Distribution of somatic mutations of cancer-related genes according to microsatellite instability status in Korean gastric cancer

Abstract

In studies of the molecular basis of gastric cancer (GC), microsatellite instability (MSI) is one of the key factors. Somatic mutations found in GC are expected to contribute to MSI-high (H) tumorigenesis. We estimated somatic mutation distribution according to MSI status in 52 matched pair GC samples using the Ion Torrent Ion S5 XL with the AmpliSeq Cancer Hotspot panel.

Seventy-five (9.8%) somatic variants consisting of 34 hotspot mutations and 41 other likely pathogenic variants were identified in 34 GC samples. The TP53 mutations was most common (35%, 26/75), followed by EGFR (8%, 6/75), HNF1A (8%, 6/75), PIK3CA (8%, 6/75), and ERBB2 (5%, 4/75). To determine MSI status, 52 matched pair samples were estimated using 15 MSI markers. Thirty-nine MS stable (S), 5 MSI-low (L), and 8 MSI-H were classified. GCs with MSI-H tended to have more variants significantly compared with GCs with MS stable (MSS) and MSI-L (standardized J-T statistic  =  3.161 for number of variants; P  =  .002). The mean number of all variants and hotspot mutations per tumor samples only in GCs with MSI-H were 3.9 (range, 1–6) and 1.1 (range, 0–3), respectively. Whereas, the mean number of all variants and hotspot mutations per tumor samples only in GCs with MSS/MSI-L were 1 (0–5)/0.8 (0–1) and 0.5 (0–3)/0.8 (0–1), respectively.

In conclusion, GC with MSI-H harbored more mutations in genes that act as a tumor suppressor or oncogene compared to GC with MSS/MSI-L. This finding suggests that the accumulation of MSIs contributes to the genetic diversity and complexities of GC. In addition, targeted NGS approach allows for detection of common and also rare clinically actionable mutations and profiles of comutations in multiple patients simultaneously. Because GC shows distinctive patterns related to ethnics, further studies pertaining to different racial/ethnic groups or cancer types may reinforce our investigations.

1. Introduction

Gastric cancer (GC) is currently the third leading cause of global cancer-related death, and is particularly prevalent in Asia.^[1] Gastric adenocarcinoma, the most common type of GC, is heterogeneous and its incidence and cause varies widely with geographical regions, gender, ethnicity, and diet. Achieving a detailed molecular understanding of GC pathogenesis is pivotal to improving patient outcomes. The identification of genomic alterations may provide insight into the mechanisms for oncogenic gastric pathways and is important to identify a tumor marker in GC.^[2–6]

Updated studies on GC molecular profiling have revealed heterogeneous characteristics for GC and have defined the subtypes on the basis of genomic basis, gene expression, and amplification patterns.^[6–8] The Cancer Genome Atlas recently published results of a comprehensive study of GC, which provides an invaluable resource upon which to interpret other related GC findings. A molecular classification scheme in that study defined 4 major genomic subtypes of GC: Epstein–Barr virus (EBV)-infected tumors, microsatellite instability (MSI) tumors, genomically stable tumors, and chromosomal instability (CIN) tumors. Identification of these subtypes links the molecular alterations with clinical phenotypes and disease progression, providing a roadmap for patient stratification and trials of targeted therapies. According to gene expression patterns, some recent studies have classified GCs into molecular subtypes that show differences in the molecular and genetic features and chemotherapeutic sensitivity.^[9,10] Importantly, the distinct salient genomic features of molecular subtypes will hopefully provide a guide to targeted agents that should be evaluated in clinical trials for distinct populations of GC patients. Several comprehensive studies describe that GC is associated with driver alterations, including gene mutations,^[11,12] somatic copy number alterations,^[13,14] structural variants, epigenetic changes,^[15] and transcriptional changes.^[16–18] Certain driver alterations can also be associated with specific GC subtypes and the major forms of genomic instability observed in GC are CIN and MSI tumors.^[19] MSI is defined as alterations in the lengths of microsatellites due to deletion or insertion of repeating units to produce novel length alleles in tumor DNA when compared with the normal/germline DNA from the same individual. When mismatch repair (MMR) genes, including mutL homolog 1 (MLH1) and mutS homolog 2 (MSH2), are inactivated, replication errors include insertions or deletions of bases within microsatellite regions.^[20]

