Skip to main content
Download PDF

Predicting tumor invasion depth in gastric cancer: developing and validating multivariate models incorporating preoperative IVIM-DWI parameters and MRI morphological characteristics

European Journal of Medical Research volume 29, Article number: 431 (2024) Cite this article

Abstract

Introduction

Accurate assessment of the depth of tumor invasion in gastric cancer (GC) is vital for the selection of suitable patients for neoadjuvant chemotherapy (NAC). Current problem is that preoperative differentiation between T1-2 and T3-4 stage cases in GC is always highly challenging for radiologists.

Methods

A total of 129 GC patients were divided into training (91 cases) and validation (38 cases) cohorts. Pathology from surgical specimens categorized patients into T1-2 and T3-4 stages. IVIM-DWI and MRI morphological characteristics were evaluated, and a multimodal nomogram was developed. The MRI morphological model, IVIM-DWI model, and combined model were constructed using logistic regression. Their effectiveness was assessed using receiver operating characteristic (ROC) curves, calibration curves, decision curve analysis (DCA), and clinical impact curves (CIC).

Results

The combined nomogram, integrating preoperative IVIM-DWI parameters (D value) and MRI morphological characteristics (maximum tumor thickness, extra-serosal invasion), achieved the highest area under the curve (AUC) values of 0.901 and 0.883 in the training and validation cohorts, respectively. No significant difference was observed between the AUCs of the IVIM-DWI and MRI morphological models in either cohort (training: 0.796 vs. 0.835, p = 0.593; validation: 0.794 vs. 0.766, p = 0.79).

Conclusion

The multimodal nomogram, combining IVIM-DWI parameters and MRI morphological characteristics, emerges as a promising tool for assessing tumor invasion depth in GC, potentially guiding the selection of suitable candidates for neoadjuvant chemotherapy (NAC) treatment.

Introduction

In 2020, gastric cancer (GC) accounted for approximately 1.09 million new cases and 769,000 deaths worldwide, ranking as the fifth most common malignant tumor and the third leading cause of cancer-related death [1]. While resection remains the sole curative treatment for GC, neoadjuvant chemotherapy (NAC) has demonstrated efficacy in enhancing radical tumor resection rates and improving disease-free and overall survival [2]. As per the Chinese Society of Clinical Oncology (CSCO) guidelines for GC, NAC is now the gold standard for treating T3-4N + GC. To mitigate NAC's adverse effects, it is essential to differentiate between patients in the T1-2 stage and those in the T3-4 stage [3]. Therefore, accurate assessment of tumor invasion depth in GC is critical for selecting suitable candidates for NAC therapy.

In a prospective, multi-institutional study conducted by the Japanese Clinical Oncology Group, approximately 23% of gastric cancers classified as stage T3-4 by EUS and CT were actually T1-2 stages [3]. Advancements in magnetic resonance imaging (MRI) technology, including rapid imaging, respiratory compensation, anti-peristaltic medication, and the advent of functional MRI, have significantly enhanced its efficacy as a diagnostic tool for gastrointestinal cancers [4]. A comparative study of MRI, EUS, and CT for T staging in gastric cancer, revealed that MRI demonstrated superior overall accuracy (82.9%) compared to EUS (67.8%) and CT (71.5%) [5].

However, MRI is renowned for providing both quantitative and morphological data. Diffusion-weighted imaging (DWI) has been recognized for its ability to evaluate the aggressiveness, treatment responsiveness, and prognosis of gastric cancer [6]. Intravoxel incoherent motion (IVIM) imaging, an extension of DWI, quantifies capillary perfusion and Brownian motion using a biexponential model with multiple b-values [7, 8]. The IVIM signal comprises a perfusion-influenced pseudo-diffusion fraction and a true diffusion fraction, characterized by three parameters: true diffusion coefficient (D), perfusion coefficient (f), and pseudo-diffusion coefficient (D*) [7]. Numerous studies have confirmed the utility of IVIM-DWI in assessing HER2 status, pathologic subtypes, response to neoadjuvant chemotherapy (NAC), lymphovascular invasion, perineural invasion and prognosis in gastric cancer [9,10,11,12,13,14,15], However, only one study has shown promising results regarding serosa invasion[16]. Therefore, this study focuses on comparing IVIM-DWI parameters, MRI morphological features, and their combined efficacy in differentiating between T1-2 and T3-4 stage gastric cancer cases.

Patients and methods

Patient recruitment

This prospective study, conducted in accordance with the Declaration of Helsinki, received approval from the institutional review board. Informed consent was obtained from all participants after they were provided with relevant information. From October 2016 to October 2021, a total of 226 individuals diagnosed with adenocarcinomas were enrolled based on the criteria depicted in Fig. 1. Eventually, 129 patients (mean age: 60 ± 11 years) were categorized into a training cohort (n = 91) and a validation cohort (n = 38), maintaining a ratio of 7:3.

