastric cancer. Existing recommendations are often fragmented, regionally heterogeneous, or focused on limited aspects of staging, leading to substantial interobserver variability and inconsistency in clinical decision-making [13,14,15,16]. In particular, interpretation of early-stage disease, lymph node (LN) metastasis, peritoneal seeding, post-chemotherapy changes, and vascular anatomic variations remains highly dependent on individual experience rather than evidence-based consensus.

To address these unmet needs, the Korean Society of Abdominal Radiology (KSAR) convened a panel of expert radiologists specializing in gastric cancer imaging to develop a consensus-based guideline for comprehensive CT evaluation of gastric cancer. This document represents the official position of KSAR and aims to provide practical, evidence-based, and clinically actionable recommendations that can be readily implemented in daily practice. The guideline focuses on standardized CT protocols and interpretation strategies for initial diagnosis, TNM staging, evaluation of treatment response, re-staging after chemotherapy, assessment of vascular anatomy, and imaging surveillance after curative surgery.

ETHICAL REVIEW AND APPROVAL PROCESS

This document is a consensus recommendation developed through a structured expert-panel process using a modified Delphi method, and did not involve prospective acquisition of patient data or any new investigation involving human participants. Accordingly, formal Institutional Review Board approval for the consensus development was not required. The illustrative figures presented in this manuscript are fully de-identified retrospective clinical images selected from the routine practice of the participating institutions, used solely for educational illustration of the recommendation statements, and contain no patient-identifiable information.

ORGANIZATION OF THE EXPERT STUDY GROUP

The KSAR study group for gastric cancer CT was composed of nine board-certified abdominal radiologists from seven academic institutions in South Korea. All members were KSAR members with expertise in abdominal CT interpretation, including gastric cancer imaging using CT (median year of experience: 14 years; range of experience: 7–24 years). Each member was assigned to one or more of the following predefined topics: gastric distention, T staging, N staging, M staging, arterial anatomic variation, treatment response evaluation, re-staging after chemotherapy, and postoperative surveillance. All study group members participated in two structured online consensus meetings to discuss the scope, structure, and key contents of the gastric cancer CT imaging guideline. For each predefined topic, one or more candidate statements were developed and subsequently subjected to a modified Delphi consensus process. To ensure high methodological quality, the consensus process was structured following the framework of the Appraisal of Guidelines for Research and Evaluation (AGREE) II instrument.

LITERATURE REVIEW

A literature search was conducted using MEDLINE (PubMed) to identify relevant articles for each predefined topic. Members assigned to each topic performed the literature review, developed candidate statements, analyzed the voting results, and drafted the corresponding sections of the recommendations. The literature was reviewed with a focus on CT acquisition protocols, interpretation criteria, diagnostic performance, and clinical outcomes related to gastric cancer imaging.

DEVELOPMENT OF STATEMENTS

Initially, 11 candidate statements were developed, and they were presented and discussed during the two online consensus meetings. Duplicated or redundant items were merged or removed, and items deemed essential for the CT evaluation of gastric cancer were retained. After iterative refinement, 10 consensus statements were finalized. When a single standardized reporting statement could not be formulated, a multiple-choice question format was adopted to capture expert preference.

AGREEMENT VOTING

To enhance objectivity and external validity, an additional panel of 12 gastrointestinal radiologists from 10 tertiary academic medical centers and one affiliated healthcare center in South Korea, who were not involved in the initial statement development but were active members of the KSAR, was recruited for the Delphi voting. The agreement levels reported in this guideline are based on the votes of this 12-member external Delphi panel. All voters had more than three years of experience in abdominal CT interpretation (median year of experience: 12 years; range of experience: 3–26 years). The nine KSAR study group members also participated in the Delphi voting; the combined agreement levels from all 21 participants are provided in Supplementary Table 1 as complementary results.

The modified Delphi voting used a six-point Likert scale (strongly agree, agree with minor reservation, agree with major reservation, disagree with minor reservation, disagree with major reservation, and strongly disagree). A priori consensus was defined as ≥80% of responses falling within “strongly agree” or “agree with minor reservation.”

For statements in which the level of agreement narrowly missed the 80% threshold during the first round, or when one option in a multiple-choice question was selected by a proportion close to the consensus level, a pre-planned second-round modified Delphi voting was conducted. Prior to re-voting, an in-depth discussion addressing the advantages, limitations, and potential clinical implications of the relevant items was held. Following this iterative process, consensus was successfully achieved.

KSAR APPROVAL OF CONSENSUS RECOMMENDATIONS

The draft recommendations were further discussed at the KSAR Meeting held on May 26, 2023, and the feedback obtained during this meeting was incorporated into the final recommendations whenever appropriate. After completion of the consensus process and manuscript revision, the final document was formally reviewed by the approval committee and approved by the KSAR before submission.

CONSENSUS RESULTS

All 10 finalized statements exceeded the predefined 80% consensus threshold, with agreement levels ranging from 91.7% to 100% (Table 1). This distribution indicates strong expert agreement and supports the clinical applicability and scientific validity of the proposed statements.

Table 1. Consensus recommendation statements and agreement levels.

Recommendation statements		Agreement level*
Gastric distention
	Statement 1: Adequate gastric distention is essential for the CT evaluation of gastric cancer. Both CO2 gas and water are acceptable distention agents; however, CO2 gas distention is required for 3D reconstruction including virtual gastroscopy	100%
T staging
	Statement 2: On MDCT, T1a gastric cancer may be suggested when abnormal enhancement or thickening is confined to the innermost enhancing layer of the gastric wall with preservation of the low-attenuation submucosal layer, or when no definite lesion is detectable on both 2D and 3D CT images at the site corresponding to endoscopically proven cancer	91.7%
	Statement 3: MDCT is effective for differentiating ≤T4a from T4b gastric cancer and plays a pivotal role in assessing tumor resectability	100%
N staging
	Statement 4: For N staging of gastric cancer, identifying distant LN metastasis in retropancreatic, \pancreaticoduodenal, peripancreatic, superior mesenteric, middle colic, and para-aortic areas is essential for determining resectability and treatment strategy	100%
	Statement 5: A flexible size cut-off (short-axis diameter of 6–8 mm) should be applied based on the LN location. Ancillary features, including strong enhancement, a high short-to-long diameter ratio (round shape), presence of necrosis, and clustering of small LNs, should also be integrated to improve diagnostic reliability	91.7%
M staging
	Statement 6: Minimal pelvic ascites detected on preoperative CT without other suggestive findings of peritoneal seeding should not be considered a definitive contraindication for curative radical resection	100%
Arterial anatomic variation
	Statement 7: When an aberrant LHA is present, its presence, subtype (replaced or accessory), and arterial diameter should be explicitly described on preoperative CT	91.7%
Treatment response evaluation after chemotherapy
	Statement 8: For the evaluation of therapeutic response in gastric cancer following chemotherapy, CT volumetry can serve as a supplementary tool to conventional size-based measurements. While functional and metabolic modalities such as PET/CT, PET/MRI, and perfusion CT show promise for predicting early response, further validation is required for their routine clinical application	100%
	Statement 9: Standard MDCT staging criteria may be inaccurate for re-staging gastric cancer following neoadjuvant or palliative chemotherapy. Therefore, dedicated MDCT criteria that account for post-treatment tissue changes should be applied to accurately assess re-staging and resectability before conversion surgery	91.7%
Imaging surveillance after curative resection
	Statement 10: Postoperative imaging surveillance following curative resection of gastric cancer should be individualized based on the risk of recurrence. A risk-stratified approach is recommended to optimize the clinical benefit while minimizing unnecessary radiation exposure and healthcare costs	91.7%