Next-generation sequencing (NGS), also known as high-throughput massively parallel sequencing, can detect multiple gene variants simultaneously, allowing for the precise diagnosis of a tumor at the genetic level. NGS generates many sequences across numerous targets and numerous patient samples in the same reaction and in a single instrument run, which reduces cost compared to Sanger sequencing. NGS promises to bridge this gap by allowing for simultaneous mutation detection in multiple exons from multiple genes in multiple patient samples.^[21,22] NGS platforms offer an increased breadth of testing at a lower cost and without compromising assay performance and turn-around times in the clinical setting.^[23,24] NGS analysis have shown that 15% to 20% of GCs are characterized by MSI.^[25]

Several studies^[11,16,26] have characterized mutation profiles in Korean GC; however, somatic mutations of cancer-related genes according to MSI status have not been studied in clinical samples. Here, we present our experience with targeted semiconductor using the Ion Torrent Ion S5 XL with the AmpliSeq Cancer Hotspot panel v2 to estimate somatic mutation distribution according to MSI status in GC; this assay covers 2855 COSMIC-cited hot spot mutations in 50 cancer-related genes.

2. Materials and methods

2.1. Samples and DNA isolation

A total of 52 primary gastric tumor tissues and paired normal tissues were obtained from surgically dissected patients with GC between 1999 to 2016 at the Daejeon St. Mary's Hospital, Daejeon, Republic of Korea. The results of immunohistochemical staining for MMR protein expression, such as MLH1 and MSH2, revealed that 13 of these tumors exhibited defective DNA MMR, while 39 presented proficient MMR proteins.^[27] Each area of the tissue representing the “tumor” (highest numbers of cancer cells identified) and “normal” (no malignant tissue present) was microdissected separately, not to be contaminated by each other. Genomic DNA was extracted from microdissected tissue samples using the QIAmp DNeasy Blood and Tissue kit (Qiagen, Hilden, Germany) according to tissue protocol. Nucleic acid quality and quantity were assessed using a Nano-Drop 1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE) and agarose gel electrophoresis. DNA concentration was determined using the Broad range Qubit DNA kit and Qubit 2.0 Fluorometer (Life Technologies, Carlsbad, CA). The study protocol was approved by the Institutional Review Board of The Catholic University of Korea, including written informed consent for clinical and molecular analyses.

2.2. Library preparation

AmpliSeq libraries were generated using the Ion AmpliSeq Library Kit 2.0 and the Ion AmpliSeq Cancer Hotspot Panel v2 (Life Technologies). This panel contains 207 primer pairs in a single tube and surveys hotspot regions including up to 2855 COSMIC mutations of 50 oncogenes and tumor suppressor genes, with wide coverage of the KRAS, BRAF, and EGFR genes. Included in this panel are primers for the amplification of regions of the following 50 genes: ABL1, AKT1, ALK, APC, ATM, BRAF, CDH1, CDKN2A, CSF1R, CTNNB1, EGFR, ERBB2, ERBB4, EZH2, FBXW7, FGFR1, FGFR2, FGFR3, FLT3, GNA11, GNAS, GNAQ, HNF1A, HRAS, IDH1, JAK2, JAK3, IDH2, KDR, KIT, KRAS, MET, MLH1, MPL, NOTCH1, NPM1, NRAS, PDGFRA, PIK3CA, PTEN, PTPN11, RB1, RET, SMAD4, SMARCB1, SMO, SRC, STK11, TP53, and VHL. Multiplex PCR was amplified using 10 ng genomic DNA with a premixed primer pool and Ion AmpliSeq HiFi master mix (Ion AmpliSeq Library Kit 2.0). The PCR amplicons were treated with 2 μL FuPa reagent to partially digest the primer sequences and phosphorylate the amplicons. The amplicons were ligated to adapters with the diluted barcodes of the Ion Xpress Barcode Adapters kit (Life Technologies). The adapter-ligated amplicons (library) were purified using the Agencourt AMPure XP reagent (Beckman Coulter, Brea, CA). Amplified libraries were assessed for quality (size and concentration) using a model 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA) following the equipment's standard protocol. The ideal concentration for 1 sequence reaction on the One Touch instrument was between 8 and 16 pmol/L.

