The correlation between preoperative total lesion glycolysis and lymphovascular invasion based on ¹⁸F-FDG PET/CT and its predictive value in primary gastric cancer: a cross-sectional study

Abstract

Background

Lymphovascular invasion (LVI) is a critical factor in the lymphatic spread of tumor cells, and is closely associated with local recurrence and distant metastasis in gastric cancer. The study aimed to evaluate the correlation and predictive value of preoperative total lesion glycolysis (TLG) in patients with primary gastric cancer as measured by a combination of ¹⁸F-labeled fluoro-2-deoxyglucose (¹⁸F-FDG) positron emission tomography (PET) and computed tomography (CT) for LVI.

Methods

A retrospective analysis of the demographic and ¹⁸F-FDG PET/CT data of 177 patients with gastric cancer diagnosed by postoperative pathology at The Third Affiliated Hospital of Soochow University between January 2014 and August 2021 was conducted. The cohort comprised 124 males and 53 females. The ¹⁸F-FDG PET/CT data analyzed included the primary lesion location and size, lymph node metastasis (LNM) status, and metabolic parameters [i.e., the maximum standardized uptake value (SUVmax), mean standardized uptake value (SUVmean), metabolic tumor volume (MTV), and TLG]. The patients were categorized into LVI-positive (LVI+) and LVI-negative (LVI−) groups based on postoperative pathological findings. Differences in the clinical data between the two groups were analyzed. Univariate and multivariate logistic regression models were employed to assess the correlation between preoperative TLG and LVI. A generalized additive model (GAM) was applied for curve fitting, and receiver operating characteristic (ROC) curves were plotted to evaluate the predictive efficacy of preoperative TLG for LVI.

Results

Among the 177 patients, 71 (40.1%) were LVI+, and 106 (59.9%) were LVI−. Significant differences were observed between the two groups in terms of the primary lesion size, LNM status, age, SUVmax, SUVmean, MTV, and TLG (all P<0.05 or P<0.001). A per standard deviation (SD) increase in TLG was associated with a 59.1% increased risk of LVI [odds ratio (OR) =1.591; 95% confidence interval (CI): 1.142–2.216; P=0.006]. After adjusting for confounders, TLG remained significantly associated with an increased risk of LVI (OR per SD: 1.428; 95% CI: 1.018–2.002; P=0.039). Treating TLG as a categorical variable produced consistent results (P for trend =0.014). In the clinical lymph node metastasis positive (cLNM+) subgroup, a TLG value ≥53.3 predicted LVI with a sensitivity of 81.4% (35/43), a specificity of 37.8% (14/37), and an accuracy of 61.3% (49/80). In the clinical lymph node metastasis negative (cLNM−) subgroup, a TLG value ≥41.9 predicted LVI with a sensitivity of 82.1% (23/28), a specificity of 53.6% (37/69), and an accuracy of 61.9% (60/97).

Conclusions

Preoperative TLG exhibits a positive linear correlation with the risk of LVI in primary gastric cancer. TLG shows high sensitivity for predicting LVI, making it a valuable imaging biomarker for assessing LVI risk in gastric cancer, especially in cLNM− but LVI+ patients.

Introduction

In China, gastric cancer is one of the most prevalent malignancies of the digestive system (1). Lymphovascular invasion (LVI) is a critical factor in the lymphatic spread of tumor cells, and is closely associated with local recurrence and distant metastasis in gastric cancer (2,3). The presence of LVI significantly affects the prognosis of gastric cancer patients (4,5). The incidence of LVI is particularly high in patients with lymph node metastasis (LNM), and can even reach up to 7.9% in early-stage gastric cancer patients without LNM (6). Therefore, the accurate detection of LVI is essential for selecting the treatment strategy and predicting the prognosis of gastric cancer patients (4,7). Currently, the diagnosis of LVI relies on postoperative pathological examination; thus, non-invasive, accurate preoperative prediction methods urgently need to be established.

¹⁸F-labeled fluoro-2-deoxyglucose (¹⁸F-FDG) positron emission tomography computed tomography (PET/CT) imaging is a valuable tool for preoperative staging, treatment planning, therapeutic response evaluation, and prognosis assessment in gastric cancer. Previous studies have reported a correlation between the maximum standardized uptake value (SUVmax) of primary gastric lesions, and both LVI and LNM (7,8). It is well known that total lesion glycolysis (TLG), a volumetric metabolic parameter calculated by multiplying the metabolic tumor volume (MTV) by the mean standardized uptake value (SUVmean), can be used to comprehensively assess the metabolic activity of the tumor. TLG has proven useful in predicting LVI in cervical cancer patients without LNM (9). Nonetheless, the diagnostic value of preoperative ¹⁸F-FDG PET/CT metabolic parameters, specifically TLG, in detecting LVI in gastric cancer remains unclear.