*Proportion of panelists who voted “strongly agree” or “agree with minor reservation” among the 12-member external Delphi panel. CT = computed tomography, CO₂ = carbon dioxide, 3D = three-dimensional, MDCT = multidetector computed tomography, 2D = two-dimensional, LN = lymph node, LHA = left hepatic artery, PET = positron emission tomography, MRI = magnetic resonance imaging

Overall, this structured consensus methodology, including external Delphi voting, discussion at the KSAR Meeting, and formal society approval, supports the reliability and clinical applicability of the final recommendations.

Gastric Distention

Statement 1: Adequate gastric distention is essential for the CT evaluation of gastric cancer. Both carbon dioxide (CO₂) gas and water are acceptable distention agents; however, CO₂ gas distention is required for 3D reconstruction including virtual gastroscopy (agreement level: 100%).

The fundamental principle of gastric CT is achieving adequate luminal distention. Incomplete distention can result in pseudo-thickening of the gastric wall, which frequently leads to both false-positive and false-negative diagnoses as well as to inaccurate tumor staging [7,17,18]. Use of either carbon dioxide (CO₂) gas or water to achieve sufficient luminal expansion is mandatory, as non-distended stomach CT may lead to missed lesions, inaccurate T staging, and unreliable interpretation (Fig. 1) [7,17,18].

Fig. 1. Clinical significance of adequate gastric distention for lesion detection. A-D: A 43-year-old male with a gastric lesion incidentally identified during screening gastroscopy. Initial axial (A) and coronal (B) portal phase CT images obtained in a collapsed state of the stomach. Due to inadequate luminal distention, the gastric wall appears pseudo-thickened, and the lesion is not demonstrable. Subsequent axial (C) and coronal (D) portal phase CT images following the administration of two packs of effervescent agents show excellent luminal expansion. A well-defined, ovoid, and homogeneously enhancing subepithelial lesion (arrows) is clearly visualized at the posterior wall of the distended gastric body. This comparison underscores that sufficient gastric distention is mandatory to avoid missed lesions and ensure reliable radiological interpretation. CT = computed tomography.

From both diagnostic and clinical perspectives, the choice between CO₂ gas and water is influenced by patient compliance, institutional settings, and diagnostic objectives. In terms of diagnostic performance, CT gastrography with CO₂ distention, facilitated by 3D reconstruction and virtual gastroscopy, has been reported to increase the detection rate of EGC to as high as 98% [7,10]. Furthermore, gas-distended 3D CT images are strongly preferred by surgeons as they provide an essential “preoperative road map.” These images offer a comprehensive spatial overview of the tumor location in relation to anatomical landmarks and anticipated resection margins, which is critical for surgical planning (Fig. 2) [9,19,20]. Reflecting these clinical advantages, approximately 70% of Korean hospitals or institutions currently utilize CO₂ gas as the primary oral contrast agent for dedicated stomach CT protocols.

Fig. 2. Clinical utility of 3D CT gastrography as a preoperative roadmap for surgical planning. A-D: A 58-year-old female with gastric cancer. Initial gastroscopy identified a suspected small EGC in the gastric lower body (not shown). Coronal reformatted CT image (A) obtained in the left posterior oblique position with CO₂ gas distention. In this 2D image, no definitive focal wall thickening or abnormal enhancement is identifiable. Corresponding 3D surface-shaded display image (B) clearly demonstrates two discrete, shallow ulcerative lesions accompanied by malignant fold convergence at the greater curvature of the gastric high (arrows) and lower (arrowheads) body. Based on the 3D spatial overview, the radiologist reported synchronous EGCs, leading to a critical modification of the surgical plan from distal gastrectomy to near-total gastrectomy. Subsequent second gastroscopy (C) confirms the presence of two shallow ulcerative lesions (arrows for high body lesion and arrowheads for lower body lesion) with abrupt fold changes, consistent with the 3D CT findings. Near-total gastrectomy pathologically (D) confirmed synchronous EGCs (pT1aN0 for high body lesion [arrows] and pT1bN0 for lower body lesion [arrowheads]). 3D = three-dimensional, CT = computed tomography, EGC = early gastric cancer, CO₂ = carbon dioxide, 2D = two-dimensional, ST = stomach.

Nevertheless, hydro-CT using water remains a valuable alternative in selected clinical scenarios. Water distention is particularly indicated for patients with intolerance to effervescent agents, in emergency settings, or during routine post-treatment follow-up where the primary objective is the evaluation of gross wall thickening or recurrence rather than detailed mucosal or 3D assessment.

TNM Staging on CT

Accurate TNM staging is a cornerstone in the management of gastric cancer, as it directly influences treatment strategy, surgical planning, and prognostic assessment [21]. While staging traditionally relies on a multimodality approach including endoscopic ultrasonography (EUS), magnetic resonance imaging (MRI), positron emission tomography (PET) or PET/CT, MDCT has emerged as the primary imaging modality due to rapid technological advances [6,7,8,9,10,11,12,22,23]. Modern MDCT allows simultaneous evaluation of the primary tumor, regional and distant LNs, and distant metastatic sites in a single, noninvasive examination, making it indispensable for both initial staging and treatment planning.

