Comparison between two amplicon-based sequencing panels of different scales in the detection of somatic mutations associated with gastric cancer

Background Sequencing data from The Cancer Genome Atlas (TGCA), the International Cancer Genome Consortium and other research institutes have revealed the presence of genetic alterations in several tumor types, including gastric cancer. These data have been combined into a catalog of significantly mutated genes for each cancer type. However, it is unclear to what extent significantly mutated genes need to be examined for detecting genetic alterations in gastric cancer patients. Here, we constructed two custom-made sequencing panels of different scales, the Selective ﻿hotspot﻿ Panel and the Comprehensive Panel, to analyze genetic alterations in 21 resected specimens endoscopically obtained from 20 gastric cancer patients, and we assessed how many mutations were detectable using these different panels. Results A total of 21 somatic mutations were identified by the Selective hotspot Panel and 70 mutations were detected by the Comprehensive Panel. All mutations identified by the Selective hotspot Panel were detected by the Comprehensive Panel, with high concordant values of the variant allelic fraction of each mutation (correlation coefficient, R = 0.92). At least one mutation was identified in 13 patients (65 %) by the Selective hotspot Panel, whereas the Comprehensive Panel detected mutations in 19 (95 %) patients. Library preparation and sequencing costs were comparable between the two panels. Conclusions Our results indicate the utility of comprehensive panel-based targeted sequencing in gastric cancer. Electronic supplementary material The online version of this article (doi:10.1186/s12864-016-3166-4) contains supplementary material, which is available to authorized users.


Background
Gastric cancer is the third-and fifth-highest cause of cancer mortality in men and women, respectively, and accounts for 8 % of total cancer cases and 10 % of total cancer-related deaths worldwide [1]. The highest incidence rates of gastric cancer are in Eastern Asia, Eastern Europe, and South America, while the lowest rates are in North America and most parts of Africa [1]. Major risk factors include Helicobacter pylori and Epstein-Barr virus infection, as well as dietary factors such as excessive salt intake [2,3].
Gastric cancer develops in a step-wise manner, involving chronic gastritis, atrophy, intestinal metaplasia, and dysplasia [4]. Early gastric cancer presents as a malignant tumor confined to the mucosa or submucosa, regardless of the presence of regional lymph node metastasis [5,6]. The detection of early gastric cancer has recently improved, following the development of endoscopic techniques [7,8]. In particular, endoscopic submucosal dissection (ESD) has enabled a high en bloc resection rate for small and large lesions, as well as in patients with scarring. Moreover, the specimens obtained by ESD can be used for a histological assessment of curability [9]. Endoscopic resection is now widely accepted as a low invasive method for the local resection of early gastric cancer with a negligible risk of lymph node metastasis [10,11]. Endoscopically-resected early gastric cancer also provides suitable material for genomic analysis to better understand the molecular and genetic features of the initial event leading to cancer development [12].
Next-generation sequencing (NGS) technology enables us to determine the sequence of the genome at a range of different scales, including whole genome, whole exome, and the targeted sequencing of multiple regions of interest. Whereas large-scale analyses are essential for discovery projects, targeted sequencing can focus on genes associated with disease and may lead to advances in the molecular diagnostics of cancer [13]. As an example, NGS has identified a subset of driver and tumor suppressor genes associated with several cancer types [14]. It can also produce thousands to millions of short sequence reads that are massively parallel, and offers a cost-effective approach for detecting genetic alterations.
Large amounts of sequencing data have been disclosed from The Cancer Genome Atlas (TCGA), the International Cancer Genome Consortium (ICGC) and other research institutes. Analyses of these data identified significantly mutated genes (SMGs) in several cancer types [15,16]. Although SMGs have been revealed by whole exome and whole genome sequencing data, it is unclear to what extent SMGs need to be examined for detecting genetic alterations in gastric cancer. In the present study, we used gastric cancer-associated SMGs to construct two sequencing panels of different scales [17][18][19][20][21][22][23]. We performed targeted sequencing and analyzed genetic alterations in gastric tumors at an early phase and assessed how many mutations were detectable using these different panels.

