Comparable genetic alteration profiles between gastric cancers with current and past Helicobacter pylori infection

Gastric cancers can develop even after Helicobacter pylori (H. pylori) eradication in 0.2–2.9% cases per year. Since H. pylori is reported to directly activate or inactivate cancer-related pathways, molecular profiles of gastric cancers with current and past H. pylori infection may be different. Here, we aimed to analyze whether profiles of point mutation and gene amplification are different between the two groups. Current or past infection by H. pylori was determined by positive or negative amplification of H. pylori jhpr3 gene by PCR, and past infection was established by the presence of endoscopic atrophy. Among the 90 gastric cancers analyzed, 55 were with current infection, and 35 were with past infection. Target sequencing of 46 cancer-related genes revealed that 47 gastric cancers had 68 point mutations of 15 different genes, such as TP53 (36%), KRAS (4%), and PIK3CA (4%) and that gene amplification was present for ERBB2, KRAS, PIK3CA, and MET among the 26 genes assessed for copy number alterations. Gastric cancers with current and past infection had similar frequencies of TP53 mutations (38% and 31%, respectively; p = 0.652) and oncogene activation (20% and 29%, respectively; p = 0.444). Gastric cancers with current and past infection had comparable profiles of genetic alterations.


Molecular profiles were similar between gastric cancers with current and past H. pylori infection.
To analyze whether molecular profiles between gastric cancers with current and past H. pylori infection are different, frequencies of the somatic point mutations and gene amplifications were compared between the two groups. Both groups had similar frequencies of TP53 mutations (38% and 31% in gastric cancers with current and past infection, respectively; p = 0.652), KRAS mutations (2% and 9%; p = 0.295), and PIK3CA mutations (4% and 6%; p = 0.641) (Fig. 3a). As for gene amplifications, gastric cancers with current and past infection also had similar frequencies of ERBB2 amplification (9% and 3%; p = 0.398) and KRAS amplification (2% and 3%; p = 1.000) (Fig. 3b, Table 3). These results showed that gastric cancers with current and past infection had comparable profiles of genetic alterations.

Discussion
Gastric cancers with current and past H. pylori infection had comparable profiles of genetic alterations, namely somatic point mutations and gene amplification. Even when activation of known oncogenes, such as ERBB2 and PIK3CA, by either a point mutation or gene amplification was analyzed, both groups had similar frequencies.
Since genetic activation of these genes has been clinically utilized in molecular targeted therapy 21,22 , it was considered that similar therapeutic strategies can be applicable for both gastric cancers with current and past infection.
It is known that H. pylori can directly activate oncogenic pathways, such as the MEK-ERK pathway and WNT pathway, and inactivate tumor-suppressive pathways, such as the p53 pathway, by injecting CagA into epithelial cells 3 . Therefore, it was considered that the alteration mechanisms of cancer-related signaling pathways might be different between gastric cancers with current infection and those with past infection. However, both groups had similar frequencies of alterations of genes involved in these cancer-related pathways. This suggested that direct activation or inactivation of cancer-related pathways by H. pylori has limited influence on genetic alterations.
Approximately 47% and 46% of gastric cancers with current H. pylori infection and past infection, respectively, had no genetic alterations of known cancer-related genes. In such gastric cancers, repression of tumor-suppressive pathways, such as cell cycle regulation and the p53 pathway, and activation of oncogenic pathways, such as the WNT pathway, are known to be frequently caused by epigenetic alterations, namely aberrant DNA methylation 23 . Therefore, it was considered that epigenetic alterations might be important in both gastric cancers with current H. pylori infection and past infection.
Somatic point mutations were analyzed by next-generation target sequencing, which covered 190 regions of 46 cancer-related genes. Although this panel covered almost all of the mutation hot spots of oncogenes, such as KRAS, PIK3CA, and CTNNB1, it covered limited regions of tumor-suppressor genes, such as TP53 (55.3%), CDH1 (7.5%), and MLH1 (2.6%). In addition, this panel did not cover several genes known to be mutated in 10% or more of gastric cancers, such as ARID1A, CREBBP, ERBB3, SMARCA4, and TGFBR2. Gene amplification was analyzed for 26 genes, including both oncogenes and tumor-suppressor genes, but was detected only in oncogenes, supporting the methodological validity. Approximately 9% of gastric cancers are known to be affected by Epstein-Barr virus (EBV), but EBV infection status was not analyzed in this study. EBV-positive gastric cancers are reported to have recurrent mutations of PIK3CA, ARID1A, and BCOR and amplifications of JAK2, PD-L1, and PD-L2 24  www.nature.com/scientificreports/ Eradication of H. pylori is known to prevent the progression of gastric atrophy or intestinal metaplasia (IM) 25 , and almost all patients with gastric cancers are known to have gastric atrophy or IM. Actually, also in this study, most patients with past H. pylori infection had atrophy (Supplementary Figure S1). Although information on clinical history will improve the data quality, we consider that the number of patients with H. pylori eradicated before the development of gastric atrophy or IM would be small.
In conclusion, gastric cancers with current H. pylori infection and those with past infection had comparable profiles of genetic alterations.

Methods
Clinical samples. Surgically resected and fresh-frozen samples of 96 pairs of gastric cancers and corresponding non-cancerous tissues were obtained from the National Cancer Center Biobank. Twenty-one pairs of gastric cancers and corresponding non-cancerous tissues were collected for our previous study 23 , and also used for this study. This study was approved by the Institutional Review Boards of the National Cancer Center, Japan (2012-305 and 2018-024), and written informed consents were obtained from all the patients. All methods were carried out in accordance with relevant guidelines and regulations. Genomic DNA was extracted from gastric cancers and corresponding non-cancerous tissues by the phenol/chloroform method.