Fig. 1
figure 1

Flowchart Illustrating Patient Inclusion and Exclusion Criteria

Full size image

MRI protocol

Prior to MRI examinations, patients were instructed to fast for at least 10 h to ensure an empty stomach and also underwent breath-holding training. To minimize artifacts from intestinal peristalsis, patients without contraindications (such as glaucoma, asthma, enlarged prostate, or severe heart disease) received a 20 mg intramuscular injection of butylscopolamine bromide (10 mg; Chengdu NO.1 Drug Research Institute Company Limited, Chengdu, China). Subsequently, they consumed 500–800 ml of water to dilate the stomach before MRI scanning.

The examinations were conducted using a 1.5-Tesla magnetic resonance (MR) scanner (MAGNETOM Aera, Siemens Healthcare, Erlangen, Germany) with a standard 32-channel, phased-array body coil. IVIM-DWI was performed in the transverse plane using a single-shot echo-planar imaging sequence in three orthogonal directions. Diffusion gradients with eight b-values (0, 50, 100, 150, 200, 400, 800, 1600 s/mm2) were applied. During contrast enhancement, a 0.1 mmol/kg contrasting dose of gadolinium-based contrast agent (gadodiamide; Omniscan, Nanjing, China) was given, followed by a 15–20 ml saline flush at a rate of 3 ml/s. Eighteen seconds after the contrast agent was given, the arterial phase was acquired, followed by the portal phase at 55 s and the equilibrium phase at 120 s. Additional gastric routine sequences and detailed scanning parameters can be found in Table 1.

Table 1 IVIM-DWI and routine sequence parameters
Full size table

Image interpretation

Gastric MRI DICOM files were transferred to MITK Diffusion, an open-source software developed by the German Cancer Research Center (version 2018.09, [http://www.mitk.org/]), for post-processing and image analysis. Two board-certified radiologists in gastrointestinal (GI) radiology, with 4 and 20 years of experience, independently reviewed the MRI images in a blinded manner. Each radiologist conducted their own independent assessment, and when differences of opinion developed, they discussed the results until a compromise was achieved. Image quality was assessed using a 4-point scale: 1 indicating no artifacts or distortion, 2 for mild, 3 for moderate, and 4 for severe artifacts. Images consistently rated as 4 by both radiologists were excluded.

For IVIM parameter measurements, T2-weighted imaging (T2WI), IVIM-DWI with b = 800 s/mm2 and Dixon-VIBE T1WI were utilized. Radiologists manually delineated a region of interest (ROI) around the GC lesion's outer boundary on the largest axial plane using a freehand technique. Apparent diffusion coefficient (ADC) values were calculated using two different b-values. The IVIM parameters were derived using the following equations:

$$S \left(b\right)/S\left(0\right)= {f\cdot exp}^{-\text{b}\cdot {\text{D}}^{*}}+ {(1-f)\cdot exp}^{-\text{b}\cdot D}$$

D stood for the slow component of diffusion, which was measured in mm2/s and was able to reflect the actual diffusion of water molecules. The pseudo diffusion coefficient, denoted by D*, was used to indicate the changes in blood perfusion. The fraction of microcirculatory volume perfused, denoted as f.

MRI morphological parameters were also meticulously documented. Maximal tumor thickness was measured perpendicular to the largest tumor diameter [17]. Prior studies [18, 19] have identified a low-signal-intensity band at the interface between surrounding fat tissues and the water-filled stomach wall in out-of-phase imaging. In MRI T1-weighted images (MRT1), there was no abnormal signal intensity across the stomach wall. MRT2 featured a distinct, continuous low-signal-intensity band encircling the lesion. MRT3 was characterized by an irregular or zigzag low-signal-intensity band around the lesion, while MRT4 was marked by the absence of a low-signal-intensity band and the spread of high signal intensity to adjacent structures. The corresponding MRT staging diagram is available in Fig. 2. Thus, extra-serosal invasion was classified as either MRT3 or MRT4, and its absence as MRT1 or MRT2.

Fig. 2
figure 2

MRT staging diagram

Full size image

Clinical and pathological data collection

Pathological and clinical data, encompassing gender, age, tumor location, differentiation, Lauren classification, tumor markers, and T stage, were meticulously collected. The depth of tumor invasion was histopathologically determined according to the eighth edition of the AJCC's TNM cancer staging system[20]. As a result, the gastric neoplasms in our study were classified into T1-2 and T3-4 stages.

Statistical analyses

Statistical analyses were conducted using SPSS software (IBM, version 26.0) and the R software package (R Foundation for Statistical Computing, version 4.2.2). The process involved the following steps:

  1. 1.

