Clinical significance of gastritis cystica profunda and its association with Epstein-Barr virus in gastric cancer

Authors


Abstract

BACKGROUND:

Gastritis cystica profunda (GCP) is a relatively rare disorder characterized by hyperplastic and cystic down growth of gastric glands into the submucosa. In the current study, the authors attempted to clarify the clinical and pathologic features of GCP in patients with gastric cancer.

METHODS:

The records of 10,728 patients with gastric cancer who underwent gastric cancer surgery were reviewed. The clinicopathologic features of patients who had GCP (n = 161) were compared with the features of patients without GCP (n = 10,567). In situ hybridization to determine Epstein-Barr virus (EBV) positivity was performed in cancer tissues from patients with (n = 119) and without (n = 503) GCP.

RESULTS:

GCP was associated significantly with older age, male gender, proximal tumor location, differentiated histology and Lauren intestinal type compared with non-GCP. GCP also was present more frequently in remnant and multiple gastric cancers. Patients who had GCP presented with earlier tumor stages in terms of depth of invasion and lymph node metastasis, and they had less lymphatic and perineural invasion than patients without GCP; however, the presence of GCP was not an independent prognostic factor. The EBV-positive rate was significantly higher in the GCP group (31.1%) than in the non-GCP group (5.8%).

CONCLUSIONS:

Patients with gastric cancer who had GCP had clinicopathologic features that differed from the features observed in patients without GCP. GCP was associated significantly with EBV-positive gastric cancers, and its possible role as a premalignant lesion needs to be clarified. Cancer 2012. © 2012 American Cancer Society.

INTRODUCTION

Gastritis cystica profunda (GCP) is a relatively rare, benign lesion characterized by polypoid hyperplasia and cystic dilatation of the gastric glands extending into the submucosa of the stomach.1 Since its first description by Littler and Gleibermann in 1972, there have been several reports on GCP, but its characteristics and clinical significance are not well recognized. It has been suggested that GCP represents a manifestation of hyperplastic and metaplastic responses to mucosal injury caused by several factors, such as chronic inflammation, ischemia, gastric surgery, and the presence of foreign materials.2, 3 An interruption of the muscularis mucosae may be caused by erosion of the gastric mucosa in chronic gastritis or ischemia4; and the effects of previous surgery, such as disruption of the integrity of muscularis mucosa or the presence of suture material, could cause mucosa to migrate into the submucosa.5 It also has been postulated that gastrojejunostomy with subsequent reflux of bile and small intestinal contents may cause atrophic gastritis and intestinal metaplasia, which are well known precancerous lesions6 and may predispose individuals to the development of GCP. However, its pathogenesis is still unclear, and it remains controversial whether GCP is a precancerous lesion or a paracancerous lesion.

Gastric cancer is the most common cancer in Korea and is the third most common cause of death from cancer in Korea.7 After surgery for gastric cancer, GCP has been identified incidentally in surgical specimens, but it is difficult to understand its clinical significance. Several reports also have mentioned that GCP is associated with high-grade dysplasia or adenocarcinoma,8-10 suggesting its possible role as a premalignant lesion.

The current study is the largest clinicopathologic study of GCP to date, and, for this report, we compared the clinical/pathologic characteristics of patients with and without GCP to clarify its clinical significance and to evaluate its possible role as a premalignant lesion. In addition, we compared Epstein-Barr virus (EBV)-positive cancer tissues from patients with and without GCP to characterize the biologic features of the groups.

MATERIALS AND METHODS

Patients

Prospectively collected data were analyzed in a retrospective study of 10,728 patients with gastric cancer who underwent gastric cancer surgery at the Department of Surgery, Samsung Medical Center between January 2000 and December 2009; the study comprised 161 patients with GCP and 10,567 patients without GCP. The patient characteristics and pathology findings were reviewed from the medical records. The study group included 7112 men (66.3%) and 3616 women (33.7%). The mean (±standard deviation) patient age was 56.3 ± 11.9 years, and the mean follow-up duration was 40.7 ± 29.9 months. Surgery was regarded as curative when all visible tumors were resected completely and the resection margins were free of cancer cells on microscopic examination. Adjuvant chemotherapy usually was recommended except for patients who had T1N0 or T2N0 disease after surgery. Staging was carried out according to the American Joint Committee on Cancer (AJCC) Cancer Staging Manual (seventh edition, 2010). Survival data were obtained from patients' medical records and the Korean cancer registry.