High-resolution isotropic datasets combined with multiplanar reformations (MPR) in axial, coronal, and sagittal planes enable more accurate assessment of tumor invasion depth, perigastric fat infiltration, and involvement of adjacent organs [24,25,26]. In addition, advanced volume-rendering techniques, including transparent and surface-rendered images, facilitate detection of subtle mucosal and submucosal abnormalities and provide a global overview of tumor extent within the stomach [9,27]. These 3D visualization tools can complement conventional 2D imaging by improving lesion conspicuity and spatial understanding, particularly in early-stage disease [9,27].

Moreover, optimized MPR images assist in differentiating metastatic LNs from small perigastric vessels and normal anatomic structures, thereby improving the reliability of N staging [25,26,28]. As a result, contemporary MDCT has demonstrated improved diagnostic performance for both T and N staging compared with earlier CT techniques, while maintaining a central role in the evaluation of M staging.

Given these advantages, MDCT plays a pivotal role in the integrated TNM staging of gastric cancer. In this guideline, CT-based assessment of T, N, and M categories is discussed in detail, with an emphasis on standardized imaging techniques, interpretation criteria, and clinically relevant pitfalls.

T Staging

Statement 2: On CT, T1a gastric cancer may be suggested when abnormal enhancement or thickening is confined to the innermost enhancing layer of the gastric wall with preservation of the low-attenuation submucosal layer, or when no definite lesion is detectable on both 2D and 3D CT images at the site corresponding to endoscopically proven cancer (agreement level: 91.7%).

Accurate T staging is important for treatment planning in gastric cancer, particularly for early-stage disease, because differentiation between T1a and ≥T1b tumors directly influences the choice between endoscopic resection and surgical gastrectomy. T1a tumors, which are confined to the mucosa, have a very low incidence of LN metastasis and are therefore suitable candidates for endoscopic submucosal dissection (ESD).

With adequate gastric distention and optimal contrast enhancement, the normal gastric wall can be visualized on MDCT as a layered structure [23]. Although exact histologic correlation of each CT layer remains debated, the innermost strongly enhancing layer is generally accepted to correspond to the mucosa [23,29,30]. Consequently, T1a gastric cancer typically manifests as focal enhancement or thickening limited to this innermost layer, with preservation of the underlying low-attenuation submucosal layer (Fig. 3) [23,31]. In contrast, disruption or obliteration of the submucosal layer suggests at least T1b disease.

Fig. 3. Multidetector CT findings of T1a gastric cancer in a 59-year-old male. A: Axial portal phase CT image obtained in the prone position using water as a distention agent (hydro-CT). The gastric lumen is well-distended, allowing clear visualization of the three distinct gastric wall layers (arrowheads). A focal, enhancing wall thickening (arrow) is localized within the innermost mucosal layer of the gastric cardia, while the underlying low-attenuation submucosal layer remains intact. B: Corresponding gastroscopic image reveals a focal shallow ulcerative lesion (arrows) at the gastric cardia. Total gastrectomy confirmed an early gastric cancer with T1a stage (invading the muscularis mucosae) (not shown). CT = computed tomography, ST = stomach.

Nevertheless, the detectability of T1a gastric cancer on CT remains inherently limited due to its superficial depth of invasion. Multiple studies have shown that T1a lesions are frequently occult on both 2D and 3D CT images or only seen on 3D images, even when dedicated stomach CT protocols are applied (Fig. 4) [31,32]. In such cases, absence of a demonstrable lesion on CT or selective visibility only on 3D images at a site confirmed by endoscopy strongly favors T1a disease rather than representing a limitation of CT interpretation (Fig. 4). Using this approach, previous studies have reported diagnostic accuracies exceeding 95% for identifying T1a gastric cancer [7,31].

Fig. 4. Enhanced detection of T1a gastric cancer using 3D virtual gastroscopy in a 57-year-old female. A: 2D axial portal phase CT image showing adequate gastric distention with CO₂. Despite the use of a dedicated stomach CT protocol, there is no demonstrable wall thickening or abnormal enhancement in the gastric body, including the posterior wall side, rendering the lesion occult on 2D axial imaging. B: Corresponding 3D virtual gastroscopic image clearly demonstrates a subtle mucosal ulceration (arrow) associated with an abrupt cut-off of adjacent gastric folds (arrowheads), characteristic of malignant fold changes. C: Gastroscopic image confirms the focal ulceration (arrow) with disrupted adjacent folds (arrowheads). Note that the 3D reconstructed image provides a near-identical representation of the mucosal surface and fold patterns compared to the actual gastroscopic view. Distal gastrectomy confirmed an early gastric cancer with T1a stage. This case illustrates that the selective visibility of a lesion only on 3D imaging at a site confirmed by endoscopy strongly favors a diagnosis of T1a stage. 3D = three-dimensional, 2D = two-dimensional, CT = computed tomography, CO₂ = carbon dioxide, ST = stomach.

EUS provides excellent spatial resolution for local tumor assessment by enabling visualization of the five histologically correlated layers of the gastric wall and allows evaluation of regional LN involvement [33,34,35,36,37]. Indeed, regarding the capacity of EUS to distinguish between T1a versus T1b cancers, the meta-analysis of 20 studies (n = 3,321) showed that the summary sensitivity and specificity were 0.87 (95% confidence interval [CI] 0.81 to 0.92) and 0.75 (95% CI 0.62 to 0.84) respectively [33]. In addition, for the N staging, the meta-analysis of 44 studies (n = 3,573) showed that the summary sensitivity and specificity were 0.83 (95% CI 0.79 to 0.87) and 0.67 (95% CI 0.61 to 0.72), respectively [33]. Nevertheless, its role is inherently limited to local and regional staging, and it is suboptimal for the assessment of extragastric disease, particularly distant lymphadenopathy, hepatic metastases, and peritoneal dissemination [38]. Furthermore, EUS is operator-dependent and may be technically challenging in cases of stenotic or diffusely infiltrative tumors. Recent meta-analyses and comparative studies have shown that MDCT’s diagnostic performance is comparable to EUS, with T staging accuracies ranging from 77% to 89% for MDCT and 65% to 92% for EUS [39]. Therefore, EUS and MDCT should be regarded as complementary rather than competing modalities for gastric cancer staging.

Statement 3: MDCT is effective for differentiating ≤T4a from T4b gastric cancer and plays a pivotal role in assessing tumor resectability (agreement level: 100%).

