Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA
Share the Article
Gu YC, Yang TY, Zhu K, Zhao CL, Zang XJ, Xie YD. Predictive value of multi-slice spiral computed tomography tumor diameter and necrotic ratio for very high-risk gastrointestinal stromal tumors. World J Gastrointest Surg 2026; 18(10): 121503 [DOI: 10.4240/wjgs.121503]
Yang-Chun Gu, Tian-Yu Yang, Ke Zhu, Xiao-Jin Zang, Ying-Di Xie, Department of Radiology, Binhai County People’s Hospital; Jiangsu Vocational College of Medicine, Yancheng 224500, Jiangsu Province, China
Cheng-Li Zhao, Department of Gastroenterology, Binhai County People’s Hospital; Jiangsu Vocational College of Medicine, Yancheng 224500, Jiangsu Province, China
Co-corresponding authors: Xiao-Jin Zang and Ying-Di Xie.
Author contributions: Gu YC and Yang TY contributed equally to study design, data collection and manuscript drafting and they contribute equally to this study as co-first authors; Zang XJ and Xie YD supervised the research and revised the manuscript and they contribute equally to this study as co-corresponding authors; Zhu K and Zhao CL participated in data analysis; and all authors approved the final manuscript.
AI contribution statement: No AI tools were utilized in drafting, data analysis or revision of this manuscript. All contents were independently completed by authors, who bear full responsibility for the integrity and accuracy of the whole work.
Supported by Jiangsu Medical Vocational College Yancheng Health Commission Collaborative Innovation Research Project, No. 202491013.
Institutional review board statement: This study was reviewed and approved by the Institutional Review Board of Binhai County People’s Hospital (Approval No. 2025BYKYLL030).
Informed consent statement: All study participants and their legal guardians provided written informed consent before recruitment.
Conflict-of-interest statement: The authors declare no conflict of interest.
Data sharing statement: No additional data are available.
Corresponding author: Ying-Di Xie, Associate Chief Physician, Research Fellow, Department of Radiology, Binhai County People’s Hospital; Jiangsu Vocational College of Medicine, No. 299 Binhai Seaside Avenue, Yancheng 224500, Jiangsu Province, China. xieyingdi2020@163.com
Received: April 24, 2026 Revised: June 11, 2026 Accepted: June 26, 2026 Published online: October 27, 2026 Processing time: 177 Days and 22 Hours
Abstract
BACKGROUND
Accurate preoperative risk stratification of gastrointestinal stromal tumors (GISTs) is essential for individualized treatment planning and prognostic evaluation. However, reliable non-invasive imaging biomarkers for identifying very high-risk GISTs remain limited.
AIM
To evaluate multi-slice spiral computed tomography (MSCT)-derived maximum tumor diameter and necrotic area ratio (NAR) for predicting very high-risk GISTs.
METHODS
We enrolled 158 patients with primary GIST treated at Binhai County People’s Hospital (January2025 to October 2025). The patients underwent preoperative plain abdominal MSCT and triple-phase enhanced scanning. Based on postoperative pathology, the patients were categorized as very high-risk or non-very high-risk. Univariate and multivariate logistic regression analyses identified independent predictors of very high-risk GISTs. The diagnostic efficacy of the combined predictive model was evaluated using receiver operating characteristic (ROC) curves.
RESULTS
Among 158 patients with primary GIST, 35 (22.2%) were classified as very high-risk. Multivariate logistic regression identified maximum tumor diameter and NAR as independent predictors of very high-risk GISTs (both P < 0.01). The combined predictive model demonstrated excellent diagnostic performance [area under the ROC curve (AUC) = 0.968], yielding a significantly higher AUC than that of either parameter alone.
CONCLUSION
MSCT-derived maximum tumor diameter and NAR are reliable independent predictors of very high-risk GISTs. Their combined predictive model facilitates accurate, non-invasive preoperative risk stratification, guiding individualized clinical decision-making.