Epstein-Barr Virus-Encoded RNA in Situ Hybridization

Tissue microarray blocks were used for EBV-encoded RNA in situ hybridization. Three-micrometer-thick sections were cut from each block and mounted on Superfrost+ slides (Thermo Scientific, Waltham, Mass). The entire procedure was performed on a fully automatic system (Bond-Max) for in situ hybridization with an EBV-encoded RNA probe (both from Leica, Newcastle, United Kingdom) according to the manufacturer's instructions. Only sections that had a strong signal within almost all tumor cell nuclei were considered positive.

EBV status was reviewed in cancer tissues with GCP that were available for evaluation. Starting in August 2009, the EBV status of gastric cancers has been checked in a routine manner at our center, and we reviewed 503 consecutive cancers without GCP from August 2009 to the end of the study period.

Statistical Analysis

The clinical and pathologic findings were compared statistically between patients with and without GCP using the chi-square test and the Student t test. Patients who were lost to follow-up and those who died from causes other than gastric cancer were censored in the survival analysis. Univariate survival analysis was performed using the Kaplan-Meier method and the log-rank test. Multivariate analysis was carried out using a Cox proportional hazards model. All statistical analyses were performed using the statistical software program PASW Statistics 18 (SPSS Inc, Somers, NY). A P value < .05 was considered significant.

RESULTS

Among 10,728 patients with gastric cancer, 161 patients (1.5%) had GCP. Gastric cancer with GCP was associated significantly with older age, men, proximal tumor location, differentiated type of histology, and Lauren intestinal type tumors versus gastric cancer without GCP. GCP also was present more frequently in remnant and multiple gastric cancers. The incidence of GCP was 9.1% (10 of 110 patients) in those with remnant gastric cancer and 1.4% (151 of 10,618 patients) in those with nonremnant gastric cancer. Patients with GCP presented with earlier tumor stages in terms of depth of invasion and lymph node metastasis and had less lymphatic and perineural involvement than patients without GCP. However, we did not observe any significant difference in mean tumor size, distant metastasis, venous invasion, or surgical curability between the 2 groups (Table 1).

Table 1. Comparison of Clinicopathologic Features in Patients With Gastric Cancers With and Without Gastritis Cystica Profunda
 No. of Patients (%) 
VariableGCP, n = 161Non-GCP, n = 10,567P
  • Abbreviations: GCP, gastritis cystica profunda; SD, standard deviation.

  • a

    Some data were missing because of unclassified histologic type and unreported Lauren type; only available data were included in the analysis.

Age: Mean±SD, y60.8±9.856.3±11.9<.001
Sex   
 Men153 (95)6959 (65.9)<.001
 Women8 (5)3608 (34.1) 
Remnant cancer   
 Remnant10 (6.2)100 (0.9)<.001
 Nonremnant151 (93.8)10467 (99.1) 
Multiplicity   
 Single142 (88.2)10,221 (96.7)<.001
 Multiple19 (11.8)346 (3.3) 
Location of tumor   
 Lower44 (27.3)5492 (52)<.001
 Middle56 (34.8)3651 (34.6) 
 Upper58 (36)1281 (12.1) 
 Whole3 (1.9)143 (1.4) 
Size: Mean±SD, cm4.1±2.54.7±3.1.085
Histologic typea   
 Differentiated95 (60.1)4405 (42.3)<.001
 Undifferentiated63 (39.9)6004 (57.7) 
Lauren typea   
 Intestinal type101 (64.3)4809 (46.7)<.001
 Diffuse46 (29.3)5009 (48.7) 
 Mixed10 (6.4)475 (4.6) 
Depth of invasion   
 T1119 (73.9)5718 (54.1)<.001
 T219 (11.8)1257 (11.9) 
 T318 (11.2)2228 (21.1) 
 T45 (3.1)1364 (12.9) 
Lymph node metastasis   
 N0123 (76.4)6694 (63.3).004
 N114 (8.7)1202 (11.4) 
 N212 (7.5)977 (9.2) 
 N312 (7.5)1694 (16) 
Distant metastasis   
 M0159 (98.8)10,324 (97.7).373
 M12 (1.2)243 (2.3) 
Stage   
 I127 (78.9)6212 (58.8)<.001
 II13 (8.1)1884 (17.8) 
 III19 (11.8)2230 (21.1) 
 IV2 (1.2)241 (2.3) 
Lymphatic invasion   
 Absent134 (83.2)7066 (66.9)<.001
 Present27 (16.8)3501 (33.1) 
Venous invasion   
 Absent152 (94.4)9824 (93).639
 Present9 (5.6)743 (7) 
Perineural invasion   
 Absent148 (91.9)8699 (82.3).001
 Present13 (8.1)1868 (17.7) 
Curability   
 Curative159 (98.8)10,152 (96.1).096
 Noncurative2 (1.2)415 (3.9) 