Accurate differentiation between ≤T4a and T4b gastric cancer is of paramount clinical importance, as direct invasion of adjacent organs (T4b) substantially influences treatment strategy. Patients with ≤T4a gastric cancer are generally candidates for curative radical gastrectomy, often combined with perioperative or neoadjuvant chemotherapy. In contrast, while some aggressive surgical approaches involving multivisceral resection have been proposed for T4b gastric cancers [40,41], adjacent organ involvement by gastric cancers often leads to a shift toward palliative chemotherapy in current Korean practice due to the high surgical morbidity and controversial survival gains [42]. Therefore, the accurate radiological differentiation of T4b is crucial: overstaging may deprive patients of an opportunity for curative-intent surgery, whereas understaging may result in futile laparotomy or incomplete (R1/R2) resection. Consequently, radiologists must meticulously evaluate the fat plane between the gastric tumor and adjacent organs to avoid both unnecessary morbidity and the loss of potentially curative treatment options. However, assessment of direct invasion into adjacent organs, particularly the pancreas, can be challenging in AGC due to close anatomic proximity and physiologic effacement of fat between the stomach and adjacent organs.

The diagnostic performance of CT for evaluating adjacent organ invasion has improved substantially with the introduction of MDCT and MPR techniques. In a landmark study by Kim et al. [24], the addition of coronal and sagittal MPR images to conventional axial CT significantly enhanced the prediction of adjacent organ invasion in patients with pathologically confirmed T3 and T4 gastric cancer. In per-organ analyses, the area under the receiver operating characteristic curve (AUC) for identifying T4b disease increased from 0.88 to 0.95 for reader 1 (P = 0.010) and from 0.88 to 0.98 for reader 2 (P < 0.001) when adding MPR images to axial CT images. Similarly, in per-patient analyses, AUC values improved from 0.79 to 0.91 (P = 0.006) for reader 1 and from 0.78 to 0.96 (P < 0.001) for reader 2. Notably, diagnostic accuracy with an addition of MPR images was significantly improved for invasion of the pancreas and transverse colon or mesocolon, which represent the most frequently involved adjacent organs in T4b gastric cancer [24].

Despite these advances, static imaging alone may remain inconclusive in selected cases, particularly when the fat plane between the gastric tumor and adjacent organs is effaced. To address this limitation, Jeon et al. [43] introduced the concept of a positional “sliding sign” using right down decubitus CT imaging, which allows dynamic assessment of the relative mobility between the tumor and adjacent organs (Figs. 5, 6). When the tumor does not invade adjacent organs (≤T4a), gravitational repositioning results in a demonstrable positional change between the tumor and organs such as the pancreas or liver (Fig. 5). In contrast, absence of such movement suggests fixation due to direct invasion, consistent with T4b disease (Fig. 6). In their study, the addition of right down decubitus CT images to conventional imaging significantly improved diagnostic performance for differentiating ≤T4a from T4b disease. The AUC values increased from 0.667 to 0.917 for reviewer 1 and from 0.850 to 0.894 for reviewer 2 (P < 0.001). In a subgroup analysis focusing on pancreatic invasion, AUC values improved from 0.654 to 0.893 for reviewer 1 (P < 0.001) and from 0.786 to 0.814 for reviewer 2 (P = 0.035) [43]. These findings indicate that right down decubitus CT provides incremental value, particularly in excluding T4b disease by demonstrating preserved mobility between the tumor and adjacent organs.

Fig. 5. Utility of the “CT sliding sign” in excluding T4b stage by demonstrating preserved organ mobility. A, B: A 51-year-old male patient with advanced gastric cancer at the posterior wall of the gastric body. Axial portal phase CT image (A) obtained in the left posterior oblique position shows the tumor (arrow) widely abutting the pancreatic body (P). The intervening fat plane is completely obliterated (arrowheads), making it inconclusive to differentiate between T3/T4a and T4b stage based on this static image alone. Delayed phase CT image (B) obtained in the right down decubitus position. During the positional change, the stomach and the tumor (arrow) show free gravitational movement, while the pancreas remains fixed in the retroperitoneum. Consequently, a clear fat plane (arrowheads) newly appears between the tumor and the pancreas, indicating the presence of a positive sliding sign and the absence of direct invasion. Surgical exploration confirmed the absence of pancreatic involvement, and the final pathologic stage was pT3N0 (not shown). CT = computed tomography, ST = stomach, P = pancreas.

Fig. 6. Representative case of T4b gastric cancer demonstrated by the absence of the “CT sliding sign.” A-D: A 63-year-old female with advanced gastric cancer. Axial arterial (A) and portal (B) phase CT images obtained in the left posterior oblique position show diffuse, heterogeneously enhancing wall thickening in the gastric body (arrows). The tumor is broadly abutted to the tail of the pancreas (P) with complete effacement of the intervening fat plane (red arrows). Delayed phase CT image (C) obtained in the right down decubitus position for dynamic assessment. Despite gravitational repositioning, the tumor (arrows) remains fixed and attached to the pancreatic tail (P, red arrows), indicating the absence of the sliding sign. Based on these dynamic CT findings, the lesion was radiologically staged as cT4b with pancreatic invasion. Corresponding diagnostic laparoscopic view (D) confirms that the gastric tumor (Tm) is directly invaded and fixed to the pancreas (P, arrows). The final pathologic stage was confirmed as pT4b. CT = computed tomography, P = pancreas, Tm = tumor.

Taken together, these studies support a stepwise CT-based approach for differentiating ≤T4a from T4b gastric cancer: initial evaluation using high-quality MDCT with MPR images to assess interface characteristics, followed by right down decubitus CT with assessment of the sliding sign in equivocal cases. This combined strategy improves staging accuracy and facilitates optimal, individualized treatment planning. When MDCT findings remain indeterminate for adjacent organ invasion or when peritoneal dissemination is suspected but not definitively demonstrated, staging laparoscopy should be considered to avoid futile laparotomy.

N Staging

Accurate nodal staging is a critical component of gastric cancer evaluation, as LN metastasis is one of the strongest prognostic factors [21,44] and directly influences treatment strategy, the extent of lymphadenectomy, and the indication for neoadjuvant or palliative chemotherapy.

Current nodal staging follows the 8th edition of the American Joint Committee on Cancer (AJCC) manual, which is based on the numerical count of metastatic LNs [21]. However, surgical extent is primarily guided by the 6th edition of the Japanese Gastric Cancer Association (JGCA) (Fig. 7) and the 5th edition of the Korean Gastric Cancer Association (KGCA) guidelines [16,42]. While all three guidelines generally share the definition of regional LN stations (No. 1-12), subtle discrepancies exist regarding the boundary between regional and distant nodes, necessitating meticulous radiological interpretation.

