Methylation-based subtypes of gastric cancer provide distinct prognostic impact in patients who receive chemotherapy

doi:10.21203/rs.3.rs-5280044/v1

High levels of cancer-specific aberrant promoter hypermethylation provide distinct clinicopathological features in gastric cancer (GC) patients. This study explored the molecular subtypes based on the cancer-specific aberrant promoter hypermethylation to predict prognosis of advanced stage GC treated by the chemotherapy. Using five methylated in tumors (MINT) loci (MINT1, 2, 12, 25, and 31), cancer-specific aberrant promoter hypermethylation was characterized in 186 advanced stage GCs receiving chemotherapy (stages II, III and IV diseases). We found that the both methylation-high (4 or 5 markers positive) and negative (one or 0 marker positive) cases presented significantly worse overall survival (OS) and progression free survival (P = 0.003, 0.009, respectively). Multivariate survival analysis using Cox's regression model demonstrated that cases with stage IV and methylation-negative were significantly associated with worse OS (stage IV, hazard ratio [HR] = 4.33, 95% confidence interval [CI] = 2.26–8.31, P < 0.0001; methylation-negative, HR = 2.43, 95%CI = 1.29–4.59, P = 0.006;), while cases in the middle third was also associated with better OS (HR = 0.48, 95%CI = 0.24–0.98, P = 0.045;). Our result show that the cancer-specific epigenetic anomaly provides prognostic impact in patients with GC treated by chemotherapy.

Gastric cancer

prognosis

overall survival

chemotherapy

cancer-specific aberrant promoter hypermethylation

MINT

Gastric cancer (GC), which is one of the most common gastrointestinal malignancies, is ranked fifth regarding the incidence rate among human malignancies and ranks third in mortality [1]. Despite advance in strategy for early detection, many patients still have advanced disease at diagnosis. Since the prognosis of patients with advanced tumor is poor [2, 3], improved treatment outcomes in patients with advanced GC would be required to further decrease in mortality.

Chemotherapy is now recognized as the most effective treatment for patients especially with

unresectable advanced or metastatic GC. To date, many clinical trials have evaluated its efficacy and the safety [4–7]. Other than unresectable cases, neoadjuvant or adjuvant chemotherapy can be also considered for potentially resectable cases to further improve their outcomes. Several studies have evaluated the potential usefulness of neoadjuvant and adjuvant chemotherapy in locally advanced GC [8–11]. However, the clinical outcomes of patients with GC receiving chemotherapy are still significantly different due to the great difference in tumor sensitivity to chemotherapy agents [12]. Predicting treatment response is therefore am important clinical issue in the hope to further improve chemotherapy outcomes in patients with GC.

Transcriptional inactivation by promoter CpG island (PCGI) methylation is important mechanism in human carcinogenesis [13, 14]. High levels of cancer-specific aberrant promoter hypermethylation, first described in colorectal cancer [15], provides distinct clinicopathological and molecular features in cancers in other tissue types such as gastric [16], brain [17], breast [18] and kidney [19]. The cancer-specific aberrant promoter hypermethylation is documented in methylated in tumors (MINT) loci and other several genes including p16INK4A, CSPG2, BNIP3 and p57KIP2 [15, 16]. Importantly, high levels of cancer-specific hypermethylation (methylation-high) in GC is associated with several unique characteristics, including Epstein–Barr virus (EBV)-positive GC, microsatellite instability (MSI) and epigenetic silencing of the mismatch repair gene MLH1 [16], suggesting that methylation-high GC arises from a process that is different from methylation-negative GCs.

In colorectal cancer, methylation-high cases are usually associated with a better prognosis [20]; however, patients with methylation-high colorectal cancers do not benefit from 5-FU (Fluorouracil)-based adjuvant chemotherapy [21]. Since information about potential utility of

cancer-specific aberrant promoter hypermethylation as a molecular marker in GC is currently limited in patients who receive chemotherapy, we have explored the molecular subtypes based on the cancer-specific aberrant promoter hypermethylation to predict prognosis of GC treated by the chemotherapy. We showed that the cancer-specific epigenetic anomaly provide distinct clinicopathological features and prognostic impact.