The overall 5-year survival rates did not differ significantly between patients with GCP (86.3%) and those without GCP (83.4%) (Fig. 1). Univariate analysis revealed that the following parameters were associated significantly with survival: age, tumor multiplicity, tumor location, tumor size, differentiation, Lauren type, depth of invasion, lymph node metastasis, distant metastasis, TNM staging, lymphatic invasion, venous invasion, perineural invasion, and surgical curability (Table 2). Multivariate Cox regression analysis revealed that age, tumor size, depth of invasion, lymph node metastasis, TNM staging, lymphatic invasion, and surgical curability were independent prognostic indicators (Table 3).

Figure 1.

These are survival curves for patients with gastric cancer combined with gastritis cystica profunda (GCP) and those without GCP. There was no difference in overall survival between groups.

Table 2. Univariate Analysis of Prognostic Factors in Patients With Gastric Cancer
Variable5-Year Survival, %P
  1. Abbreviation: GCP, gastritis cystica profunda.

GCP  
 Present86.3.276
 Absent83.4 
Sex  
 Men83.8.068
 Women82.7 
Age, y  
 <6085.4<.001
 ≥6080.7 
Remnant cancer  
 Remnant79.9.159
 Nonremnant83.5 
Multiplicity  
 Single83.3.028
 Multiple88.2 
Tumor location  
 Lower86.4<.001
 Middle82.3 
 Upper75.9 
 Whole45.1 
Size, cm  
 <592.5<.001
 5-1071.8 
 >1041.4 
Histologic type  
 Differentiated89.5<.001
 Undifferentiated78.5 
Lauren type  
 Intestinal89.4<.001
 Diffuse78.7 
 Mixed75.6 
Depth of invasion  
 T197.2<.001
 T289.6 
 T368.9 
 T440.6 
Lymph node metastasis  
 N095.6<.001
 N183.9 
 N270 
 N340.5 
Distant metastasis  
 M085<.001
 M116.5 
Stage  
 I97.4<.001
 II82.9 
 III50.9 
 IV16.3 
Lymphatic invasion  
 Absent89.6<.001
 Present68.6 
Venous invasion  
 Absent85.1<.001
 Present57.5 
Perineural invasion  
 Absent87.3<.001
 Present60.3 
Surgical curability  
 Curative86.1<.001
 Noncurative15.6 
Table 3. Multivariate Analysis of Prognostic Factors
VariableHR (95% CI)Pa
  • Abbreviations: CI, confidence interval; HR, hazard ratio.

  • a

    Cox proportional hazards model.

Age: ≥60 y/<60 y1.420 (1.262-1.597)<.001
Tumor size  
 5-10 cm/<5 cm1.072 (0.926-1.241).354
 ≥10 cm/<5 cm1.440 (1.189-1.745)<.001
Depth of invasion  
 T2/T11.405 (0.947-2.084).091
 T3/T12.426 (1.567-3.756)<.001
 T4/T13.573 (2.257-5.658)<.001
Lymph node metastasis  
 N1/N01.570 (1.199-2.056).001
 N2/N02.058 (1.440-2.940)<.001
 N3/N03.972 (2.784-5.667)<.001
TNM stage  
 II/I2.602 (1.640-4.129)<.001
 III/I3.128 (1.635-5.983).001
 IV/I3.853 (1.947-7.625)<.001
Lymphatic invasion: Present/absent1.164 (1.025-1.323).020
Surgical curability: Noncurative/curative2.484 (2.016-3.060)<.001

In situ hybridization for EBV revealed that the EBV-positive rate was significantly greater in cancer tissues with GCP (37 of 119 patients; 31.1%) than in those without GCP (29 of 503 patients; 5.8%; P < .001) (Fig. 2). In a comparison of EBV-positive cancers with and without GCP, the tumors with GCP were associated significantly with men, earlier tumor stage in terms of depth of invasion, and less lymphatic involvement (Table 4).

