Prediction model of lymph node metastasis for early gastric cancer: a better choice than computed tomography

DOI: https://doi.org/10.21203/rs.3.rs-1984319/v1

Abstract

Objective: The purpose of this study was to establish and validate a nomogram for predicting lymph node metastasis in early gastric cancer and to compare it with the predictive power of computed tomography (CT).

Methods: Patients with early gastric cancer (2016-2021) from the First Affiliated Hospital of Nanchang University were included in the study. A nomogram was constructed according to stepwise regression analysis and logistic regression analysis.

Results: In the validation cohort, the incidence of lymph node metastasis was 15.67%. Multivariate logistic regression revealed that 7 variables are associated with lymph node metastasis in early gastric cancer. According to stepwise regression analysis, 5 variables were screened to construct a nomogram, including T stage, total bilirubin (TB), Lauren typing, γ-glutamyl transpeptidase (γ-GT), vascular invasion. the AUCs of the ROC for the nomograms in the training cohort and the validation cohort is 0.795 (95% CI: 0.754–0.837) and 0.729 (95% CI: 0.655–0.803), respectively, higher than the AUCs of the CT in the training cohort and the validation cohort.

Conclusion: The constructed nomogram has good performance and discrimination, which is better than CT, and successfully visualizes risk factors associated with LN metastasis in early gastric cancer.

Introduction

Gastric cancer (GC) is the third leading cause of cancer death and the fifth most common cancer in the world(1). In East Asia, the incidence of gastric cancer is the highest, but with the improvement of endoscopic technology in recent years and the timely eradication of Helicobacter pylori, the incidence has declined(2, 3). Early gastric cancer (EGC) is mainly confined to the mucosa and submucosa, with or without lymph node metastasis(4). For well-differentiated early gastric cancer (T1), endoscopic resection (ER) has been considered to be a more advantageous treatment than traditional surgery(5). ER mainly includes endoscopic mucosal resection (EMR) and endoscopic submucosal dissection (ESD). ESD has many characteristics such as being minimally invasive and more economical. International guidelines recommend ESD as the first-line treatment of EGC(6, 7).

Although endoscopic resection has many advantages, it cannot dissect already metastatic lymph nodes. Kim et al (8) included 3926 gastric cancer patients for univariate and multivariate analysis. The results showed that lymph node metastasis was the most important prognostic factor (P < 0.001). A previous study found that only 20% of EGC patients had lymph node metastases after radical gastrectomy and D2 lymphadenectomy. This means that most patients are overtreated(9). It is very necessary to evaluate the status of lymph node metastasis in early gastric cancer. The existing methods for evaluating tumor lymph node metastasis mainly include contrast-enhanced CT, B-ultrasonography, and Magnetic Resonance Imaging (MRI)(10, 11). Previous studies reported that traditional CT techniques were only 60% accurate in predicting LNM status (11–13)because traditional CT is difficult to detect small lymph node metastases and is prone to misdiagnosis when there is a lymph node enlargement caused by inflammation. To assess the status of lymph node metastasis more accurately, a predictive model based on individual information needs to be established.

The nomogram has been used in the construction of various prediction models due to its simple operation and intuitive features(14–16). So far, many predictive models for LNM in patients with early gastric cancer have been constructed(17–20). But none of the studies have explored whether the prediction effect of the prediction model is higher than that of CT, so this article aims to establish a new prediction model through the First Affiliated Hospital of Nanchang University (NCU1h) cohort (2016–2021) and compare it with CT, to determine whether this predictive model is more conducive for clinicians to assess risk.

Methods

Patients

This study included data(n = 1062) from patients after NCU1H surgery from January 2016 to December 2021. The inclusion criteria were as follows: (i) patients older than 18 years old and (ii) diagnosed with early gastric cancer. (iii) Patients who received radical gastrectomy and D1+/D2 regional lymph node dissection. (iv) No preoperative treatment. The exclusion criteria were as follows: (i) patients with missing data concerning age, sex, differentiation, Lauren typing, TNM category, Vascular invasion, CT diagnosis, laboratory examinations. (ii) Gastric stump cancer, and Multiple cancers (Fig. 1). This study was approved by the Ethics Committee of NCU1h. All patients provided written informed consent before.