Patients, survival and response evaluation using different criteria

We examined the genomic DNA samples from 186 primary GC and matched normal gastric tissues, attending the endoscopic center of the Fujita Health University Hospital between April 2003 to July 2017. All the DNA samples were obtained from frozen specimens of the endoscopic biopsies.

Study cohort was consecutively recruited from GC patients with stage II, III and IV diseases who received chemotherapy during the study period. The exclusion criteria were stage I diseases that are not usually treated by the chemotherapy. None of the GCs had been treated with chemotherapy or radiation before sample collection. All GC cases were diagnosed histologically. Detailed information about the histologic subtype, anatomic location, lymph node status, and other metastasis and peritoneal dissemination information was also obtained according to the Japanese classification of gastric carcinoma (Japanese Gastric Cancer Association, 2011). Through histological examination of the biopsy specimens taken almost in the same area where biopsy specimens were taken to obtain DNA samples, we confirmed that all biopsies from cancerous tissue contained more than 70% of cancer cells.

The H. pylori infection status was assessed by histological analysis of both the biopsy specimens from the greater curvature of the gastric antrum and upper corpus using antibodies. The Epstein–Barr Virus status was evaluated using the in-situ hybridization (ISH) of the Epstein–Barr encoding region (EBER).

Overall survival (OS), defined as the time from gastrectomy, or start of the initial administration of chemotherapy to the date of cancer-related death, was determined for all patients. Progressionfree survival (PFS), defined as the time from gastrectomy, or start of initial the administration of chemotherapy to the date of tumor progression or cancer-related death, was also determined in the same cases. Patients with no confirmation of progression or cancer-related death were censored at the date of the last objective tumor assessment.

The Ethics Committee of Fujita Health University School of Medicine approved the protocol, and written informed consent was obtained from all the subjects.

Molecular study

To define methylation status in GC, we used conventional five panels of markers (MINT1, 2, 12, 25, and 31) [15, 16]. Methylation status of the MLH1 promoter, a marker for mismatch repair deficiency [15, 16], was also evaluated. Bisulfite treatment of DNA was carried out with BislFast DNA Modification Kit (Toyobo) according to the manufacturer's protocol. Then, the methylation analysis was performed using the bisulfite pyrosequencing with the PSQ24 system and Pyro-Gold reagent Kit (QIAGEN). The results were analyzed using PyroMark Q24 software (QIAGEN). Considering the distributions of methylation value among all cases (Supporting Information, Fig.S1), we used the cut-off of 20% methylation for MINT1, 25 and 31,15% methylation for MINT 2and 12, and 6% methylation for MLH1 to define positive of each panel, respectively. The methylation-high was defined as having positive of four or more of these markers, while the methylation-low was defined as having positive of two or three of these markers, respectively. Methylation-negative was defined as having positive of one or none of these markers. Methylation status of MLH1 promoter was also examined in all GC tissues. Using the genomic DNA from GC tissues and matched normal gastric tissues, coding regions of TP53 gene (exon2 to exon11) were examined using the Sanger Sequencing. The sequence chromatograms were visually inspected with DNA Dynamo Sequence Analysis Software (Blue Tractor Software, Llanfairfachan, Wales, UK). All mutations were confimed by independent PCR and sequencing reactions. We only consider the nucleotide variation as the mutations, shown in the tumor samples, not in the normal tissues. For non-silent single nucleotide substitutions, SIFT (sorting intolerant from tolerant) analysis was performed to predict whether an amino acid substitution affects protein function. The primers used for Sanger sequencing and pyrosequencing are listed in Supporting Information (Table S1).

List of TP53 mutations in our data set are listed in Supporting Information (Table S2).

Statistical analysis

The statistical significance of the categorical and continuous values was evaluated by the Chi square test and the Student's t-test, respectively. Overall survival and progression-free survival, in relation to the molecular features, were assessed using the Kaplan–Meier method and log-rank test. Multivariate survival analysis using Cox's regression model was also used to calculate the hazard ratio (HR) and 95% confidence interval (CI) with the adjustment of clinicopathological factors and molecular subtypes. A P-value < 0.05 was considered to be statistically significant.

