Fax: (011) 82-2-765-5600
Mucinous gastric carcinomas
Clinicopathologic and molecular analyses
Article first published online: 28 MAY 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 15, pages 3581–3590, 1 August 2009
How to Cite
Choi, J. S., Kim, M. A., Lee, H. E., Lee, H. S. and Kim, W. H. (2009), Mucinous gastric carcinomas. Cancer, 115: 3581–3590. doi: 10.1002/cncr.24422
- Issue published online: 20 JUL 2009
- Article first published online: 28 MAY 2009
- Manuscript Accepted: 16 JAN 2009
- Manuscript Revised: 8 JAN 2009
- Manuscript Received: 7 NOV 2008
- Ministry of Education, Science, and Technology of Korea. Grant Number: FG08-11-03/21C
- stomach neoplasms;
- mucinous adenocarcinoma;
- microsatellite instability;
- erbB-2 gene;
- epidermal growth factor receptor;
- tissue array analysis;
- survival analysis
Mucinous gastric carcinoma (MGC) is characterized by substantial mucous lakes within tumors and comprises 3% of gastric carcinomas at the authors' institute.
The authors analyzed the clinicopathologic characteristics, mucin gene expression profiles, microsatellite instability (MSI), and status of the human epidermal growth factor receptor 2 (HER-2) and epidermal growth factor receptor (EGFR) genes in 133 MGCs and compared them with the same variables in nonmucinous gastric carcinomas (NMGCs). In addition, the prognostic implications of clinicopathologic parameters were evaluated.
Patients who had MGC had deeper invasion (P = .003), more frequent lymph node metastasis (P < .001), more advanced pathologic stage (P < .001), more frequent lymphatic invasion (P < .001), and lower disease-specific survival rates (P < .0001) than patients who had NMGC. However, a mucinous histology per se was not identified as an independent prognostic factor. Negative mucin 1, cell surface associated (MUC1) status (P < .001); positive mucin 2, oligomeric mucus/gel-forming (MUC2) status (P < .001); negative mucin 5AC, oligomeric mucus/gel-forming (MUC5AC) status (P = .036); and negative mucin 6, oligomeric mucus/gel-forming (MUC6) status (P < .001) were more frequent in MGCs. The frequency of MSI in MGC was not significantly different from that in NMGC. MGCs had a significantly lower incidence of HER-2 protein overexpression (P = .046), HER-2 gene amplification (P = .009), and EGFR protein overexpression (P = .017) than NMGCs; and multivariate analysis identified EGFR overexpression as a factor associated with a poor prognosis (P = .047). Patients with MGC who had a predominance of signet ring cells in mucin pools had poorer disease-specific survival than patients who had MGC with predominant tubular differentiation (P = .017).
The clinicopathologic and molecular characteristics of MGCs differed from those of NMGCs. Furthermore, the results indicated that EGFR overexpression and histologic subtyping by predominant tumor cell type in mucin pools may be helpful for predicting clinical outcome in patients with MGC. Cancer 2009. © 2009 American Cancer Society.
Mucinous gastric carcinoma (MGC) constitutes only 2% to 6% of all stomach cancers.1 MGC is defined by the World Health Organization (WHO) as a gastric adenocarcinoma with a substantial amount of extracellular mucin (≥50% of tumor volume) within tumors.2 The clinicopathologic characteristics and prognostic significances as well as prognostic markers of MGCs have not been well determined. Some studies have suggested that MGC is associated with a poorer prognosis than nonmucinous gastric carcinoma (NMGC).1, 3 Conversely, others have insisted that the prognoses of MGC and NMGC are similar.4
Mucins, the main components of MGCs, are high-molecular-weight glycoproteins that consist of a mucin core protein and O-linked oligosaccharides. To our knowledge, to date, 21 mucin genes (designated MUC) have been identified in humans.5 In the normal stomach, mucin 1, cell surface associated (MUC1); mucin 5AC, oligomeric mucus/gel-forming (MUC5AC); and mucin 6, oligomeric mucus/gel-forming (MUC6) demonstrate cell type-specific expression. However, mucin 2, oligomeric mucus/gel-forming (MUC2) is not expressed in the normal stomach but is expressed in goblet cells in intestinal metaplasia. It is known that mucin expression is altered during the carcinogenesis of gastric carcinoma, suggesting that the mucin genes play important roles during cell differentiation and carcinogenesis.6
It is widely accepted that the developments and progression of various human cancers, including gastrointestinal cancers, involve 2 processes: microsatellite instability (MSI) and chromosomal instability. Some recent studies have reported that MSI was observed more frequently in mucinous than nonmucinous colon cancer, whereas the latter more frequently had chromosomal instability.7 However, to our knowledge, little is known regarding MSI status in MGC, and we considered that this warranted an evaluation of MSI in MGC.