Multi-row spiral CT scan

The patient was scanned with a SIEMENS SOMATOM Definition AS 128-layer spiral CT scanner. The patients fasted for 6–8 hours before the examination, and the scanning range was from the top of the diaphragm to the level of the iliac spine. 80-100ml of iohexol was injected through the median cubital vein, and CT-enhanced scanning was performed in the active venous phase and the equilibrium phase, and the scan slice thickness was 5.0 mm. Drink 500-1000ml of water 10 minutes before the CT scan to distend the stomach. One highly qualified radiologist and two or more gastrointestinal surgeons predict tumor TNM staging by enhancing the arterial, venous, and equilibrium phases of CT images. The presence of lymph node metastasis is considered when the lymph nodes show a short diameter > 0.6 cm and high enhancement or uneven enhancement on CT images.

Characteristics of variables

All clinical and pathological data of the patient were collected. These variables include age, gender, vascular invasion, tumor size (cm), Lauren classification (intestinal type, diffuse, hybrid), T stage, N stage, degree of differentiation (well-differentiated, mediated differentiated, poorly differentiated/undifferentiated), CT diagnosis, TB (µmol/L), γ- GT (U/L), uric acid (µmol/L), and blood sugar(mmol/L). Among them, the T/N stage was classified based on postoperative pathological examination according to the International Union for Cancer Control/American Joint Committee on Cancer 7th edition TNM staging system (2010). T stage was described as T1a and T1b. Tumor size was divided into two groups: <3cm, >=3cm. TB determines the cut-off values based on the ROC curve and divides them into two groups: <7.85µmol, >=7.85µmol. γ- GT determines the cut-off values based on the ROC curve and divides them into two groups: <11.5U/L, >=11.5U/L. Uric acid determines the cut-off values based on the ROC curve and divides them into two groups: <281.25µmol/L, >=281.25µmol/L. Blood sugar determines the cut-off values based on the ROC curve and divides them into two groups: <5.745mmol/L, >=5.745mmol/L. In addition, In the postoperative pathological examination results, the degree of tumor differentiation was divided into three categories: poorly differentiated tumors, moderately differentiated tumors, and well-differentiated tumors.

Statistical analysis

SPSS 26.0 for Windows (SPSS, IL, USA) was used for statistical analysis of the data to determine risk factors, and the chi-square test was used to determine significance among categorical variables. Calculate the cut-off value of a continuous variable from the Receiver operating characteristic (ROC) curves(21). Patients were randomized using the 'caret' package in R version 4.0.2, with 70% of the training cohort (n = 743) and 30% of the validation cohort (n = 319). In the training cohort, variables that were significant in univariate logistic regression analysis (P < 0.05) were included in multivariate logistic regression analysis. According to the multivariate logistic regression analysis results, the odd ratio (OR, using the ratio of exposed to non-exposed people in the case group divided by the ratio of exposed to non-exposed people in the control group) and 95% CI of each independent risk factor were obtained(22). Screening of the best predictors using stepwise regression analysis in the training cohort. According to the results of stepwise regression analysis, use the rm software package in R studio to make a nomogram for the filtered variables. The ROC curve was plotted using the “pROC” package, decision curves analysis (DCA) was plotted using the "rmda" function package. Calibration curves are plotted using the "RMS" function package. ROC curves and DCA were used to validate the clinical predictive effect of nomograms.

