Lymph Node Ratio Predicts Overall Survival in pT4 Gastric Cancer Patients who Underwent Curative Surgery

Background: Lymph node ratio (LNR), the ratio of metastatic lymph nodes to the total number of examined lymph nodes, has been considered a prognostic factor in gastric cancer (GC) patients. However, the predictive values of LNR vary as characteristics of patients are different. We conrmed its prognostic value and investigated the differences between patients with different pathological features. Methods: Totally, 211 GC patients who underwent curative gastrectomy between October 2004 and November 2016 were retrospectively investigated. Patients were classied into LNR 0, 1 and 2, using cutoff values 0.05 and 0.2. Cutoff values were calculated by Youden index. Relationships between LNR and clinicopathological characteristics were investigated by chi-square test or Fisher's exact test. Survival analysis were based on Cox proportional hazard regression model and Kaplan-Meier method. Results: LNR were related to tumor size (p = 0.003), Lauren classication (p = 0.007), grade of differentiation (p = 0.041), pT stage (p = 0.050) and pN stage (p < 0.001). In multivariable analysis, LNR was an independent prognostic factor (HR = 7.023, p < 0.001). In stratication analysis, LNR showed independent prognostic value (HR = 4.852, p < 0.001) in pT4 patients, but not in pT3 patients (p = 0.361) and pT1-2 patients (p = 0.123). Conclusions: LNR is an independent predictor of overall survival in pT4 GC patients who underwent curative surgery.


Background
Gastric cancer (GC) is the fth most common cancer and the third leading cause of cancer-related death around the world [1]. Although neoadjuvant therapy has been recommended by many guidelines, surgery with or without adjuvant therapy is still the main treatment in gastric cancer [2]. Lymph node metastasis often show a great predictive value for the prognosis, no matter whether the patients underwent neoadjuvant therapy and adjuvant therapy [3,4]. Therefore, lymph node status is considered one of the most important prognostic factors in gastric cancer.
Nevertheless, in the AJCC TNM system, only the number of metastatic lymph nodes is considered to determine the pathological lymph node stage (pN) [5]. Although to determine pN stage, at least 16 lymph nodes need to be cleared, some studies supported the number of removed lymph nodes should be larger than 30 [6]. Moreover, some studies also found the number of examined lymph nodes is related to the prognosis [7]. These studies hint that not only the number of metastatic lymph nodes is important, the number of retrieved lymph nodes also plays an important role to the prognosis.
Lymph node ratio (LNR) is the ratio of positive lymph nodes to the total lymph nodes removed. It contains information about both metastatic lymph nodes and retrieved lymph nodes. Many studies prove that LNR is related to the prognosis [8,9] and it could be used to determine the appropriate adjuvant chemotherapy regimen [10,11]. However, the predictive values of LNR are not the same when other characteristics of patients are different [12][13][14], and the cutoff value of LNR is still controversial [15,16]. Even, in some studies, LNR was not related to the prognosis [17]. Therefore, the aim of this study is to verify the prognostic value of LNR, to add an evidence for the possible cutoff value based on our data, and to explore the differences of the predictive value of LNR among patients with different characteristics.

Patients
Patients who underwent curative resection for gastric cancer between October 2004 and November 2016 at our hospital were identi ed. Inclusion criteria were the following: (1) pathological con rmed gastric adenocarcinoma; (2) R0 curative gastrectomy with or without adjuvant; (3) no less than a D2 lymph node dissection; Exclusion criteria were (1) identi able distant metastasis in the liver, peritoneum, and so on; (5) recurrent gastric carcinoma or remnant gastric carcinoma; (6) incomplete information about staging or treatment. After selected, 211 patients were included in this retrospective study.

Reevaluation of clinicopathological characteristics
All clinicopathological characteristics about included patients were extracted from our databases. All slices or blocks indicating surgical specimens and lymph nodes were reevaluated by two pathologists who had no knowledge about other information of the patients. Pathological staging was according to 8 th American Joint Committee on Cancer (AJCC) guideline.
LNR was de ned as the ratio of the number of metastatic regional lymph nodes to the total number of lymph nodes examined. Based on the method of Youden index, the optimal cutoff values of LNR were set as 0.05 and 0.2. Therefore, LNR in this study was divided into three groups, LNR0: LNR less than 0.05; LNR1: LNR more than or equal to 0.05 but less than 0.2; and LNR2: LNR more than or equal to 0.2.

Statistical methods
The associations between LNR and other categorical variables were calculated using the chi-square test or Fisher's exact test. Univariable and multivariable analyses for survival were analyzed by Cox proportional hazard regression model. Variables with p < 0.05 were considered signi cant. The Kaplan-Meier method was used to obtain survival curves, and log-rank test was used to compare the differences. The cutoff values of LNR were obtained by Youden index. All patients were followed up every three months during the rst two years, every six months for the following three years and yearly thereafter. Overall survival (OS) was de ned as the time from the day of operation to the day of death from any cause or last follow-up day. Data was processed by SPSS ver. 25.0 and R 3.6.1 software.