    Assessment of observer agreement in measuring ADC, D, D*, and f values using the intraclass correlation coefficient (ICC) with a 95% confidence interval (CI). ICC values interpretation: 0.00–0.20 indicates weak correlation; 0.21–0.40 satisfactory; 0.41–0.60 moderate; 0.61–0.80 strong; 0.81–1.00 outstanding.

  2. 2.

    Normality of data distributions was evaluated using the Kolmogorov–Smirnov test. Continuous variables were analyzed with t-tests or U-tests, and categorical variables with chi-square tests or Fisher’s exact tests.

  3. 3.

    Single-variable and multivariable logistic regression models were applied to explore associations between variables and tumor invasion depth. Univariate analyses identified potential risk factors differentiating T1-2 from T3-4 stages. The multivariate model included variables with a bivariate association with tumor invasion depth at p < 0.1.

  4. 4.

    Screening for collinearity among independent variables, excluding those with variance inflation factors (VIF) > 10.

  5. 5.

    Development of a stepwise nomogram model for estimating tumor invasion depth using multivariate logistic regression, employing a backward step-down selection procedure.

  6. 6.

    Validation of the nomogram's performance in a validation cohort. Receiver operating characteristic (ROC) curves, calibration curves, decision curves, and clinical impact curves were produced and analyzed. The area under the curve (AUC) and optimal cut-off values were calculated using the Youden index. AUC comparisons were performed using the DeLong et al. test [21].

Results

General clinical and imaging data

The demographic data, endoscopic biopsy results, tumor markers, morphological characteristics, and IVIM-DWI parameters of gastric cancer patients in the training and validation cohorts are summarized in Table 2. Overall, among the 129 patients, 68.2% were men, and 48.1% were under 60 years of age. Pathologically, 31.8% were classified as pT1-T2 stage, and 68.2% as pT3-T4 stage. The baseline data showed no significant statistical difference between the cohorts (p = 0.058–1.000), indicating a reasonable and randomized grouping.

Table 2 baseline information of 129 patients with gastric cancer
Full size table

In the training cohort of 91 patients, 62.6% were men and 48.4% were under 60 years. Pathologically, 30.8% were pT1-T2 stage, and 69.2% were pT3-4 stage. The positive rates for serum CEA, CA199, and CA724 were 20.9%, 24.2%, and 24.2%, respectively. According to the Lauren classification, there were 38 cases of intestinal, 27 of diffuse, and 26 of mixed gastric cancer. Tumoral differentiation was moderate to well in 38 cases and poor in 53 cases. Tumors were located in the upper (28.6%), middle (17.6%), lower (49.5%), and whole (4.4%) stomach regions. Extra-serosal invasion was present in 67% of cases. Median maximum tumor thickness, ADC, D, f, and D* values, along with their interquartile ranges (IQR), were 1.6 cm (1.2–2.0 cm), 1.0 × 10–3 mm2/s (0.9–1.2 × 10–3 mm2/s), 1.0 × 10–3 mm2/s (0.9–1.3 × 10–3 mm2/s), 0.3 (0.2–0.4), and 18.9 × 10–3 mm2/s (11.4–82.2 × 10–3 mm2/s), respectively.

Interobserver agreement

The agreement level among the readers' evaluations ranged from moderate to outstanding, with ICC values for ADC, D, D*, f, and maximum tumor thickness recorded as 0.828 (0.765–0.876), 0.900 (0.862–0.929), 0.641 (0.527–0.732), 0.777 (0.698–0.837), and 0.948 (0.927–0.963), respectively.

Development and validation of the combined model, and comparative analysis of MRI morphological and IVIM-DWI models

Three models were developed: an MRI morphological model, an IVIM-DWI model, and a comprehensive model (the nomogram). Multivariate and univariate logistic analyses of all factors were conducted (Table 3). Relevant variables from the univariate analysis were used to construct ROC curves, determining cutoff values to differentiate between T1-2 and T3-4 stage gastric cancer patients. The effectiveness of these models was compared using Decision Curve Analysis (DCA) and Clinical Impact Curve (CIC).

Table 3 univariate and multivariate logistic regression analysis of differentiating T1-2 and T3-4 stage tumors in patients with gastric cancer in training cohort
Full size table

Key findings from the univariate logistic regression analysis, including demographic information, endoscopic biopsy results, tumor markers, morphological characteristics, and IVIM-DWI variables, were integrated into the multivariate analyses. The independent predictors for distinguishing between T1-2 and T3-4 stages were identified as: (1) extra-serosal invasion (OR = 0.101, CI 0.027–0.377, P < 0.001); (2) maximum tumor thickness (OR = 5.251, CI 1.460–18.884, P = 0.011); (3) D value (OR = 0.019, CI 0.002–0.230, P = 0.002). These risk indicators were used to construct a nomogram (Fig. 3). Using the nomogram, influencing factors were assigned scores, which were totaled to determine the probability of T3-4 stage (Figs. 4, 5).