Of the innumerable oncogenes that participate in chromosomal instability, human epidermal growth factor receptor 2 (HER-2) and epidermal growth factor receptor (EGFR), also known as HER-1, are of practical interest, because monoclonal antibodies against them entered clinical trials for stomach cancer. It has been demonstrated that HER-2 and EGFR are indicators of a poor prognosis in patients with stomach cancer,8 but their prognostic and clinical implications have not been investigated comprehensively in MGC to date.
In the current study, we analyzed clinicopathologic and molecular characteristics of MGCs by evaluating the expression of MUC1, MUC2, MUC5AC, and MUC6; MSI; human mutL homolog 1 (hMLH1) and human mutS homolog 2 (hMSH2); and gene amplification and protein expression of HER-2 and EGFR. In addition, we evaluated the relations between these biomolecules and prognosis as well as histogenesis in MGC.
MATERIALS AND METHODS
The files of 133 and 310 consecutive patients with MGCs and NMGCs who underwent surgical resection between 1989 and 2000 or between January 1995 and June 1995, respectively, were retrieved from the Department of Pathology, Seoul National University College of Medicine and were reviewed. This study was approved by the Institutional Review Board of Seoul National University Hospital (C0511-519-163).
The age, sex, clinical outcomes, and tumor size, location, and pathologic stage were obtained from medical charts, from pathologic records, and from National Statistical Office death statistics. Hematoxylin and eosin glass slides were reviewed to determine histologic types (according to the Lauren and the WHO classifications), differentiation, and the presence of lymphatic and venous invasion. Patients who were lost during follow-up and those who died to any cause other than gastric cancer were censored during survival analysis.
Microsatellite Instability Analysis
DNA was obtained from formalin-fixed, paraffin-embedded surgical sections. To reduce the possibility that the genetic abnormalities of tumor cells would be masked by normal cells, tumor areas were selected and microdissected under a light microscope. DNA was extracted from harvested tumor cells by standard proteinase-K digestion and phenol/chloroform extraction.
The presence of MSI in tumor tissues was determined using the BAT26 and BAT25 mononucleotide markers, as described previously.9 In brief, polymerase chain reaction (PCR) was carried out using 10-μL reaction volumes with fluorescent dye, 5′-end labeled primers, and MSI status was determined using GeneScan software in an ABI PRIMS 3100 Genetic Analyzer (Perkin Elmer Applied Biosystems, Foster City, Calif). Tumors that had abnormal band patterns were defined as exhibiting MSI.
Core tissue biopsy specimens (2 mm in greatest dimension) were obtained from individual paraffin-embedded tumors (donor blocks) and arranged in new recipient blocks (tissue microarray blocks) using a trephine apparatus (Superbiochips Laboratories, Seoul, Korea). Three separate core samples per tumor were obtained to counter the effects of tumor heterogeneity.
Sections (4 μm) were cut from each tissue microarray block, deparaffinized, and dehydrated. Immunohistochemical staining for MUC1 (mouse monoclonal antibody NCL-MUC-1), MUC2 (mouse monoclonal antibody NCL-MUC-2), MUC5AC (mouse monoclonal antibody NCL-MUC-5AC), and MUC6 (mouse monoclonal antibody NCL-MUC-6; all from Novocastra, Newcastle, United Kingdom) were performed using a streptavidin-peroxidase procedure (labeled streptavidin-biotin) after microwave antigen retrieval. Immunohistochemical staining results were considered positive when >10% of tumor cells were stained.
Immunohistochemical staining for hMLH1 and hMSH2 was performed as described previously.10 The antibody to hMSH2 (clone FE11; 0.5 μg/mL; Oncogenes Science, Cambridge, Mass) is a mouse monoclonal antibody that was generated with the carboxy-terminal fragment of hMSH2 protein, whereas hMLH1 antibody (clone G168-728; 1 μg/mL; Pharmingen, San Diego, Calif) is a mouse monoclonal antibody that was generated with the full-length hMLH1 protein. Normal mouse sera were used as negative controls. Infiltrating lymphocytes and normal epithelium adjacent to tumors served as internal positive controls.