Core Tip: Quantitative multi-slice spiral computed tomography imaging biomarkers may improve the preoperative assessment of gastrointestinal stromal tumor (GIST) aggressiveness. This study demonstrated that maximum tumor diameter combined with necrotic area ratio achieved high predictive performance for identifying very high-risk GISTs, providing practical support for individualized clinical management.
Citation: Gu YC, Yang TY, Zhu K, Zhao CL, Zang XJ, Xie YD. Predictive value of multi-slice spiral computed tomography tumor diameter and necrotic ratio for very high-risk gastrointestinal stromal tumors. World J Gastrointest Surg 2026; 18(10): 121503
Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the digestive tract. They originate from the interstitial cells of Cajal or their precursor cells and account for approximately 1%-3% of all gastrointestinal malignant tumors. These tumors exhibit a high degree of heterogeneity, presenting with diverse clinical manifestations and highly variable prognoses[1,2]. The National Institutes of Health (NIH) consensus criteria are widely utilized globally for GIST risk classification. These criteria primarily rely on the detailed evaluation of postoperative surgical specimens, encompassing tumor size, mitotic count, tumor location, and other pathological parameters. However, this approach has inherent limitations because it relies on postoperative pathology, it cannot provide preoperative decision-making support for clinicians[3,4]. Therefore, developing an accurate, non-invasive preoperative risk grading tool for GISTs is clinically significant for guiding treatment and predicting patient prognosis.
Multi-slice spiral computed tomography (MSCT)—a widely used imaging modality with high spatial resolution and multi-phase enhanced scanning capabilities—can clearly delineate a gastrointestinal tumor's morphology, size, location, and internal structure[5,6]. Specific MSCT GIST features—such as large tumor volume, irregular shape, ill-defined margins, ulceration or necrosis, heterogeneous enhancement, and the presence of metastatic foci—correlate closely with the degree of malignancy[7,8]. In particular, the correlation between tumor size and prognosis is well established.
However, clinical studies regarding GIST size are frequently limited to qualitative descriptions or simple semi-quantitative grading, lacking rigorous quantitative analysis[9,10]. Consequently, this study investigated MSCT-derived maximum tumor diameter and necrotic area ratio (NAR) and analyzed their predictive value for very high-risk GISTs to provide robust imaging support for individualized clinical management.
MATERIALS AND METHODS
Study population
Following approval from the Institutional Review Board of Binhai County People’s Hospital (Approval No. 2025BYKYLL030), 158 patients with primary GISTs treated at the hospital between January and October 2025 were enrolled in this study.
Inclusion criteria: (1) Primary GIST confirmed by postoperative histopathology; (2) Availability of preoperative standard abdominal unenhanced and triple-phase contrast-enhanced MSCT scans; (3) Provision of informed consent by the patients or their families; and (4) Availability of complete clinical and follow-up data.
Exclusion criteria: (1) Preoperative antitumor therapy, including targeted therapy, radiotherapy, or chemotherapy; (2) Tumor diameter < 1.0 cm, which precludes accurate imaging evaluation; (3) Concurrent malignant tumors; and (4) Inability to tolerate or cooperate with the MSCT examination.
MSCT examination protocol
Patients fasted for at least 6-8 hours and restricted water intake for 2 hours before the examination. As a neutral contrast agent, patients consumed 500-800 mL of warm water 30 minutes before scanning and an additional 200 mL immediately beforehand. Standardized breath-holding training was also provided. Full abdominal scanning was performed using a 64-slice GE Optima CT660 scanner (GE Healthcare, United States). The scanning parameters were as follows: Tube voltage, 120 kV; automatic tube current modulation with a noise index of 12; detector configuration, 128 mm × 0.625 mm; pitch, 0.99; and rotation time, 0.5 second/rot. The slice thickness and reconstruction interval were both set to 5.0 mm. The scanning range extended from the diaphragm to the inferior margin of the pubic symphysis, encompassing the entire abdomen and pelvis.