2.3. Emulsion PCR

The clonal amplification of the barcoded DNA library (The AmpliSeq libraries) onto the ion spheres (ISPs) was carried out using emulsion PCR following standard Ion Torrent protocols and the subsequent isolation of ISPs with DNA was conducted using Ion OneTouch 200 Template Kit v2 DL and Ion OneTouch ES (Life Technologies) according to the manufacturer's instructions. The polyclonal percentage and quality of the enriched, template-positive ISPs was determined using the Ion Sphere Quality Control Kit (Life Technologies). Samples with polyclonal percentage <30% and enriched, template-positive ISPs > 80% were subjected for sequencing on the Ion Torrent Ion S5 XL.

2.4. Sequencing

Enriched ISPs were loaded onto 540 chips taking 26 matched tumor and normal samples on a single chip per sequencing run. Ion Torrent Ion S5 XL sequencing was performed on the Ion 540 Kit-Chef (2 sequencing runs per initialization) and following the standard protocol. Sequencing was performed using 500 flow runs that generated approximately 200 bp reads.

2.5. Bioinformatic analysis

Raw signal data from sequencing runs from the Ion Torrent Ion S5 XL were automatically transferred to the Torrent Server Hosting the Torrent Suite Software that processed the raw voltage semiconductor sequencing data into DNA base calls. The pipeline included signaling processing, base calling, quality score assignment, adapter trimming, read mapping to 19 reference human genomes, quality control of mapping quality, coverage analysis with downsampling, and variant calling. Identification of variants was performed by the Ion Torrent Variant Caller plug-in and Ion Reporter software v5.2 (Life Technologies). Coverage maps were generated using the coverage analysis plug-in.

Torrent Variant Caller v5.2 was used for alignment and variant detection. The variant caller parameter setting was AmpliSEq Exome Tumor Normal v1 (5.2) Total Variants. Following data analysis, annotation of single-nucleotide variants, insertions, deletions, and splice site alterations was performed by the Ion Reporter Software (Life Technologies). Sequence data were visually confirmed with the Integrative Genomics Viewer (IGV) and any sequence, alignment, or variant call error artifacts were discarded. Hot spot mutations found in the 52 matched tumor and normal samples were compared to validate mutations listed in the Catalogue Of Somatic Mutations In Cancer (COSMIC). In addition, widely established computational prediction methods were applied to other than Hot spot mutations in this study. Evolution-based sequence information included the Sorting Intolerant From Tolerant (SIFT, http://sift.jcvi.org/) based on the degree of conservation of amino acid residues in sequence alignments derived from closely related sequences, collected through PSI-BLAST and the Grantham score that attempts to predict the distance between 2 amino acids, in an evolutionary sense.^[28] Polymorphism Phenotyping v2 (PolyPhen-2, http://genetics.bwh.harvard.edu/pph2/) was used to predict the possible impact of an amino acid substitution on the structure and function of a human protein using straightforward physical and comparative considerations.