Immunohistochemical staining for HER-2 (DAKO, Glostrup, Denmark) was performed using an avidin-biotin complex method (labeled biotin) after microwave antigen retrieval. EGFR PharmDx kits were used to detect EGFR expression (DAKO). Immunohistochemical expression of HER-2 was scored using DAKO-HercepTest kits as follows: score 0, no membrane staining at all or membrane staining in <10% of tumor cells; score 1+, faint/barely perceptible partial membrane staining in >10% of tumor cells; score 2+, weak to moderate staining of entire membranes in >10% of tumor cells; and score 3+, strong staining of entire membranes in >10% of tumor cells. Scores of 0 and 1+ were considered negative for HER-2 overexpression, and scores of 2+ and 3+ were considered positive. EGFR immunopositivity was scored by using the instructions supplied with EGFR PharmDx kits, and scores of 2+ and 3+ indicated overexpression.11
Fluorescence In Situ Hybridization
Dual-color Vysis kits (Vysis, Downers Grove, Ill) were used for fluorescence in situ hybridization (FISH). Briefly, 2-μm sections of deparaffinized and dehydrated tissue microarray slides were incubated in 20% sodium bisulfate/2 × standard saline citrate at 4°C for 20 minutes. After washing in 2 × standard saline citrate, the slides were treated with proteinase K at 37°C for 25 minutes. Denaturation, hybridization, and posthybridization washing were performed according to the manufacturer's instructions. Slides were then counterstained with 4′,6-diamidine-2′-phenylindole dihydrochloride in antifade solution and examined under a fluorescence microscope (Olympus, Tokyo, Japan) equipped with a Triple Bandpass Filter Set (Vysis). After counting at least 20 tumor cell nuclei per case, gene amplification was defined according to the presence of tight gene clusters or when the FISH ratio (either the red signal of HER-2/green signal of the centromere of chromosome 17 or the red signal of EGFR/green signal of the centromere of chromosome 7) was ≥2.0.
The chi-square test or the Fisher exact test (2-sided) was used to compare discontinuous variables, whereas the Student t test was used to compare continuous variables. Differences were considered statistically significant when P values were <.05. Disease-specific survival curves were drawn using the Kaplan-Meier product-limit method, and the significance of differences between survival curves were determined using the log-rank test. Multivariate survival analysis was performed using a Cox proportional hazards model. All statistical analyses were conducted using SPSS statistical software (version 11.0) for Windows (SPSS, Chicago, Ill).
Comparisons of Clinicopathologic Features of MGC and NMGC
MGCs (133 cases) constituted 3% of resected gastric adenocarcinomas (3924 cases) resected at our institute between 1989 and 2000; and, as indicated in Table 1, their clinicopathologic characteristics differed from those of NMGCs (Table 1). In particular, MGC tumors were larger, more frequently had an antral location, invaded deeper, more frequently metastasized to lymph nodes, had a more advanced pathologic stage, more frequently were of the intestinal type (by the Lauren classification), and more frequently invaded the lymphatic vessels.
|Clinicopathologic Feature||No. of Patients (%)||P|
|Mean age, y||56||54||NS|
|Men||103 (77.4)||210 (67.7)|
|Women||30 (22.6)||100 (32.3)|
|Mean size, cm||6.99||5.10||<.001|
|Antrum||92 (69.2)||158 (51)|
|Body and fundus||29 (21.8)||108 (34.8)|
|Cardia||10 (7.5)||42 (13.5)|
|Whole||2 (1.5)||2 (0.6)|
|T1||7 (5.3)||105 (33.9)|
|T2||61 (45.9)||142 (45.8)|
|T3||62 (46.6)||59 (19)|
|T4||3 (2.3)||4 (1.3)|
|Lymph node classification||.003|
|N0||28 (21.1)||121 (39)|
|N1||58 (43.6)||97 (31.3)|
|N2||31 (23.3)||56 (18.1)|
|N3||16 (12)||36 (11.6)|
|I||23 (17.3)||140 (45.2)|
|II||35 (26.3)||56 (18.1)|
|III||49 (36.8)||66 (21.3)|
|IV||26 (19.5)||48 (15.5)|
|Intestinal||89 (66.9)||122 (39.4)|
|Diffuse||44 (33.1)||183 (59)|
|Mixed||0 (0)||5 (1.6)|
|Absent||63 (47.4)||224 (72.3)|
|Present||70 (52.6)||86 (27.7)|
Human Mucin Gene Expression and Mucin Phenotypes in MGC and NMGC
Mucin gene expression profiles differed in MGCs and NMGCs (Table 2), and MUC2 was related distinctly to MGC and was expressed in 95.5% of MGCs, whereas it was expressed in only 33.4% of NMGCs (P < .001). MGCs were characterized by MUC1 negativity (P < .001), MUC2 positivity (P < .001), MUC5AC negativity (P = .036), and MUC6 negativity (P < .001) (Fig. 1).