Results

The basic characteristic of patients

This study included 1062 patients in NCU1h, including 743 patients in the training cohort and 319 patients in the validation cohort. The basic clinical characteristics of all patients are summarized in Table 1. As shown in the table, there are 837 patients over 50 years old, accounting for 78.81% of all patients, and 686 (64.60%) male patients, A small number of patients had lymph node metastasis (18.46%), in the Lauren classification, more patients had intestinal type (67.61%). Regarding the T stage, there were 514 patients with T1a stage and 548 patients with T1b stage. Most of the patients had well-differentiated tumors (50.47%), and 683 patients (64.31%) had tumors smaller than 3 cm. In laboratory examinations, all continuous variables were determined with cut-off values using ROC curves (Supplementary Fig. 1). We found that there were 726 patients (68.36%) with TB ≥ 7.85 µmol/L, 892 patients (83.99%) with γ-GT ≥ 11.5U/L, and 634 patients (59.70%) with uric acid ≥ 281.25 µmol/L), 785 patients (73.92%) had blood sugar < 5.745mmol/L. All items had no significant difference between the training cohort and the validation cohort (P > 0.05).

Table 1

Basic characteristics of patients diagnosed as early gastric cancer

Variables

Total (%)

Training set, n

(%)

Validation set, n (%)

P Value

n

1062

743(70%)

319(30%)

  

Age

      

0.623

< 50

225(21.19%)

154(21.73%)

71(22.26%)

  

>=50

837(78.81%)

589(79.27%)

248(77.74%)

  

Sex

      

0.331

Male

686(64.60%)

473(63.66%)

213(66.77%)

  

Female

376(35.40%)

270(36.34%)

106(33.23%)

  

Lymph node Metastasis

      

0.126

Negative

866(81.54%)

597(80.35%)

269(84.33%)

  

Positive

196(18.46%)

146(19.65%)

50(15.67%)

  

Tumor size

      

0.471

< 3cm

683 (64.31%)

483(65.01%)

200(62.70%)

  

>=3cm

379(35.69%)

260(34.99%)

119(37.30%)

  

Lauren typing

      

0.390

Intestinal type

718(67.61%)

493(66.35%)

225(79.94%)

  

Diffuse

178(16.76%)

128(17.23%)

50(15.67%)

  

Hybrid

166(15.63%)

122(16.42%)

44(13.79%)

  

T Stage

      

0.958

T1a

514(48.40%)

360(48.45%)

154(48.28%)

  

T1b

548(51.60%)

383(51.55%)

165 (51.72%)

  

Grade

      

0.689

Well-differentiated

536(50.47%)

380(51.14%)

156(48.90%)

  

Mediated differentiated

472(44.44%)

324(43.61%)

148(46.40%)

  

Poorly differentiated/undifferentiated

54(5.08%)

39(5.25%)

15(4.70%)

  

Vascular invasion

      

0.351

No

969(91.24%)

674 (90.71%)

295(92.48%)

  

Yes

93(8.76%)

69(9.29%)

24(7.52%)

  

CT diagnosis

      

0.464

No

918(86.44%)

646(86.94%)

272(85.27%)

  

Yes

144(13.56%)

97(13.06%)

47(14.73)

  

Total bilirubin

      

0.245

< 7.85 µmol/L

336(31.64%)

227(30.55%)

109(34.17%)

  

>=7.85 µmol/L

726(68.36%)

516(69.45%)

210(65.83%)

  

γ-GT

      

0.991

< 11.5 U/L

170(16.01%)

119(16.02%)

51(15.99%)

  

>=11.5 U/L

892(83.99%)

624(83.98%)

268(84.01%)

  

Uric acid

      

0.458

< 281.25 µmol/L

428(40.30%)

294(39.57%)

134(42.01%)

  

>=281.25 µmol/L

634(59.70%)

449(60.43%)

185(57.99%)

  

Blood sugar

      

0.563

< 5.745 mmol/L

785(73.92%)

553(74.43%)

232(72.73%)

  

>=5.745 mmol/L

277(26.08%)

190(25.57%)

87(27.27%)

  

