Ethnic disparities in demographic, clinicopathologic and biological behaviours and prognosis of gastric cancer in northwest China

Abstract This retrospective study aimed to investigate ethnic disparities in demographic, clinicopathologic, and biological behaviours of gastric cancer (GC) in a high GC incidence area of China. There were 5022 GC patients, including 3987 Han (79.4%) and 987 Hui (14.4%) patients from Northwest China. All patient data were retrieved from 2009 to 2017. Median survival was estimated using the Kaplan‐Meier method and compared using the log‐rank test. A Cox proportional hazards model was used to assess the impact of covariates. Similarly, low 5‐year OS rates were observed in both the Hui and Han groups (23.8% and 24.2% respectively). Hui patients with stage T1 or N0 or with tumours <5 cm had 2.144‐fold, 1.426‐fold and 1.305‐fold increased risks of poor prognosis compared with Han patients with these characteristics respectively (all p < 0.05). Further, Hui patients had 1.265‐fold, 1.364‐fold and 1.401‐fold increased risks of poor prognosis compared with Han patients among those with high expression of Ki67, EGFR and VEGF respectively (all p < 0.05). There are ethnic disparities in the prognosis of GC patients in Northwest China. Understanding the effects of ethnicity on GC will guide reasonable evaluations of prognosis and future interventions to equalise access to high‐quality care for GC patients of different ethnicities in China.


| INTRODUCTION
Accumulating data suggest that geographical factors, such as ethnic density (the proportion of ethnic minority populations in a geographic area), and geographical segregation may be related to various cancer-associated outcomes. 1,2 Therefore, identifying ethnic disparities and establishing ethnic survival models may offer a variety of social benefits, including greater availability of social support and community resources.
Gastric cancer (GC) has been prevalent in China for decades, with half of the world deaths (estimated 390,182 in 2018) occurring in China. 3 Despite improvements in endoscopic, surgical and systemic treatments, the overall survival (OS) rates and prognosis of GC patients are still dismal. The 5-year OS rate for GC is above 50% in Japan and Korea but only approximately 20% in China. [4][5][6] Factors including older age, advanced TNM stage, poor differentiation and lymphatic metastasis have all been shown to be unfavourable for the prognosis of GC patients among different racial or ethnic groups. [7][8][9] Some other factors, including environmental, as well as genetic background due to diversity in race/ethnicity, were also indicated to affect an individual's cancer spectrum. Two recent studies specifically indicated that race and ethnicity diversity played an important role in the prognosis of patients with GC. 10,11 Asian or Pacific Islander (API) GC patients were found to have a 1.262-fold increased risk of poor prognosis compared to non-API GC patients. Similarly, American Indian/Alaskan Natives have a 2.215-fold increased risk of poor prognosis compared to white patients, who had a survival similar to that of API patients. 11 Although ethnic differences in survival may partly be explained by inconsistent exposure to GC-associated risk factors, such as unique clinical features, elevated rates of Helicobacter pylori infection, inappropriate treatment or cultural differences in lifestyle, [12][13][14] there have been too few studies of ethnic density, especially among Asian ethnicities, to draw any meaningful conclusions with respect to cancer mortality. China has 56 officially recognised ethnicities and is becoming more ethnically diverse, although many ethnic minorities are geographically segregated in certain areas. 15 To more fully understand the distinct contextual and compositional effects associated with different ethnicities, better data on risk factors, including relevant biological markers of cancer risk or outcomes, must be collected. 16 In this study, patients with GC from two Chinese ethnic groups, namely Han and Hui, were retrospectively selected from one of the highest GC incidence areas: the Ningxia Hui Autonomous Region (hereafter called Ningxia), located on the Loess Plateau in Northwest China. Data from 5022 GC patients (3987 Han patients and 987 Hui patients) were reviewed, and their demographic, clinicopathologic and biological features, including Ki67, VEGF and EGFR were compared. These three cancer-related biomarkers are routinely tested indexes and are clinically used to determine the degree of malignancy of GC and/or evaluate the prognosis of patients with GC. 17,18 To the best of our knowledge, this is the first study to examine the ethnic disparity in outcomes among GC patients between two main Chinese ethnic groups and evaluate prognosis and future interventions to equalise access to high-quality care for GC patients of different ethnicities.