This study aimed to examine the correlation between preoperative TLG measured by ¹⁸F-FDG PET/CT and LVI in primary gastric cancer, particularly in those who were LNM-negative (LNM−), and its predictive value. The evaluation of LVI status can aid in patient stratification, the selection of suitable treatment plans, and the prediction of patient prognosis. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-644/rc).

Methods

Study population

We conducted a retrospective study of patients who underwent preoperative ¹⁸F-FDG PET/CT imaging and were subsequently diagnosed with gastric cancer by pathology following radical gastrectomy at The Third Affiliated Hospital of Soochow University between January 2014 and August 2021. The data collected included patient demographics [gender, age, and body mass index (BMI)], primary lesion characteristics (location and size), metabolic parameters, LMN and LVI status, and pathological type. None of the patients included in the study had received any anti-tumor treatment before undergoing ¹⁸F-FDG PET/CT imaging. Patients were excluded from the study if they met any of the following exclusion criteria: (I) had not undergone radical gastrectomy with D2 lymph node dissection, or had less than 16 lymph nodes excised; (II) had distant metastasis or multiple cancers; (III) had an interval between surgery and PET/CT imaging >15 days; (IV) had poor-quality images rendering analysis impossible, or had no significant ¹⁸F-FDG uptake in the primary gastric lesion; and/or (V) had a severe infection or diabetes. Ultimately, 177 patients met the study inclusion criteria. Based on the postoperative pathology results, the patients were categorized into LVI-positive (LVI+) and LVI-negative (LVI−) groups, which comprised 71 and 106 patients, respectively (Figure 1).

Figure 1.

Flowchart of inclusion process for study patients. ¹⁸F-FDG PET/CT, ¹⁸F-labeled fluoro-2-deoxyglucose positron emission tomography computed tomography; LVI, lymphovascular invasion.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The Third Affiliated Hospital of Soochow University (No. [2020] K104), and the requirement of informed consent was waived due to the retrospective nature of the study.

18F-FDG PET/CT imaging

¹⁸F-FDG PET/CT imaging was performed using a Siemens Biograph mCT [64] PET/CT scanner (Siemens Healthineers, Erlangen, Germany), covering the region from the base of the skull to the mid-thigh, with a scan time of two minutes per bed position. The patients fasted for at least six hours before the examination to ensure that their fasting blood glucose levels were ≤11.1 mmol/L. Under resting conditions, the patients received an intravenous injection of ¹⁸F-FDG at a dose of 4.44 MBq/kg and rested quietly for approximately one hour. Ten minutes before imaging, the patients ingested 500 mL of water to distend the stomach. The CT scans were acquired with a tube voltage of 100 kV, a tube current of 80 mA, and a slice thickness of 5 mm. PET imaging was conducted using a three-dimensional acquisition mode, and the images were reconstructed using the ordered subset expectation maximization algorithm; the CT data were used for attenuation correction.

Measurement of the metabolic parameters of primary lesions

All the PET/CT images were uploaded in Digital Imaging and Communications in Medicine format to LIFEx software (version 6.0, http://www.lifexsoft.org). Regions of interest (ROIs) were delineated using a threshold of SUVmax ≥40% to obtain the metabolic parameters of the primary lesion, including the SUVmax, SUVmean, MTV, and TLG. Two experienced nuclear medicine physicians independently reviewed and measured the images. Any discrepancies were resolved through discussion until a consensus was reached.

Determination of LNM

The lymph nodes were classified as clinical LNM-positive (cLNM+) if their short-axis diameter was ≥10 mm on the CT images, or if the ¹⁸F-FDG uptake in the lymph nodes on PET images was ≥ the ¹⁸F-FDG uptake of the liver (10), and otherwise, as clinical LNM-negative (cLNM−). Postoperative pathological evaluation was the gold standard for the diagnosis of LNM; the presence or absence of LNM was recorded as pathological LNM-positive (pLNM+) or pathological LNM-negative (pLNM−), respectively.

Determination of LVI

The surgical specimens were routinely processed for histopathological examination and stained with hematoxylin and eosin (H&E). The determination of LVI was based on the “TNM Classification of Malignant Tumours. 8^th ed.” (11).