2.6. MSI analysis

To determine the MSI status in 52 GC samples, MSI analysis was performed using 15 different mono- and dinucleotide microsatellite markers. Eight mononucleotide markers consist of BAT25, BAT26, BAT40, BAT-RII, NR21, NR22, NR24, and NR27. Seven dinucleotide markers include D2S123, D5S346, D17S250, D17S261, D17S520, D18S34, and D18S58. Fifteen MSI markers were divided into groups of 5 and coamplified in 3 reaction tubes (1 group of 5 per tube). Allelic sizes to match the tumor and normal samples were compared and considered to be MSI unstable if there was a shift of 3 bp or more in the tumor allele.^[29] All tumors with 1 or more unstable markers were regarded as carrying some degree of instability and were defined as MSI and MS stable (MSS) when there were no unstable markers. The cutoff for classification was applied on the basis of the threshold of about 40% that is commonly used to discriminate MSI-H (high) and MSI-L (low) tumors.^[30]

2.7. Statistical analyses

Only patients with valid, exact clinicopathological characteristics were included in the statistical analysis. Since the tumor size, depth of tumor invasion, lymph node (LN) involvement, and disease stage was treated as an ordinal variable, the Jonckheere–Terpstra test was used to compare the ordinal variables. Overall survival (OS) was defined as the time between diagnosis and death by any cause. The statistical analyses were then carried out using the MedCalc ver. 12.7.2 (MedCalc software, Mariakerke, Belgium) and a P < .05 was considered to be statistically significant.

3. Results

3.1. Clinicopathological characteristics

The median age of the patients was 65.5 years (range 34–84) and the male to female ratio was 1.65. The most common tumor site in the stomach was distal (n  =  28), followed by middle (n  =  16), and proximal (n  =  6). All tumors were diagnosed as GC; the vast majority were Borrmann 3 (n  =  31), followed by EGC IIb (n  =  7) and Borrmann 4 (n  =  4) based on gross type. The most common Lauren classification was intestinal (n  =  21), followed by mixed (n  =  14), and diffuse (n  =  12) type. According to the Japanese classification, tubular (n  =  25) and tubulopapillary (n  =  23) types were frequent. No metastasis was found; the disease stage was I (n  =  17), II (n  =  14), and III (n  =  21). Clinicopathological characteristics of the 52 GCs are provided in Table 1.

Table 1.

Clinicopathological characteristics of 52 gastric cancers according to microsatellite instability status.

3.2. Results of interest regions covered by the AmpliSeq Cancer Hotspot panel

The AmpliSeq Cancer Hotspot panel consists of 207 amplicons, which examines 2855 mutations in 50 commonly mutated oncogenes and tumor suppressor genes. Presently, the average of Q20 bases per sample were sequenced (Q20  =  Phred quality score of 20; base call accuracy of 99%) and the median of >Q20 bases was 170,640,400 (range, 69,833,168–348,107,992). The median total numbers of bases was 186,030,778 (range, 75,324,153–376,373,688), the median of reads was 1,627,422 (range, 659,686–3,296,967), the median of mapped reads was 1,612,050 (range, 650,180–3,258,476), the median of mean read length was 114 bp (range, 104–121), the median of on target rate (%) was 97.1 (range, 80.9–99.4), the median of mean depth was 7,246 (range, 2,712–15,153), and the median of mean coverage uniformity (>20% mean coverage, %) was 99.9 (range, 90.1–100).