Following an unenhanced full abdominal scan, contrast-enhanced scanning was performed. A non-ionic iodinated contrast agent (iohexol, 350 mg I/mL) was administered via the antecubital vein at a flow rate of 3.5 mL/s (dose: 1.5 mL/kg body weight; maximum volume: 100 mL), followed by a 30-mL saline flush at the same rate. Automatic bolus tracking was employed, with a region of interest (ROI) placed in the abdominal aorta at the level of the celiac trunk; scanning was triggered automatically when the attenuation in the ROI reached 100 Hounsfield units (HUs). Multi-phase acquisition was timed as follows: The arterial phase occurred 10-30 seconds after the trigger, whereas the portal venous and delayed phases occurred 60-70 seconds and 120-180 seconds after the initiation of contrast injection, respectively. Transverse images were reconstructed from the raw data using a standard algorithm. Both 5-mm and 1-mm thin-slice images of all phases were transferred to an advanced post-processing workstation (Philips IntelliSpace Portal, version 10) and the hospital picture archiving and communication system for storage and subsequent analysis.
Imaging analysis
Two senior radiologists with over 10 years of experience in abdominal imaging independently analyzed all images, blinded to the pathological results and clinical groupings. Any discrepancies were resolved through discussion until reaching a consensus. Maximum tumor diameter was measured on the largest cross-section across the axial, coronal, and sagittal planes using multiplanar reformation. The NAR was calculated using the phase that provided optimal visualization of the tumor parenchyma and necrosis (typically, the portal venous phase). The entire tumor contour was manually delineated slice by slice on axial images—carefully excluding adjacent fat, vessels, and the intestinal lumen—to calculate the total area (TA) of each slice automatically. All non-enhancing or markedly hypo-enhancing necrotic or cystic regions (attenuation < 30 HUs) were manually segmented to compute the necrotic area (NA). The NAR was then calculated as the sum of the NA across all slices divided by the sum of the TA across all slices, multiplied by 100%. Other evaluated MSCT features included tumor location (e.g., stomach, small intestine, or colorectum), shape (regular or irregular), margins (well-defined or ill-defined), enhancement pattern (homogeneous or heterogeneous), the presence of calcification, adjacent organ invasion, and distant metastasis.
Pathological evaluation and grouping
Postoperative histopathology served as the reference standard, from which tumor size, mitotic count, and primary site were documented. Two senior pathologists classified all GISTs into risk categories according to the 2017 modified NIH criteria. For this study, patients presenting with tumor rupture or highly aggressive pathological features based on these criteria were categorized into the very high-risk group. Whereas, patients not meeting these criteria—encompassing high-, intermediate-, low-, and very low-risk cases—were assigned to the non-very high-risk group.
Clinical data collection
Collected clinical data included demographic characteristics (e.g., age, sex, body mass index, and underlying comorbidities such as hypertension and diabetes) and initial clinical manifestations (e.g., abdominal pain, discomfort, gastrointestinal bleeding, and palpable mass). Preoperative laboratory parameters, obtained within 1 week before surgery, included hemoglobin levels, platelet counts, neutrophil-to-lymphocyte ratio (NLR), and serum albumin levels.
Statistical analysis
Statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, United States). Continuous variables are presented as mean ± SD or median and interquartile range, depending on the normality of the data distribution. Between-group comparisons were performed using the independent-samples t-test or Mann-Whitney U-test, as appropriate. Categorical variables are reported as n (%), and comparisons were made using the χ2 test or Fisher’s exact test. Univariate and multivariate binary logistic regression analyses were performed to identify factors independently associated with very high-risk GISTs, and odds ratios (ORs) and 95% confidence intervals (95%CIs) were calculated. Variables with a P value < 0.1 in the univariate analysis were entered into the multivariate model. Receiver operating characteristic (ROC) curve analysis was applied to assess the predictive performance of each independent factor and the combined model derived from logistic regression probabilities. The area under the ROC curve (AUC) was calculated, and comparisons between AUC values were assessed using the DeLong test. A two-sided P value < 0.05 was considered statistically significant.