Fig. 3
figure 3

a Preoperative Nomogram for the Combined Model, Utilized for Distinguishing Between T1-2 and T3-4 Stage Tumors in Gastric Cancer Patients. b Training Cohort ROC Curves for the MRI Morphological Model, IVIM-DWI Model, and Combined Model. c Validation Cohort ROC Curves for the MRI Morphological Model, IVIM-DWI Model, and Combined Model

Full size image
Fig. 4
figure 4

Multi-parametric Maps in a T1-2 Stage Gastric Cancer Case. a DWI map (b = 800 s/mm2) showing an irregular mass at the esophagus-cardia junction in a 66-year-old woman with histopathologically confirmed gastric cancer. b Freehand ROI drawn along the tumor's edge. The maximum thickness was 0.96 cm. c True diffusion coefficient (D) image with a D value of 1.24 × 10–3 s/mm2. d Dixon out-of-phase image showing a smooth and clear low-signal band around the lesion (white arrow), indicating no invasion of the subserosa and serosa. e Calculated scores: D value (56 points), maximum tumor thickness (51 points), and Dixon sequence evaluation (60 points), totaling 167 points. This corresponds to a predicted invasion depth of T1-2 with a probability of 0.124. Postoperative pathology confirmed a T2 stage

Full size image
Fig. 5
figure 5

Multi-parametric Maps in a T3-4 Stage Gastric Cancer Case. a DWI map (b = 800 s/mm2) showing an irregular mass in the gastric antrum of a 71-year-old man with histopathologically confirmed gastric cancer. b Freehand ROI drawn along the tumor’s edge. The maximum thickness was 1.91 cm. c True diffusion coefficient (D) map with a D value of 0.84 × 10–3 s/mm2. d Dixon inverse-phase image showing a disrupted and interrupted low-signal band around the lesion (white arrow), indicative of subserosa and serosa invasion. e Calculated scores: D value (67.5 points), maximum tumor thickness (61.5 points), and Dixon sequence evaluation (80 points), totaling 209 points. This corresponds to a predicted invasion depth of T3-4 with a probability of 0.973. Postoperative pathology confirmed a T4a stage

Full size image

C-indices for the combined model were superior to those of individual models, yielding 0.901 (95% CI 0.834–0.969) in the training cohort and 0.852 (95% CI 0.728–0.977) in the validation cohort (Table 4, Fig. 3). The combined model showed significant improvement over the IVIM-DWI and MRI morphological models in the training cohort (P = 0.049 and P = 0.012, respectively). However, no significant difference was observed between the AUCs of the IVIM-DWI and MRI morphological models in either cohort (training: 0.796 vs. 0.835, P = 0.593; validation: 0.794 vs. 0.766, P = 0.79). Representative multi-parametric maps for T1-2 and T3-4 stage gastric cancer cases are illustrated in Figs. 4, 5. The calibration curves (Fig. 6a,b) indicated that prediction accuracy for tumor invasion depth was slightly underestimated at a threshold value of 0.3–0.7 in both cohorts, but overall, the calibration results were acceptable. The DCA (Fig. 6c, d) showed significant deviations of the combined model from standard reference lines, indicating its substantial influence. The combined model demonstrated the greatest clinical benefit across most threshold probability ranges, as evidenced by the CIC (Fig. 6e, f), confirming its high clinical value.

Table 4 Performance of different models in training and validation cohorts
Full size table
Fig. 6
figure 6

a Calibration Curve Analysis for the Nomogram in the Training Cohort. b Calibration Curve Analysis for the Nomogram in the Validation Cohort. c Decision Curve Analysis for the MRI Morphological Model, IVIM-DWI Model, and Combined Model in the Training Cohort. d Decision Curve Analysis for the MRI Morphological Model, IVIM-DWI Model, and Combined Model in the Validation Cohort. e Clinical Impact Curve Analysis for the Nomogram in the Training Cohort. f Clinical Impact Curve Analysis for the Nomogram in the Validation Cohort

Full size image

Discussion

In this study, we analyzed five categories of potential variables in 129 gastric cancer (GC) patients with varying invasion depths. These variables included endoscopic biopsy, demographic data, tumor markers, morphological characteristics, and IVIM-DWI parameters. We constructed and compared three models: an MRI morphological model, an IVIM-DWI model, and a combined model (the nomogram), to predict differentiation between T1-2 and T3-4 stages in GC. The comprehensive model demonstrated the highest diagnostic accuracy, with an AUC of 0.901% in the training set and 0.88% in the validation set. These results suggest that the newly developed nomogram is a promising tool for noninvasively differentiating between T1-2 and T3-4 stage tumors, potentially guiding the selection of suitable candidates for neoadjuvant chemotherapy (NAC) treatment.

