Poorly differentiated is more significant than signet ring cell component for lymph node metastasis in mixed-type early gastric cancer: a retrospective study from a large-volume hospital

Objective The purpose of this study was to explore the role of different undifferentiated components in the lymph node metastasis (LNM) of early mixed gastric cancer. Methods A total of 1596 patients with EGC who underwent gastrectomy were divided into four types: pure differentiated (PD), pure poorly differentiated (Poorly D), pure signet ring cell carcinoma (SRC), and mixed. Mixed type was classified into four subtypes: differentiated-predominant type mixed with poorly differentiated component (MD-P), poorly differentiated-predominant type mixed with differentiated component (MP-D), differentiated-predominant type mixed with SRC component (MD-S), and poorly differentiated-predominant type mixed with SRC component (MP-S). We analyzed the clinicopathological differences between all types and the rates of LNM of patients fulfilling endoscopic submucosal dissection (ESD) criteria. Results LNM was more common in mixed (21.9%) than in PD (7.5%, P < 0.001) or SRC (11.3%, P < 0.001). When analyzed the subgroups of mixed type, LNM was more frequent in MD-P (15.4%) than in PD (7.5%, P = 0.003). LNM in MD-S (7.4%, P = 1.000) was not higher than in PD. MP-S (24.5%) showed a higher rate of LNM than SRC (11.3%, P < 0.001) rather than Poorly-D (22.7%, P = 0.681). For lesions satisfying ESD criteria, MD-S (0%, P = 1.000), and MD-P (5.9%, P = 0.12) did not have higher rates of LNM than PD (1.4%). Conclusion The presence of poorly differentiated component but not SRC increases the LNM rate of mixed type. ESD is recommended for the treatment of MD-S and MD-P consistent with ESD criteria.

has been reported that mixed gastric cancer is an LNM-predictive factor for submucosal EGC rather than for mucosal EGC [11][12][13]. In the presence of an undifferentiated component, the treatment of mixed gastric cancer needs to be cautious, while the guideline is still incomplete and controversial. Some studies have suggested that endoscopists need to judge the indications more careful when using ESD to treat mixed gastric cancer considering the low curative resection rate and high incidence of local recurrence of it [14][15][16][17][18]. However, Min et al. reported that ESD can be an effective therapy for differentiated-predominantly mixed gastric cancer that meets the absolute or expanded criteria [19]. In fact, undifferentiated carcinomas are mainly composed of Signet ring cell carcinoma (SRC) and poorly differentiated cancer (Poorly-D), and studies have shown that early SRC has a lower rate of LNM and a better survival rate [20]. However, relevant studies have not totally distinguished the undifferentiated components in mixed cancer. Thus, given the putative role of SRC in LNM, we focused on its role in LNM from mixed cancer. To achieve this goal, we divided the MD type into differentiated-predominant type mixed with SRC component (MD-S) and differentiated-predominant type mixed with poorly differentiated component (MD-P). We classified undifferentiated cancers into pure poorly differentiated cancer, pure signet ring cell cancer, poorly differentiatedpredominant (MP) type mixed with differentiated component (MP-D), and poorly differentiated-predominant type mixed with SRC component (MP-S). All the groups were taken into the analysis.
The purpose of this study was to clarify the role of different undifferentiated components in the LNM of mixed gastric cancer. In addition, we investigated whether mixed gastric cancer satisfying the ESD criteria is suitable for ESD treatment.

Patients
A total of 1596 surgically treated patients with EGC treated in Renji Hospital, Medical College of Shanghai Jiao Tong University, between July 2008 and December 2018 were enrolled in this retrospective study. Patients with incomplete pathological information, multiple gastric cancers, other malignancies, and invasion deeper than the submucosa were excluded. All data access was ethically approved.
Clinicopathologic features were obtained. These included age, gender, tumor size, location, macroscopic type, ulcer, depth of invasion, presence of LNM, N stage, and histology type.

Definition and classification of mixed-type EGC
Histologically, all lesions were cut into three to five micron sections, and each section was stained with hematoxylin and eosin to observe the cell morphology and the presence of LNM. All the sections were retrospectively analyzed by two pathologists and divided into four types according to the 2010 World Health Organization (WHO) classification: PD type (consisting of well and moderately differentiated papillary adenocarcinoma and tubular adenocarcinoma), Poorly-D type (poorly differentiated adenocarcinoma), SRC type (signet ring cell carcinoma), and mixed gastric cancer (mixed) type. The mixed type was further classified into 4 subtypes according to the major component. These included MD-P (differentiated-predominant mixed-type with poorly differentiated component, which accounted for less than 50%), MP-D type (poorly differentiated-predominant mixed type with differentiated component, which accounted for less than 50%), MD-S type (differentiated-predominant mixed type with signet ring cell component, which accounted for less than 50%), and MP-S type (poorly differentiatedpredominant mixed-type with signet ring cell component, which accounted for less than 50%). Three histological mixed type were not included in our research because of the small sample size. Macroscopic cancer was classified by the Japanese Gastric Cancer Association (JGCA) guidelines. These classes include the elevated types (I, IIa), the flat type (IIb), and the depressed types (IIc, III) [6]. Tumor location was identified as being in the upper third, middle third, lower third, or residual stomach. After lymph node dissection, the AJCC Cancer Staging Manual (eighth edition) was used to classify the N stage of the cancer.