RESULTS
Comparison of baseline and clinical characteristics
Of the 158 patients with primary GISTs, 35 (22.2%) were categorized into the very high-risk group, while the remaining 123 (77.8%) were categorized into the non-very high-risk group. Compared with the non-very high-risk group, the very high-risk group exhibited significantly higher rates of palpable abdominal mass, tumor rupture, and non-gastric tumor origin, as well as higher platelet counts, NLRs, mitotic counts, and Ki-67 indices (all P < 0.05). Conversely, the proportion of asymptomatic cases detected during physical examination, as well as the hemoglobin and albumin levels, was significantly lower in the very high-risk group (all P < 0.05; Table 1).
Table 1 Comparison of baseline data and clinical characteristics between the two groups.
Compared with the non-very high-risk group, patients in the very high-risk group demonstrated more aggressive MSCT imaging characteristics, including mixed intra- and extraluminal growth, irregular or lobulated morphology, ill-defined margins, heterogeneous enhancement, adjacent organ invasion, and distant metastasis. In addition, maximum tumor diameter and NAR were markedly greater in the very high-risk group, suggesting that a larger tumor burden and more extensive intratumoral necrosis were closely associated with a higher malignant potential (Table 2).
Table 2 Comparison of multi-slice spiral computed tomography features between the two groups.
Multivariate logistic regression analysis of predictors for very high-risk GIST
Preoperatively available clinical and imaging variables with P < 0.1 in the univariate analysis were entered into a multivariate binary logistic regression model. To avoid multicollinearity, only the imaging-derived maximum tumor diameter was included. Independent variables were appropriately coded before the analysis. The multivariate analysis demonstrated that maximum tumor diameter (OR = 1.30; 95%CI: 1.11-1.52; P = 0.001) and NAR (OR = 1.12; 95%CI: 1.06-1.19; P < 0.001) were independent predictors of very high-risk GISTs (Tables 3 and 4).
Table 3 Variable assignment table for multivariate logistic regression analysis.
Variable type
Variable name
Coding/definition
Dependent variable
GIST risk classification
0 = non-very high-risk group; 1 = very high-risk group
The AUC for predicting very high-risk GISTs was 0.894 (95%CI: 0.832-0.956) for maximum tumor diameter, 0.912 (95%CI: 0.861-0.963) for the NAR, and 0.968 (95%CI: 0.941-0.995) for the combined model. The DeLong test confirmed that the AUC of the combined model was significantly higher than that of either maximum tumor diameter (Z = 2.841, P = 0.004) or NAR (Z = 2.104, P = 0.035) alone (Figure 1).
Figure 1 Receiver operating characteristic curves of maximum tumor diameter, necrotic area ratio, and their combination for predicting the very high-risk group of gastrointestinal stromal tumor.
AUC: Area under the receiver operating characteristic curve.
DISCUSSION
GISTs are the most common mesenchymal tumors of the digestive tract, known for their highly variable biological behavior. While some tumors exhibit indolent growth, others demonstrate highly aggressive clinical courses, posing substantial diagnostic and prognostic challenges[11,12]. Their inaccurate preoperative risk stratification can lead to insufficient surgical resection and increased perioperative risk, or cause patients with locally advanced disease to miss opportunities for tumor downsizing and downstaging via neoadjuvant targeted therapy[13,14]. In this study, 35 of 158 patients with primary GISTs (22.2%) were categorized into the very high-risk group, highlighting the clinical importance of accurate preoperative identification. Although tumor rupture, mitotic count, and the Ki-67 index are key components of the reference standard for the postoperative pathological diagnosis of very high-risk GISTs[15,16], these indicators are exclusively available postoperatively and cannot guide timely preoperative decision-making. Therefore, identifying effective preoperative predictors for very high-risk GISTs is of substantial clinical value.