| Study population and data collection
A total of 5022 patients diagnosed with GC in the General Hospital of Ningxia Medical University, Ningxia, Northwest China from January 2009 to December 2017 were recruited for this retrospective study. Diagnosis was based on endoscopy and confirmatory histopathology. First, demographic data, including age at admission, sex, ethnicity, alcohol consumption and cigarette smoking, were collected from medical records. Those subjects who had three times or more alcoholic drinks a week for more than 6 months were defined as alcohol drinkers, and those who smoked one cigarette per day for more than 1 year were defined as smokers. Second, clinicopathologic characteristics, including T stage, N stage, M stage, tumour size, tumour location, Borrmann's type, Lauren's type, clinical staging and radical gastrectomy were further obtained. Patients with other previous cancers were excluded. T stage, N stage and M stage were assessed based on the 7th TNM stage system, and the clinical staging was assessed based on the TNM stage.
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The study was approved by the Ethical Committee of Ningxia Medical University prior to conducting the study. All the patients provided written informed consent for the review and use of their medical records according to the regulations.

| Follow-up
The information about the survival status, death time and cause of death was collected for all eligible GC patients from the hospital database, medical records or by telephone calling or text messages to patients or their family members. All patients were followed up through January 2019. Patients who were followed from treatment initiation to the date of death due to GC were considered as reaching the end point. Those patients who still cannot be contacted after three attempts or were noticed to be with wrong phone numbers were regarded as lost to follow-up. Those who were alive, were lost to follow-up or died from other diseases by the end of data collection were censored. The survival data were available for 3872 of 5022 GC patients (77.1%). The OS duration was calculated from treatment initiation data to either the date of death due to GC or the last follow-up date.

| Immunohistochemical analysis
For further immunohistochemical analysis, all the biopsy specimens were deparaffinised and serially sectioned into 4-μm thick sections. The expression of Ki67, VEGF and EGFR in GC tissues was examined using the S-P immunohistochemical method. Briefly, after routine deparaffinisation and rehydration, slides were treated with 1% hydrogen dioxide and then heated in EDTA (pH 8.0) for antigen retrieval. Following blocking in 10% goat serum, tissue sections were then incubated with mouse antibodies against human Ki67, VEGF and EGFR (ZSGB-BIO, Beijing, China) at 4°C overnight. After rinsing, sections were subsequently incubated with goat anti-mouse biotin-conjugated IgG for 15 min and then with streptavidin-peroxidase conjugate for 15 min. The signal was developed with diaminobenzidine, and slides were counterstained with 5% haematoxylin. The brown signals located in the cytoplasm represented positive staining for the respective proteins. Sections not treated with monoclonal antibody were used as negative controls. The expression of Ki67, VEGF and EGFR was assessed independently based on the proportion of positive cells by two pathologists who were blinded to the clinical data. The staining was scored on a scale from 0 to III as follows: 0, fewer than 10% of cells were stained; I, 10-25% of cells were stained; II, 26-50% of cells were stained; and III, >50% of cells were stained. A score of III was classified as high expression, whereas scores of 0-II were considered low expression. 19-21

| Statistical analysis
Statistical analysis was performed using the Statistical Package for the Social Sciences 17.0 (SPSS, Inc., Microsoft, Chicago IL, USA). The distributions of the general demographic and clinicopathologic characteristics of the GC patients were analysed using the χ 2 test or Fisher's exact test for categorical variables, as appropriate. Cumulative survival rates were calculated from the treatment initiation date to the date of death or the last follow-up. Median survival was estimated using the Kaplan-Meier method and compared using the log-rank test. Cox regression analysis was used to identify factors associated with survival in each population, and significant factors in the univariate analysis were included in the multivariate analysis. Relative risks were estimated by calculating hazard ratios (HRs) and 95% confidence intervals (CIs). Statistical significance was set at p < 0.05, and all statistical tests were two-sided.