Statistical analysis

The statistical analysis was performed using R software (version 3.4.3, https://www.R-project.org). The normally distributed quantitative data were presented as the mean ± standard deviation (SD; x±s), while the non-normally distributed data were expressed as the median (quartile 1, quartile 3) [M (Q1, Q3)]. The categorical data were reported as the frequency (%). Differences in terms of age, metabolic parameters, gender, primary lesion location and size, LNM status, and pathological type between the LVI+ and LVI− groups were assessed using the independent sample t-test, Mann-Whitney U test, χ² test, or Fisher’s exact test as appropriate. The Kappa value was used to analyze the concordance rate between the presence or absence of LNM between PET/CT imaging and postoperative pathology. To investigate the association between primary lesion TLG and LVI in gastric cancer, univariate and multivariate logistic regression analyses were conducted. The following three models were developed: an unadjusted model, a partially adjusted model, and a fully adjusted model. The relationship between primary lesion TLG and LVI was visualized using a generalized additive model (GAM) (12). The diagnostic performance of TLG for predicting LVI was evaluated by receiver operating characteristic (ROC) curve analysis. A P value of <0.05 was considered statistically significant.

Results

Comparison of clinical data between the LVI+ and LVI− groups

The clinical data of the LVI+ and LVI− groups were compared (Table 1). Among the 177 gastric cancer patients, 71 (40.1%) were classified as LVI+ and 106 (59.9%) were classified as LVI−. In the entire study cohort, 97 (54.8%) were cLNM−, of whom 28 (28.9%) were LVI+, while 80 (45.2%) were cLNM+, of whom 43 (53.8%) were LVI+. In the entire study cohort, 75 patients (42.4%) were pLNM−, of whom 15 (20%) were LVI+, while 102 patients (57.6%) were pLNM+, of whom 56 (54.9%) were LVI+. Significant differences were observed between the LVI+ and LVI− groups in terms of the primary lesion size, LNM status, and age (all P<0.05 or P<0.001). The consistency of LNM between the PET/CT findings and pathological results was assessed using the kappa value, with the resulting value of 0.67 indicating good agreement.

Table 1. Comparison of clinical data between the LVI+ and LVI− groups.

Variables	LVI− (n=106)	LVI+ (n=71)	t/χ2	P
Gender (n)			0.34‡	0.56
Male	76	48
Female	30	23
BMI (kg/m2; $\overline{x}\pm s$)	23.0±3.1	22.5±3.2	−	0.231
Age (years; $\overline{x}\pm s$)	64±10	67±9	−2.14†	0.034
Primary lesion location (n)			2.89‡	0.089
Upper 1/3 and middle 1/3	73	40
Lower 1/3	33	31
LNM status (n)
cLNM			0.53‡	<0.001
cLNM+	37	43
cLNM−	69	28
pLNM			0.78‡	<0.001
pLNM+	46	56
pLNM−	60	15
Primary lesion size (n)			4.44‡	0.035
<3 cm	31	11
≥3 cm	75	60
Pathological type (n)			0.27‡	0.601
Adenocarcinoma	93	61
Other types	12	10

Other pathological types include mucinous adenocarcinoma, signet ring cell carcinoma, neuroendocrine carcinoma, and adenocarcinoma with signet ring or mucinous components. ^†, t value; ^‡, χ² value. +, positive; −, negative. BMI, body mass index; cLNM, clinical lymph node metastasis; LVI, lymphovascular invasion; LNM, lymph node metastasis; pLNM, pathological lymph node metastasis.

Comparison of the metabolic parameters of the primary lesions between LVI+ and LVI− groups

The metabolic parameters of the primary lesions between the LVI+ and LVI− groups were compared as shown in Table 2. The preoperative SUVmax, SUVmean, MTV, and TLG of the primary gastric lesions differed significantly between the LVI+ and LVI− groups (all P<0.05 or P<0.001). The measurements of the PET metabolic parameters (i.e., the SUVmean, SUVmax, MTV, and TLG) for the primary lesions demonstrated high consistency and repeatability with intraclass correlation coefficients ranging from 0.919 to 0.998 (all P<0.001).

Table 2. Comparison of metabolic parameters of primary lesions between the LVI+ and LVI− groups.