|Expression||No. of Patients (%)||P|
|Negative||119 (89.5)||205 (68.6)|
|Positive||14 (10.5)||94 (31.4)|
|Negative||6 (4.5)||193 (66.6)|
|Positive||127 (95.5)||97 (33.4)|
|Negative||85 (63.9)||161 (53)|
|Positive||48 (36.1)||143 (47)|
|Negative||130 (97.7)||245 (80.9)|
|Positive||3 (2.3)||58 (19.1)|
|Gastric phenotype||3 (2.3)||91 (33.6)|
|Mixed phenotype||46 (34.6)||46 (17)|
|Intestinal phenotype||81 (60.9)||62 (22.9)|
|Unclassified phenotype||3 (2.3)||72 (26.6)|
On the basis of mucin gene expression profiles,12 MGCs and NMGCs were categorized into 4 mucin phenotypes: gastric, mixed, intestinal, and unclassified (Table 2). In MGCs, the intestinal phenotype was the most common mucin phenotype (60.9%), and the gastric phenotype was rare (2.3%). In contrast, the most common mucin phenotype in NMGCs was the gastric phenotype. In MGCs, patients with the gastric or mixed phenotype had shorter median survivals (37 months) than patients with the intestinal phenotype (57 months), although the difference was not significant (log-rank test).
MSI and Mismatch Repair Genes in MGC and NMGC
MSI was observed in 9.4% of MGCs (12 of 128 tumors) and in 8.4% of NMGCs (25 of 299 tumors), which was not a significant difference (Table 3). Expressional losses of hMLH1 or hMSH2 were observed in 8.2% and 1.6% of MGCs, respectively, and these differences also were not significantly different from the losses in NMGCs (Table 3). MGC with MSI was associated with older age (aged <65 years vs ≥65 years; P = .014), less lymph node metastasis (P = .015), and lower pathologic stage (P = .010) than MGC without MSI (Table 4). However, no significant difference was observed between MGCs with and without MSI in terms of sex, invasion depth, differentiation, location, tumor size, lymphatic or vascular invasion, Lauren classification, or disease-specific survival rates.
|Variable||No. of Patients (%)||P|
|Positive||12 (9.4)||25 (8.4)|
|Negative||116 (90.6)||274 (91.6)|
|Loss||10 (8.2)||27 (10.6)|
|Expression||112 (91.8)||227 (89.4)|
|Loss||2 (1.6)||4 (1.4)|
|Expression||120 (98.4)||291 (98.6)|
|Clinicopathologic Feature||No. of Patients (%)||P|
|MGC With MSI||MGC Without MSI|
|Mean age, y||64||55||.014|
|Men||9 (75)||90 (77.6)|
|Women||3 (25)||26 (22.4)|
|Mean size, cm||5.55||7.10||NS|
|T1||1 (8.3)||6 (5.2)|
|T2||9 (75)||50 (43.1)|
|T3||2 (16.7)||57 (49.1)|
|T4||0 (0)||3 (2.6)|
|Lymph node classification||.015|
|N0||6 (50)||20 (17.2)|
|N1||6 (50)||51 (44)|
|N2||0 (0)||31 (26.7)|
|N3||0 (0)||14 (12.1)|
|I||6 (50)||15 (12.9)|
|II||3 (25)||32 (27.6)|
|III||2 (16.7)||46 (39.7)|
|IV||1 (8.3)||23 (19.8)|
|Intestinal||8 (66.7)||77 (66.4)|
|Diffuse||4 (33.3)||39 (33.6)|
|Mixed||0 (0)||0 (0)|
|Absent||8 (66.7)||50 (43.1)|
|Present||4 (33.3)||66 (56.9)|
HER-2 and EGFR Protein Overexpression and Gene Amplification in MGC Versus NMGC
HER-2 protein expression and gene amplification were observed in 1.5% and 1.5% of MGCs, respectively (Table 5) (Fig. 2), rates that were significantly lower than those observed in NMGCs (6.2% and 7.9%, respectively; P = .046 and P = .009, respectively). EGFR protein overexpression was observed in 10.7% of MGCs, which also was significantly lower than that observed in NMGCs (21.8%; P = .006) (Table 5) (Fig. 2). However, the frequency of EGFR gene amplification did not differ significantly between MGCs and NMGCs. Both HER-2 protein overexpression and EGFR protein overexpression, as determined by immunohistochemistry, were correlated with gene amplification, as determined by FISH in all gastric carcinomas that were examined (data not shown).