Statistical analysis
The χ 2 test and multivariate analyses were performed to evaluate the effect of different mixed types on LNM. Multivariate analyses tested by logistic regression were used to compare groups. P < 0.05 was considered statistically significant. SPSS 23.0 for Windows was utilized for data analysis. Table 1 shows the clinical characteristics of all patients. The median age of the patients was 62 years (range 17-88), with 1069 males (67%) and 527 females (33%). According to the WHO classification, pathological types were divided into the following 4 groups: 723 (45.3%) PD, 181 (11.3%) Poorly-D, 203 (12.7%) SRC, and 359 (30.6%) mixed type. The overall rate of LNM was 14.1%, and the highest rate was found in patients with poorly differentiated gastric cancer (22.7%). The LNM rates of the mixed type were higher than those of the PD type (mixed versus PD, 21.9% versus 7.5%, respectively, P < 0.001) and the SRC type (mixed versus SRC, 21.9% versus 11.3%, respectively, P < 0.001), but were not different from those of the Poorly-D type (mixed versus Poorly-D, 21.9% versus 22.1%, respectively, P = 0.831).

Comparisons of LNM among different histological types
To exclude other factors which may be involved in the LNM of EGC, age, gender, tumor size, tumor morphology, tumor location, ulcer, invasion depth, and pathological classification were integrated into a multivariate regression analysis. As shown in Table 2 To investigate whether different types of mixed EGC had different risks for LNM, we divided them into 4 subgroups, including 130 (26.6%) MD-P, 144 (29.4%) MP-D, 188 (38.4%) MP-S, and 27 (5.5%) MD-S (Table 3). We compared the different mixed types with the pure histology type (Table 4). When we compared the LNM rate of SRC with that of Poorly-D and MP-S, MP-S showed higher LNM (MP-S versus SRC, 24.5% versus 11.3%, respectively, P < 0.001). MP-S significantly predicted LNM (OR 1.923; 95% CI 1.075-3.441; P = 0.028). Although Poorly-D was more likely to have LNM (Poorly-D versus SRC, 22.7% versus 11.3%, respectively, P = 0.003), multivariate analysis showed that Poorly-D had no effective impact on LNM (OR 1.529; 95% CI 0.834-2.801, P = 0.17).
The LNM of MD-S was not higher than that of PD (MD-S versus PD, 7.4% versus 7.5%, respectively, P = 1.000), which was confirmed by multivariate analysis (OR 1.284; 95% CI 0.273-6.054; P = 0.752). Despite the fact that the Chi-square test showed that the LNM rate of SRC was not different from that of PD (SRC versus PD, 11.3% versus 7.5%, respectively, P = 0.078), multivariate analysis suggested SRC as an independent risk factor of LNM (OR 2.112; 95% CI 1.175-3.797; P = 0.013).

LNM rates in EGCs fulfilling the endoscopic resection criteria
We extracted the cases with tumor size less than 20 mm and with absence of ulcer and presence of mucosal invasion. As shown in Table 5, there was no difference in LNM rate between Poorly-D and PD (Poorly-D versus PD, 3.0% versus 1.4%, respectively, P = 1.000). When compared with PD type, LNM was more common in the Mixed type (Mixed versus PD, 11.8% versus 1.4%, respectively, P < 0.001) and the SRC type (SRC versus PD, 7.9% versus 1.4%, respectively, P < 0.003). However, such a distinction was not observed between the Mixed type and the Poorly-D type (Mixed versus Poorly-D, 11.8% versus 3.0%, respectively, P = 0.241) or the SRC type (Mixed versus SRC, 11.8% versus 7.9%, respectively, P = 0.333). Next, we analyzed different subgroups of mixed gastric cancer. The MD-P type (MD-P versus PD, 5.9% versus 1.4%, respectively, P = 0.12) and the MD-S type (MD-S versus PD, 0% versus 1.4%, respectively, P = 1.000) did not have any impact on the LNM. However, MP-D (MP-D versus PD, 13.3% versus 1.4%, respectively, P = 0.003) and MP-S (MP-S versus PD, 16.1% versus 1.4%, respectively, P < 0.001) did evidently increase the incidence of LNM.