Variables	LVI− (n=106)	LVI+ (n=71)	z	P
SUVmax	7.95 (5.03, 13.63)	10.04 (7.15, 16.31)	−2.49	0.013
SUVmean	4.53 (2.72, 7.62)	5.7 (3.90, 9.15)	−2.16	0.03
MTV (cm3)	10.35 (6.07, 17.22)	15.30 (8.05, 23.04)	−3.00	0.003
TLG (g)	50.55 (26.76, 91.48)	84.77 (55.09, 129.28)	−3.53	<0.001

Data are presented as median (quartile 1, quartile 3). +, positive; −, negative. LVI, lymphovascular invasion; MTV, metabolic tumor volume; SUVmax, maximum standardized uptake value; SUVmean, mean standardized uptake value; TLG, total lesion glycolysis.

Univariate and multivariate analyses of the effect of TLG on LVI

Univariate and multivariate analyses were performed to investigate the effect of TLG on LVI. As detailed in Table 3, the unadjusted model showed that the risk of LVI significantly increased with per SD increase in TLG [odds ratio (OR) =1.591; 95% confidence interval (CI): 1.142–2.216; P=0.006]. When TLG was considered a categorical variable, the OR for LVI increased progressively across the TLG tertiles, with the highest OR observed in the group with the highest TLG (OR =4.194; 95% CI: 1.883–9.341; P<0.001). In the partially adjusted model (adjusted for gender, age and BMI), the OR for LVI continued to increase across the TLG tertiles (P for trend =0.002), and a per SD increase in TLG corresponded to an OR of 1.505 for LVI (95% CI: 1.078–2.100; P=0.016). In the fully adjusted model (adjusted for gender, age, BMI, primary lesion location, and size), a per SD increase in TLG was associated with an OR of 1.428 for LVI (95% CI: 1.018–2.002; P=0.039). The findings for the analysis in which TLG was treated as a categorical variable were consistent with the findings from the analysis in which TLG was treated as a continuous variable (P for trend =0.014).

Table 3. Univariate and multivariate analyses of the effect of TLG on LVI.

Variables	Unadjusted model			Partially adjusted model			Fully adjusted model
	OR (95% CI)	P		OR (95% CI)	P		OR (95% CI)	P
TLG (per SD)	1.591 (1.142–2.216)	0.006		1.505 (1.078–2.100)	0.016		1.428 (1.018–2.002)	0.039
Groups
Low TLG group	1			1			1
Medium TLG group	2.788 (1.250–6.217)	0.012		2.681 (1.184–6.070)	0.018		2.175 (0.896–5.278)	0.086
High TLG group	4.194 (1.883–9.341)	<0.001		3.681 (1.625–8.336)	0.002		3.078 (1.287–7.361)	0.012
P for trend		<0.001			0.002			0.014

Unadjusted model: no variables adjusted; partially adjusted model: adjusted for gender, age and BMI; fully adjusted model: adjusted for gender, age, BMI, primary lesion location and size. BMI, body mass index; CI, confidence interval; LVI, lymphovascular invasion; OR, odds ratio; SD, standard deviation; TLG, total lesion glycolysis.

Curve fitting

As Figure 2 shows, the GAM results revealed a positive linear correlation between the TLG of the primary gastric lesion and the risk of LVI after fully adjusting for confounding factors, such that as TLG increased, the risk of LVI also increased.

Figure 2.

Curves depicting the relationship between TLG and the risk of LVI. The solid red line shows that TLG was linearly associated with LVI, and the blue dotted lines above and below represent the 95% CIs. CI, confidence interval; LVI, lymphovascular invasion; TLG, total lesion glycolysis.

ROC curve analysis of TLG for predicting LVI in the cLNM+ and cLNM− subgroups

For the gastric cancer patients with or without cLNM, the area under the curve (AUC) of the ROC curve for TLG in predicting LVI was 0.657 (95% CI: 0.576–0.738; P<0.001). When TLG was ≥55.3, its sensitivity, specificity, and accuracy in predicting LVI were 77.5% (55/71), 52.8% (56/106), and 62.7% (111/177), respectively (Figure 3A). A subgroup analysis based on cLNM status showed that the AUC for predicting LVI was 0.590 (95% CI: 0.464–0.716; P=0.168) in the cLNM+ subgroup and 0.668 (95% CI: 0.553–0.784; P=0.010) in the cLNM− subgroup. In the cLNM+ subgroup, when TLG was ≥53.3, its sensitivity, specificity, and accuracy in predicting LVI were 81.4% (35/43), 37.8% (14/37), and 61.3% (49/80), respectively (Figure 3B). In the cLNM− subgroup, when TLG was ≥41.9, its sensitivity, specificity, and accuracy in predicting LVI were 82.1% (23/28), 53.6% (37/69), and 61.9% (60/97), respectively (Figure 3C). Representative cases are illustrated in Figures 4,5.