Relationships between LNR and other clinicopathological characteristics
The associations between LNR and other clinicopathological characteristics could be seen in Table 1.
Patients who had a larger tumor tended to have a worse LNR (p = 0.003). Patients who had a tumor of diffuse or mixed classi cation (p = 0.007) or had a poor differentiated tumor (p = 0.041) also showed a worse LNR. In addition, pathological T stage (p = 0.050) and N stage (p < 0.001) also showed relationships with LNR ( Table 1). The adjuvant therapy showed no relationships with LNR (p = 0.661) (data not shown).

Prognostic value of LNR
The analyses for overall survival were shown in Table 2. In the univariable analysis, tumor size (HR = 2.545, p < 0.001), Lauren classi cation (HR = 1.924, p = 0.016), pathological T stage (pT) (HR = 5.679, p = 0.004), pathological N stage (HR = 16.302, p < 0.001) and LNR (HR = 7.023, p < 0.001) were related to the prognosis. Because pN and LNR had great collinearity, these two factors were included in the multivariable analysis respectively. In multivariable analysis including pN, pT (HR = 3.602, p = 0.043) and pN (HR = 11.986, p = 0.001) were independent prognostic factors. Similarly, in multivariable analysis including LNR, pT (HR = 3.694, p = 0.041) and LNR (HR = 4.852, p < 0.001) were independent prognostic factors ( Table 2). The survival curves of pN and LNR were shown in Fig. 1. Pathological N stage (p < 0.001) and LNR (p < 0.001) were related to prognosis. In addition, ROC curves of pN stage and LNR were made, and the area under curves were 0.686 and 0.687, respectively, with no statistical signi cance (data not shown).

Strati cation analysis on LNR according to pathological T stage
Because pT was also an independent prognostic factor besides pN and LNR, strati cation analysis according to pT was performed. The survival curves of LNR strati ed by pT were shown in Fig. 2. In pT1-2 (p = 0.043) patients and pT4 (p <0.001) patients, LNR was related to overall survival. However, in pT3 patients, LNR was not related to the prognosis (p = 0.374).

Discussion
In this study, LNR was a predictor for the prognosis. This result is in line with many other studies. Kano et al.
[8] suggested LNR was an independent predictor for the recurrence-free survival; however, their cutoff values of LNR were 0, 0.1 and 0.25. Huang et al.[18] proved that LNR was related to the 5-year OS rates, especially when it was used together with pathological T stage, though they used the cutoff values of 0.03, 0.08, 0.3 and 0.7. In the study published by Hwang et al. [10], LNR was associated with disease-free survival, especially in the stage III patients, based on the cutoff values of 0.1 and 0.25. Bilici et al. [19] also proved that the LNR was a useful indicator to determine the prognosis and recurrence; however their patients were all in pN3 stage, and the cutoff value they used was 0.75.
On the other hand, LNR has the potential to improve the AJCC system. Zhou et al. [20] suggested that LNR had a better predictive value than pN stage. Huang et al. [18] suggested that a modi ed AJCC system based on LNR had a better prognostic value than the one based on pN stage. Lee et al. [21] and Alatengbaolide et al. [22] suggested that AJCC system based on LNR has better stability than that based on pN stage. Nevertheless, Nakagawa et al. [14] suggested that LNR was not superior to pN stage in evaluating the prognosis, though their study was based on patients with remnant gastric cancer. In our study, the prognostic predictive value of pN stage and LNR had no signi cant difference, which is similar with the conclusion from Bouliaris et al.
Many studies supported that LNR was an important indicator for the choice of appropriate adjuvant chemotherapy regimen [10,11,23]. Hwang et al. [10] suggested that if stage III gastric cancer patients with lymphovascular invasion had a high LNR, these patients tended to bene t from platinum-based adjuvant chemotherapy. Another study published by Kim et al. [11] suggested that patients with high LNR tented to have a better prognosis if they underwent adjuvant chemoradiation rather than adjuvant chemotherapy. Shin et al. [23] suggested that patients with a high LNR could bene t more from XELOX over S-1 in adjuvant therapy. However, in our study, there were no associations among LNR and adjuvant therapy. In [10]. In our study, LNR was also related to Lauren classi cation. To verify this result, more researches based on larger sample size are needed.
There are some limitations in this study. This is a retrospective study, which means potential bias might exist. The sample size is small, which could cause too large or too small statistics and could hinder further hierarchical analysis. The cutoff values were different from other studies, which might hinder the comparison among studies. Nevertheless, this study is based on a certain group of patients, and has proved the predictive value of LNR changes with other characteristics, especially pT stage. Besides, this study adds an evidence to the possible cutoff value, which might contribute to the uni cation of the LNR.

Conclusions
In conclusion, LNR was an independent prognostic factor for GC patients who underwent curative surgery and adjuvant therapy, especially for patients in pT4 stage.

Declarations
Ethics approval and consent to participate The study was reviewed and approved by the Faculty of Science Ethics Committee at Shengjing Hospital of China Medical University. The use of hospital databases was approved by the ethics committee. This is a retrospective study and all patients have signed the informed consent prior to treatment. All methods were performed in accordance with the relevant guidelines and regulations.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this published article.