Actionable gene-based classification toward precision medicine in gastric cancer

Abstract

Background: Intertumoral heterogeneity represents a significant hurdle to identifying optimized targeted therapies in gastric cancer (GC). To realize precision medicine for GC patients, an actionable gene alteration-based molecular classification that directly associates GCs with targeted therapies is needed.

Methods: A total of 207 Japanese patients with GC were included in this study. Formalin-fixed, paraffin-embedded (FFPE) tumor tissues were obtained from surgical or biopsy specimens and were subjected to DNA extraction. We generated comprehensive genomic profiling data using a 435-gene panel including 69 actionable genes paired with US Food and Drug Administration-approved targeted therapies, and the evaluation of Epstein-Barr virus (EBV) infection and microsatellite instability (MSI) status.

Results: Comprehensive genomic sequencing detected at least one alteration of 435 cancer-related genes in 194 GCs (93.7%) and of 69 actionable genes in 141 GCs (68.1%). We classified the 207 GCs into four The Cancer Genome Atlas (TCGA) subtypes using the genomic profiling data; EBV (N = 9), MSI (N = 17), chromosomal instability (N = 119), and genomically stable subtype (N = 62). Actionable gene alterations were not specific and were widely observed throughout all TCGA subtypes. To discover a novel classification which more precisely selects candidates for targeted therapies, 207 GCs were classified using hypermutated phenotype and the mutation profile of 69 actionable genes. We identified a hypermutated group (N = 32), while the others (N = 175) were sub-divided into six clusters including five with actionable gene alterations: ERBB2 (N = 25), CDKN2A, and CDKN2B (N = 10), KRAS (N = 10), BRCA2 (N = 9), and ATM cluster (N = 12). The clinical utility of this classification was demonstrated by a case of unresectable GC with a remarkable response to anti-HER2 therapy in the ERBB2 cluster.

Conclusions: This actionable gene-based classification creates a framework for further studies for realizing precision medicine in GC.

Keywords: Actionable gene; Gastric cancer; Gene panel; Next-generation sequencing; Precision medicine.

Fig. 1

The Cancer Genome Atlas molecular subtypes in 207 Japanese gastric cancers (GCs). a The 207 Japanese GCs are classified into four molecular subtypes; Epstein-Barr virus-positive (EBV, red), tumor with microsatellite instability (MSI, blue), tumors with chromosomal instability (CIN, purple), and genomically stable tumors (GS, green). Color tiles indicate pathological and molecular characteristics of of GC. Tumors are ordered by mutation rate and the red dots indicate hypermutated tumors with a mutation rate of more than 18.5 counts/Mb, which was the lowest value in the MSI subtype. b Classification of 207 Japanese GCs into TCGA molecular subtypes. The cutoff value of somatic copy number alterations (SCNAs) is defined as four loci according to the status of TP53 mutation and histological type (Additional file 3: Figure S1). c Distribution of molecular subtypes in 207 Japanese (upper) and TCGA (lower) GCs. d Landscape of frequently observed actionable gene alterations (5% or more) in 207 Japanese GCs classified by TCGA subtype. Alteration color indicates the class of gene alterations

Fig. 2

Mutation rates in Japanese and The Cancer Genome Atlas gastric cancers. a Mutation rate from Japanese GCs was determined by the number of non-synonymous SNVs in the 435-gene panel. Red, MSI-H; gray, MSS; HM, hypermutation. b WES data from TCGA GCs was down-sampled to the content of the 435-gene panel. Red, MSI-H; gray, MSS; HM, hypermutation. c Correlation between mutation rates determined using the 435-gene panel and WES data of TCGA GCs

Fig. 3

Clustering based on 69 actionable gene alterations. According to co-alteration patterns of the 69 actionable gene subset, 175 non-hypermutated tumors are divided into six clusters, and the gene alteration spectrum and TCGA molecular subtype of each tumor is demonstrated as color tiles. Alteration in color indicates the class of gene alteration. The 32 hypermutated tumors are shown on the right (brown). The specific gene alteration and proposal for targeted therapy are presented under each cluster. Fourteen genes which have no alterations in 207 tumors are not included in this figure

Fig. 4

Details of ERBB2 alterations and a case of remarkable response to anti-HER2 therapy in the ERBB2 cluster. a Mutations of ERBB2 identified in 207 Japanese GCs are aligned to the protein domain. Patient samples are further plotted by mutation status. b Abdominal enhanced CT before trastuzumab therapy (upper panel) demonstrated extensive metastases in abdominal (arrow head) and paraaortic (arrow) lymph nodes. All the metastatic lymph nodes were remarkably reduced in size after trastuzumab therapy (lower panel). c Abdominal lymph nodes were reduced in size, and paraaortic lymphadenectomy was performed (upper panel). Histological examination revealed no viable tumor cells in the abdominal and paraaortic lymph nodes (lower panel, hematoxylin and eosin, original magnification × 40). d Genomic sequencing in biopsy specimen obtained before treatment (left panel, hematoxylin and eosin; right panel, anti-HER2 antibody) identified ERBB2 and ZNF217 amplification and TP53 mutation. However, concomitant alterations in the ERBB2 downstream pathway were not observed in the tumor of this patient