Study cohort

Clinic-pathological characteristics of subjects and information about treatment are shown in

Table 1. Among the 186 patients, 109 patients were considered as operable and received chemotherapy before or after gastrectomy with a D2 lymph node dissection (Neoadjuvant or Adjuvant Chemotherapy). For the remaining 77 cases, chemotherapy alone was used for the treatment (Chemotherapy alone).Regarding the chemotherapy agents, majority of the cases (n=149, 80%) received S-1 (TS-1; tegafur, gimeracil, oteracil potassium) based chemotherapy (S-1+Cisplatin, n=75; S-1 alone, n=49; S-1+Oxaliplatin+Cisplatin, n=10; S-1+Docetaxel, n=2; S-1+Paclitaxel, n=2; S-1+Trastuzumab, n=2;), while other subset of cases received Paclitaxel (n=13), Tegafur Uracil (n=7), Capecitabine+Cisplatin+Trastuzumab (n=6), Docetaxel (N=2) and other agents (n=9) as the first line treatment.

Clinicopatological characteristics of methylation-negative, low and high gastric cancer cases

We have characterized the methylation status with five panels of markers (MINT1, 2, 12, 25, and 31) and MLH1 promoter in 186 GC cases. We found 18 cases (9.7%) to be methylation-high, 64 cases (34.4%) to be methylation-low, and 104 cases (55.9%) to be methylation-negative. We also found 17 cases (9.1%) to be MLH1 methylated cases. We investigated the clinicopatological characteristics of methylation-negative, low and high GC cases (Table 2). We found that methylation-negative cases were more frequent in the cardia, while methylation-high cases were less, and more frequent in the middle and lower thirds, respectively. We also found that more advanced stage GC such as stage IV disease was tended to be frequent in methylation-negative and high cases compared to that in low cases. In the comparison of methylation-low and high cases, distant metastasis was more frequent in methylation-high compared to that in low cases. In addition, EB virus positive cases was more frequent in methylation-high cases compared to others, while MLH1 methylated cases was more frequent in both methylation-high and low cases compared to negative cases.

Methylation subtypes and its prognostic impact in GC patients who received chemotherapy

We investigated the association between methylation status, OS and PFS in GC patients who received chemotherapy. We found that the both methylation-high and negative cases presented significantly worse OS than the methylation-low cases by the log-rank test (P = 0.003, Fig.1). Similarly, the same groups presented significantly worse PFS than the others (P = 0.009, Fig.1). We further investigated whether the methylation status of GC would be associated with prognosis of GC dividing patients into Neoadjuvant or Adjuvant Chemotherapy groups and Chemotherapy alone group. However, we did not found stronger association than that observed in all cases, probably due to smaller cases (Supporting Information, Fig.S2). To confirm the association between methylation status and and prognosis in GC patients who received chemotherapy, we next conducted multivariate survival analysis using Cox's regression model to determine whether the methylation status would be associated with OS as an independent factor (Table 3). Age, gender, H. pylori status, location, histology, staging, EB virus positive, TP53 mutation, hMLH1 methylation and methylation subtypes were included in this analysis. This analysis demonstrated that cases with stage IV and methylation-negative were significantly associated with worse OS (stage IV, HR=4.33, 95%CI=2.26-8.31, P<0.0001; methylation-negative, HR=2.43, 95%CI=1.29-4.59, P=0.006;), while cases in the middle third was also associated with better OS (HR=0.48, 95%CI=0.24-0.98, P=0.045;).

Methylation-high GC provides distinct subgroup in GC such as EBV-positive, MSI and epigenetic silencing MLH1 [16, 23, 24], while the prognostic value of cancer-specific DNA methylation for GC has not been conclusive especially in patients who received chemotherapy. Our result showed that methylation status based on the conventional five marker panels (MINT1, 2, 12, 25 and 31) was significantly associated with prognosis of GC treated chemotherapy. The methylation-negative group, defined as having positive of one or none of above markers was significantly associated with worth OS as an independent factor. The association between low methylation in GC and poor prognosis has been reported in several studies [25, 26], suggesting that lower methylation subtype can be considered as an alarm sign associated with worse prognosis regardless their treatment.