|Variable||No. of Patients (%)||P|
|Negative||130 (98.5)||273 (93.8)|
|Positive||2 (1.5)||18 (6.2)|
|No amplification||130 (98.5)||220 (92.1)|
|Amplification||2 (1.5)||19 (7.9)|
|Negative||117 (89.3)||226 (78.2)|
|Positive||14 (10.7)||63 (21.8)|
|No amplification||131 (99.2)||238 (97.9)|
|Amplification||1 (0.8)||5 (2.1)|
Histologic Patterns in MGCs
We subdivided MGCs into 2 histologic subtypes based on the predominant tumor cell type in mucin pools: 1) MGC with predominant signet ring cells (MGC-SRC), in which ≥50% of tumor cells were signet ring cells; and 2) MGC with predominant tubular differentiation (MGC-T), in which ≥50% tumor cells in mucin pools had tubular differentiation (Fig. 3). The mean tumor size of MGCs-SRCs was greater than that of MGCs-Ts (9.42 cm vs 6.68 cm, respectively) (Table 6), and MGC-SRCs metastasized to distant sites more frequently than MGC-Ts (P = .018).
|Clinicopathologic Feature||No. of Patients (%)||P|
|Mean age, y||51||57||NS|
|Men||12 (100)||90 (75)|
|Women||0 (0)||30 (25)|
|Mean size, cm||9.42||6.68||.006|
|T1||0 (0)||7 (5.8)|
|T2||5 (41.7)||56 (46.7)|
|T3||6 (50)||56 (46.7)|
|T4||1 (8.3)||1 (0.8)|
|Lymph node classification||NS|
|N0||2 (16.7)||26 (21.7)|
|N1||4 (33.3)||53 (44.2)|
|N2||5 (41.7)||26 (21.7)|
|N3||1 (8.3)||15 (12.5)|
|M0||8 (66.7)||111 (92.5)|
|M1||4 (33.3)||9 (7.5)|
|I||1 (8.3)||22 (18.3)|
|II||1 (8.3)||34 (28.3)|
|III||6 (50)||43 (35.8)|
|IV||4 (33.3)||21 (17.5)|
|Absent||6 (50)||56 (46.7)|
|Present||6 (50)||64 (53.3)|
Disease-specific Survival for MGC and NMGC
Disease-specific survival rates differed significantly between MGCs and NMGCs (P < .001) (Fig. 4). The 5-year survival rates were 35% and 66.7% in patients with MGC and NMGC, respectively. However, a mucinous histology per se was not identified as an independent prognostic factor in multivariate analysis using a Cox proportional hazards model (Table 7).
|Prognostic Factor||HR||95% CI||P*|
|Pathologic stage: III/IV vs I/II||6.031||4.021-9.045||<.001|
|EGFR overexpression: Presence vs absence||1.644||1.119-2.414||.011|
|Mucinous histology: MGC vs NMGC||1.361||0.915-2.025||NS|
|MSI: Presence vs absence||0.894||0.328-2.437||NS|
Among the patients with MGC, those with deeper tumoral invasion (P = .0004), lymph node metastasis (P = .0002), more advanced pathologic stage (P < .0001), and MGC-SRC (P = .0169) had significantly poorer survival (Fig. 4). MGC-SRC was associated with a shorter median survival than MGC-T (16 months vs 44 months). However, sex, Lauren classification, lymphatic invasion, mucin type, and MSI did not influence the survival of patients with MGC. Patients who had MGC that overexpressed EGFR protein had a tendency to have poorer survival than patients who had MGC that did not over express EGFR protein, although the difference had borderline significance (P = .0560) (Fig. 4). However, no difference in survival was observed between patients who had MGC with and without EGFR gene amplification.