Figure 2.

These are photomicrographs of gastric carcinoma associated with Epstein-Barr virus (EBV) and gastritis cystica profunda (GCP) (original magnification, ×40). (A) Gastric carcinoma is combined with GCP, which is characterized by the cystic down growth of gastric glands into the submucosa. (B) EBV-encoded RNA in situ hybridization demonstrates a strong signal within almost all tumor cell nuclei.

Table 4. Comparison of Clinicopathologic Features of Gastric Cancers With and Without Gastritis Cystica Profunda Among Patients With Epstein-Barr Virus-Positive Cancers
 No. of Patients (%) 
VariableGCP, n = 37Non-GCP, n = 29P
  • Abbreviations: GCP, gastritis cystica profunda; SD, standard deviation.

  • a

    Some data are missing because of unclassified histologic type and unreported Lauren type; only available data were included in the analysis.

Age: Mean±SD, y58.4±9.057.4±9.6.662
Sex   
 Men37 (100)25 (86.2).033
 Women0 (0)4 (13.8) 
Remnant cancer   
 Remnant4 (10.8)0 (0).125
 Nonremnant33 (89.2)29 (100) 
Multiplicity   
 Single31 (83.8)28 (96.6).124
 Multiple6 (16.2)1 (3.4) 
Location of tumor   
 Lower2 (5.4)1 (3.4).421
 Middle12 (32.4)14 (48.3) 
 Upper23 (62.2)14 (48.3) 
Tumor size: Mean±SD, cm3.5±2.04.0±2.6.480
Histologic typea   
 Differentiated15 (40.5)8 (42.1).910
 Undifferentiated22 (59.5)11 (57.9) 
Lauren typea   
 Intestinal22 (61.1)11 (52.4).777
 Diffuse9 (25)7 (33.3) 
 Mixed5 (13.9)3 (14.3) 
Depth of invasion   
 T130 (81.1)14 (48.3).047
 T24 (10.8)8 (27.6) 
 T32 (5.4)4 (13.8) 
 T41 (2.7)3 (10.3) 
Lymph node metastasis   
 N033 (89.2)21 (72.4).306
 N10 (0)1 (3.4) 
 N22 (5.4)3 (10.3) 
 N32 (5.4)4 (13.8) 
Distant metastasis   
 M036 (97.3)29 (100)1.000
 M11 (2.7)0 (0) 
Stage   
 I32 (86.5)20 (69).071
 II0 (0)4 (13.8) 
 III4 (10.8)5 (17.2) 
 IV1 (2.7)0 (0) 
Lymphatic invasion   
 Absent34 (91.9)20 (69).017
 Present3 (8.1)9 (31) 
Venous invasion   
 Absent35 (94.6)28 (96.6)1.000
 Present2 (5.4)1 (3.4) 
Perineural invasion   
 Absent33 (89.2)24 (82.8).450
 Present4 (10.8)5 (17.2) 

DISCUSSION

Our study demonstrated that the incidence of GCP was 1.5% in patients with gastric cancer; however, there may have been a potential underestimation of the incidence of GCP in other inflamed tissues apart from the main cancer. In the current study, only GCP lesions that coexisted with cancers were investigated. Patients with GCP had clinicopathologic features that were distinct from those observed in patients without GCP, in that they were associated significantly with older age, men, early stage tumors, and differentiated type of histology, which also are characteristics of multiple gastric cancers.11 In our study, the incidence of GCP in patients with multiple cancers was significantly greater than in those with single cancers, and there also have been several reports of GCP associated with multiple gastric cancers.10, 12 It appears that GCP frequently develops in the background of chronic inflammation, which may cause multicentric carcinogenesis as well as an aberration of the gastric glands.