Several reports demonstrated that methylation phenotype including methylation-high and/or methylation low GC was associated with favorable OS compared to CIMP-negative cases [16, 23, 24]. Our result from patients receiving chemotherapy showed methylation-low cases presented better OS and PFS compared to methylation-negative cases, which was in line with other studies. However, the survival curve based on the Kaplan–Meier method showed that methylation-high had rather worse OS and PFS like methylation-negative cases. This paradoxical association between methylation-high and low cases with prognosis may be attributed by the distinct clinicopathological and molecular characteristics of methylation phenotypes especially the methylation-high cases. In 18 cases with methylation-high GS in our dataset, more than half of cases (12/18, 66.7%) were EB virus positive or hMLH1 methylated tumors. The hMLH1 methylated tumors are closely associated with MSI positive in both colorectal and gastric cancer [15, 16, 27, 28].

Regarding the association between molecular subtypes in digestive cancer and prognosis of patients receiving chemotherapy, colon cancer patients with MSI and/or methlation-high show resistance to 5-Fluorouracil (5-FU), a fluoropyrimidine-based drug [21, 29], while MSI positive colon cancer show sensitivity to immunosuppressive checkpoint inhibitors [30]. MSI positive colorectal cancer showed both overexpressed immune checkpoints (mRNA level) and activated tumor-infiltrating lymphocytes, which may explain the satisfying response of ICIs [31]. Similarly, EB virus positive GC showed remarkable tumor-infiltrating lymphocytes and over expression of programmed death ligand-1(PD-L1) [28, 32]. These evidences suggest that methylation-high GC have distinct immunological feature which should be distinguished from methylation-low cases. It is reasonable to expect that differences in immunological feature may be relevant to prognosis especially in patients receiving chemotherapy. Majority of GC patients in our dataset received S-1 based chemotherapy, while subset of other patients received Tegafur Uracil or Capecitabine, all of these are members of fluoropyrimidine-based drugs. S-1 is comprising tegafur, a prodrug of 5-FU and potassium oxonate, which minimises toxicity [33]. Both Tegafur Uracil and Capecitabine also contain prodrug of 5-FU [33]. It is possible that methylation subtypes in both GC and colon cancer share phenotypic feature in relation to the response to fluoropyrimidine-based drugs.

Metastatic GCs have been treated with chemotherapy, but the landscape of cancer treatment is rapidly shifting towards immune-based therapies. Combination therapy with fluorouracil, platinum, and nivolumab, an ICI, is now recognized as the standard first-line chemotherapy for metastatic GC [34], while the introduction of immune checkpoint inhibitors as neoadjuvant and adjuvant treatments is also currently underway, indicating a significant paradigm shift in the treatment strategies [34]. Whether methylation-based molecular subtypes can predict prognosis of patients treated with ICIs would be of great interest. Further study needs to clarify this issue. Our finding emphasize the importance of methylation based subtypes in GC patients treated chemotherapy but it has been also shown that both genetic and epigenetic abnormalities occurs simultaneously in GC. For example, PIK3CA and ARID1A gene, but not the TP53 mutation status have been linked to methylation phenotype in GC [22], suggesting that the genetic and epigenetic abnormalities would reflect different carcinogenetic pathway to promote GC progression, dividing cases into distinct subgroups related to their prognosis. Additional well-designed studies will be also needed for the establishment of molecular subtypes as the prognostic indicator in GC especially in patients treated by chemotherapy.

Our study show that conventional methylation panels would be useful to predict prognosis of GC treated by chemotherapy. We believe that our result will provide meaningful information

Author contributions

The original draft was prepared and written by Tomomitsu Tahara. Sayumi Tahara, Jumpei Yamazaki, Takuya Shijimaya, Tomomitsu Tahara, and Tomoyuki Shibata were in charge of the experiments and data curation. Naohiro Nakamura, and Yu Takahashi, reviewed and revised the all figures and the article.Takashi Tomiyama, Toshiro Fukui, Tomoyuki Shibata, Masakatsu Nakamura, Tomiyasu Arisawa and Makoto Naganuma supervised the project. All authors contributed to the article and approved the submitted version.

Funding

This work was supported by JSPS KAKENHI Grant-in-Aid for Scientists (C) Grant Number　22K08089.

Data availability

Raw data were generated at Kansai Medical University. Derived data supporting the findings of this study are available from the corresponding author T.T on request.

Competing interests

The authors declare no competing interests.