Construction and validation of EGC lymph node metastasis prediction model

Logistic univariate and multivariate regression analysis was established according to the training cohort. The results showed that tumor size, Lauren typing (P < 0.001, OR 2.818, 95%CI 1.624–4.889), T stage (P < 0.001, OR 3.493, 95%CI 2.146–5.685), vascular invasion (P < 0.001, OR 8.111, 95%CI 4.460–14.150), TB (P = 0.004, OR 0.526 95%CI 0.339–0.816), γ-GT (P = 0.017 OR 0.516, 95%CI 0.300-0.888), and blood sugar were independent risk factors for LN metastasis. Table 2 shows OR values and 95% CI for all independent risk factors. Subsequently, these related factors of LN metastasis were included in the stepwise regression analysis, and the results showed that the model constructed by T stage, Lauren classification, vascular invasion, γ-GT, and TB had the best predictive effect (R2 = 0.233). Select these variables to construct a nomogram of lymph node metastasis (Fig. 2). The characteristics of each variable have a corresponding score. According to the nomogram, the total score can be calculated and the corresponding probability of LN metastasis can be obtained.

Table 2

Risk factors affecting early gastric cancer lymphatic metastasis, according to univariate and multivariate Logistic analyses.

Variables

Univariate analysis

P Value

Multivariate analysis

P Value

Age

        

< 50

-

-

-

-

>=50

1.014(0.648–1.586)

0.953

-

-

Sex

        

Female

-

-

-

-

Male

0.510(0.354–0.736)

0.000

0.762(0.476–1.218)

0.256

Tumor size

        

< 3cm

-

-

-

-

>=3cm

1.714(1.186–2.477)

0.004

1.546(1.008–2.372)

0.046

Grade

        

Well-differentiated

-

-

-

-

Mediated differentiated

7.432(0.998–55.315)

0.050

3.152(0.411–24.172)

0.269

Poorly differentiated/undifferentiated

12.139(1.644–89.642)

0.014

3.735(0.481–28.978)

0.207

Lauren typing

        

Intestinal type

-

-

-

-

Diffuse

1.514(0.926–2.474)

0.098

1.672(0.875–3.196)

0.120

Hybrid

3.309(2.125–5.154)

0.000

2.818(1.624–4.889)

0.000

T Stage

        

T1a

-

-

-

-

T1b

4.554(2.968–6.988)

0.000

3.493(2.146–5.685)

0.000

Vascular invasion

        

No

-

-

-

-

Yes

12.387(7.151–21.456)

0.000

8.111(4.460–14.750)

0.000

Total bilirubin

        

< 7.85 µmol/L

-

-

-

-

>=7.85 µmol/L

0.577(0.397–0.841)

0.004

0.526(0.339–0.816)

0.004

γ-GT

        

< 11.5 U/L

-

-

-

-

>=11.5 U/L

0.469(0.302–0.728)

0.001

0.516(0.300-0.888)

0.017

Uric acid

        

< 281.25 µmol/L

-

-

-

-

>=281.25 µmol/L

1.065(0.735–1.545)

0.738

-

-

Blood sugar

        

< 5.745 mmol/L

-

-

-

-

>=5.745 mmol/L

0.604(0.383–0.952)

0.030

0.561(0.327–0.961)

0.035

The calibration curves of the nomogram and CT in the training cohort and the validation cohort are summarized in Fig. 3, indicating that the predicted results of the nomogram are in better agreement with the actual results. Subsequently, we created a DCA plot of nomogram and CT (Fig. 4). DCA plots show that nomograms are more predictive than CT in both the training and validation cohorts. Similarly, the AUCs of the ROC for the nomograms in the training and validation cohorts in Fig. 5 are 0.795 and 0.729, respectively, while the CTs are 0.564 and 0.543, respectively. Taken together, these results show a stronger predictive performance of the nomogram in the clinic.

Discussion

This paper builds a new predictive model and compares it with the predictive power of CT. The establishment of predictive models makes these risk factors more intuitive and more visible. The high AUC value of the nomogram and the good calibration performance shown by the calibration curve both demonstrate the better predictive power of the nomogram than the CT. These results give us greater certainty that predictive models can better inform clinicians about treatment decisions.