3.3. Spectrum of pathogenic and likely pathogenic somatic variants

In total, we obtained 767 variants from 52 matched pair samples. Of these, 692 variants were identified in both tumor and normal samples. The mean number of all variants per tumor and normal samples were 14.8 (range, 8–22) and 13.3 (range, 7–20), respectively. The mean number of hotspot mutations per tumor and normal samples was 2 (range, 0–5) and 1.3 (range, 0–5), respectively. Ion AmpliSeq v2 revealed somatic mutations in 34 of the 52 GC matched pair samples. After filtering out variants identified only in tumor samples, 75 (9.8%) variants comprising 34 hotspot mutations and 41 pathogenic variants that were likely other than hotspot mutations were selected (Tables 2 and 3). Seventy-five somatic variants in 21 genes were detected with a mean of 2.2 variants (range, 1–6) per sample. Categorized according to mutated genes, TP53 mutations were most common (35%, 26/75), followed by EGFR (8%, 6/75), HNF1A (8%, 6/75), PIK3CA (8%, 6/75), ERBB2 (5%, 4/75), ATM (4%, 3/75), FGFR2 (4%, 3/75), CDKN2A (4%, 3/75), FLT3 (3%, 2/75), PDGFRA (3%, 2/75), PTEN (3%, 2/75), RB1 (3%, 2/75), STK11 (3%, 2/75), APC (1%, 1/75), FBXW7 (1%, 1/75), MET (1%, 1/75), NOTCH1 (1%, 1/75), RET (1%, 1/75), SMAD4 (1%, 1/75), SMO (1%, 1/75), and UBALD1 (1%, 1/75). The mutations detected in EGFR, ERBB2, PIK3CA, and STK11 were considered potentially actionable.^[31,32] Such actionable mutations were identified in 12 patients, most commonly in EGFR (8%, 6/75), PIK3CA (8%, 6/75), ERBB2 (5%, 4/75), and STK11 (3%, 2/75). Classified according by mutation types, the majority of pathogenic and likely pathogenic variants identified only in tumor samples were missense (65%, 49/75) followed by frameshift (28%, 21/75), nonsense (4%, 3/75), and silent (3%, 2/75) mutations (Fig. 1). We also identified the cooccurrence of the most frequently altered and clinically significant genes. Not surprisingly, TP53 mutations cooccurred most commonly with mutations of other genes. Among 6 patients with EGFR mutations, 3 also had a coexisting PIK3CA (patient I37), ERBB2 (patient I08), and STK11 (patient I41) mutations. Interestingly, the 3 patients harbored compound heterozygous mutations of PIK3CA (patient I37), TP53 (patient I45), and ERBB2 (patient I10).

Table 2.

Hotspot somatic mutations in 34 matched pair gastric cancers.

Table 3.

Other likely pathogenic somatic variants than hotspot mutations in 34 matched pair gastric cancers.

Figure 1.

Distribution of pathogenic and likely pathogenic somatic variants identified in 34 gastric cancers.

3.4. Associations between clinicopathological characteristics and MSI status

To determine MSI status, 52 matched pair samples were estimated using 15 MSI markers. Thirty-nine MSS and 13 unstable MSI were classified. Out of the 13 samples with defective MMR protein expression, 8 (62%) were classified as MSI-H with instability at ≥6 markers and 5 (38%) were MSI-L with instability at ≤5 markers. Jonckheere–Terpstra test results showed that when MSI status went from stable to highly unstable, tumor size, depth of tumor invasion, LN involvement, and disease stage tended to decrease. These observations were not statistically significant (standardized J-T statistic  =  −0.587 for tumor size, 0.274 for depth of tumor invasion, −1.485 for LN involvement, and −0.881 for stage; P  =  .557, .784, .138, and .378, respectively). However, GCs with MSI-H tended to have significantly more variants compared with GCs with MSS/MSI-L (3.161 for number of variants; P  =  .002) (Fig. 2). The mean number of all variants and hotspot mutations per tumor samples only in GCs with MSI-H were 3.9 (range, 1–6) and 1.1 (range, 0–3), respectively (Table 4). Meanwhile, 36 patients were alive and 6 had died at the time of the analysis. Ten were lost to follow-up during the study period. The median follow-up duration was 16.3 months (range, 8.7–216.1). Within the cohort of survival outcome that were available, all 6 with MSI-H were alive, while 5 with MSS and 1 with MSI-L had died.

Figure 2.

Spectrum of pathogenic and likely pathogenic somatic variants according to MSI status. Blue box, frameshift mutation; green box, silent mutation; red box, nonsense mutation; yellow box, missense mutation.