Patients and sample preparation
This study included 20 patients who were diagnosed with gastric cancer (16 males and four females; age 60-87 years) at our hospital (Yamanashi, Japan), one of whom had two tumors. Informed consent was obtained from all subjects. This study was approved by the Institutional Review Board at our hospital and complied with Declaration of Helsinki principles. Peripheral blood samples were obtained from gastric cancer patients and DNA extraction was performed as previously described [24]. Briefly, peripheral blood samples were centrifuged at 820 × g at 25°C for 10 min, and buffy coats were isolated and stored at −80°C until required for DNA extraction. Buffy coat DNA was extracted using the QIAamp DNA Blood Mini QIAcube Kit (Qiagen, Hilden, Germany) with the QIAcube (Qiagen). The concentration of DNA was determined using the Nano Drop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA).

Laser capture microdissection and histology
Tumor samples were fixed using 10 % buffered formalin. Serial sections of 10-μm-thick, formalin-fixed, paraffinembedded (FFPE) tissue were stained with hematoxylin and eosin, and then microdissected using an ArcturusXT laser capture microdissection system (Thermo Fisher Scientific) using ESD-resected specimens. Tumor cells from endoscopic biopsy samples were obtained from 25 serial sections because of the high tumor content. Tumor DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen).

DNA quality analysis
The integrity of purified DNA from FFPE samples was assessed using the TaqMan RNase P Detection Reagents kit and the FFPE DNA QC Assay v2 on the ViiA 7 Real-Time PCR System (Thermo Fisher Scientific). Human control genomic DNA included in the TaqMan RNase P Detection Reagents Kit was diluted to create a five-point serial dilution for a standard curve, and absolute DNA concentrations were determined. DNA fragmentation was estimated as the ratio of DNA (relative quantification; RQ) obtained for the long amplicon to the short amplicon. High RQ values indicated that the genomic DNA was intact and high quality.

Selecting genes and primer design
We searched the literature and selected genes based on the following criteria (Additional file 1: Table S1): (a) SMGs relative to the background mutation rates analyzed by MutSigCV analysis tool [17]; (b) genes involved in signaling pathways and potential therapeutic targets in gastric cancer; and (c) known drivers of gastric carcinogenesis reported by TCGA [17] and other projects [18][19][20][21][22]. We examined the hotspot mutation site of each gene in gastric cancer from the COSMIC database (http:// cancer.sanger.ac.uk/cancergenome/projects/cosmic).
We selected 20 genes for the Selective hotspot Panel, which comprises a subset of SMGs and genes related to receptor tyrosine kinases (RTKs) and RAS signaling pathway based on the TCGA project [17]. To expand and cover more SMGs, we selected 58 genes (which include the 20 genes in the Selective hotspot Panel) based on published data from TCGA and another research institute [17][18][19][20][21][22][23]. Ion AmpliSeq designer software (Thermo Fisher Scientific) was used to design two custom sequencing panels: the Selective hotspot Panel targeting 20 genes in gastric cancer and the Comprehensive Panel targeting 58 genes [17][18][19][20][21][22][23] (Table 1). A total of 376 and 3515 primer pairs were contained within the Selective hotspot Panel (covering 38.01 kb) and the Comprehensive Panel (covering 351.05 kb), respectively.

Targeted sequencing
Targeted sequencing was performed as previously described [25]. Multiplex polymerase chain reaction (PCR) of these panels was performed using the Ion AmpliSeq Library Kit 2.0 (Thermo Fisher Scientific). Primer sequences were digested with FuPa reagent (Thermo Fisher Scientific), and then barcoded using Ion Xpress Barcode Adapters (Thermo Fisher Scientific). Purification was carried out by Agencourt AMPure XP reagents (Beckman Coulter, Brea, CA). The library concentration was determined using an Ion Library Quantitation Kit (Thermo Fisher Scientific); each library was diluted to 10 pM, and the same amount of libraries was pooled for one sequence reaction. Emulsion PCR was carried out using the Ion OneTouch System and Ion PGM Template OT2 200 kit or Ion PI Template OT2 200 Kit v3 (Thermo Fisher Scientific). Template-positive Ion Sphere Particles were then enriched using the Ion OneTouch ES system (Thermo Fisher Scientific), and purified Ion Sphere particles were loaded on an Ion 318 Chip v2 or PI Chip (Thermo Fisher Scientific). Massively parallel sequencing was carried out on Ion PGM or Ion Proton systems (Thermo Fisher Scientific).