MRI effectively delineates the location, extent, and depth of lesion invasion, offering superior soft tissue resolution and multiple sequences without radiation exposure. Prior MRI studies assessing tumor invasion depth in gastric cancer (GC) primarily focused on morphology [18, 19, 22,23,24,25,26,27]. Our study identified larger maximum tumor thickness and absence of extra-serosal invasion as independent predictors for distinguishing between T1-2 and T3-4 stages in GC. Tumors with greater thickness tend to be more aggressive and have a higher likelihood of deeper invasion. The study by M. Matsushita et al. [18, 19] also reported a low-intensity band surrounding abdominal tissue in Dixon sequences. This band's irregularity or disappearance, indicating extra-serosal invasion, is due to chemical-shift misregistration or phase cancellation, which occurs when the resonance frequency of water protons is 220 Hz higher than that of fat protons in a 1.5 T magnetic field. Spatial misregistration of the fat signal can occur, creating signal attenuation through interference between water and fat signals. Our MRI morphological model's sensitivity and specificity (0.873 and 0.786, respectively) differed from previous studies (0.93 and 0.63, respectively) [19], possibly due to variations in sample compositions and the experience of different radiologists.

In addition to morphological insights, MRI also provides valuable quantitative information. Several studies have indicated that gastric neoplasms with lower ADC tend to have deeper invasion [28,29,30], a phenomenon attributed to the ability of DWI with ADC to depict tumor heterogeneity. In our research, however, we discovered that a lower D value, rather than ADC, served as an independent marker for differentiating between T1-2 and T3-4 stages in GC in the IVIM-DWI model. This discrepancy may arise because ADC values are influenced by both tissue microstructure and microcirculation. Consequently, ADC may not accurately represent the real diffusion coefficient, failing to fully capture the complex diffusion characteristics of water molecules in tissues. D values, derived from IVIM-DWI with varying b values, more accurately reflect the hindrance to water molecule diffusion, both intracellularly and extracellularly, caused by cellular components such as membranes, macromolecules, and fibers. Tumors in T3-4 stages typically exhibit higher cell density, reduced free water levels, and more compact extracellular spaces compared to T1-2 stage tumors. By eliminating the confounding effects of blood perfusion, D values more effectively represent the diffusion properties of water molecules within cells, providing a reliable indicator of tumor cell proliferation, including nuclear atypia and nuclear-cytoplasmic ratio. Our study showed strong interobserver agreement for ADC, D value, and maximum tumor thickness measurements, with moderate agreement for f and D* values. This aligns with other research demonstrating relatively lower repeatability of D* values[31]. This result might be the explanation of nonsignificant findings of D* and eventually bring about the downfall of the application of D in clinical practice. The IVIM-DWI model yielded good diagnostic precision; AUC = 0.796, sensitivity = 0.746, and specificity = 0.821. These decent but not optimal results are reflective of the occurrence of a non-negligible overlap in the IVIM parameters' values between GC in the T1-2 stage and the T3-4 stage.

To enhance the accuracy of estimating tumor invasion depth in gastric cancer (GC), we developed a detailed model combining preoperative IVIM-DWI parameters and MRI morphological characteristics, yielding satisfactory results. This contrasts with previous CT-based studies, where visual evaluations using routine CT achieved accuracies ranging from 0.770 to 0.792 in differentiating between T1-2 and T3-4 stages [32, 33]. Wang et al. [17] reported AUCs of 0.899 and 0.825 in the arterial and portal phases, respectively, in their radiomics analysis. Tan et al.[34] found that the integration of radiomics and pathomics features has resulted in a powerful radiopathomics nomogram(AUC, training cohort: 0.937; test cohort: 0.792), providing a promising tool for precise staging of gastric cancer. Liu et al. [35]found that two two-layer MLPs identified through the NAS approach were employed to predict tumor stage, demonstrating greater discrimination with an average accuracy (ACC) of 0.646 for five T stages compared to traditional methods with an ACC of 0.543 (P = 0.034).Our comprehensive model, integrating preoperative IVIM-DWI parameters and MRI morphological characteristics, demonstrated superior diagnostic performance with an AUC of 0.901, a sensitivity of 0.794, and a specificity of 0.893 in the training cohort at the optimal cutoff value. The AUC for our model surpassed those of the MRI morphological and IVIM-DWI models. Consequently, this novel nomogram emerges as a promising tool for assessing tumor invasion depth in GC, potentially guiding the selection of suitable candidates for neoadjuvant chemotherapy (NAC) treatment.