Our findings suggested that T stage, total bilirubin, Laurent classification, vascular invasion, blood sugar, tumor size, and γ-GT were independent risk factors for LNM. Mu et al.(23) constructed a predictive model for LNM based on clinical patient data and found that tumor size, vascular invasion, degree of differentiation, and invasion depth were considered high-risk factors for LNM through univariate and multivariate analysis. While the authors emphasize that vascular invasion plays a crucial role in the assessment of LNM as it accounts for a large proportion of the nomogram score, which is consistent with our findings. Previous studies have also reported that vascular invasion is one of the important steps in lymph node metastasis(24), which also explains the high risk of vascular invasion. Therefore, the inclusion of vascular infiltration in the model can greatly improve the prediction accuracy, making the model more consistent with the actual situation.

As a reliable classification method for gastric cancer, the Lauren classification has been used in clinical treatment guidelines, which categorize gastric cancer into intestinal type, diffuse type, and mixed type(25). In general, compared with the diffuse type, the intestinal type has a lower risk of LN metastasis and is more suitable for ER. The multivariate results of this article show that the diffuse type has a higher odd ratio, but it is not statistically significant. The mixed type not only has a higher risk of LN metastasis than the intestinal type but also has a significant statistical significance. Pyo et al.(26) included 5309 EGC patients in the study, of whom 495 (9.3%) had LN metastases. The positive rate of mixed type (15.4%) was significantly higher than that of intestinal type (7.2%) and diffuse type (10.6%). This may be because mixed tumors are commonly found in the upper third of the stomach, are larger, and are prone to submucosal infiltration. Therefore, in the assessment of LNM metastasis, more attention should be paid to mixed gastric cancer.

Moreover, T stage and tumor size have also been considered by many studies as risk factors for LN metastasis in EGC,(23, 27, 28). Poorer T stage and larger tumor size make it easier for lesions to invade the submucosa, resulting in a higher risk of LN metastasis. It is also the reason why the prediction model of this article is included in the T stage.

Laboratory examinations of patients with EGC were also considered, and TB and γ-GT were incorporated into the prediction model. In this article, patients with TB ≥ 7.85 µmol/L (OR = 0.526, 95%CI 0.339–0.816, P = 0.004) and patients with γ-GT ≥ 11.5 U/L (OR = 0.516, 95%CI 0.300-0.888, P = 0.017) had a lower probability of LN metastasis. To find out the relationship between TB and LNM, we found that serum bilirubin had anticancer activity through related experiments and clinical research(29–31). It has been reported that oxidative stress is associated with the development and prognosis of cancer(32). As an antioxidant, bilirubin is likely to prevent the development of cancer(33). which explained our research conclusion to a certain extent. Wei et al.(34) found that total bilirubin levels, as well as direct and indirect bilirubin, were significantly reduced in gastric cancer patients. Another study constructed a survival-related nomogram by including variables such as total bilirubin and albumin in 778 gastric cancer patients and found that patients with lower total bilirubin and albumin had a shorter five-year overall survival rate (P < 0.01)(35). As for γ-GT, according to a multivariate analysis of a clinical study(36), high levels of γ-GT (> 14U/L, HR = 1.004, 95%CI 1.002–1.007, P = 0.001) was associated with lower overall survival in advanced gastric cancer. This is inconsistent with our conclusions, most likely because of heterogeneity of data from different institute.

Finally, our results found that blood sugar was also a risk factor for LN metastasis in EGC patients. The study by Deng et al.(37) found that Glucose-derived AGEs (Advanced glycation end products) were highly expressed in tumor tissue and blood of GC patients. The depth of tumor invasion and lymph node metastasis were associated with AGEs. This also explains why hyperglycemic EGC patients have a higher risk of LN metastasis.

Of course, this study has certain limitations. First, this is a single-center retrospective study, and to improve the accuracy of the nomogram, we should validate it with more diverse populations. Second, this paper does not incorporate genetic information such as P53, HER-2, and molecular factors of the tumor microenvironment associated with LNM, which may make the nomogram more accurate.

Conclusion

In conclusion, we constructed a new prediction model for EGC LN metastasis and presented the risk factors in a nomogram. Subsequent comparisons with the predictive power of CT demonstrated the higher predictive performance of the predictive model, thereby helping clinicians determine more appropriate treatment modalities.