Multivariate analysis in patients with MGC revealed that pathologic stage and EGFR overexpression were independent prognostic factors (Table 8). However, histologic subtypes of MGC-SRC and MGC-T revealed a difference that had borderline significance (P = .091) in multivariate analysis.
|Prognostic Factor||HR||95% CI||P*|
|Pathologic stage: III/IV vs I/II||3.730||1.662-8.372||.001|
|EGFR overexpression: Presence vs absence||2.334||1.010-5.394||.047|
|Histologic subtype: MGC-SRC vs MGC-T||1.973||0.898-4.335||.091|
The clinicopathologic characteristics and prognostic significance of MGCs have not been studied with large series patients. In the current study, MGCs presented at a more advanced stage, more frequently had lymphatic invasion, and were associated with poorer disease-specific survival than NMGCs. However, a mucinous histology per se was not identified as an independent prognostic factor in multivariate analysis, which means that the poorer prognosis of MGC is not related to a mucinous histology but, rather, to an advanced stage. The survival of patients with MGCs versus NMGCs of the same pathologic stage at presentation also did not differ significantly (data not shown). These results concur with recent literature on MGCs.1, 13
It is unclear why MGCs aggressively invade the gastric wall and frequently present at an advanced tumor stage. It is possible that extracellular mucin acts as an infiltrating medium into surrounding stroma and assists the penetration of deeper layers by tumor cells. In colonic mucinous carcinoma, which demonstrates a poorer prognosis than nonmucinous carcinoma in most series, Perez et al proposed that extracellular mucin may lead to tissue inbibition and facilitate cancer cell dispersion.14 Conversely, Papadopoulos et al proposed that mucin interferes with inflammatory response and the immunologic recognition of tumor cells.15
In the current study, MGC was characterized by a larger tumor size than NMGC. The majority of studies on this topic have reported the same finding,3, 13, 16 although some have reported no difference in size between the 2.17 In terms of tumor locations, results have varied. It has been reported that MGCs were located more frequently in the lower third3 (which we also observed in the current study), or in the upper third,16 or that they had no locational preference.1
In the current study, the Lauren classification characterized MGC more frequently as intestinal type than NMGC, which may correspond to the frequency of MUC2 expression. Of the mucin types, MUC2 was strongly related to MGC. Most MGCs (95.5%) revealed MUC2 expression, whereas only 33.4% of NMGCs did so (P < .001), suggesting that MUC2 is closely related to a mucinous histology and that it may play a role in the histogenesis of MGC. MUC2 is expressed in normal colonic and small intestinal mucosa, but is not expressed in normal gastric mucosa. However, when intestinal metaplasia occurs in the stomach, MUC2 is expressed in goblet cells. Our study demonstrates a close relation between a mucinous histology and MUC2 expression in gastric carcinoma. Few reports have addressed mucin gene expression in MGC, and the reported MUC2 expression frequencies in MGC range from 0% (0 of 4 tumors)18 to 100% (11 of 11 tumors),19 but the numbers of the examined cases were small. To date, the current study is the largest to investigate human mucin gene expression in MGC versus NMGC. In our study, MGC was characterized by MUC1 negativity, MUC2 positivity, MUC5AC negativity, and MUC6 negativity compared with NMGC. MGCs were categorized as intestinal mucin phenotype in 60.9%, mixed phenotype in 34.6%, and gastric phenotype in 2.3%; whereas NMGCs were categorized as intestinal mucin phenotype in 22.9%, mixed phenotype in 17%, and gastric phenotype in 33.6%. In conventional gastric carcinoma, it has been established that the gastric mucin phenotype has pronounced tendencies toward deeper invasion and a poorer prognosis than the intestinal phenotype.12 In the current study, patients with MGC who had the gastric or mixed phenotype had a shorter median survival (37 months) than patients who had the intestinal phenotype (57 months), although the survival curves were not significantly different (data not shown).