Table 4.

Mean number of variant per matched-pair samples according to MSI status in 52 matched pair gastric cancers.

4. Discussion

GC is characterized by a high level of biological heterogeneity, with each patient exhibiting a distinct genetic and molecular profile. Studies of the molecular basis of GC have led to the recognition of 4 major genomic subtypes of GC.^[10] MSI is one of the key factor in several cancers, including colorectal, endometrial, and GC. Microsatellite mutations found in these cancers are expected to contribute to MSI-H tumorigenesis.^[33] Depending on the type and number of MSI markers analyzed, widely variable results have been reported for the frequency of MSI-H in GC: 11.7% to 33.8% in Asian^[34–36] and 16.3% to 25.2% in European^[37–40] population. Recent exome sequencing of gastric adenocarcinoma showed that samples with unstable MSI had an average of 31.61 somatic mutations per megabase of DNA, whereas MSS GCs had an average of 3.29, a difference of approximately 10-fold.^[41] Moreover, a recent comprehensive genome- and transcriptome-wide study of Korean GCs supported the suggestion that mutations in 3′ untranslated regions (UTRs) influence gene expression in MSI-H tumors.^[16] Several cis- or trans-elements of the UTRs might fail to regulate gene functions like stability and activity if the mutations alter RNA sequences or structure.^[16] This finding suggests that aberrant expression of genes may create a growth or survival advantage for MSI-H GC. In our study, the survive analysis between MSI-H and MSS/MSI-L was not suitable because of the small sample size (n  =  52) and short median follow-up duration of 16.3 months. A recent meta-analysis for GC with unstable MSI with good prognosis, although heterogeneity in recent studies was present.^[42] MSI-H indicated by double negativity of MLH-1 and MSH-2 was not implicated in the pathogenesis of the early-onset GCs.^[43] Altered epidemiology and effects of chemotherapy are potential causes of the heterogeneity. On the other hand, MSI-H tumors were associated with a good prognosis in Stage II and III GC when patients were treated by surgery alone, and the benefits of MSI-H status were attenuated by chemotherapy.^[44]

Postreplication MMR is an important mechanism for maintaining microsatellite stability through the correction of base change mismatches and insertion/deletion events. MSI has length specificity, and the instability was driven mostly by deletion rather than insertion.^[16] Premutational intermediates are identified and processed by heterodimers of the MutS and MutL family of proteins.^[45] The number of candidate genes with deletion mutations is significantly greater than previously reported, implying that the accumulation of MSIs contributes to the genetic complexities of GC.^[16] Colorectal cancer (CRC) patients with defective MMR tumors have distinct clinical and pathologic features, thus this finding highlights the importance of MSI testing in early-stage disease where patients can be potentially cured by surgery alone or combined with adjuvant chemotherapy.^[46] Important in the biology of CRC are somatic mutations in the KRAS and BRAF oncogenes and the status of the DNA MMR system.^[46,47] Recently, Gurzu et al^[48] identified the molecular and immunohistochemical criteria that can be used to recognize the possible serrated pathway. Interestingly, we characterized GC with MSI-H containing more somatic variants than that with MSS and MSI-L (3.9 vs 1 and 0.8). The detection of multiple, cooccurring, potentially actionable mutations in an individual tumor represents an advance in molecular pathology with possible significant clinical, therapeutic, and research implications based on the different combinations of mutations.^[32] Even though the mutational spectra for genes with high mutational frequencies are quite different between MSI-H GC and CRC, our analysis of mutations associated with MSI in GCs will provide further information about discrete molecular pathways, which may explain the difference. The number of candidate cancer-related genes mutated in cases of GC with MSI-H is significantly greater than those in GCs with MSS or MSI-L, implying that the accumulation of MSIs contributes to the genetic complexities of GC. The present findings will enhance our understanding of gastric tumorigenesis in MSI-H cancers. Establishing a consensus for defining MSI in GC is a laudable aim for further studies.