Univariate and multivariate logistic regression analyses consistently demonstrated that maximum tumor diameter (OR = 1.30) and NAR (OR = 1.12) are independent predictors of very high-risk GISTs (P < 0.05). MSCT enables rapid scanning, short acquisition times, and prolonged enhancement durations. It serves as the primary modality for the preoperative localization, characterization, and staging of GISTs, while multi-phase contrast-enhanced scanning further elucidates the internal tumor architecture and hemodynamic features[17,18]. With the advent of precision radiology, MSCT is increasingly utilized to evaluate tumor biological behavior by converting subjective imaging findings into objective, quantifiable biomarkers[19,20]. Maximum tumor diameter is the most intuitive quantitative parameter and a widely recognized adverse prognostic factor in GISTs[21,22], directly reflecting macroscopic tumor growth and invasiveness. Larger tumors typically exhibit higher proliferative activity, greater metastatic potential, and more severe adjacent organ invasion, corresponding to elevated clinical risk. Intratumoral necrosis is a hallmark of rapidly growing malignant tumors outstripping their blood supply. The extent of necrosis strongly correlates with tumor growth rate, perfusion, and metabolic activity, reflecting an invasive biological nature[23,24]. As an emerging imaging indicator, the NAR quantifies the proportion of intratumoral necrosis, with higher values indicating rapid, disorganized growth and a hostile internal microenvironment. Intratumoral necrosis or cystic degeneration on contrast-enhanced computed tomography is a significant predictor of very high-risk GISTs[25]. Similarly, intratumoral cystic change or necrosis serves as an independent imaging feature predicting high mitotic counts and risk stratification[26]. In this study, precise slice-by-slice segmentation and area calculations were utilized to quantify this morphological feature into reproducible NAR values, transforming subjective impressions into objective data for accurate and reliable clinical evaluation.
The combined model achieved an AUC of 0.968, yielding a diagnostic performance significantly higher than that of either indicator alone (P < 0.05). Accumulating clinical evidence highlights the inherent limitations of relying solely on tumor size. Preoperative imaging may underestimate the actual pathological size of certain gastric GISTs, and risk thresholds vary across different anatomical locations[27]. Furthermore, some large GISTs may behave indolently, whereas certain small tumors can be highly aggressive owing to elevated mitotic counts[28,29]. Therefore, integrating the NAR—which reflects internal tumor characteristics—effectively compensates for the deficiencies of using maximum tumor diameter alone, thereby improving predictive accuracy. Maximum tumor diameter and NAR characterize the malignant potential of GISTs from complementary dimensions: Maximum diameter reflects the macroscopic tumor burden and growth duration, whereas NAR reveals internal biological invasiveness. Their combination establishes a comprehensive evaluation system[30], providing robust support for clinicians in formulating individualized treatment strategies.
This study has some limitations. First, as a single-center retrospective study with a relatively small sample size and an imbalance between the risk groups, the statistical stability and generalizability of the results may be limited. Second, the potential for selection bias cannot be entirely excluded. Therefore, future multicenter studies with larger cohorts are warranted for validation.
CONCLUSION
In conclusion, the MSCT-derived quantitative parameters of maximum tumor diameter and NAR serve as effective imaging biomarkers for predicting very high-risk GISTs. Their combination offers superior predictive value, facilitating accurate, non-invasive preoperative risk assessment to support individualized clinical decision-making.