Fig. 7. Illustration of Japanese Gastric Cancer Association nodal station. Stations #1-12 represent the regional LN stations. According to the American Joint Committee on Cancer 8th edition, distant nodes (red circles) include the retropancreatic (#13), mesenteric root (#14), transverse mesocolic (#15), and para-aortic (#16) areas. LN = lymph node.

From a treatment perspective, the identification of nodal metastasis is particularly important in EGC because the presence or absence of metastatic LNs determines whether endoscopic resection is sufficient or surgical resection with lymphadenectomy is required. Furthermore, the 5th KGCA guideline recommends D1+ dissection for EGC, whereas the 6th JGCA guideline advises D2 dissection if nodal metastasis is clinically suspected in EGC [16,42]. Thus, identifying even a single suspected LN in EGC is critical as it influences the selection of treatment options between ESD and radical surgery and can mandate a more extensive surgical approach from D1+ to D2 dissection. Although the overall incidence of LN metastasis is low in pT1a disease (3%–5%), a non-negligible proportion (15%–24%) of patients with pT1b tumors harbor nodal metastasis, emphasizing the need for careful nodal evaluation in early-stage disease [45,46,47]. Conversely, in AGC where D2 dissection is routine, the clinical focus shifts from counting regional nodes to identifying “bulky” nodes or distant nodal metastasis. For instance, conglomerate LNs around the celiac trunk or vascular structures may necessitate neoadjuvant or palliative chemotherapy rather than immediate upfront surgery [16].

N staging on CT should be approached as a risk stratification process rather than an attempt at exact pathologic nodal counting. Emphasis on clinically actionable findings, particularly the detection of distant or bulky metastatic nodes, enhances the value of CT in guiding multidisciplinary management of gastric cancer.

Statement 4: For N staging of gastric cancer, identifying distant LN metastasis in retropancreatic, pancreaticoduodenal, peripancreatic, superior mesenteric, middle colic, and para-aortic areas is essential for determining resectability and treatment strategy (agreement level: 100%).

According to the AJCC 8th edition, distant nodes include the retropancreatic, pancreaticoduodenal, peripancreatic, superior mesenteric, middle colic, and para-aortic areas (Fig. 7) [21]. Notably, the hepatoduodenal LN (No. 12), previously classified as distant in the 7th edition, is now regarded as regional. A key discrepancy exists for the superior mesenteric vein LN (No. 14v): it is classified as distant by AJCC but regional by JGCA [16,21].

Radiologists must explicitly report these distant LNs because their presence usually shifts the treatment goal toward palliation, except for specific para-aortic (No. 16 a2/b1) involvements between the celiac axis and the inferior mesenteric artery that may be managed with neoadjuvant chemotherapy followed by D2 with para-aortic LN dissection. Furthermore, identifying splenic hilar LNs (No. 10) is now more important; since the 5th JGCA guideline removed No. 10 dissection from the definition of standard D2 dissection during total gastrectomy, its radiological presence may uniquely alter the surgical plan to include splenic hilar LN dissection (Fig. 8).

Fig. 8. Clinical impact of preoperative detection of splenic hilar LNs (station #10) in a 71-year-old female with advanced gastric cancer. A, B: Preoperative axial portal phase CT images show focal enhancing wall thickening at the greater curvature of the gastric high body (arrows). Although enhancing soft tissue lesions were present in the splenic hilar area (arrowhead), they were initially misinterpreted as accessory spleens and were not reported as metastatic LNs. Based on the preoperative staging, the patient underwent total gastrectomy with standard D2 LN dissection. However, following the 5th Japanese Gastric Cancer Association guidelines, splenic hilar LN (#10) dissection was omitted from the standard D2 procedure. C: Follow-up CT obtained 1 month postoperatively reveals enlarged, residual metastatic LNs at the splenic hilum (arrowheads). These missed station #10 LNs resulted in incomplete resection and subsequent dismal oncologic outcome. This case illustrates that according to current guidelines, the radiological identification of station #10 LNs is critical, as their presence uniquely alters the surgical plan to include specific splenic hilar dissection. LN = lymph node, CT = computed tomography.

Statement 5: A flexible size cut-off (short-axis diameter of 6–8 mm) should be applied based on the LN location. Ancillary features, including strong enhancement, a high short-to-long diameter ratio (round shape), presence of necrosis, and clustering of small LNs, should also be integrated to improve diagnostic reliability (agreement level: 91.7%).

Unlike T staging, where imaging criteria are relatively well established, N staging on CT remains challenging due to the variable size and morphology of metastatic LNs. Indeed, the diagnostic performance of CT for nodal staging remains modest compared with T staging. The diagnostic accuracy of MDCT for metastatic LNs is approximately 64%–80% [30,48,49], a significant improvement over conventional CT (48%–60%) [50,51]. While the AJCC 8th edition suggests a 10 mm short-axis threshold [21], pathologic studies reveal that only 21% of metastatic LN had a long diameter of ≥10 mm and 38%–55% of metastatic nodes are micrometastatic (<5 mm) [52,53]. Using a strict 8 mm cut-off in T1/T2 cancers yields high accuracy (80%) but poor sensitivity (29%), which is problematic given the increasing prevalence of EGC [32].

Recent evidence suggests that location-specific cut-offs can be more reliable [28,30,48,54]. For example, a shorter threshold (6–7 mm) is appropriate for the perigastric region (6 mm for the greater curvature and 7 mm for the lesser curvature), while larger reactive nodes are common in extra-perigastric areas such as suprapancreatic or portocaval regions, necessitating a higher threshold (8–9 mm) to maintain specificity (Fig. 9) [28,30,48,54]. Adoption of flexible, location-specific size thresholds provides a more balanced trade-off between sensitivity and specificity than a uniform size cut-off.