Discussion
In our study, SRC was first extracted from the undifferentiated component in early mixed gastric cancer, and mixed cancer was further classified into four types: MD-P, MD-S, MP-D, and MP-S, to be taken into analysis. The results of our research showed that LNM was more likely to occur in the MD-P type than in the MD type, but not in the MD-S type. MP-D and MP-S did not increase the risk of LNM from the MP type. However, MP-S significantly increased the LNM of SRC. Next, we compared the patients who met ESD criteria and demonstrated that there was no difference in the LNM rate between the MD type and the PD type. On the contrary, MP-D and MP-S are more likely to cause LNM than PD. Generally, undifferentiated cancer consists of poorly differentiated cancer and signet ring cell cancer. Many studies have reported that early SRC type tended to be confined to the mucosal layer and had a lower LNM rate and a higher 5-year survival rate than the non-SRC type [21][22][23]. On the contrary, the Poorly-D type is more likely to invade the submucosal layer, which may contribute to LNM. Several studies have shown that the MD type has higher LNM rates than the PD type, although there was no significant difference.  However, these studies do not distinguish the undifferentiated components of the MD type [24,25]. Considering these differences in histology heterogeneity, we hypothesized that the undifferentiated components of the MD type have different impacts on LNM and need to be evaluated and managed independently. The current study classified the MD type into an MD-S type and an MD-P type and revealed that the MD-P type has more aggressive behavior, as proven by the higher submucosal invasion rate and prevalence of LNM when compared with the PD type, while the MD-S type did not show these features. Meanwhile, our data also suggested that the MD-S type cannot be an effective predictor of LNM. Studies on early SRC and the effect of the Poorly-D component on SRC remain controversial. A few studies have reported the LNM rate of early SRC ranges from 5 to 13.9%. Hu et al. reported a higher LNM rate of both the Poorly-D type and the MP-S type than SRC, although there was no statistical difference [17,[26][27][28][29]. Our study gave results consistent with this, showing that the LNM rate of SRC is 11.3%, lower than that of the MP-S type (21.9%) and the Poorly-D type (22.7%). In addition, to exclude factors other than the MP-S type, such as tumor size, submucosal invasion and ulceration, which may play a role in generating similar results, we performed a multivariate regression analysis and confirmed that MP-S was an independent risk factor compared to SRC, while Poorly D was despite its high LNM.
To see whether histologic predominance could affect the risk of LNM, we performed increased proportion of a poorly differentiated component in the mixed type, the LNM increased quickly, which validated that being poorly differentiated itself was an important factor that played a decisive role in the LNM of the mixed type, while SRC was not. However, this conclusion may remain controversial, as Seo's study suggested that differences in composition of the mixed type do not cause differences in LNM [10]. Such inconsistencies may be due to the existence of a large amount of MP-D with submucosal invasion and small amounts of MD-S with mucosal invasion in our study.
At present, ESD is considered an effective method to treat early differentiated gastric cancer with diameter less than 2 cm limited to the mucosal layer without ulceration. Undifferentiated cancer with the same description is also included in the expanded criteria for ESD [6]. To investigate whether ESD can be used in the treatment of mixed gastric cancer, we screened out gastric cancer that meets the absolute or expanded criteria and found that the LNM rates from the mixed type and the SRC type were both higher than from the PD type, but not from the Poorly D type. We further analyzed the role of different subtypes of mixed gastric cancer and found that the differentiated-predominantly mixed type was not an independent risk factor for LNM for EGC limited in the mucosa when compared with PD. In contrast, the poorly differentiated-predominantly mixed type can effectively predict the rate of LNM in gastric cancer patients whose lesions are confined to the mucosal layer. The difference in the rate of LNM between surgical resected MD-P and MD-P meeting the ESD criteria may be attributed to the higher submucosal infiltration in the former, which is consistence with the previous study that poorly differentiated component is an independent risk factor for LNM in submucosal EGC [30]. Currently, a few studies have reported that a Poorly D component in the SRC type can significantly increase the rate of LNM and non-curative dissection of EGC. Even the MP-S type meets the ESD treatment criteria [15,17]. In this study, although there was no statistical difference, the MP-S type that met the criteria were more prone to LNM than Poorly D and SRC. Additionally, considering the high rates of LNM of MP-S, surgical resection may be a better treatment for MP-S than ESD.
There are some limitations in this study. First, as this study is a retrospective study, there may be some data omissions and errors in the procedures of collecting data. However, all lesions were evaluated and pathologically diagnosed by two experienced pathologists to minimize errors. Second, only the specimens of EGC removed by surgery were included in this study, while the specimens of patients receiving ESD treatment were not included, which may lead to bias in selection. Third, we did not investigate the clinical outcome of patients. However, as for EGC, patients usually have a good prognosis. Fourth, we did not investigate the patient's history of Helicobacter pylori. As previously reported, H. pylori infection is not uncommon in undifferentiated cancer and may have an association with low LNM in SRC [31].
In conclusion, mixed gastric cancer is an independent risk factor for LNM in EGC. As for the undifferentiated components in differentiated-predominant EGC, these data suggest that the degree of malignancy of the tumor is significantly increased by poorly differentiated cancer when compared with SRC. In addition, for patients meeting the ESD treatment criteria, ESD treatment is recommended for differentiated-predominantly mixed cancer, while additional surgery is required for poorly differentiated-predominantly mixed cancer.
Funding Funding sources were not involved in the design, analysis or collection and interpretation of data.

Compliance with ethical standards
Disclosures Jin-Nan Chen, Qi-Wen Wang, Qing-Wei Zhang, Zhao-Rong Tang and Xiao-Bo Li have no conflict of interest or financial ties to disclose. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, 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 changes were made. 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://creat iveco mmons .org/licen ses/by/4.0/.