Data analysis
Sequence data were processed using standard Ion Torrent Suite Software running on the Torrent Server. Raw signal data were analyzed using Torrent Suite version 4.4. The data processing pipeline involved signaling processing, base calling, quality score assignment, adapter trimming, PCR duplicate removal, read alignment to the human genome 19 reference (hg19), quality control of mapping quality, coverage analysis, and variant calling. Following data analysis, the annotation of single nucleotide variants, insertions, and deletions was performed by the Ion Reporter Server System (Thermo Fisher Scientific), and peripheral blood DNA was used as a control to detect variants in tumors (Tumor-Normal pairs). We used the following filtering parameters for variant calling: the minimum number of variant allele reads was ≥5, the coverage depth was ≥10, and the variant allele fraction was ≥10 %. If somatic mutations were called using either the Selective hotspot Panel or Comprehensive Panel, sequence data were visually confirmed with the Integrative Genomics Viewer and any sequence, alignment, or variant call error artifacts were discarded.

Quality assessment of extracted FFPE DNA
We examined 21 FFPE tumor samples collected from 20 patients (early stage, 19 patients; advanced stage, one patient) who had not previously undergone chemotherapy or radiotherapy. Matched peripheral blood lymphocytes were included as a control. Of the 21 FFPE tumor samples, 19 tumors had been resected by ESD and two by endoscopic biopsy. ESD-resected tumor tissue was dissected by laser capture microdissection with an average cutting area of 29.4 mm 2 (range, 12.4-51.5 mm 2 )    Fig. 1 and Additional file 1: Table S2). Endoscopic biopsy samples were not microdissected because of the high tumor content.

Targeted sequencing analysis
To identify genetic alternations in gastric cancer, we reviewed cancer genome sequences from TCGA, ICGC, and COSMIC databases, and selected all SMGs associated with gastric cancer. We constructed two custommade gastric cancer panels. The Selective hotspot Panel spans 38,010 nucleotides, covers 20 SMGs, and mainly targets hotspot regions ( Table 1). The Comprehensive Panel spans 354,050 nucleotides, and 58 of the genes contained within this panel overlapped with the Selective hotspot Panel (Table 1).
We performed targeted sequencing using the two panels with a next-generation sequencer (Ion Proton or Ion PGM, Thermo Fisher Scientific). The percentage of mapped reads aligned to target regions was 98.7 % (97.6-99.3 %) in the Selective hotspot Panel and 97.0 % (95.0-98.6 %) in the Comprehensive Panel, suggesting that all FFPE-derived DNAs had been successfully subjected to library preparation following sequencing analysis ( Table 2).
The mean coverage depth of tumors was 610× (range, 207-1502) by the Selective hotspot Panel, and 516× (range, 145-923) by the Comprehensive Panel ( Table 2). The two approaches identified a total of 21 and 70 somatic mutations in tumors, respectively ( Fig. 2a and Table 3). All 21 mutations identified by the Selective hotspot Panel were also confirmed by the Comprehensive Panel (Fig. 2a). The variant allelic fraction values were significantly correlated between the two panels (Fig. 2b).

Running costs
Primer costs for the Comprehensive Panel were higher than those of the Selective hotspot Panel (Comprehensive Panel: $26363 vs. Selective hotspot Panel: $2820). However, the total cost of library preparation, emersion PCR, and massively parallel sequencing was comparable between the two panels at $200-250 per sample. Use of the Barcode Xpress toolkit enabled multiple samples to be simultaneously sequenced in 4-5 h and allowed us to obtain high-depth sequence data using the Ion PGM or Ion Proton system.