Fig. 9. CT criteria for diagnosing metastatic LNs in gastric cancer. A: Axial portal phase CT image of a patient with gastric cancer manifesting as enhancing wall thickening at the posterior wall of the gastric body (arrows). An enlarged LN (13 mm in short-axis diameter) at the left gastric area (arrowhead, station #7) exhibits heterogeneous enhancement with central necrosis and spiculated borders, which are highly suggestive of metastasis. Notably, another small LN (5 mm in short-axis diameter) in the same station (thin arrow) demonstrates intense enhancement and central necrosis. Both LNs were pathologically confirmed as metastatic despite the latter being below the conventional size threshold. B: Coronal portal phase CT image of another patient showing focal enhancing wall thickening at the lesser curvature of the gastric angle and antrum (arrows). Several small (<6 mm) clustered and intensely enhancing LNs are identified along the lesser curvature of the gastric body (arrowheads, station #3). These clustered small LNs were confirmed as metastatic after surgery, illustrating that ancillary features such as clustering and strong enhancement can improve diagnostic reliability even in the absence of significant enlargement. CT = computed tomography, LN = lymph node.

Beyond size, ancillary features provide crucial diagnostic clues. Metastatic LNs often exhibit strong enhancement, with attenuation values exceeding 85–100 Hounsfield unit in the portal venous phase [26,30,50], a greater short-to-long axis ratio of 0.6–0.8 (round appearance) (Fig. 9) [26,30,50,55]. In addition, the presence of multiple clustered LNs in a single station significantly increases the likelihood of malignancy regardless of individual node size (Fig. 9) [26,49,50].

M Staging

The presence of distant metastasis (M1) is a definitive contraindication for radical surgery in gastric cancer, fundamentally altering treatment strategy. Distant metastasis primarily occurs through hematogenous, lymphatic, and peritoneal routes, with the liver and peritoneum representing the most frequent sites of metastatic disease at initial presentation [56]. Accordingly, preoperative imaging plays a pivotal role in identifying metastatic disease and preventing non-therapeutic laparotomy.

Statement 6: Minimal pelvic ascites detected on preoperative CT without other suggestive findings of peritoneal seeding should not be considered a definitive contraindication for curative radical resection (agreement level: 100%).

MDCT findings that increase suspicion for peritoneal metastasis include nodular or plaque-like soft tissue thickening along the peritoneal surfaces, infiltrative stranding of the omental or mesenteric fat, parietal peritoneal thickening with enhancement, distortion or fixation of small bowel loops, and ascites [57,58]. The diagnostic performance of MDCT for peritoneal seeding varies significantly with the size of the implants; sensitivity is reported at 43% for nodules <5 mm but increases to 89% for those ≥5 mm [59]. Among various imaging features, ascites is one of the most specific indicators. Yajima et al. [60] reported that the presence of ascites in patients with AGC has a specificity of 97% for diagnosing peritoneal metastases. This high specificity implies that clinically significant or unexplained ascites is a strong predictor of peritoneal carcinomatosis.

However, a clinical dilemma arises with “minimal” fluid. Previous investigations have shown that small amounts of fluid in the pelvic cavity can be physiologically observed in both healthy men and post-menopausal women [61,62]. Therefore, if such minimal ascites is the sole finding without other high-risk features, it is rarely associated with true carcinomatosis. On MDCT, “minimal” ascites is typically defined as a localized fluid collection with a fluid volume of <10 mL, primarily confined to the most dependent portion of the pelvic cavity, such as the Douglas pouch (rectovesical or rectouterine pouch) below the level of the S3 vertebra. High-risk features increasing the likelihood of peritoneal dissemination include large tumor size (≥3 cm), T3 or T4 disease, N2 or N3 disease, and Borrmann type III or IV morphology [61,63,64].

Therefore, while larger amounts of ascites are highly specific for peritoneal seeding, minimal pelvic fluid alone should not preclude the possibility of curative resection. In such ambiguous cases, further evaluation via diagnostic laparoscopy or peritoneal cytology is recommended rather than immediate exclusion of surgical candidacy.

Arterial Anatomic Variation

Understanding arterial anatomic variation is an essential component of preoperative CT evaluation for gastric cancer, as vascular anatomy directly influences surgical safety, LN dissection, and postoperative hepatic function. The increasing adoption of minimally invasive surgeries, such as laparoscopic and robotic gastrectomy, has further heightened the clinical demand for precise preoperative evaluation of perigastric vascular anatomy. Among vascular variations, an aberrant left hepatic artery (LHA) arising from the left gastric artery (LGA) is of particular clinical importance because it courses through the operative field of gastrectomy. From a surgical perspective, MDCT is no longer merely a tool for staging but serves as a ‘preoperative road map.’ Accurate identification of aberrant LHA through high-resolution MDCT may prevent unintended intraoperative hemorrhage and facilitate meticulous LN dissection.

Statement 7: When an aberrant LHA is present, its presence, subtype (replaced or accessory), and arterial diameter should be explicitly described on preoperative CT (agreement level: 91.7%).

A large-scale anatomical and surgical study has reported that aberrant LHA originating from the LGA is encountered in approximately 14% of patients overall. Specifically, replaced LHAs, which supply the entire left lateral segment of the liver, account for approximately 8%, whereas accessory LHAs, supplying a portion of the left hepatic lobe, account for approximately 6% of cases [65].

Ligation or preservation of aberrant LHA is of great concern to surgeons because it affects the patient’s postoperative liver function, safety, and oncologic outcome. However, the management of these arteries during gastric cancer surgery remains controversial. Several surgical series have demonstrated that ligation of an aberrant LHA can generally be performed safely without major postoperative morbidity [66,67,68]. However, transient postoperative elevation of liver enzymes has been reported, particularly when the diameter of the ligated aberrant LHA exceeds approximately 1.5 mm, regardless of whether the artery is of the replaced or accessory type [68]. Ischemic changes in the left hepatic lobe, although usually reversible, have also been documented on postoperative imaging (Fig. 10). From a surgical perspective, preservation of an aberrant LHA may be technically challenging, especially in laparoscopic or robotic gastrectomy, because complete LN dissection around the LGA must be balanced against the risk of vascular injury and oncologic outcome. Therefore, accurate preoperative depiction of arterial anatomy using MDCT is crucial for surgical planning and risk stratification.

Fig. 10. Clinical significance of identifying aberrant LHA variations on preoperative multidetector CT in a 61-year-old male with gastric cancer. A: Preoperative axial arterial phase CT image reveals a prominent aberrant LHA (arrowhead) traversing the fissure for the ligamentum venosum, located between the LLS of the liver and the caudate lobe (S1). B: Coronal maximum intensity projection image clearly delineates the anatomical course of a replaced LHA (arrowheads) originating from the left gastric artery (arrow). The patient subsequently underwent total gastrectomy, during which the replaced LHA was ligated along with the left gastric artery. C: Following an elevation of liver enzymes on postoperative day 4, a follow-up portal phase CT was performed. CT image demonstrates a geographic perfusion defect (*) in the left lateral hepatic segment, consistent with acute ischemic change resulting from the arterial ligation. This case emphasizes that accurate preoperative assessment of the origin, subtype, and diameter of aberrant LHAs is crucial for predicting postoperative hepatic complications and planning meticulous surgical dissection. LHA = left hepatic artery, CT = computed tomography, LLS = left lateral segment.