Limitations

This study has several limitations. First, the small patient population may constrain the model's performance, necessitating further external validation by multi-center cooperation. Although we have found that the nomogram shows promise as a tool for differentiating between T1-2 and T3-4 stage tumors in GC, the T staging should be sufficiently detailed to further differentiate between T1-2-3-4 respectively. Second, our IVIM acquisition utilized only eight b values, focusing on initial pseudo-diffusion and molecular diffusion decays without exploring more complex protocols with additional b values. Incorporating a wider range of b values in IVIM sequences could potentially enhance the diagnostic accuracy for GC [36]. Lastly, we did not employ advanced techniques such as machine learning algorithms (for example radiomics, habitat imaging, SVM, PCA) in model development, nor did we assess their potential contribution in differentiating between T1-2 and T3-4 stages in GC.

Conclusion

In summary, we developed and validated combined models incorporating IVIM-DWI parameters and MRI morphological characteristics to distinguish between T1-2 and T3-4 stages in GC. These models outperformed their individual counterparts. The nomogram demonstrated exceptional predictive ability, presenting itself as a noninvasive and practical tool for differentiating between T1-2 and T3-4 stage patients. Ultimately, this nomogram shows promise in identifying patients who may benefit from NAC treatment.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

No datasets were generated or analysed during the current study.

Abbreviations

IVIM:

Intravoxel incoherent motion

DWI:

Diffusion weighted imaging

ADC:

Apparent diffusion coefficient

GC:

Gastric cancer

NAC:

Neoadjuvant chemotherapy

ROI:

Region of interest

ROC:

Receiver operating curve

AUC:

Area under curve

DCA:

Decision curve analysis

CIC:

Clinical impact curve

CI:

Confidence interval

OR:

Odds ratio

CEA:

Carcinoembryonic antigen

CA19-9:

Carbohydrate antigen19-9

CA72-4:

Carbohydrate antigen72-4

ICC:

Intraclass correlation coefficient

References

  1. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Global cancer statistics, et al. GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2020. https://doi.org/10.3322/caac.21660.

    Article  Google Scholar 

  2. Biondi A, Lirosi MC, D’Ugo D, Fico V, Ricci R, Santullo F, et al. Neo-adjuvant chemo(radio)therapy in gastric cancer: Current status and future perspectives. World J Gastrointest Oncol. 2015;7(12):389–400.

    Article  PubMed  PubMed Central  Google Scholar 

  3. Fukagawa T, Katai H, Mizusawa J, Nakamura K, Sano T, Terashima M, et al. A prospective multi-institutional validity study to evaluate the accuracy of clinical diagnosis of pathological stage III gastric cancer (JCOG1302A). Gastric Cancer. 2018;21(1):68–73.

    Article  PubMed  Google Scholar 

  4. Mortensen MB. Novel imaging strategies for upper gastrointestinal tract cancers. Expert Rev Gastroenterol Hepatol. 2015;9(3):295–303.

    Article  CAS  PubMed  Google Scholar 

  5. Seevaratnam R, Cardoso R, McGregor C, Lourenco L, Mahar A, Sutradhar R, et al. How useful is preoperative imaging for tumor, node, metastasis (TNM) staging of gastric cancer? A meta-analysis Gastric Cancer. 2012;15(Suppl 1):S3-18.

    Article  PubMed  Google Scholar 

  6. Giganti F, Tang L, Baba H. Gastric cancer and imaging biomarkers: Part 1 - a critical review of DW-MRI and CE-MDCT findings. Eur Radiol. 2019;29(4):1743–53.

    Article  PubMed  Google Scholar 

  7. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeantet M. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology. 1986;161(2):401–7.

    Article  PubMed  Google Scholar 

  8. Le Bihan D. Intravoxel incoherent motion imaging using steady-state free precession. Magn Reson Med. 1988;7(3):346–51.

    Article  PubMed  Google Scholar 

  9. Song XL, Kang HK, Jeong GW, Ahn KY, Jeong YY, Kang YJ, et al. Intravoxel incoherent motion diffusion-weighted imaging for monitoring chemotherapeutic efficacy in gastric cancer. World J Gastroenterol. 2016;22(24):5520–31.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Ji C, Zhang Q, Guan W, Guo T, Chen L, Liu S, et al. Role of intravoxel incoherent motion MR imaging in preoperative assessing HER2 status of gastric cancers. Oncotarget. 2017;8(30):49293–302.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Zuo HD, Zhang XM. Could intravoxel incoherent motion diffusion-weighted magnetic resonance imaging be feasible and beneficial to the evaluation of gastrointestinal tumors histopathology and the therapeutic response? World J Radiol. 2018;10(10):116–23.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Li HH, Sun B, Tan C, Li R, Fu CX, Grimm R, et al. The value of whole-tumor histogram and texture analysis using intravoxel incoherent motion in differentiating pathologic subtypes of locally advanced gastric cancer. Front Oncol. 2022;12: 821586.