Abbreviations

Abbreviations                             Full name

  EGC                                     Early gastric cancer

  γ-GT                                    γ-glutamyl transpeptidase

  GC                                      Gastric cancer

  ER                                      Endoscopic resection

  LNM                                    Lymph node metastasis

  ROC                                     Receiver operating characteristic

  OR                                      Odd ratio

  DCA                                     Decision curve analysis

  TB                                      Total bilirubin

  EMR                                    Endoscopic mucosal resection

  ESD                                     Endoscopic submucosal dissection

  Advanced glycation end products              AGEs

Declarations

Ethics approval and consent to participate

All case data included in the study were obtained from the NCU1h Center for Human Genetic Resources. It was approved by the NCU1h Ethics Committee(Ethical number: (2022)CDYFYYLK(06-025)). All research involving human data is conducted in accordance with the Declaration of Helsinki. All patients provided written informed consent.

Consent for publication

Not appliable

Availability of data and materials

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

Competing interests

The authors disclose no conflicts of interest.

Funding

The Foundation of Jiangxi provincial department of Science and Technology (grant No. 20201ZDG02007 and No.20161ACG70014, PI: You-xiang Chen; grant NO. 20202BAB206051, PI: Chunyan Zeng); This study was supported by grants from the National Natural Science Foundation of China (Grant No. 81560398, PI: Youxiang Chen, Grant No. 81660404, PI: Chunyan Zeng). All funders provided support to authors and paid the fee for statistical analysis.

Author contributions

Peng Wang and Chaotao Tang: experiment performing, data analysis, and manuscript writing. Jun Li and RuiRi Jin: sample collecting and data analysis. Chunyan Zeng and Youxiang Chen: project development.

 