MGCs and NMGCs had similar rates of MSI (9.4% vs 8.4%). The number of previous studies conducted on MSI in MGC is limited. In most studies, the frequency of MSI in MGCs and NMGCs was similar, which we also observed in the current study. Those studies reported MSI frequencies in MGCs of 0% (0 of 7 tumors),20 5% (1 of 20 tumors),21 and 14% (6 of 43 tumors)22; whereas we reported an MSI frequency in MGCs of 9.4%. Conversely, 1 previous study that included only 10 patients reported an MSI frequency of 60% (6 of 10 MGCs).23 MSI reportedly was more frequent in mucinous colorectal carcinomas than in nonmucinous colorectal carcinomas.24 However, MSI reportedly was less frequent in pancreatic mucinous carcinomas than in ordinary pancreatic duct carcinomas.25 In view of the findings described above and our own results, it appears that carcinogenetic pathways of mucinous carcinomas in the digestive systems differ, although mucinous carcinomas similarly produce abundant extracellular mucins regardless of organ type.
In the current study, MGC with MSI was associated with older age (P = .014), lower prevalence of lymph node metastasis (P = .015), and lower pathologic stage (P = .010) than MGC without MSI, although no survival difference was observed between patients who had MGCs with or without MSI. Conversely, in patients with conventional gastric carcinoma, MSI-positive tumors have been associated with a lower prevalence of lymph node metastasis, lower pathologic stage,21 and better survival.23
Recent developments of therapeutic agents that target oncogenes have attracted much attention. In particular, trastuzumab, an anti-HER-2 monoclonal antibody, and cetuximab, an anti-EGFR monoclonal antibody, have reached the clinical trial stage for gastric cancer treatments. Therefore, the status of both HER-2 and EGFR has become clinically relevant. In the current study, the protein expression levels of HER-2 and EGFR and HER-2 gene amplification were significantly lower in MGCs than in NMGCs, suggesting that the pathogenesis of MGC does not involve HER-2 or EGFR. However, among the patients with MGC, those who had tumors that overexpressed EGFR protein had poorer disease-specific survival than those without overexpression; in addition, EGFR overexpression was identified as an independent prognostic factor in multivariate analysis, suggesting that EGFR overexpression is likely to be 1 of the potential prognostic markers in MGC. In addition, patients who had NMGC with EGFR overexpression also had poorer survival than patients who had NMGCs without EGFR overexpression (P = .0093). When we subdivided MGCs into MGC-SRC and MGC-T according to tumor cell histology in mucin pools, patients who had MGC-SRC had a significantly poorer prognosis than patients who had MGC-T (P = .0169), suggesting that histologic subtyping by predominant tumor cell type in mucin pools may be used to predict clinical outcome in patients with MGC.
In summary, the results of the current study demonstrated that MGC is characterized by deeper invasion, more frequent lymph node metastasis, a more advanced pathologic stage, more frequent lymphatic invasion, and poorer survival than NMGC; however, a mucinous histology per se was not identified as an independent prognostic factor of survival. Furthermore, MUC2 overexpression was identified as a characteristic feature of MGC. The frequency of MSI in MGC and in NMGC did not differ significantly. However, MGCs expressed significantly less HER-2 and EGFR protein and had less frequent HER-2 gene amplification compared with NMGCs. Finally, we suggest that EGFR overexpression in MGCs and histologic subtyping by tumor cell type in mucin pools may be useful for predicting clinical outcome in patients with MGC.
Conflict of Interest Disclosures
Supported by Grant FG08-11-03/21C (Frontier Functional Human Genome Project) from the Ministry of Education, Science, and Technology of Korea.
- 2HamiltonSR, AaltonenLA, eds. World Health Organization Classification of Tumours. Pathology and Genetics. Tumours of the Digestive System. Lyon, France: IARC Press; 2001.
- 5Role of MUC genes and mucins in pancreatic neoplasia. Am J Gastroenterol. 2006; 101: 2330-2332..Direct Link:
- 9A National Cancer Institute Workshop on Microsatellite Instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer. Cancer Res. 1998; 58: 5248-5257., , , et al.
- 11A proposal for diagnostically meaningful criteria to classify increased epidermal growth factor receptor and c-erbB-2 gene copy numbers in gastric carcinoma, based on correlation of fluorescence in situ hybridization and immunohistochemical measurements. Virchows Arch. 2004; 445: 255-262., , , et al.