Rengo M, Onori A, Caruso D, Bellini D, Carbonetti F, De Santis D, Vicini S, Zerunian M, Iannicelli E, Carbone I, Laghi A. Development and Validation of Artificial-Intelligence-Based Radiomics Model Using Computed Tomography Features for Preoperative Risk Stratification of Gastrointestinal Stromal Tumors.J Pers Med. 2023;13:717.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 18][Reference Citation Analysis (2)]
Doita S, Watanabe M, Ogawa T, Tanakaya K, Aoki H. Utility of Follow-Up Computed Tomography (CT) and Fluorodeoxyglucose-Positron Emission Tomography/CT (FDG-PET/CT) in Diagnosing Gastrointestinal Stromal Tumor (GIST) Presenting as Spontaneous Hemoperitoneum: A Case Report.Cureus. 2024;16:e69799.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 2][Reference Citation Analysis (1)]
Gu T, Li T, Zhang J, Zhou Q, Yang D, Xu J, Li Q, Xu Z, Li F, Xu H. Using body composition to predict treatment-related adverse events and disease-free survival in patients with gastrointestinal stromal tumors treated with imatinib: a retrospective cohort study.Front Immunol. 2025;16:1576834.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 1][Reference Citation Analysis (1)]
Pozas J, Lindsay D, Wall A, Mavroeidis L, Thway K, Smith M, Hayes A, Strohaeker J, Benson C, Messiou C, Huang P, Jones RL, Napolitano A. A novel risk score after neoadjuvant imatinib predicts relapse-free survival in patients with gastrointestinal stromal tumours (GIST).Eur J Surg Oncol. 2025;51:110473.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 1][Cited by in RCA: 2][Article Influence: 2.0][Reference Citation Analysis (0)]
Kim SM, Kim EY, Cho JW, Jeon SW, Kim JH, Kim TH, Moon JS, Kim JO; Research Group for Endoscopic Ultrasound of the Korean Society of Gastrointestinal Endoscopy. Predictive Factors for Differentiating Gastrointestinal Stromal Tumors from Leiomyomas Based on Endoscopic Ultrasonography Findings in Patients with Gastric Subepithelial Tumors: A Multicenter Retrospective Study.Clin Endosc. 2021;54:872-880.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 18][Cited by in RCA: 19][Article Influence: 3.8][Reference Citation Analysis (1)]
Sharma AK, de la Torre J, IJzerman NS, Sutton TL, Zhao B, Khan TM, Banerjee S, Cui C, Nguyen V, Alkhuziem M, Snaebjornsson P, van Boven H, Bruining A, Tang CM, Yoon H, De la Fuente A, Kato S, Patel H, Heinrich MC, Corless CL, Horgan S, Burgoyne AM, Fanta P, Mesirov JP, Blakely AM, Davis JL, Mayo SC, van Houdt WJ, Steeghs N, Sicklick JK. Location of Gastrointestinal Stromal Tumor (GIST) in the Stomach Predicts Tumor Mutation Profile and Drug Sensitivity.Clin Cancer Res. 2021;27:5334-5342.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 19][Cited by in RCA: 22][Article Influence: 4.4][Reference Citation Analysis (1)]
Xie W, Zhang Z, Sun Z, Wan X, Li J, Jiang J, Liu Q, Yang G, Fu Y. A machine learning model integrating clinical-radiomics-deep learning features accurately predicts postoperative recurrence and metastasis of primary gastrointestinal stromal tumors.Insights Imaging. 2025;16:135.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 7][Reference Citation Analysis (3)]
Cannella R, Tabone E, Porrello G, Cappello G, Gozzo C, Incorvaia L, Grignani G, Merlini A, D'Ambrosio L, Badalamenti G, Regge D, Bartolotta TV. Assessment of morphological CT imaging features for the prediction of risk stratification, mutations, and prognosis of gastrointestinal stromal tumors.Eur Radiol. 2021;31:8554-8564.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 29][Cited by in RCA: 27][Article Influence: 5.4][Reference Citation Analysis (1)]
Wang Y, Bai G, Liu Y, Huang M, Chen W, Wang F. Interpretable machine learning model based on CT semantic features and radiomics features to preoperatively predict Ki-67 expression in gastrointestinal stromal tumors.Sci Rep. 2024;14:29336.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 9][Reference Citation Analysis (1)]