Because distinct somatic mutations have been identified in each human cancer, it is essential to classify and characterize the molecular alterations underlying GC for improving more personalized and precision therapies.^[49] Once the molecular profile of a tumor is known, the appropriate use of targeted clinical therapies or eligibility for clinical trials can be determined. It is desirable to have the ability to analyze several genes simultaneously to assess for the presence of a known clinically actionable variant in a tumor. To understand and develop new therapeutics and treat GC more effectively, it is essential to profile the individual cancer genome and MSI status and dissect the oncogenic mechanisms that regulate the progression of GC, which may form the foundation for individualized, tailored therapy. The molecular profiling using NGS technologies offers advantages to detect somatic cancer genome alterations in accuracy, sensitivity, and speed that can make a significant impact, enabling the assessment of all potentially causative genes at the same time.^[21] In this study, we estimated the utility of the Ion Torrent Ampliseq technology for clinical genotyping of GC. Using this targeted NGS panel in our cohort, we frequently identified TP53 (48%, 25/52 vs 4.6% and 27%) and PIK3CA (10%, 5/52 vs 5.1% and 5.6%) mutations similar to previous reports about Korean GCs,^[11,26] whereas higher frequency of EGFR (12%, 6/52 vs 0% and 0%) and HNF1A (12%, 6/52 vs 0% and 0%) mutations were evident. This may be due to different GC subtypes, histological type, the TNM staging, degree of metastasis, use of matched pair samples, or the design of gene panel even though the same Korean ethnicity was studied. We identified rare mutations that would be specific therapeutic significance. Examples include detection of EGFR c.278delT (p.Leu93fs), c.2399A>G (p.Asp800Gly), c.2405delT (p.Val802fs); ERBB2 c.2264T>C (p.L755S), c.2524G>A (p.V842I), c.2408_2409delATinsTC (p.Tyr803Phe); PIK3CA c.3140A>G (p.H1047R), c.1636C>G (p.Q546E), c.1637A>G (p.Q546R), c.3140A>T (p.H1047L), c.1637A>G (p.Q546R), c.3140A>G (p.H1047R), and STK11 c.1096T>A (p.Phe366Ile). There were no recurrent mutations in the BRAF, EGFR, ERBB2, PDGFRA, PTEN, and RET genes, for which targeted drug therapies are available.

Although NGS technology is useful for identifying and characterizing the somatic mutations that accrue in GC and provides novel potential targets for molecular therapies, there are several limitations for GC.^[50] First, the previous reports were studied at single institutes using specific subtypes of GC in small sample sizes, rather than all types of GCs. Thus, a large-scaled multiinstitutional study for GC using optimized NGS for GC is required. Second, functional study on potential driver or actionable genes discovered by NGS should be conducted to prove the functional consequences of genomic alterations.^[51,52] To date, 2 candidate driver mutated genes (TP53 and ARID1A) have been simultaneously identified by exome sequencing.^[41,53,54] Despite these limitations, NGS remains a powerful molecular profiling approach, which enables ultra-deep sequencing of the primary tumor lesion to detect rare subclones, and low-depth sequential tumor characterization to identify dominant clones.^[55,56]

In conclusion, GC with MSI-H harbored more mutations in genes that act as a tumor suppressor or oncogene compared to GC with MSS/MSI-L. This finding suggests that the accumulation of MSIs contributes to the genetic diversity and complexities of GC. In addition, targeted NGS approach allows for detection of common and also rare clinically actionable mutations and profiles of comutations in multiple patients simultaneously. Because GC shows distinctive patterns related to ethnics, further studies pertaining to different racial/ethnic groups or cancer types may reinforce our investigations.

Acknowledgments

We would like to specially thank Chang Hyeon Lee and Keun-Joon Park of Thermo Fisher Scientific for providing technical support about Ion Torrent NGS works in this study.