Discussion
The identification of oncogenic driver genes has led to the development of potent molecular targeting drugs together with companion diagnostics. The advent of NGS has also resulted in the identification of a subset of cancer-related genes in several tumors [14,15], including hundreds of genes mainly associated with tumor development [28]. TCGA, ICGC, and other research institutes have revealed a tumor mutational landscape and produced a catalog of somatic mutations associated with tumors. Information from this catalog has enabled the analysis of recurrently mutated genes by targeted sequencing [29]. This is a useful, cost-effective method for identifying variants in dozens to hundreds of genes, and is fairly readily available for routine diagnosis in a clinical setting as well as for research purposes.
In this study, we constructed two amplicon-based targeted panels of different scales to analyze the genetic alterations associated with gastric cancer. In our cohort, 20 out of 21 tumors (95 %) were shown to carry at least one mutation by the Comprehensive Panel. Thus, our panel-based approach enabled us to detect somatic mutations in gastric cancer, suggesting that it has the potential to obtain robust data and to detect genetic  Progress in endoscopic technology has led to the curative resection of gastric cancer at an early stage. However, although ESD is widely performed to resect early gastric cancer, the genetic alterations occurring in such tumors are not fully understood, even though this would provide us with an insight into the mechanisms of tumorigenesis. Here, we performed targeted sequencing using ESD-resected early gastric cancers, together with endoscopically-resected biopsies of advanced cancer. A total of 70 somatic mutations were identified in 19 patients, and an average 3.2 mutations were found in early gastric cancer. The most recurrent mutation was identified in TP53 gene (43 %, 9/21). In line with this observation, previous studies have shown that TP53 mutations occur in early gastric cancer as well as in high-grade intraepithelial neoplasia [30]. These observations indicate that TP53 is a key molecule for the progression of gastric tumorigenesis.
In this study, somatic mutations in TP53 (43 %), APC (29 %), MUC6 (33 %), and SYNE1 (24 %) were frequently observed (identified in over 20 % of tumors). These frequencies are almost consistent with previous studies , and SYNE1 (20 %). Less common mutations were observed in CTNNB1 (5 %) and KRAS (5 %) genes in our study, but these gene mutations (CTNNB1 S45F and KRAS G13D) are well-known hotspot driver mutations [31]. Previous data also showed that CTNNB1 (1-9 %) and KRAS (5-6 %) mutations were relatively uncommon in gastric cancer. These results indicated that our designed panels validated the data of previous reports. The TCGA project demonstrated there are four major subtypes of gastric cancer based on the genomic analysis, i.e., chromosomal instability (CIN), genomically stable (GS), Epstein-Barr virus-positive and microsatellite instability [17]. According to this molecular classification, TP53 mutation mostly occurs in the CIN category and intestinal histology. Consistent with this, we examined ESD-resected gastric tumors and most were intestinal type gastric cancer (data not shown). Additionally, the GS subtype is classified as diffuse histology and frequently shows CDH1 and RHOA mutations and CLDN18-ARHGAP fusion. Again, in our series, one advanced gastric cancer was diffused type histology and had a CDH1 splice site mutation (Case 20 in Table 3). Collectively, our data reinforced the molecular classifications of gastric cancer.
Analyses that include a large number of SMGs are important for several reasons. First, analyzing additional SMGs will detect more somatic alterations in tumors. In this study, we were unable to identify any mutations in seven patients using the Selective hotspot Panel, compared with only one using the Comprehensive Panel (Table 3). A recent study reported newly identified SMGs including NRG1, ERBB4, XIRP2, NBEA, COL14A1, CNBD1, ITGAV, and AKAP6 [32,33] that should be included in the mutational spectrum analyzed in all patients with gastric cancer. Second, from a cost perspective, covering more SMGs is beneficial, as shown by the comparable library preparation and sequencing running costs between the two panels used in this study. Third, including more primer pairs in the design of the panel enables more high-resolution copy number data to be examined [34]. Previous bioinformatics analysis combined with variant allelic fraction and copy number alteration data revealed the cellular prevalence of tumor heterogeneity [35]. Together, these findings suggest that SMG-based sequencing analysis is a useful method for further investigating tumor heterogeneity in clinical samples.

Conclusions
In the present study, use of the Comprehensive Panel covering SMGs associated with gastric cancer enabled the analysis of genetic alterations in patients with early gastric cancer.