GCP is a well recognized entity that occurs several years after previous gastric surgery,1, 5, 6, 13, 14 but it also has been reported that GCP can develop in patients without previous surgery.4, 8, 15, 16 The current study also revealed that, although GCP was identified more frequently in the setting of previous gastric surgery, it also could develop in patients who did not undergo previous surgery.

Although all of the patients in our study who had GCP had it in continuity with adenocarcinoma, it is still not clear whether GCP is a precancerous lesion or just a combined phenomenon in the background of carcinogenic circumstances, such as chronic atrophic gastritis. Several reports have suggested that GCP may be a precancerous lesion,3, 17 and it is known that the foveolar gland in GCP is hyperplastic and highly proliferative.18 Some investigators have reported several cases of GCP accompanied by carcinoma, and their immunohistochemical staining studies revealed enhanced expressions of Ki-67, p53, and p21 in the mucosa of GCP.9, 19 Those findings suggest that the underlying mechanisms of epithelial cell proliferation and p53-dependent p21 expression in DNA-damaged cells in GCP may be associated with the carcinogenic process. Other investigators have suggested that the disruption of potassium voltage-gated channel, lsk-related family, member 2 (KCNE2), 1 of the important subunits of apical potassium channels in parietal cells, is associated with the development of GCP and gastric neoplasia.20 Those authors demonstrated that 1-year-old Kcne2−/− mice exhibit a severe gastric preneoplastic phenotype, including GCP, polypoid adenomas, and neoplastic growth.

Our study revealed that the EBV-positive rate was significantly higher in cancers with GCP (31.1%) than in those without GCP (5.8%). EBV infection also can be detected in noncarcinomatous gastric epithelium. It was reported that DNA in situ hybridization detected EBV genomes in 10.5% (2 of 19) of chronic atrophic gastritis tissues,21 and it seems like that prolonged EBV persistence may contribute to the development of gastric carcinoma. It is also well known that the clinicopathologic and molecular characteristics of EBV-associated gastric carcinoma (EBVaGC) are quite different from those of conventional gastric carcinoma.22, 23 EBVaGC, which comprises nearly 10% of all gastric cancers, has clinicopathologic features like a preponderance of men, proximal location, diffuse Lauren type, frequent association with multiple and remnant gastric cancers, a lower frequency of lymph node metastasis, a relatively favorable prognosis, and aberrant concordant methylation of multiple genes.

The patients with GCP-associated cancers in our study shared similar clinicopathologic features with EBVaGC, such as a preponderance of men, proximal location, earlier tumor stages, and frequent association with multiple and remnant gastric cancers. In addition, the current results confirmed that gastric cancer with GCP is strongly associated with EBV infection. In this regard, the clinicopathologic features of GCP may be explained in part by the association between GCP and EBV infection. However, in our subgroup analysis of EBV-positive cancers, the presence of GCP still was associated significantly with men, earlier tumor stage in terms of depth of invasion, and less lymphatic involvement compared with the absence of GCP. It is not clear whether the association between gastric cancer with GCP and EBV infection was a coincidental finding or a cause-effect relation. It appears that chronic inflammation may induce GCP and facilitate EBV infection, and these 2 factors may contribute to the development of gastric cancer.24

However, it is difficult to confirm the specificity of GCP for gastric cancer because of the rare incidence of GCP in gastric cancers and the requirement for transmural tissue evaluation. Further research is needed to clarify the exact mechanism of gastric carcinogenesis associated with GCP and/or EBV with regard to other factors, such as Helicobacter pylori, molecular markers, etc. Clinically, we should be cautious about patients who have GCP, because GCP may be combined with malignancy or may have a potential role as a premalignant lesion; thus, endoscopic resection or close follow-up is recommended after a definite diagnosis.5, 15

In conclusion, the current study demonstrated distinct clinicopathologic features in patients who had gastric cancers with and with GCP, suggesting they the 2 groups have different pathways of gastric carcinogenesis. Further studies are needed to clarify the pathogenesis of GCP along with its possible link in the development of gastric carcinoma.

FUNDING SOURCES

This study was supported by a grant from the Korea Healthcare Technology R&D Project, Ministry for Health and Welfare Affairs, Republic of Korea (A092255).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary