Abstract
Background/Aim: Although our understanding of the molecular mechanisms of gastric cancer (GC) development and progression is steadily deepening, the clinical outcome of GC patients remains inadequate. The identification of molecules associated with GC will help improve prognosis. We aimed to identify the molecules involved in GC progression and metastasis. Materials and Methods: Transcriptome analysis was performed on surgically resected gastric tissue from patients with hepatic metastasis. Fourteen cell lines and 230 pairs of primary GC tissues and their corresponding normal adjacent tissues were included in the mRNA expression analysis. Results: Adhesion molecule with Ig like domain 2 (AMIGO2) was identified as a gene of interest. The levels of AMIGO2 mRNA positively correlated with those encoding FOXC2, NODAL, GEMIN2 and negatively correlated with TFPI2. Patients with high AMIGO2 expression experienced significantly shorter disease-free survival and overall survival. High levels of AMIGO2 were associated with poor prognosis. Conclusion: Patients with GC with high AMIGO2 mRNA levels experienced significantly shorter survival, suggesting that AMIGO2 may serve as a prognostic biomarker for GC.
Gastric cancer (GC) is a significant public health concern worldwide, particularly in East Asia. Despite declines in incidence and mortality, GC remains the third leading cause of cancer-related death (1-4). Unfortunately, patients with GC have a high rate of metastasis specific to the liver and peritoneum, and prognosis is poor in the absence of curative treatment (5-7). Improved detection and personalized therapies are required to overcome these hurdles.
The tumor-node-metastasis (TNM) classification is the most widely-accepted clinical tool for staging and prognostication of GC; however, patients with GC diagnosed with an identical TNM stage occasionally present an entirely different clinical course. Furthermore, although molecular markers such as carcinoembryonic antigen (CEA) and cancer antigen 19-9 (CA19-9) are useful for monitoring response to treatment and detecting recurrence, their diagnostic and prognostic value are limited (8). It is necessary to identify molecules involved in the progression and metastasis of GC and serve as biomarkers that accurately reflect the malignancy.
For this purpose, we herein conducted transcriptome analysis of gastric cancers with an aggressive phenotype that culminates in distant metastasis. This analysis identified the gene encoding adhesion molecule with Ig like domain 2 (AMIGO2) through its differential expression in primary GC tissues of patients with hepatic metastasis. Although evidence indicates that AMIGO2 is involved in the genesis and progression of GC, its association with prognosis is unknown. Therefore, herein we aimed to determine whether AMIGO2 expression will serve as a significant prognostic biomarker for GC and to identify the molecular mechanisms responsible for the contribution of AMIGO2 to GC.
Materials and Methods
Ethics statement. This study complied with the World Medical Association Declaration of Helsinki Ethical Principles for Medical Research Involving Human Subjects and was approved by the Institutional Review Board of Nagoya University, Japan (Research Approval Number 2014-0043). Written informed consent for the use of clinical samples and data was obtained from all patients.
Transcriptome analysis. Surgically resected gastric tissues from four patients with metastatic GC were subjected to transcriptome analysis. Global expression profiling was performed using the HiSeq platform (Illumina, San Diego, CA, USA) to compare the levels of 57,749 mRNAs expressed in primary GC tissues with those of the gastric mucosa in corresponding noncancerous areas (9, 10).
Cell lines. Cell lines established from gastric cancers (AGS, GCIY, IM95, KATOIII, MKN1, MKN7, MKN45, MKN74, N87, NUGC2, NUGC3, NUGC4, OCUM1 and SC-6-JCK) were purchased from the American Type Culture Collection (ATCC; Manassas, VA) or the Japanese Collection of Research Bioresources Cell Bank (JCRB; Osaka, Japan). The cell lines were cultured in RPMI-1640 (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% fetal bovine serum at 37°C in an atmosphere containing 5% CO2.
Quantitative real-time reverse-transcription polymerase chain reaction (qRT-PCR) and PCR array analysis. The levels of AMIGO2 mRNA in clinical samples were evaluated in triplicate using RT-PCR analysis, as described previously (11, 12). AMIGO2 primer sequences were as follows: forward 5’-GTGACAGACAC GGACAGACG-3’ and reverse 5’-CAGCCTCCCACCAGTGAA-3’. The level of the mRNA encoding glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as a standard. To identify genes correlating with AMIGO2 in gastric cancer cell lines, we used the Human EMT RT2 Profiler PCR Array (Qiagen, Hilden, Germany), which includes 84 genes that are associated with the epithelial-to-mesenchymal transition (EMT).
Clinical samples. Between 2001 and 2017, 230 pairs of primary GC tissues in pathological Stage I-III and corresponding normal adjacent tissues were collected from surgical specimens of patients who underwent curative gastrectomy for GC without neoadjuvant treatment at the Department of Gastroenterology, Nagoya University Hospital. Tissue specimens were frozen in liquid nitrogen immediately after collection and stored at –80°C. The specimens were histologically classified using the International Union for the Control of Cancer (UICC) Classification, 7th edition. Patients were pathologically diagnosed with stages I-III GC, and relevant clinicopathological parameters were obtained from medical records. Since 2006, all patients with UICC stages II-III GC have been treated with adjuvant chemotherapy with S-1 (oral fluorinated pyrimidine), except when contraindicated by the patient’s condition (13).
Validation dataset. To validate our experimental data, we queried an integrated microarray dataset comprising the data of 1,065 patients from three major cancer research Centres (Berlin, Bethesda and Melbourne datasets: http://kmplot.com/analysis/) (14).
Statistical analysis. The significance of the associations between AMIGO2 mRNA levels and clinicopathological factors was analysed using the χ2 test, and numerical variables between the two groups were compared using the Mann-Whitney test. The significance of the difference between two variables was calculated using Spearman’s rank correlation coefficient. Survival analysis was conducted using the Kaplan–Meier method. Cox proportional hazards models were used to calculate hazard ratios (HRs) and for multivariate regression analysis. JMP Pro 15 software (SAS Institute Inc., Cary, NC, USA) was used for all statistical analyses. p<0.05 indicates a significant difference.
Results
Identification of AMIGO2 as a putative driver gene of gastric cancer. We analysed the transcriptomes of primary GC tissues of patients with hepatic metastases to identify driver genes involved in a higher-grade malignant phenotype. We identified 26 genes that were differentially expressed at higher levels in GC tissues compared to adjacent normal tissues (Table I). A literature review of these 26 genes did not uncover information indicating the utility of AMIGO2 as a biomarker for gastric cancer. Furthermore, AMIGO2 tends to be highly expressed in hepatic metastasis tissue, one of the specific metastatic sites of GC, as well as in primary tissues, suggesting that AMIGO2 expression may be a candidate molecule for predicting liver metastasis. We therefore selected AMIGO2 for subsequent analyses.
List of candidate genes expressed at increased levels in gastric cancer tissues from patients with hepatic metastasis.
AMIGO2 mRNA levels in gastric cancer cell lines and their correlation with EMT-related genes. The levels of AMIGO2 mRNA in GC cell lines are shown in Figure 1A. Each cell line expressed significantly different levels of AMIGO2 mRNA regardless of cell differentiation (Figure 1A). We analysed PCR arrays to detect associations between AMIGO2 expression with genes that encode proteins associated with the EMT that may contribute to GC tumorigenesis, progression and metastasis. We found that the levels of AMIGO2 mRNA positively correlated with those of mRNAs encoding forkhead box C2 (FOXC2), nodal growth differentiation factor (NODAL), gem nuclear organelle associated protein 2 (GEMIN2) and negatively correlated with the levels of tissue factor pathway inhibitor 2 (TFPI2) (Figure 1B).
Analysis of AMIGO2 expression in 14 gastric cancer cell lines. (A) The levels of AMIGO2 mRNA in gastric cancer cell lines. (B) PCR array analysis.
Clinical significance of AMIGO2 expression in gastric cancer tissues. The median age of the 230 patients was 67 years (range=26-96 years); female-to-male ratio, 65:165. There were 50, 71 and 109 patients with GC stages I, II and III, respectively. AMIGO2 mRNA levels were significantly higher in GC tissues compared to the corresponding noncancerous adjacent tissues. Moreover, AMIGO2 expression levels increased with disease progression (Figure 2). Patients were assigned to high (n=115) or low (n=115) AMIGO2 expression groups according to median AMIGO2 mRNA levels in GC tissues. The clinicopathological characteristics of both groups are listed in Table II. There were significant differences between the two groups in their levels of CA19-9, tumour location, infiltrative growth, lymph node metastasis and UICC stage. Kaplan–Meier curve analysis showed that high levels of AMIGO2 mRNA were significantly associated with shorter disease-free survival (DFS) and overall survival (OS) [DFS: HR=2.20, 95% confidence interval (CI)=1.35-3.61, p=0.0012. OS: HR=2.40, 95% CI=1.35-4.28, p=0.0022] (Figure 3A). Multivariate analysis revealed that high levels of AMIGO2 mRNA were associated with the second-highest HR values (after lymph node metastasis) among candidate risk factors (HR=2.02, 95% CI=1.09-3.74, p=0.0262) (Table III).
Comparison of AMIGO2 mRNA levels in normal adjacent tissues and GC tissues as a function of UICC stage. *p<0.05.
Clinical significance of AMIGO2 mRNA expression. (A) Kaplan–Meier analysis of disease-free survival (DFS) and overall survival (OS) rates of the institutional and external validation cohorts. (B) Frequencies of initial recurrence after curative gastrectomy in patients according to AMIGO2 mRNA levels. *p<0.05.
Association between the expression of AMIGO2 mRNA and clinicopathological parameters of 230 patients with stage I-III gastric cancer.
Prognostic factors for overall survival of patients with stage I-III gastric cancer (n=230).
To further evaluate the clinical influence of AMIGO2 expression on survival in the external cohort, we used prognostic data from an external validation dataset comprising of 444 patients with stages I-III gastric cancer. The results are consistent with those of our institutional cohort (Figure 3A). Peritoneal metastasis was significantly more common in the high AMIGO2-expressing group; however, there was no significant difference in hepatic metastasis (Figure 3B).
Discussion
Herein we identified AMIGO2 as a likely driver of GC that contributes to the establishment of a highly aggressive malignant phenotype. Moreover, analyses of internal and external datasets support the conclusion that high expression of AMIGO2 correlated with poor clinical outcomes, indicating that AMIGO2 expression may serve as a predictor of survival of patients with GC.
AMIGO was initially discovered as a novel sequence induced in neurons by the neurite-promoting protein amphoterin (15, 16). AMIGO2, which resides on human chromosome 12q13, is a member of AMIGO gene family that encodes type 1 transmembrane proteins. AMIGO2 is differentially overexpressed in melanoma tissues, and its loss-of-function significantly compromises the growth of melanoma cells (17). Further, AMIGO2 is differentially expressed in human gastric adenocarcinoma, and the silencing of AMIGO2 expression in a mouse model of subcutaneous tumours leads to chromosomal instability, decreased cell adhesion and migration and inhibition of tumour growth (18).
In our analysis of the association of AMIGO2 expression with clinical characteristics of patients with GC, high levels of AMIGO2 mRNA were significantly related to infiltrative growth and lymph node metastasis. Further, its expression levels in patients with stages II-III gastric cancer were significantly higher than that of noncancerous and stage I cancer tissues. Together, these findings support the conclusion that the aggressive phenotype of GC cells is significantly associated with the expression of AMIGO2. These findings may indirectly explain the basis of AMIGO2 function in GC.
The mechanisms of regulation of gastric carcinogenesis and progression of GC are unknown. We, therefore, conducted PCR array analysis to identify genes co-expressed with AMIGO2 during the EMT. Consequently, we found that the levels of the mRNAs encoding FOXC, NODAL and GEMIN2 positively correlated with those of AMIGO2 mRNA and that the levels of TFPI2 mRNA correlated negatively. It has been noticed that these molecules participate in EMT-inducing signal transduction pathways (19-23). Notably, AMIGO2 and FOXC2 are related to cancer progression through activation of the AKT signalling pathway that is required for processes such as cell proliferation, survival and angiogenesis (16, 24). Further, TFPI2 silencing contributes to tumour invasion via the AKT pathway (23). These observations highlight the possibility that AMIGO2 is involved in AKT signalling pathway and its importance in GC progression.
In this study, AMIGO2 was identified as a gene that is differentially expressed at high levels in cancer tissues of patients with GC with hepatic metastasis. Kanda et al. identified AMIGO2 as a regulatory molecule that increases liver endothelial cell adhesion and liver metastasis, which was revealed through sequential in vivo selection of metastatic liver foci that were formed following intrasplenic injection of GC cells (25). Together, these findings suggest that AMIGO2 may contribute to the malignant potential of GC as well as to the formation of hepatic metastases.
CEA and CA19-9 are widely used as biomarkers for GC (8, 26, 27); however, our multivariable analysis shows that increased levels of these biomarkers did not predict survival. Our present findings indicate that investigating AMIGO2 expression will serve to predict prognosis and recurrence of GC.
In clinical practice, AMIGO2 expression in surgical specimens may indicate that more intensive postoperative adjuvant therapy, such as using docetaxel plus S-1 (28), will benefit patients with pathological stage II GC who express high levels of AMIGO2 mRNA, as well as those with pathological stage III. Although the benefit of neoadjuvant chemotherapy has been investigated in Japan, its effectiveness has not been definitively established (29, 30). In determining eligibility for preoperative chemotherapy, AMIGO2 mRNA levels in preoperative biopsy tissue obtained using endoscopic surveillance may be one of the indicators of patients with expected poor survival outcome.
This study has several limitations. First, this was a single institutional-based study that involved a relatively small number of patients. We, therefore, performed survival analysis using expression data from an external cohort and obtained similar results. Second, the detailed molecular mechanism of AMIGO2-mediated tumour progression is unknown. Further research into the signalling pathways associated with AMIGO2 function is required to establish the oncogenic role of AMIGO2 in gastric cancer.
In conclusion, patients with gastric cancer with high AMIGO2 mRNA levels experienced significantly shorter survival, suggesting that AMIGO2 expression may serve as a prognostic biomarker for GC.
Acknowledgements
The Authors thank Edanz Group (https://en-author-services.edanzgroup.com/ac) for editing a draft of this manuscript.
Footnotes
Authors’ Contributions
SN and MK conceived the study concept and design, analyzed data and wrote the manuscript. MK, DS, MK and YK contributed to data acquisition and interpretation. KO contributed to statistical analysis. CT, IY, NH, MH, SY, GN, MK and YK revised the draft. All Authors have read and approved the final version of the manuscript.
Conflicts of Interest
The Authors have no potential conflicts of interest with regard to the present study.
- Received October 13, 2020.
- Revision received October 27, 2020.
- Accepted October 29, 2020.
- Copyright © 2020 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