| General demographic characteristics
The present study included 5022 GC patients aged 15-93 years diagnosed between 2009 and 2017 and followed through 2019. All patients' characteristics are presented in sex between Han and Hui groups (all p > 0.05). Statistical differences were found between two ethnic groups in cigarette smoking, alcohol consumption, occupation, TNM stage, tumour location, Lauren's type, radical gastrectomy and radiotherapy (all p < 0.05). Hui patients were more likely than Han patients to be nonsmokers, nondrinkers and farmers. They were also more likely to present at a more advanced T stage, N stage and clinical staging. Compared with Han patients, Hui patients were less likely to receive radical gastrectomy or radiotherapy.  Table S2. Univariate log-rank analysis identified age, T stage, N stage, M stage, tumour size, tumour location, differentiation, Borrmann's type, clinical staging, radical gastrectomy and palliative status as significant prognostic factors for longterm OS among both Han and Hui GC patients (all p < 0.05). Among patients with high expression of Ki67, EGFR or VEGF, the prognosis of Hui patients was much worse than that of Han patients (all p < 0.05). The details are shown in Table 2, Table S3, Figures 1, and 2.

| Survival and prognostic factors for OS between two ethnic groups
Given that the clinical staging was assessed based on the T stage, N stage and M stage, the collinearity diagnosis   23,24 live in the Ningxia region of Northwest China, which has the highest incidence and mortality of GC. In this study, compared with Han patients, Hui patients were less likely to smoke or consume alcohol, but they were more likely to present at a more advanced T stage, N stage and clinical staging. Although no statistically significant difference in OS was  The similar low OS rates of two ethnic groups and the survival disparities between Hui and Han people may be explained by several factors. First, Ningxia is one of the economically underdeveloped regions in China. The lagged education level, the limited medical resources and low income in this area may all contribute to the poor survival we observed in both the Hui and Han GC populations. Second, most Hui people live in rural mountain areas within a closed circle for historical reasons, which limits their acceptance of modern medical knowledge, especially the importance of health care and early treatment. Even worse, many Hui people, who are subsistence farmers, belong to the low-income class and have difficulty affording medical expenses. Last, some Hui people, who are influenced by their religious beliefs, are unwilling to accept surgery or other treatments.
Ethnicity may also reflect genetic biological effects in addition to origin, history and cultural identity. An understanding of the biological and molecular basis of GC will enable oncologic treatment or prognostic assessments to be tailored to the unique characteristics of individual tumours. Some reports have showed that high expression of Ki67, EGFR or VEGF was an unfavourable factor for GC prognosis; 18,26,27 our findings showed that, among patients with high expression of Ki67, EGFR or VEGF, the prognosis of Hui patients was much worse than that of Han patients, which indicated that Hui patients exhibiting high expression levels of three cancer-related genes seemed to be more susceptible to a poor prognosis. Although we do not know the underlying reasons for this phenomenon, several studies have demonstrated hereditary genetic diversity between Hui and Han populations. Li et al. systemically investigated the effects of exogenous additives on the genetic diversity of drug absorption, distribution, metabolism and excretion between the Han majority and Hui minority in Northwest China and reported significant differences between the Han and Hui populations. 28 Our previous study also showed that the distribution of genetic variants in cancer-related genes varies between Han and Hui people. 24 These studies highlight that genetic background diversity in ethnicities, such as the Han and Hui populations, may affect an individual's cancer spectrum. These results have significant implications for assessing cancer susceptibility, sensitivity to treatments and prognosis among different ethnic groups.
The limitations of the present study need to be considered in the interpretation of the obtained results. All of the participants were enrolled from one region, and the subjects may not represent all GC patients in China. Therefore, further investigations with multiple populations warrant confirmation. Another limitation of this study is the lack of information on socioeconomic status, which may also affect the prognosis of GC patients. 29

| CONCLUSIONS
In conclusion, there are ethnic disparities in the survival status of Hui and Han GC patients in Northwest China, even after hundreds of years of living together. Hui GC patients were more likely to have a poorer prognosis than Han patients among those with early stage disease or high expression of Ki67, EGFR or VEGF. An understanding of the effects of ethnicity on GC will prompt public health experts and policy makers to identify, intervene and eventually alleviate the root causes of GC disparities.