Ethical approval

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.

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Siegel RL, et al. Cancer statistics, 2022. CA Cancer J Clin. 2022;72(1):7–33.
Nashimoto A et al. Gastric cancer treated in 2002 in Japan: 2009 annual report of the JGCA nationwide registry. Gastric Cancer. 2013;16(1):1–27.
Sexton RE, et al. Gastric cancer: a comprehensive review of current and future treatment strategies. Cancer Metastasis Rev. 2020;39(4):1179–203.
Koizumi W, et al. S-1 plus cisplatin versus S-1 alone for first-line treatment of advanced gastric cancer (SPIRITS trial): a phase III trial. Lancet Oncol. 2008;9(3):215–21.
Yamaguchi K, et al. Efficacy and safety of capecitabine plus cisplatin in Japanese patients with advanced or metastatic gastric cancer: subset analyses of the AVAGAST study and the ToGA study. Gastric Cancer. 2013;16(2):175–82.
Yoshida K, et al. Addition of docetaxel to oral fluoropyrimidine improves efficacy in patients with stage III gastric cancer: interim analysis of JACCRO GC-07, a randomized controlled trial. J Clin Oncol. 2019;37(15):1296–304.
Sahin U, et al. FAST:a randomised phase II study of zolbetuximab (IMAB362) plus EOX versus EOX alone for first-line treatment of advanced CLDN18.2-positive gastric and gastro-oesophageal adenocarcinoma. Ann Oncol. 2021;32(5):609–19.
Kodera Y, et al. A feasibility study of postoperative chemotherapy with S-1 and cisplatin (CDDP) for gastric carcinoma (CCOG0703). Gastric Cancer. 2010;13(3):197–203.
Tsuburaya A, et al. Phase II trial of paclitaxel and cisplatin as neoadjuvant chemotherapy for locally advanced gastric cancer. Cancer Chemother Pharmacol. 2013;71(5):1309–14.
Hirakawa M, et al. A phase II study of neoadjuvant combination chemotherapy with docetaxel, cisplatin, and S-1 for locally advanced resectable gastric cancer: nucleotide excision repair (NER) as potential chemoresistance marker. Cancer Chemother Pharmacol. 2013;71(3):789–97.
Yoshikawa T, et al. Neoadjuvant chemotherapy for gastric cancer in Japan: a standing position by comparing with adjuvant chemotherapy. Surg Today. 2014;44(1):11–21.
Sun J, et al. The sensitivity prediction of neoadjuvant chemotherapy for gastric cancer. Front Oncol. 2021;11:641304.
Jones PA, Baylin SB. The epigenomics of cancer. Cell. 2007;128(4):683–92.
Ushijima T, Sasako M. Focus on gastric cancer. Cancer Cell. 2004;5(2):121–5.
Toyota M, et al. CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci. 1999;96(15):8681–6.
Kusano M et al. Genetic, epigenetic, and clinicopathologic features of gastric carcinomas with the CpG island methylator phenotype and an association with Epstein-Barr virus. Cancer. 2006;106(7):1467–79.
Noushmehr H, et al. Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer Cell. 2010;17(5):510–22.
Fang F, et al. Breast cancer methylomes establish an epigenomic foundation for metastasis. Sci Transl Med. 2011;3(75):75ra25.
Arai E, et al. Alterations of the spindle checkpoint pathway in clinicopathologically aggressive CpG island methylator phenotype clear cell renal cell carcinomas. Int J Cancer. 2015;137(11):2589–06.
Ogino S, et al. CpG island methylator phenotype, microsatellite instability, BRAF mutation and clinical outcome in colon cancer. Gut. 2009;58(1):90–6.
Jover R, et al. 5-Fluorouracil adjuvant chemotherapy does not increase survival in patients with CpG island methylator phenotype colorectal cancer. Gastroenterology. 2011;140(4):1174–81.
Kakiuchi M, et al. Recurrent gain-of-function mutations of RHOA in diffuse-type gastric carcinoma. Nat Genet. 2014;46(6):583–7.
Enomoto S, et al. Lack of association between CpG island methylator phenotype in human gastric cancers and methylation in their background non-cancerous gastric mucosae. Cancer Sci. 2007;98(12):1853–61.
Tahara T, et al. Molecular subtyping of gastric cancer combining genetic and epigenetic anomalies provides distinct clinicopathological features and prognostic impacts. Hum Mutat. 2019;40(3):347–54.
Tahara S, et al. Helicobacter pylori infection associated DNA methylation in primary gastric cancer significantly correlates with specific molecular and clinicopathological features. Mol Carcinog Mol Carcinog. 2024;63(2):266–74.
Xiang R, Fu T. Gastrointestinal adenocarcinoma analysis identifies promoter methylation-based cancer subtypes and signatures. Sci Rep. 2020;10(1):21234.
The Cancer Genome Atlas Research Network. Comprehensive molecular characterization of human colon and rectal cancer. Nature. 2012;487(7407):330–7.
Cancer Genome Atlas Research Network. Comprehensive molecular characterization of gastric adenocarcinoma. Nature. 2014;513(7517):202–9.
Sargent DJ, et al. Defective mismatch repair as a predictive marker for lack of efficacy of fluorouracil-based adjuvant therapy in colon cancer. J Clin Oncol. 2010;8(20):3219–26.
Yaghoubi N, et al. PD-1/ PD-L1 blockade as a novel treatment for colorectal cancer. Biomed Pharmacother. 2019;110:312–8.
Luo Y, et al. Genomic and immunological features of microsatellite instability in colon cancer. Gene. 2021;781:145534.
Cheng N, et al. Is gastric lymphoepithelioma-like carcinoma a special subtype of EBV-associated gastric carcinoma? New insight based on clinicopathological features and EBV genome polymorphisms. Gastric Cancer. 2015;18(2):246–55.
García-Alfonso P, et al. Oral drugs in the treatment of metastatic colorectal cancer. Ther Adv Med Oncol. 2021;13:17588359211009001.
Takahari D, Nakayama I. Perioperative immune checkpoint inhibitor therapy for gastric and gastroesophageal junction cancers: a review of current approaches and future perspectives. Int J Clin Oncol. 2023;28(11):1431–41.

