Prognostic Model of D2 Radical Gastrectomy Combined with Neoadjuvant Chemotherapy for Gastric Cancer

Introduction

Nowadays, neoadjuvant chemotherapy and gastrectomy combined with D2 lymphadenectomy has been widely used in the treatment of advanced gastric cancer.^1–3 Compared with surgery merely, combined neoadjuvant chemotherapy can prolong survival rate, reduce tumor staging, and improve radical resection (R0 resection) rate. The timing and method of surgery can be selected according to the reaction degree after neoadjuvant chemotherapy, so as to achieve D2 lymphadenectomy as far as possible.^4,5 However, patients have different prognosis due to individual differences, and affect postoperative treatment decisions.^6–8 Therefore, patients receiving neoadjuvant chemotherapy combined with D2 radical gastrectomy for gastric cancer need more accurate prognosis prediction to guide treatment. Many previous studies have utilized clinicopathological factors as prognostic factors, but most prognostic factors are still limited and controversial.^9–14 AJCC (the American Joint Committee on Cancer) ypTNM (post-neoadjuvant pathologic stage group) staging is the most authoritative guideline prognostic system for gastric cancer patients receiving neoadjuvant chemotherapy.^2,15 However, this system was established based on patients with lymphadenectomy scope less than D2 and did not include ypT0N0 patients with pathologically complete response (PCR), so it may not be suitable for patients with D2 radical gastrectomy or pathologically complete response. In addition, other clinicopathological prognostic factors were not incorporated.^16,17 Therefore, more generalized models are needed to predict the prognosis of these patients. Nowadays, nomogram can contain a variety of prognostic factors, and can be quantified and visualized according to the proportion of each factor, which has been widely used in the study of gastric cancer.^7,13–16 In this study, we retrospectively analyzed the dual center data of 838 gastric cancer patients who received neoadjuvant chemotherapy combined with D2 radical gastrectomy, and established and confirmed a prognostic model.

Materials and Methods

Case Selection

From January 2010 to January 2023, dual center of 838 patients with gastric cancer who received neoadjuvant chemotherapy and D2 radical gastrectomy were selected. These patients were divided into the training cohort (671 patients, the Affiliated Hospital of Qingdao University), and validation cohort (167 patients, Qingdao West Coast New Area Central Hospital). Inclusion criteria: 1) Preoperative pathological diagnosis of gastric cancer (gastric adenocarcinoma); 2) The AJCC clinical stage of T3-4aN1-3M0 or T4bN0-3M0 as evaluated by preoperative examination were included in the study. 3) Patients were evaluated as tolerable for neoadjuvant chemotherapy after Multi-Disciplinary Team (MDT), and were treated with two-drug combination (SOX (Tegafur Timeracil Oteracil Potassium Capsule (S-1) plus oxaliplatin)/XELOX (Oxaliplatin plus Capecitabine) regimen). 4) According to the patient’s requirements and preoperative conditions, D2 radical gastrectomy is taken as the standard treatment mode, and R0 resection is performed. Exclusion criteria: 1) Patients with a history of other malignant tumors and who cannot tolerate neoadjuvant chemotherapy and surgery; 2) Clinical data are incomplete. 3) There was previous history of abdominal or gastric surgery. This is a retrospective study that follows the Declaration of Helsinki and good clinical practice guidelines, and was approved by the Ethics Committee of the Affiliated Hospital of Qingdao University and the Qingdao West Coast New Area Central Hospital The committee waived the need for informed consent to be obtained because this case series is a retrospective study. The medical records and data accessed in the study contained no personal or identifying information.