Acknowledgments

Not appliable

References

  1. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68:394–424.
  2. Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray F. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer. 2015;136:E359-386.
  3. Kumar S, Metz DC, Ellenberg S, Kaplan DE, Goldberg DS. Risk Factors and Incidence of Gastric Cancer After Detection of Helicobacter pylori Infection: A Large Cohort Study. Gastroenterology. 2020;158:527–536 e527.
  4. Van Cutsem E, Sagaert X, Topal B, Haustermans K, Prenen H. Gastric cancer. The Lancet. 2016;388:2654–2664.
  5. Isomoto H, Shikuwa S, Yamaguchi N, Fukuda E, Ikeda K, Nishiyama H, Ohnita K, Mizuta Y, Shiozawa J, Kohno S. Endoscopic submucosal dissection for early gastric cancer: a large-scale feasibility study. Gut. 2009;58:331–336.
  6. Pimentel-Nunes P, Dinis-Ribeiro M, Ponchon T, Repici A, Vieth M, De Ceglie A, Amato A, Berr F, Bhandari P, Bialek A, Conio M, Haringsma J, Langner C, Meisner S, Messmann H, Morino M, Neuhaus H, Piessevaux H, Rugge M, Saunders BP, Robaszkiewicz M, Seewald S, Kashin S, Dumonceau JM, Hassan C, Deprez PH. Endoscopic submucosal dissection: European Society of Gastrointestinal Endoscopy (ESGE) Guideline. Endoscopy. 2015;47:829–854.
  7. Ono H, Yao K, Fujishiro M, Oda I, Nimura S, Yahagi N, Iishi H, Oka M, Ajioka Y, Ichinose M, Matsui T. Guidelines for endoscopic submucosal dissection and endoscopic mucosal resection for early gastric cancer. Digestive endoscopy: official journal of the Japan Gastroenterological Endoscopy Society. 2016;28:3–15.
  8. Kim JP, Kim YW, Yang HK, Noh DY. Significant prognostic factors by multivariate analysis of 3926 gastric cancer patients. World journal of surgery. 1994;18:872–877; discussion 877–878.
  9. Kawata N, Kakushima N, Takizawa K, Tanaka M, Makuuchi R, Tokunaga M, Tanizawa Y, Bando E, Kawamura T, Sugino T, Kusafuka K, Shimoda T, Nakajima T, Terashima M, Ono H. Risk factors for lymph node metastasis and long-term outcomes of patients with early gastric cancer after non-curative endoscopic submucosal dissection. Surgical endoscopy. 2017;31:1607–1616.
  10. Cardoso R, Coburn N, Seevaratnam R, Sutradhar R, Lourenco LG, Mahar A, Law C, Yong E, Tinmouth J. A systematic review and meta-analysis of the utility of EUS for preoperative staging for gastric cancer. Gastric cancer: official journal of the International Gastric Cancer Association and the Japanese Gastric Cancer Association. 2012;15 Suppl 1:S19-26.
  11. Kim AY, Kim HJ, Ha HK. Gastric cancer by multidetector row CT: preoperative staging. Abdominal imaging. 2005;30:465–472.
  12. Kim HJ, Kim AY, Oh ST, Kim JS, Kim KW, Kim PN, Lee MG, Ha HK. Gastric cancer staging at multi-detector row CT gastrography: comparison of transverse and volumetric CT scanning. Radiology. 2005;236:879–885.
  13. Saito T, Kurokawa Y, Takiguchi S, Miyazaki Y, Takahashi T, Yamasaki M, Miyata H, Nakajima K, Mori M, Doki Y. Accuracy of multidetector-row CT in diagnosing lymph node metastasis in patients with gastric cancer. European radiology. 2015;25:368–374.
  14. Shariat SF, Karakiewicz PI, Suardi N, Kattan MW. Comparison of nomograms with other methods for predicting outcomes in prostate cancer: a critical analysis of the literature. Clinical cancer research: an official journal of the American Association for Cancer Research. 2008;14:4400–4407.
  15. International Bladder Cancer Nomogram C, Bochner BH, Kattan MW, Vora KC. Postoperative nomogram predicting risk of recurrence after radical cystectomy for bladder cancer. Journal of clinical oncology: official journal of the American Society of Clinical Oncology. 2006;24:3967–3972.
  16. Wierda WG, O'Brien S, Wang X, Faderl S, Ferrajoli A, Do KA, Cortes J, Thomas D, Garcia-Manero G, Koller C, Beran M, Giles F, Ravandi F, Lerner S, Kantarjian H, Keating M. Prognostic nomogram and index for overall survival in previously untreated patients with chronic lymphocytic leukemia. Blood. 2007;109:4679–4685.
  17. Zhang M, Ding C, Xu L, Feng S, Ling Y, Guo J, Liang Y, Zhou Z, Chen Y, Qiu H. A nomogram to predict risk of lymph node metastasis in early gastric cancer. Sci Rep. 2021;11:22873.
  18. Sui W, Chen Z, Li C, Chen P, Song K, Wei Z, Liu H, Hu J, Han W. Nomograms for Predicting the Lymph Node Metastasis in Early Gastric Cancer by Gender: A Retrospective Multicentric Study. Frontiers in oncology. 2021;11:616951.
  19. Zhang Y, Liu Y, Zhang J, Wu X, Ji X, Fu T, Li Z, Wu Q, Bu Z, Ji J. Construction and external validation of a nomogram that predicts lymph node metastasis in early gastric cancer patients using preoperative parameters. Chinese journal of cancer research = Chung-kuo yen cheng yen chiu. 2018;30:623–632.
  20. Wang Z, Liu J, Luo Y, Xu Y, Liu X, Wei L, Zhu Q. Establishment and verification of a nomogram for predicting the risk of lymph node metastasis in early gastric cancer. Revista espanola de enfermedades digestivas: organo oficial de la Sociedad Espanola de Patologia Digestiva. 2021;113:411–417.
  21. Heagerty PJ, Lumley T, Pepe MS. Time-dependent ROC curves for censored survival data and a diagnostic marker. Biometrics. 2000;56:337–344.
  22. Meurer WJ, Tolles J. Logistic Regression Diagnostics. Jama. 2017;317.
  23. Mu J, Jia Z, Yao W, Song J, Cao X, Jiang J, Wang Q. Predicting lymph node metastasis in early gastric cancer patients: development and validation of a model. Future oncology (London, England). 2019;15:3609–3617.
  24. Alpaugh ML, Tomlinson JS, Kasraeian S, Barsky SH. Cooperative role of E-cadherin and sialyl-Lewis X/A-deficient MUC1 in the passive dissemination of tumor emboli in inflammatory breast carcinoma. Oncogene. 2002;21:3631–3643.
  25. Lauren P. The Two Histological Main Types of Gastric Carcinoma: Diffuse and So-Called Intestinal-Type Carcinoma. An Attempt at a Histo-Clinical Classification. Acta pathologica et microbiologica Scandinavica. 1965;64:31–49.
  26. Pyo JH, Lee H, Min BH, Lee JH, Choi MG, Lee JH, Sohn TS, Bae JM, Kim KM, Yeon S, Jung SH, Kim JJ, Kim S. Early gastric cancer with a mixed-type Lauren classification is more aggressive and exhibits greater lymph node metastasis. J Gastroenterol. 2017;52:594–601.
  27. Tang CT, Chen SH. Higher Lymph Node Metastasis Rate and Poorer Prognosis of Intestinal-Type Gastric Cancer Compared to Diffuse-Type Gastric Cancer in Early-Onset Early-Stage Gastric Cancer: A Retrospective Study. Front Med (Lausanne). 2021;8:758977.
  28. Hirasawa T, Gotoda T, Miyata S, Kato Y, Shimoda T, Taniguchi H, Fujisaki J, Sano T, Yamaguchi T. Incidence of lymph node metastasis and the feasibility of endoscopic resection for undifferentiated-type early gastric cancer. Gastric cancer: official journal of the International Gastric Cancer Association and the Japanese Gastric Cancer Association. 2009;12:148–152.
  29. Jiraskova A, Novotny J, Novotny L, Vodicka P, Pardini B, Naccarati A, Schwertner HA, Hubacek JA, Puncocharova L, Smerhovsky Z, Vitek L. Association of serum bilirubin and promoter variations in HMOX1 and UGT1A1 genes with sporadic colorectal cancer. Int J Cancer. 2012;131:1549–1555.
  30. Zucker SD, Horn PS, Sherman KE. Serum bilirubin levels in the U.S. population: gender effect and inverse correlation with colorectal cancer. Hepatology (Baltimore, Md). 2004;40:827–835.
  31. Ioannou GN, Liou IW, Weiss NS. Serum bilirubin and colorectal cancer risk: a population-based cohort study. Alimentary pharmacology & therapeutics. 2006;23:1637–1642.
  32. Chew SH, Toyokuni S. Malignant mesothelioma as an oxidative stress-induced cancer: An update. Free radical biology & medicine. 2015;86:166–178.
  33. Tomaro ML, Batlle AM. Bilirubin: its role in cytoprotection against oxidative stress. The international journal of biochemistry & cell biology. 2002;34:216–220.
  34. Wei TT, Wang LL, Yin JR, Liu YT, Qin BD, Li JY, Yin X, Zhou L, Zhong RQ. Relationship between red blood cell distribution width, bilirubin, and clinical characteristics of patients with gastric cancer. Int J Lab Hematol. 2017;39:497–501.
  35. Sun H, He B, Nie Z, Pan Y, Lin K, Peng H, Xu T, Chen X, Hu X, Wu Z, Wu D, Wang S. A nomogram based on serum bilirubin and albumin levels predicts survival in gastric cancer patients. Oncotarget. 2017;8:41305–41318.
  36. Yang S, He X, Liu Y, Ding X, Jiang H, Tan Y, Lu H. Prognostic Significance of Serum Uric Acid and Gamma-Glutamyltransferase in Patients with Advanced Gastric Cancer. Dis Markers. 2019;2019:1415421.
  37. Deng R, Mo F, Chang B, Zhang Q, Ran H, Yang S, Zhu Z, Hu L, Su Q. Glucose-derived AGEs enhance human gastric cancer metastasis through RAGE/ERK/Sp1/MMP2 cascade. Oncotarget. 2017;8:104216–104226.