Preoperative MDCT for gastric cancer should incorporate a dedicated arterial phase in addition to the portal venous phase. This dual-phase approach ensures the reliable identification of aberrant LHAs, allowing for the precise delineation of their anatomical origin, course, and luminal diameter within a single examination. Notably, an aberrant LHA originating from the LGA consistently traverses the fissure for the ligamentum venosum, situated between the left lateral segment of the liver and the caudate lobe (Fig. 10). Therefore, radiologists should meticulously examine this specific anatomical region to detect such variations.

Treatment Response Evaluation After Chemotherapy

Statement 8: For the evaluation of therapeutic response in gastric cancer following chemotherapy, CT volumetry can serve as a supplementary tool to conventional size-based measurements. While functional and metabolic modalities such as PET/CT, PET/MRI, and perfusion CT show promise for predicting early response, further validation is required for their routine clinical application (agreement level: 100%).

Accurate assessment of tumor response is crucial for optimizing treatment strategies, enabling early decisions such as the discontinuation of ineffective regimens or a timely switch to alternative therapies as well as determining eligibility for curative-intent surgery including conversion surgery [69,70]. Although CT is the primary modality for monitoring treatment response, the Response Evaluation Criteria in Solid Tumors (RECIST) often faces limitations when applied to primary gastric lesions [69,71,72]. Due to the hollow-organ nature and irregular morphology of the gastrointestinal tract, primary gastric tumors are frequently categorized as “non-measurable” or “non-target” lesions under RECIST [69,71,72].

To overcome these limitations, CT volumetry has been investigated as a more reliable and reproducible alternative for primary tumor assessment [73,74,75,76,77]. Previous studies have demonstrated that post-treatment tumor volume and the volume reduction rate correlate significantly with histopathologic T- and N-stages [75]. Lee et al. [74] also reported that the volume reduction rate of the primary tumor has shown a stronger correlation with histopathologic tumor regression grade than changes in LN diameter (Fig. 11). For instance, a volume reduction rate exceeding 32% to 36% at 6–8 weeks after neoadjuvant chemotherapy has been proposed as a threshold for identifying pathologic or clinical responders, achieving sensitivities approaching 100% in selected cohorts [74,76]. Despite these advantages, CT volumetry is not yet routine in clinical practice due to its reliance on dedicated 3D software, the time-consuming nature of manual segmentation, and potential overestimation caused by adjacent lymphadenopathy or variations in gastric distension [73,74,77]. Therefore, standardized distention protocols and the development of semi-automated segmentation methods are essential.

Fig. 11. Comparison of CT volumetry and conventional size-based measurements for therapeutic response evaluation. A-E: A 54-year-old male with advanced gastric cancer who underwent neoadjuvant chemotherapy. Initial portal phase stomach CT images (A, B) demonstrate diffuse, enhancing wall thickening from the gastric body to the antrum (arrows in A) and an enlarged LN (arrow in B) at the left gastric station (station #7). The baseline primary tumor volume was 11,975 mm³, and the short-axis diameter of the index LN was 11 mm. Follow-up CT images (C, D) after the FOLFOX regimen show a reduction in both gastric wall thickening (arrows in C) and LN size (arrow in D). The post-treatment tumor volume decreased to 2,815 mm³, representing a volume reduction rate of 76.5%, which corresponds to a PR. In contrast, the short-axis diameter of the index LN decreased to 8 mm, showing a reduction of only 27.3%, which would be categorized as SD according to RECIST v1.1. Microscopic examination (E, hematoxylin and eosin stain, ×100) of the surgical specimen reveals dominant fibrotic changes (*) with only sparse residual tumor cells (arrows), confirming a good pathologic response (moderate regression; ypT2N0). This case illustrates that while conventional diameter-based measurements categorized the response as SD, CT volumetry more accurately reflected the true therapeutic efficacy by identifying a PR, which was subsequently validated by histopathology. CT = computed tomography, LN = lymph node, PR = partial response, SD = stable disease, RECIST = Response Evaluation Criteria in Solid Tumors.

Furthermore, conventional morphological imaging such as CT and MRI often struggles to differentiate residual viable tumor from chemotherapy-induced fibrosis [69,78]. To address this limitation, functional and metabolic modalities are being actively explored. ¹⁸F-fluorodeoxyglucose (FDG) PET/CT has demonstrated utility in identifying histopathologic non-responders who may benefit from immediate surgical resection or modified regimens [79,80]. However, its predictive value for early neoadjuvant response remains inconclusive because inherent FDG non-avid gastric cancers are common, particularly in mucinous, signet-ring cell, or poorly differentiated gastric cancers [81]. Preliminary data on ¹⁸F-FDG PET/MRI also suggest that perfusion parameters such as K_trans and iAUC can differentiate responders from non-responders by showing higher values of those parameters in the responders [82]. Similarly, perfusion CT allows for the noninvasive assessment of neoangiogenesis, with several perfusion parameters showing potential as predictive biomarkers for therapeutic response [83,84,85]. While these functional imaging tools provide promising insights beyond anatomical changes, large-scale prospective studies are necessary to establish their clinical validity and standardized use.

Statement 9: Standard MDCT staging criteria may be inaccurate for re-staging gastric cancer following neoadjuvant or palliative chemotherapy. Therefore, dedicated MDCT criteria that account for post-treatment tissue changes should be applied to accurately assess re-staging and resectability before conversion surgery (agreement level: 91.7%).

Recent advances in chemotherapy have enabled the down-staging of initially unresectable gastric cancers, facilitating conversion surgery. This approach aims for R0 resection (complete macroscopic and microscopic tumor clearance) and has emerged as a promising strategy for improving long-term survival in clinical responders [86,87,88,89].

Consequently, the goals of imaging after neoadjuvant or palliative chemotherapy are two-pronged: first, to objectively monitor the therapeutic response for determining whether to continue, terminate, or switch the current chemotherapy regimen; and second, to triage patients who would benefit from curative-intent conversion surgery by accurately assessing resectability. Reliable imaging-based re-staging is essential not only to identify surgical candidates but also to avoid futile invasive procedures in non-responders whose disease remains extensive or has progressed despite systemic therapy.