Figure 3.

ROC curves showing the ability of TLG to predict LVI in different cLNM subgroups. (A) ROC curve showing the ability of TLG to predict LVI in gastric cancer patients with/without cLNM. (B) ROC showing the ability of TLG to predict LVI in gastric cancer patients with cLNM. (C) ROC curve showing the ability of TLG to predict LVI in gastric cancer patients without cLNM. cLNM, clinical lymph node metastasis; LVI, lymphovascular invasion; ROC, receiver operating characteristic; TLG, total lesion glycolysis.

Figure 4.

¹⁸F-FDG PET/CT images and postoperative pathology of an 81-year-old female gastric cancer patient without LNM. The CT (A), PET/CT (B), and PET (C) images showed a thickened gastric wall at the cardia of the primary lesion with abnormal tracer uptake (as indicated by the red arrows); no LNM was observed. The TLG of the lesion was 65.1. Based on a TLG threshold of >41.9 for LVI, this patient was diagnosed as LVI+. Postoperative pathology (D; H&E, ×100) revealed moderately to poorly differentiated adenocarcinoma at the lesser curvature of the cardia. No LNM was observed (0/18 at the lesser curvature; 0/2 at the greater curvature), but intravascular tumor emboli were present. +, positive; ¹⁸F-FDG PET/CT, ¹⁸F-labeled fluoro-2-deoxyglucose positron emission tomography/computed tomography; H&E, hematoxylin and eosin; LNM, lymph node metastasis; LVI, lymphovascular invasion; TLG, total lesion glycolysis.

Figure 5.

¹⁸F-FDG PET/CT images and postoperative pathology of a 68-year-old male gastric cancer patient without LNM. The CT (A), PET/CT (B), and PET (C) images showed a thickened gastric wall at the cardia of the primary lesion with abnormal tracer uptake (as indicated by the red arrows); no LNM was observed. The TLG of the lesion was 32.2. Based on a TLG threshold of <41.9 for predicting LVI, this patient was diagnosed as LVI−. Postoperative pathology (D; H&E, ×100) revealed moderately differentiated adenocarcinoma at the cardia, with no LNM (0/17 at the lesser curvature) and no intravascular tumor invasion. −, negative; ¹⁸F-FDG PET/CT, ¹⁸F-labeled fluoro-2-deoxyglucose positron emission tomography/computed tomography; H&E, hematoxylin and eosin; LVI, lymphovascular invasion; LNM, lymph node metastasis; TLG, total lesion glycolysis.

Discussion

In gastric cancer, LVI is widely considered a significant risk factor for LNM, and plays a crucial role in assessing the suitability of endoscopic submucosal dissection (ESD) in the treatment of gastric cancer. For early stage gastric cancer patients with LVI, additional surgery is generally recommended after ESD; however, research on clinical significance of LVI beyond this specific context is limited. Some studies have found that the presence of LVI is significantly associated with a higher likelihood of occult LNM in advanced gastric cancer (13). Zhang et al. found that LVI was an independent risk factor for LNM and could be used to predict the prognosis of patients who had undergone curative resection for gastric cancer (4). Thus, the accurate assessment of LVI status is critical in identifying suitable patients for adjuvant chemotherapy and guiding postoperative surveillance strategies.

This retrospective study investigated the correlation between TLG in primary gastric cancer lesions and LVI to enhance the preoperative identification of gastric cancer patients at high risk of LVI. ¹⁸F-FDG PET/CT imaging, which reflects metabolic activity at the molecular level, is a critical tool for tumor staging and has shown significant value in predicting vascular invasion in lung adenocarcinoma (14). However, its utility in assessing LVI in gastric cancer has not been extensively explored.

Using postoperative pathological results as the reference standard, we evaluated the correlation between ¹⁸F-FDG PET/CT metabolic parameters and LVI. The LVI+ group had statistically significantly higher values for the SUVmax, SUVmean, MTV, and TLG than the LVI− group (all P<0.05). This observation aligns with previous studies that reported similar differences in the SUVmax between LVI+ and LVI− groups (7,9,14). However, compared to the SUVmax, TLG, a volumetric measure obtained from PET/CT imaging, reflects the entire metabolic burden of the tumor, providing a more comprehensive assessment of tumor biology (15). Therefore, our study sought to examine the correlation between TLG and LVI, and its diagnostic efficacy.