Data Collection

Patient data are collected by two independent researchers. The collected data included clinicopathological information before neoadjuvant therapy: gender, age, age-adjusted Charlson Comorbidity Index (aCCI), body mass index (BMI), Borrmann classification, carcinoembryonic antigen (CEA), inflammatory indicators (neutrophil/lymphocyte ratio (NLR), platelet/lymphocyte ratio (PLR)), gastric surgery type and pathological date after surgery: pathologic type, lymph node positive rate, neurovascular invasion, AJCC ypTNM stage, histological differentiated type. We apply the AJCC criteria in the National Comprehensive Cancer Network (NCCN) and CSCO gastric cancer guidelines to evaluate the tumor regression grade (TRG) score.^2,3 The TRG score was divided into 4 categories based on the tumor response in pathological sections after treatment. Among them, TRG0 is defined as complete regression of the tumor; TRG1 is defined as very few tumor cell residues; TRG2 is defined as partial regression of the tumor and TRG3 is defined as basically no response in the tumor. For the Lymph node positive rate, NLR and PLR without obvious grouping criteria, the OS is used as the overall endpoint to create an ROC curve, with the cut-off value of the ROC curve as its critical value.

The first outpatient review will begin at 3 weeks after the end of treatment. The patients are followed up until January 2023 or death. From 0~2 years after the operation, follow-up will occur every 3 months. From 3~5 years after the operation, follow-up will occur every 6 months. Patients’ overall survival (OS) is recorded, which is defined as the interval between diagnosis and death or the last follow-up.

Construction and Performance Evaluation of the Nomogram

Statistical analysis is performed using SPSS 24.0 software (Chicago, IL, USA) and R software (www.r-project.org, version 3.63). 1) Construction of nomogram: Based on the data of the training cohort, univariate and multivariate COX regression analyses were performed with SPSS software and R software package “RMS”, “Foreign” and “Survival” to calculate the Hazard Rate (HR) and 95% confidence interval (CI) of the corresponding prognostic factors, select the independent prognostic factors and construct the nomogram. Kaplan–Meier survival curves are drawn to test the difference of each prognostic factor. 2) Evaluation of the performance: ROC curve, DCA curve and calibration curve are plotted by using R software package “rms” “foreign” “survival” “survivalROC” “stdca” based on the data of training cohort and validation cohort. The area under ROC curve (AUC) and consistency index (C-index) are calculated. DCA curve is used to evaluate the clinical utility clinical application value and net benefit of the model, and calibration curve is utilized to evaluate the consistency between actual and predicted results. P < 0.05 is considered statistically significant.

Result

Patient Data

Patients in this study are followed for 0–84 months, with a median follow-up of 58 months. The clinicopathological characteristics of patients in the training cohort and the validation cohort are shown in Table 1. According to Table 1, there are no statistically significant differences in other clinicopathological characteristics between the two cohorts (P > 0.05).

Table 1 Comparison of Clinicopathological Features Between the Two Cohorts (Cases (X/%))

Univariate and Multivariate COX Regression Analysis

The univariate COX regression analysis results show that aCCI, Borrmann type, PLR, ypT stage, ypN stage, lymph node positive rate and TRG are all related to the survival of patients in the training cohort, as shown in Table 2. Multivariate COX regression analysis results show that PLR, ypT and ypN staging, TRG are closely related to prognosis. The higher the value of PLR, patients have worse prognosis. Early ypT and ypN staging, low TRG suggest a good prognosis, as shown in Table 3. Kaplan-Meier method is used to plot the survival curve of prognostic factors. The results show that the overall survival rate of patients with higher PLR values, later ypT and ypN staging and higher TRG have poor prognosis. The differences are statistically significant (P < 0.05), as shown in Figure 1.

Table 2 Univariate Cox Regression Analysis of Clinicopathological Characteristics

Table 3 Multivariate Cox Regression Analysis of Clinicopathological Characteristics

Figure 1 Kaplan–Meier curves of patients with different clinicopathological features. (A) PLR: ≤165.12 and >165.12; (B) ypT stage: ypT0, ypT1, ypT2, ypT3 and ypT4; (C) ypN stage: ypN0, ypN1, ypN2 and ypN3; (D) TRG: TRG 0, TRG 1, TRG 2 and TRG 3. Abbreviations: PLR, platelet/lymphocyte ratio; TRG, tumor regression grade.

Construction of Prognostic Nomogram

The nomogram included the independent prognostic factors built by multivariate COX regression analysis, which can be used to calculate the individual scores of each prognostic factor and add them together to obtain the total score, as shown in Figure 2.