Although MDCT remains the primary tool for this assessment, the diagnostic performance of conventional staging criteria is markedly reduced in the post-chemotherapy setting [90,91,92]. Studies have reported that the accuracy of T and N staging using standard criteria after chemotherapy drops to as low as 43%–57% and 37%–44%, respectively [90,91]. Notably, radiologic T stage often fails to correlate significantly with pathologic findings, primarily due to the difficulty in differentiating viable residual tumor from chemotherapy-induced fibrosis or treatment-related inflammatory changes [91].

To address these inaccuracies, dedicated post-chemotherapy MDCT imaging criteria have recently been proposed [93]. Unlike conventional criteria, these new guidelines emphasize the differentiation of non-viable components. Specifically, non-enhancing perigastric infiltrations, non-enhancing LNs, and subtle residual soft-tissue shadows after a marked decrease in distant metastasis are interpreted as non-viable tissue rather than active T, N, or M components (Fig. 12). Preliminary results using these dedicated criteria demonstrated high specificities (up to 100%) for predicting T4b and M1 stages, alongside robust performance in assessing resectability before conversion surgery [93]. By recognizing post-treatment fibrotic changes as non-viable, radiologists can reduce false-positive interpretations of tumor invasion. However, further large-scale prospective studies are warranted to optimize and generalize these criteria for widespread clinical implementation.

Fig. 12. Application of dedicated post-chemotherapy multidetector CT criteria for accurate re-staging. A-E: A 73-year-old male with advanced gastric cancer who underwent neoadjuvant chemotherapy. Pre-treatment coronal CT images (A, B) show diffuse wall thickening with transmural enhancement in the gastric body and antrum (arrows in A). Note the exophytic, nodular tumor extension into the perigastric fat (arrowheads in A), suggesting serosal invasion (cT4a). An enlarged and enhancing LN is also noted at perigastric area, suggesting metastatic LN (cN+, arrows in B). After chemotherapy (C, D) using the FOLFOX regimen, follow-up coronal CT images demonstrate residual wall thickening (arrows in C) and perigastric fat infiltration (arrowheads in C). However, these areas show no contrast enhancement, indicating treatment-induced fibrosis rather than viable tumor. According to dedicated re-staging criteria, such non-enhancing infiltration is not considered an active T-component. The previously enlarged perigastric LN remains larger than 8 mm in short-axis diameter but shows a total absence of enhancement (arrows in D). Under the new criteria, this is assessed as a negative LN, recognizing the tissue as non-viable. Gross specimen photograph (E) shows an ulcerative mass (arrows) at the posterior wall of the gastric body and antrum. Pathologic examination confirmed a pathologic complete response (ypT0N0) with no viable tumor cells in the primary lesion or harvested LNs (not shown). This case underscores that recognizing non-enhancing post-treatment changes as non-viable tissue is essential for accurate re-staging and the assessment of resectability. CT = computed tomography, LN = lymph node.

Imaging Surveillance after Curative Resection of Gastric Cancer

Statement 10: Postoperative imaging surveillance following curative resection of gastric cancer should be individualized based on the risk of recurrence. A risk-stratified approach is recommended to optimize the clinical benefit while minimizing unnecessary radiation exposure and healthcare costs (agreement level: 91.7%).

Although definitive evidence demonstrating that postoperative surveillance prolongs overall survival or is strictly cost-effective remains limited, follow-up imaging is widely recommended by major clinical guidelines for the early detection of recurrence [16,94,95]. However, establishing an optimized surveillance strategy is imperative to balance the oncologic benefits against the psychological and economic burdens on patients, as well as the potential risks of radiation exposure and adverse reactions to iodinated contrast media [96,97,98,99,100]. Among various imaging modalities, contrast-enhanced CT and transabdominal ultrasound are preferred over MRI or PET/CT, given their high availability, cost-effectiveness, and superior performance in detecting liver and peritoneal dissemination, the two most frequent sites of recurrence in gastric cancer [101].

Effective surveillance requires a nuanced understanding of recurrence risks stratified by pathologic stage and clinical risk factors. Recent studies have consistently demonstrated that the risk of extragastric recurrence (EGR) is extremely low in specific subsets of EGC, suggesting that current surveillance protocols could be safely de-escalated for these patients.

Bae et al. [102] developed a predictive model for EGR after curative resection of EGC and categorized patients into “very-low” and “low-or-greater” risk groups. The very-low-risk group, primarily comprising patients with pT1aN0 stage disease, exhibited an exceptional 5-year EGR-free survival rate of nearly 100% in both derivation and validation cohorts [102]. Similarly, Seo et al. [103] developed a robust risk-scoring system (C-index, 0.870) based on independent risk factors, including LN metastasis, beyond ESD indications, male sex, lymphovascular invasion, and macroscopic type. In their study, a “low-risk” group (score 0–10) showed a remarkable 10-year EGR-free survival rate of 99.7%, with no recurrences occurring within the first two years after surgery. Both studies emphasize that the incidence of EGR is clinically negligible in these low-risk subsets. Therefore, postoperative imaging surveillance for EGR in patients with very low-risk EGC may be reasonably de-escalated, and omission may be considered in carefully selected patients [102,103]. In contrast, for the high-risk group identified by Seo et al. [103] (score 11–24), where recurrence was significantly higher and often occurred early (56.4% within 2 years), a more rigorous and standardized surveillance protocol remains essential to ensure the timely detection of treatable recurrences [16,94,103].

Generally, imaging surveillance for the detection of recurrence is recommended for up to 5 years following curative surgery, as the incidence of recurrence decreases significantly beyond this period [104]. Ultimately, the frequency and duration of imaging follow-up should be tailored to each patient’s specific risk profile, approaching a pragmatic balance between early detection of actionable recurrence and minimization of unnecessary examinations.

CONCLUSION

This KSAR consensus recommendation provides standardized CT acquisition and interpretation strategies for gastric cancer across initial staging, treatment response assessment, re-staging, vascular evaluation, and postoperative surveillance. These recommendations are intended to improve consistency in radiologic practice and support clinical decision-making. Further validation studies are warranted to refine and expand these recommendations.

Supplement

The Supplement is available with this article at https://doi.org/10.3348/kjr.2026.0885.