Table 1. Clinicopathological characteristics of study cohort

Variables
Age (mean +/- SEM)	66.1+/-0.92
Gender: Female / Male	52 / 134
Location: Cardia / Upper / Middle / Lower / Whole area	8 / 30 / 89 / 53 / 6
H. pylori status: Negative / Positive	44 / 127
Histologic type: Differentiated / Undifferentiated / Mixed	83 / 87 /15
Treatment: Chemotherapy alone / Neoadjuvant or Adjuvant Chemotherapy	77 / 109
Lymph node metastasis: Negative / Positive	43 / 143
Peritoneal dissemination: Negative / Positive	133 / 53
Liver metastasis: Negative / Positive	165 / 21
Other distant metastasis: Negative / Positive	154 / 32
Stage: II / III / IV	55 / 50 /81
EB virus status: Negative / Positive	179 / 7
MLH1 status: Unmethylated / Methylated	166 /17
TP53 mutation status: Wild type / Mutated	111 / 71
H. pylori status, histologic type, MLH1 and TP53 mutation status were not determined for 15, 1, 3 and 4 cases, respectively.

Table 2. Clinicopathological characteristics of methylation-negative, low and high gastric cancer cases.
Variables n (%)	Methylation-negative	Methylation-low	Methylation-high
	(n=104)	(n=64)	(n=18)
Age (mean +/- SEM)	64.9+/-1.24	66.7+/-1.6	71.1+/-2.1
Gender
Female n (%)	31 (29.8)	18 (28.1)	3 (16.7)
Male n (%)	73 (70.2)	46 (71.9)	15 (83.3)
Location
Cardia n (%)	8 (7.7)	0 (0)	0 (0)
Upper n (%)	16 (15.4)	11 (17.2)	3 (16.7)
Middle n (%)	51 (49.0)	34 (53.1)	4 (22.2)
Lower n (%)	28 (26.9)	16 (25.0)	9 (50.0)
Whole area n (%)	1 (1.0)	3 (4.7)	2 (11.1)
H. pylori status
Negative n (%)	25 (26.0)	15 (26.3)	4 (22.2)
Positive n (%)	71 (74.0)	42 (73.7)	14 (77.8)
Histologic type
Differentiated n (%)	47 (45.6)	28 (43.8)	8 (44.4)
Un differentiated n (%)	50 (48.5)	29 (45.3)	8 (44.4)
Mixed n (%)	6 (5.8)	7 (10.9)	2 (11.1)
Lymph node metastasis
Negative n (%)	26 (25.0)	14 (21.9)	3 (16.7)
Positive n (%)	78 (75.0)	50 (78.1)	15 (83.3)
Peritoneal dissemination
Negative n (%)	71 (68.3)	51 (79.7)	11 (61.1)
Positive n (%)	33 (31.7)	13 (20.3)	7 (38.9)
Liver metastasis
Negative n (%)	90 (86.5)	58 (90.6)	17 (94.4)
Positive n (%)	14 (13.5)	6 (9.4)	1 (5.6)
Other distant metastasis
Negative n (%)	82 (78.8)	59 (92.2)	13 (72.2)
Positive n (%)	22 (21.2)	5 (7.8)	5 (27.8)
Stage
II n (%)	27 (26.0)	23 (35.9)	5 (27.8)
III n (%)	27 (26.0)	20 (31.3)	3 (16.7)
IV n (%)	50 (48.0)	21 (32.8)	10 (55.6)
EB virus status
Negative n (%)	103 (99.0)	63 (98.4)	13 (72.2)
Positive n (%)	1 (1.0)	1 (1.6)	5 (27.8)