Figure 2 Prediction model of 3-year and 5-year overall survival rate based on the nomogram.

Evaluation of Nomogram

Based on the data of the training cohort and validation cohort, the differentiation and calibration of the nomogram are calculated. The results show that the C-index of the nomogram in the training cohort and validation cohort are 0.815 (95% CI = 0.750~0.880) and 0.801 (95% CI = 0.744~0.858), respectively. In the training cohort, the AUC for predicting 3-year and 5-year overall survival is 0.863 (95% CI = 0.777~0.949) and 0.84 (95% CI = 0.775~0.905). In the validation cohort, the AUC of 3 - and 5-year overall survival is 0.818 (95% CI = 0.753~0.883) and 0.796 (95% CI = 0.793~0.873), indicating that the model has good predictive value, as shown in Figure 3. The calibration curves of the 3 - and 5-year overall survival predicted by the nomogram in the training cohort and the validation cohort are shown in Figure 4. It can be seen from Figure 4 that the calibration curve is in good agreement with the 45° ideal oblique line, indicating a high consistency between the prediction and the actual results. In this study, the DCA curve shows that the nomogram showed a large positive net benefit across a wide range of death risks in both cohorts, indicating that the model has good clinical practicability, as shown in Figure 5.

Figure 3 ROC curves of 3- and 5-year overall survival of the two cohorts of patients predicted by nomogram. (A and B) ROC curves of 3- and 5-year overall survival predicted by nomogram in the training cohort; (C and D) ROC curves of 3- and 5-year overall survival predicted by nomogram in the validation cohort.

Figure 4 Calibration curves for the 3 - and 5-year overall survival of the two cohorts of patients predicted by nomogram (A and B) Calibration curves of 3- and 5-year overall survival predicted by nomogram in the training cohort; (C and D) Calibration curves of 3- and 5-year overall survival predicted by nomogram in validation cohort.

Figure 5 DCA curves for the 3 - and 5-year overall survival of the two cohorts of patients predicted by nomogram. (A and B) DCA curves of 3- and 5-year overall survival predicted by nomogram in the training cohort; (C and D) DCA curves of 3- and 5-year overall survival predicted by nomogram in the validation cohort.

Discussion

Nowadays, D2 radical resection and perioperative comprehensive therapy are the main treatment methods for advanced gastric cancer. A large number of evidence-based medical studies have confirmed that neoadjuvant chemotherapy can significantly improve the prognosis of patients. However, there are individual differences in the impact of neoadjuvant chemotherapy on the prognosis of patients,^5,6,18 Therefore, it is necessary to find relevant prognostic factors to assist in the selection of treatment. The accurate prediction of nomograms has been proved to be significant for clinical decision-making in previous studies.^10,19–21 In this multicenter study, the prognostic factors included in the nomograms, which proved to have good predictive ability.

Ever since Vichow²² proposed the “tumor-inflammation” theory in 1863, many studies have confirmed that inflammatory indicators such as NLR and PLR are related to tumor occurrence, development and prognosis. Neutrophilia is an inflammatory response factor, which starts the immune system by inhibiting the cytolysis of immune cells. Multiple previous studies have shown that NLR is an independent predictor of the prognosis of gastric cancer patients.^23,24 Both lymphocytopenia and thrombocytopenia are associated with poor prognosis of tumor.²⁵ Studies have shown that the increase of PLR in gastric cancer patients before treatment is significantly correlated with pathological types and affects the prognosis of gastric cancer.^9,20 Serum levels of various tumor markers may be elevated in patients with gastric cancer. It has been reported that the serum levels of various tumor markers (such as CEA and CA199, etc.) might rise. However, the prognostic value of these tumor markers remains controversial.^13,26 The results of this study showed that high levels of PLR before neoadjuvant chemotherapy were independent adverse prognostic factors, and the analysis of data showed that patients with low degree of pathological response were mostly accompanied by high levels of PLR. Relevant literature found that high level of PLR was also independent risk factors affecting the remission rate of neoadjuvant chemotherapy, and was related to tumor load and growth rate.^4,17,18 Therefore, the clinical application value of PLR and other inflammatory indicators should be explored in a large sample study.

Nowadays, in recent years, due to the increase in human life expectancy, the proportion of elderly patients with gastric cancer has been increasing. There are many concomitant diseases in elderly patients with gastric cancer, with high mortality. Charlson comorbidities Index (CCI) was proposed by Charlson et al in 1987 and has been widely used to assess the impact of comorbidities.²⁷ Previous studies have shown that aCCI corrects the patient’s age and improves accuracy, and aCCI has been a prognostic factor.^12,14,28 In recent years, aCCI has also been gradually applied in the study of gastric cancer.¹⁴ In this study, Univariate COX analysis showed that high aCCI was a prognostic risk factor, but multivariate COX analysis showed that it is not statistically significant, and high aCCI was not an independent prognostic factor. Since there are relatively few studies on aCCI for gastric cancer, the results of this study still remind clinicians to pay attention to the basic situation of patients, attach importance to the use of clinically relevant indexes, assist in the selection of treatment options.

AJCC ypTNM staging system is still the most authoritative guideline prediction system, although there have been many studies on the prognosis of gastric cancer.^13,24,25 The system was established based on patients undergoing lymphadenectomy (less than D2) and did not include patients with pathological complete response.¹⁶ The results of this study showed that ypT and ypN stages were independent prognostic factors of neoadjuvant chemotherapy combined with D2 radical resection for gastric cancer. The impact of ypT and ypN staging on prognosis has been widely discussed, and a consensus has been reached that later staging is associated with poorer prognosis.^21,28 Therefore, this study can be complimentary to ypTNM staging system. Recent studies have found that lymph node positive rate can also be an independent risk factor for the prognosis of gastric cancer.^11,29 However, in the multivariate COX analysis of this study, lymph node positive rate was not included in the nomogram. The reasons may be that thoroughness of lymph node dissection and sampling determines the accuracy of positive number of pathological lymph nodes,³⁰ and the number of positive lymph nodes obtained may be affected by lymph node fragmentation or confluence.³¹ For advanced gastric cancer, several previous studies have confirmed that TRG after neoadjuvant therapy is significantly correlated with survival prognosis.^32,33 In this study, TRG was also included in the nomogram. Multivariate COX analysis showed that HR changes in the TRG group were significantly higher than other prognostic factors, suggesting that patients with poor pathological reactions had a worse prognosis, which reflected that TRG had a greater impact on survival.

Based on these prognostic factors, a nomogram was constructed to predict the overall survival rate of patients. In the validation cohort, the AUC of the 3 and 5-year survival rate predicted by the nomogram was 0.818 and 0.796, respectively, and a good calibration curve was obtained, indicating that the nomogram had a good predictive value. A satisfactory DCA curve indicated that the nomogram had a high clinical value. The nomogram constructed in this study provides a more accurate personalized visual tool that can be flexibly combined with clinicopathological factors, which conform to the current concept of precision medicine and can provide a reference for evaluating the prognosis and selection of gastric cancer patients.

This study also has some limitations. As a retrospective multicenter study, due to the limited total sample size, data offset is inevitable. This study is validated only for Chinese patients and the underlying biological differences might make their applicability in other populations difficult unless also validated there. In addition, the rapid development of studies on marker genes of gastric cancer (such as human epidermal growth factor receptor 2 (HER-2)) in recent years^34,35 has diversified the studies on prognosis of gastric cancer. Although some novel factors were included in this study, most of the variables were clinical routine factors. At present, the XELOX regimen and the SOX regimen are mostly chosen for gastric cancer patients in China, and adjuvant SOX chemotherapy has similar survival benefits compared to XELOX chemotherapy in Chinese patients.³⁶ These two chemotherapy regimens are not comprehensively applied in some western countries, limiting the external application value of this study.³ Therefore, further multi-center studies are needed.

Conclusions

In conclusion, this study innovatively constructed a nomogram to predict the overall survival rate of patients with advanced gastric cancer treated with neoadjuvant chemotherapy combined with D2 radical gastrectomy, and external validation showed that the model had good predictive performance and clinical application value.