Analysis of H. pylori infection status.
The infection status of H. pylori was determined by detection of PCR products specific for H. pylori jhpr3 gene and endoscopic gastric atrophy. Sensitivity and specificity for H. pylori detection by PCR test, urea breath test and serology test are reported to be > 95% and > 95%, 95.9% and 95.7%, and 76-84% and 79-90%, respectively 26 . Therefore, the reliability of a PCR test can be considered to be comparable with the other two methods. To avoid false-negative results in PCR, the quality of genomic DNA extracted from non-cancerous tissues was first evaluated by measuring the copy number of RPPH1 using quantitative PCR (qPCR) with primers listed in Supplementary Table S3 27 . Among the 117 samples, 110 samples had 1,000 copies or more in 10 ng of genomic DNA, and were qualified for the evaluation of H. pylori infection status.
The presence of H. pylori was evaluated by qPCR using primers specific to the jhpr3 gene of H. pylori 8 (Supplementary Table S4) and 100 ng of genomic DNA from non-cancerous tissues. Samples with successful amplification of the jhpr3 gene in two independent experiments were regarded as H. pylori-positive, and those in neither experiment were regarded as negative. Samples with one positive and one negative result were excluded from the entire analysis. Among the 110 samples, 59 samples were H. pylori-positive, 36 samples were -negative, and 15 samples were excluded. Endoscopic gastric atrophy was evaluated according to the endoscopic atrophic-border scale described by Kimura Table S5).

Next-generation target sequencing. Next-generation target sequencing was conducted using an Ion
AmpliSeq Cancer Panel Kit (Thermo Fisher Scientific, Waltham, MA), as described previously 23,29 . The sequence library was prepared by a multiplex PCR, which amplified 190 regions of 46 cancer-related genes. The library DNA was loaded onto an Ion PI Chip v3 (Thermo Fischer Scientific) or Ion 318 Chip v2 (Thermo Fischer Scientific) using Ion Chef (Thermo Fischer Scientific), and was sequenced using an Ion Proton sequencer (Thermo Fischer Scientific) or an Ion PGM sequencer (Thermo Fischer Scientific). The sequences obtained were mapped onto the human reference genome hg19 with Torrent Suite 5.0 (Thermo Fischer Scientific). An amplicon with 50 reads or less was considered to have low coverage, and two samples with 10% or more amplicons of low coverage were excluded from the analysis. Finally, 55 samples with current infection and 35 samples with past infection were used for mutation and amplification analysis. A variant call was conducted using CLC Genomics Workbench 20.0 (Qiagen, Hilden, Germany) with the following criteria; (i) with an allele frequency of 10% or www.nature.com/scientificreports/ www.nature.com/scientificreports/ www.nature.com/scientificreports/ more, (ii) not in homopolymers with 3 bp or more, (iii) present in both forward and reverse reads, and (iv) with a non-synonymous amino acid change. Sequence variations registered in dbSNP Build 137 were considered as SNPs, and were excluded before Sanger sequencing.
Sanger sequencing. Genomic regions where a sequence variation was found were amplified using 20 ng of genomic DNA (gastric cancers and corresponding non-cancerous tissues) and primers listed in Supplementary  Table S6. The PCR products were purified by a DNA Clean and Concentrator-5 Kit (Zymo Research, Irvine, CA), and were sequenced by using a BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fischer Scientific) and 3730xl DNA Analyzer (Thermo Fischer Scientific). Sequence variations detected only in gastric cancers were considered as a somatic point mutation. Hotspot mutations were defined using information registered in COS-MIC. Namely, a pathogenic mutation at the specific base position whose frequency was 5% or more of all the mutations in a specific gene was defined as a hotspot mutation. Among the 154 variations detected in 72 gastric cancers (newly analyzed cases in this study) by a next-generation sequencer, 101 variations were confirmed by Sanger sequencing (54 and 47 were somatic mutations and SNPs, respectively).

Analysis of SNPs.
Six sequence variations registered in dbSNP Build 137 and twelve sequence variations confirmed as a SNP by Sanger sequencing were considered as SNPs (Supplementary Table S1). The frequencies of identified SNPs in gastric cancer patients (cases in this study) and healthy Japanese people in datasets of the Tohoku Medical Megabank Organization (ToMMo 4.7K JPN) were compared by the Fisher's exact test.
Analysis of gene amplification. Gene amplification was analyzed using a next-generation sequencer since copy number variations (CNVs) detected by next-generation sequencers are now known to be accurately confirmed by Multiplex ligation-dependent probe amplification (MLPA), the gold-standard method to evaluate CNVs (Specificity 100%) 30 . Gene amplification of 26 genes (ABL1, APC, ATM, CDH1, EGFR, ERBB2,  ERBB4, FBXW7, FGFR2, FGFR3, FLT3, KDR, KIT, KRAS, MET, PDGFRA, PIK3CA, PTEN, RB1, RET, SMAD4,  SMARCB1, SMO, STK11, TP53, and VHL), which had three PCR amplicons or more, was analyzed, as described previously 23 . For an individual sample, reading depths of 160 amplicons of the 26 genes in the sample (y-axis) and in all the samples (average, x-axis) were plotted. The amplicons were expected to be on a regression line, but amplicons of an amplified gene were outlying. The ratio of the slope of a specific gene to that of the all genes was calculated, and genes with a ratio of three or more were defined as amplified genes. Since the next-generation target sequencing of 74 gastric cancers newly collected in this study was conducted in two sequencing runs, there were two background average reading depths (Supplementary Tables S7 and S8). The origins of gastric cancer samples with gene amplification (from our previous study or new in this study) are noted (Supplementary  Table S2). www.nature.com/scientificreports/  www.nature.com/scientificreports/