MLH1 status
Unmethylated n (%)	98 (96.1)	54 (85.7)	14 (77.8)
Methylated n (%)	4 (3.9)	9 (14.3)	4 (22.2)
TP53 mutation status
Wild type n (%)	67 (66.3)	34 (54.0)	10 (55.6)
Mutated n (%)	34 (33.7)	29 (46.0)	8 (44.4)
H. pylori status, histologic type, MLH1 and TP53 mutation status were not determined for 15, 1, 3 and 4 cases, respectively. Age, Methylation-negative vs. Methylation-high, P=0.04; Location, Cardia, Methylation-negative vs. Methylation-low, P=0.03; Location, Middle, Methylation-negative vs. Methylation-high, P=0.04; Methylation-low vs. Methylation-high, P=0.03; Location, Lower, Methylation-negative vs. Methylation-high, P<0.05; Methylation-low vs. Methylation-high, P=0.04; Other distant metastasis; Methylation-low vs. Methylation-high, P=0.02; Stage, Stage IV, Methylation-negative vs. Methylation-low, P=0.052; Methylation-low vs. Methylation-high, P=0.08; EB virus status; Methylation-negative vs. Methylation-high, Methylation-low vs. Methylation-high, both P<0.0001; MLH1 status; Methylation-negative vs. Methylation-low, P=0.02; Methylation-negative vs. Methylation-high, P=0.004; Statistical analysis was performed using the Student's t-test for the comparison of age. All remaining analysis were performed using the Chi-squared test.

Table 3. Multivariate survival analysis using Cox's regression model
for adjustment of clinicopathological factors and molecular features
Variables	HR (95%CI)	P

Age	1.01 (0.99-1.03)	0.3
Gender (female)	1.06 (0.61-1.86)	0.83
H. pylori positive	1.17 (0.69-1.98)	0.57
Location (middle third)	0.48 (0.24-0.98)	0.045
Histology (undifferentiated type)	0.87 (0.36-2.09)	0.76
Staging (stage IV)	4.33 (2.26-8.31)	<0.0001
EB virus positive	0.73 (0.19-2.72)	0.64
TP53 mutated	0.80 (0.48-1.36)	0.41
hMLH1 methylated	1.44 (0.62-3.35)	0.39
Methylation-negative	2.43 (1.29-4.59)	0.006
Methylation-high	2.09 (0.84-5.24)	0.11
HR, hazard ratio; CI, confidence interval;

No competing interests reported.

Supportinginformation.docx

Methylation-based subtypes of gastric cancer provide distinct prognostic impact in patients who receive chemotherapy

Status:

Version 1

Abstract

Figures

Introduction

Materials and methods

Patients, survival and response evaluation using different criteria

Molecular study

Statistical analysis

Results

Discussion

Conclusion

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1