Our results revealed a positive linear correlation between TLG and the risk of LVI, indicating that larger tumor volumes are associated with an increased risk of LVI, which is consistent with the findings of Chen et al. (16). LVI is recognized as an independent prognostic marker for poor outcomes in gastric cancer, even in patients without LNM (17,18). Zhao et al. noted that LVI can be considered an initial step in LNM and distant metastasis, and its diagnosis indicates the presence of undetectable micrometastases, even in the absence of radiologically or pathologically confirmed LNM or distant metastatic disease (19). Moreover, most LNM may be spread by LVI (20).

This study focused on the preoperative clinical prediction of LNM and LVI. The concordance rate between the presence or absence of LNM between PET/CT imaging and postoperative pathology indicated good reliability. The incidence of LVI is significantly lower in LNM− patients than LNM+ patients (6); thus, the risk of LVI may be underestimated in LNM− patients during clinical assessments. Consequently, we then sought to evaluate the diagnostic efficacy of TLG in predicting LVI in gastric cancer patients with and without LNM. The subgroup analysis revealed that TLG had high sensitivity for diagnosing LVI in the LNM− (82.1%) subgroup, which suggests that higher TLG values in primary gastric lesions are indicative of LVI. This finding could help reduce missed diagnoses and underscores the importance of clinical monitoring.

The results of our study showed that the AUCs for predicting LVI were significant in the gastric cancer patients with or without cLNM (Figure 3A), and the gastric cancer patients without cLNM^_ subgroup (Figure 3C) (P<0.001 and P=0.010, respectively), indicating good discriminative power. The AUC for predicting LVI was non-significant (P=0.168) in the cLNM+ subgroup, the specificity (37.8%, 53.6%) and accuracy (61.3%, 61.9%) in the cLNM+ subgroups were less optimal; however, these results may be related to the limited sample size, and the complexity and variability of metabolic activity in malignant tumors. A single metabolic parameter may not fully capture the spatial heterogeneity of FDG uptake. Future studies should seek to integrate clinical factors and heterogeneity parameters to develop composite models to enhance diagnostic performance.

Previous studies have reported that LVI positivity rates range from 32% to 47.7% (5,21). In our study, the LVI positivity rate was 40.1%, which may be attributed to the sample size and the inclusion of various pathological types of gastric cancer, including pure mucinous adenocarcinoma, pure signet ring cell carcinoma, and adenocarcinoma with mixed signet ring cell and mucinous components. The ¹⁸F-FDG uptake in primary gastric lesions is influenced by the pathological type, and lower uptake is typically observed in signet ring cell carcinoma, and mucinous-rich gastric cancers. This may be due to the abundant extracellular mucin, low tumor cell density, and reduced glucose transporter 1 (GLUT1) expression (22). As the majority of cases in our study were pure gastric adenocarcinoma (87.0%), our results may be more applicable to this subgroup.

This study had several limitations. First, it was a single-center retrospective study with a relatively small sample size; thus, the results need to be validated in multi-center studies. Moreover, in our study cohort, in line with the research approach taken by Yun et al. (10), we defined lymph nodes with FDG uptake similar to or higher than that of the liver as cLNM+ on PET images. However, in real-world settings, we often see cases in which the uptake of metastatic lymph nodes is lower than that of the liver, and thus are concerned that the sensitivity of ¹⁸F-FDG PET/CT for LNM detection may be low. These factors might have led to a bias in the findings. Second, the diverse morphology of primary gastric lesions and physiological gastric uptake might have introduced errors in ROI delineation. Third, TLG had a high sensitivity but a poor specificity and accuracy in predicting LVI, and the AUC for predicting LVI was non-significant in the cLNM+ subgroup. Radiomics analysis of PET/CT images, which can quantitatively and objectively reflect tumor heterogeneity, increase the diagnostic threshold, and refine risk stratification protocols, should be explored in future research.

Conclusions

This study revealed a positive linear correlation between the preoperative TLG of primary gastric cancer lesions and the risk of LVI after adjusting for confounding factors. TLG exhibits high sensitivity in predicting LVI, and thus may aid in the risk stratification of gastric cancer patients for LVI before surgery, especially in those who are cLNM−.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The Third Affiliated Hospital of Soochow University (No. [2020] K104), and the requirement of informed consent was waived due to the retrospective nature of the study.

Data Sharing Statement

Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-644/dss

qims-15-09-7738-dss.pdf ^{(50.5KB, pdf)}

DOI: 10.21037/qims-2025-644