    Article  PubMed  PubMed Central  Google Scholar 

  13. Zhu Y, Jiang Z, Wang B, Li Y, Jiang J, Zhong Y, et al. Quantitative dynamic-enhanced mri and intravoxel incoherent motion diffusion-weighted imaging for prediction of the pathological response to neoadjuvant chemotherapy and the prognosis in locally advanced gastric cancer. Front Oncol. 2022;12: 841460.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Li J, Wang Y, Zhang HK, Xu SN, Chen XJ, Qu JR. The value of intravoxel incoherent motion diffusion-weighted imaging in predicting perineural invasion for resectable gastric cancer: a prospective study. Clin Radiol. 2023. https://doi.org/10.1016/j.crad.2023.09.014.

    Article  PubMed  Google Scholar 

  15. Li J, Yan LL, Zhang HK, Wang Y, Xu SN, Chen XJ, et al. Application of intravoxel incoherent motion diffusion-weighted imaging for preoperative knowledge of lymphovascular invasion in gastric cancer: a prospective study. Abdom Radiol. 2023;48(7):2207–18.

    Article  Google Scholar 

  16. Zeng Q, Hong Y, Cheng J, Cai W, Zhuo H, Hou J, et al. Quantitative study of preoperative staging of gastric cancer using intravoxel incoherent motion diffusion-weighted imaging as a potential clinical index. Eur J Radiol. 2021;141: 109627.

    Article  PubMed  Google Scholar 

  17. Wang Y, Liu W, Yu Y, Liu JJ, Jiang L, Xue HD, et al. Prediction of the depth of tumor invasion in gastric cancer: potential role of CT radiomics. Acad Radiol. 2020;27(8):1077–84.

    Article  PubMed  Google Scholar 

  18. Matsushita M, Oi H, Murakami T, Takata N, Kim T, Kishimoto H, et al. Extraserosal invasion in advanced gastric cancer: evaluation with MR imaging. Radiology. 1994;192(1):87–91.

    Article  CAS  PubMed  Google Scholar 

  19. Oi H, Matsushita M, Murakami T, Nakamura H. Dynamic MR imaging for extraserosal invasion of advanced gastric cancer. Abdom Imaging. 1997;22(1):35–40.

    Article  CAS  PubMed  Google Scholar 

  20. Edge SB. AJCC cancer staging manual. Berlin: Springer; 2010.

    Google Scholar 

  21. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–45.

    Article  CAS  PubMed  Google Scholar 

  22. Kim IY, Kim SW, Shin HC, Lee MS, Jeong DJ, Kim CJ, et al. MRI of gastric carcinoma: results of T and N-staging in an in vitro study. World J Gastroenterol. 2009;15(32):3992–8.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Huang Z, Xie DH, Guo L, Hu CH, Fang X, Meng Q, et al. The utility of MRI for pre-operative T and N staging of gastric carcinoma: a systematic review and meta-analysis. Br J Radiol. 2015;88(1050):20140552.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Renzulli M, Clemente A, Spinelli D, Ierardi AM, Marasco G, Farina D, et al. Gastric Cancer Staging: Is It Time for Magnetic Resonance Imaging? Cancers. 2020;12(6):1402.

    Article  PubMed  PubMed Central  Google Scholar 

  25. De Vuysere S, Vandecaveye V, De Bruecker Y, Carton S, Vermeiren K, Tollens T, et al. Accuracy of whole-body diffusion-weighted MRI (WB-DWI/MRI) in diagnosis, staging and follow-up of gastric cancer, in comparison to CT: a pilot study. BMC Med Imaging. 2021;21(1):18.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Gai Q, Li X, Li N, Li L, Meng Z, Chen A. Clinical significance of multi-slice spiral CT, MRI combined with gastric contrast-enhanced ultrasonography in the diagnosis of T staging of gastric cancer. Clin Transl Oncol. 2021;23(10):2036–45.

    Article  CAS  PubMed  Google Scholar 

  27. Yan L, Qu J, Li J, Zhang H, Lu Y, Gao J. Predicting T and N staging of resectable gastric cancer according to whole tumor histogram analysis about a non-cartesian k-space acquisition DCE-MRI: a feasibility study. Cancer Manag Res. 2021;13:7951–60.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Liu S, Wang H, Guan W, Pan L, Zhou Z, Yu H, et al. Preoperative apparent diffusion coefficient value of gastric cancer by diffusion-weighted imaging: correlations with postoperative TNM staging. J Magn Reson Imaging. 2015;42(3):837–43.

    Article  PubMed  Google Scholar 

  29. Liu S, Zhang Y, Chen L, Guan W, Guan Y, Ge Y, et al. Whole-lesion apparent diffusion coefficient histogram analysis: significance in T and N staging of gastric cancers. BMC Cancer. 2017;17(1):665.

    Article  PubMed  PubMed Central  Google Scholar 

  30. Soydan L, Demir AA, Torun M, Cikrikcioglu MA. Use of diffusion-weighted magnetic resonance imaging and apparent diffusion coefficient in gastric cancer staging. Curr Med Imaging. 2020;16(10):1278–89.

    Article  CAS  PubMed  Google Scholar 

  31. Lecler A, Duron L, Zmuda M, Zuber K, Bergès O, Putterman M, et al. Intravoxel incoherent motion (IVIM) 3 T MRI for orbital lesion characterization. Eur Radiol. 2021;31(1):14–23.

    Article  PubMed  Google Scholar 

  32. Kim AY, Kim HJ, Ha HK. Gastric cancer by multidetector row CT: preoperative staging. Abdom Imaging. 2005;30(4):465–72.

    Article  CAS  PubMed  Google Scholar 

  33. Kim HJ, Kim AY, Oh ST, Kim JS, Kim KW, Kim PN, et al. Gastric cancer staging at multi-detector row CT gastrography: comparison of transverse and volumetric CT scanning. Radiology. 2005;236(3):879–85.

    Article  PubMed  Google Scholar 

  34. Tan Y, Feng LJ, Huang YH, Xue JW, Long LL, Feng ZB. A comprehensive radiopathological nomogram for the prediction of pathological staging in gastric cancer using CT-derived and WSI-based features. Transl Oncol. 2024;40: 101864.

    Article  CAS  PubMed  Google Scholar 

  35. Liu S, Liang W, Huang P, Chen D, He Q, Ning Z, et al. Multi-modal analysis for accurate prediction of preoperative stage and indications of optimal treatment in gastric cancer. Radiol Med. 2023;128(5):509–19.

    Article  PubMed  Google Scholar 

  36. Cohen AD, Schieke MC, Hohenwalter MD, Schmainda KM. The effect of low b-values on the intravoxel incoherent motion derived pseudodiffusion parameter in liver. Magn Reson Med. 2015;73(1):306–11.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

We thank the investigators at all participating study sites.

Funding

This project has received funding from the Key Medical and Health Projects of Xiamen (No. 3502Z20234007) and the Xiamen Medical and Health Guidance (No. 3502Z20214ZD1032 and 3502Z20244ZD1069).

Author information

Author notes

  1. Yanling Hong and Xiaoqing Li have contributed equally to this article.

Authors and Affiliations

  1. Department of Radiology, Zhongshan Hospital of Xiamen University, School of Medicine, Xiamen University, Xiamen, Fujian, China

    Yanling Hong, Xiaoqing Li, Congcong Fu, Miaomiao Nie, Chenghui Chen, Hao Feng & Qiang Zeng

  2. Department of Medical Imaging Center, Zhongshan Hospital of Xiamen University, School of Medicine, Xiamen University, Xiamen, Fujian, China

    Shufen Gan

  3. Xiamen Municipal Key Laboratory of Gastrointestinal Oncology, Xiamen, Fujian, China

    Qiang Zeng

  4. Department of Pathology, Zhongshan Hospital of Xiamen University, School of Medicine, Xiamen University, Xiamen, Fujian, China

    Zhengjin Liu

Authors
  1. Yanling Hong
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Xiaoqing Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Zhengjin Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Congcong Fu
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Miaomiao Nie
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Chenghui Chen
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Hao Feng
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Shufen Gan
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Qiang Zeng
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

QZ and SG were the guarantor of integrity of entire study; YH and XL were involved in study concepts/study design or data acquisition or data analysis/interpretation; ZL, CF, MN, CC and HF were involved in clinical studies; YH contributed to statistical analysis; all authors were involved in manuscript drafting or manuscript revision for important intellectual content; all authors agree to ensure any question related to the work are appropriately resolved. All authors have read and approved the final manuscript.

Corresponding authors

Correspondence to Shufen Gan or Qiang Zeng.

Ethics declarations

Ethics approval and consent to participate

All research performed has received approval from the institutional review board of Zhongshan Hospital of Xiamen University. Informed consent was obtained from all participants.

Consent for publication

Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hong, Y., Li, X., Liu, Z. et al. Predicting tumor invasion depth in gastric cancer: developing and validating multivariate models incorporating preoperative IVIM-DWI parameters and MRI morphological characteristics. Eur J Med Res 29, 431 (2024). https://doi.org/10.1186/s40001-024-02017-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s40001-024-02017-w

Keywords

European Journal of Medical Research

ISSN: 2047-783X

Contact us