INTRODUCTION

Despite the fact that gastric cancer incidence has been recently declining, this cancer still remains a frequent cause of death-especially in certain countries such as Cost Rica, Japan and Colombia^1,2. In West Europe and USA it is the fifth or sixth commonest malignant tumor in the human.

Several pathological entities in the stomach have been implicated as precancerous or high risk conditions for cancer development. Chronic atrophic gastritis, intestinal metaplasia, pernicious anemia, Menetrier's disease, previous gastric surgery, gastric polyps and H.P. infection are some of these entities^3,8. There are convincing data in the literature concerning intestinal metaplasia, dysplasia, previous gastric surgery and other entities with precancerous potentiality. Since 1983, when H.P. was first observed in gastric antrum, it has been implicated as the causative factor for duodenal ulcer and type B antral gastritis. Nowadays, H.P. tends to be recognised as a group 1 carcinogen. In this report, the proliferative activity of gastric epithelium in relation with H.P. infection was estimated using an immunohistochemical detection of Proliferating Cell Nuclear Antigen (PCNA). PCNA is a cyclin required for cellular DNA replication and represents a useful tool for the study of tissue proliferation in oncology^9,10. It has a higher labelling index in well established rapidly proliferating tissues and can be used as a sensitive index for cell proliferation assessment^11,12.

Material and Methods

Three different groups of patients were examined:

Group A comprised 40 patients with surgically treated gastric cancer, Group B included 20 patients with H.P. gastritis and the Group C comprised 20 patients without H.P. gastric infection.

In group A, 40 consecutive patients were included; who had undergone gastrectomy for gastric cancer between January and December 1992 at the First Propaedeutic Surgical Department of Athens University Medical School. Two specimens were taken for pathology examination from each surgically removed tumor in this group. The material for the first specimen was taken from gastric mucosa adjacent to the tumor and as far as the second specimen is concerned, it was taken from gastric antrum mucosa away from the tumor. In addition, 5 cm³ of blood from each patient were taken and centrifuged; the serum was stored at -200 C.

In groups B and C, patients who has undergone gastroscopy as external patients for upper gastrointestinal tract symptoms were included. Two biopsy specimens were taken during gastroscopy in each patient. One specimen was taken from the pyloric antrum and a second one from the gastric angle. Giemsa staining was performed on all sections and if the presence of H.P. was discovered in the biopsy specimens during pathology examination, the patient was incorporated in group B; otherwise he was incorporated in group C. Any patient with a gastric disorder considered as a potential precancerous condition, was excluded from the study. We excluded patients with previous gastric surgery for ulcer disease, chronic atrophic gastritis and gastric polyps. The same procedure was followed until 20 patients were incorporated in each group. Serum samples were obtained from each patient and were stored at -200 C until use, just like samples in group A patients.

The serology of H.P. infection

Serum antibodies against H.P. were measured by an enzyme-linked immunosorbent assay (ELISA). The quantitative test of BIORAD company for identification of lgG antibodies against H.P. was used (GAP Test lgG 404 2002). In this test, we considered a titre higher than 20 U/ml as criterion for seropositivity, whereas for seronegative specimens, a titre limit lower than 12,5 U/ml was assessed. Infection with H.P. was considered doubtful in cases with values between 12.5 and 20 U/ml.

Histology

All pathology specimens were routinely processed and stained with haematoxylin-eosin and modified Giemsa for assessment of histological features and the presence of H.P. respectively (Fig. 2).

Immunohistochemical staining

The pathology specimens from all three groups (two specimens for each patient) were immunohistochemically stained with the PC10 antibody (Ab) to PCNA.

Paraffin embedded sections fixed in buffered formalin for no more than 10 hours, were submitted to immunohistochemistry, using the peroxidase-labelled, three step, avidine-biotin technique (ABC-HRP, Dako). To perform immunohistochemical staining, sections were not heated, but they were de-waxed using alcohol and then immersed in a phosphate-buffered saline solution as well as in methanol with 0.5% hydrogen peroxide to block endogenous peroxidase activity. A mouse monoclonal antibody to human proliferating cell nuclear antigen (DAKO-PCNA, clone PC10, code No M879, dilution 1:250 with overnight incubation at 40 C) was commercially obtained and the manufacturers' instructions were taken into account.

Cell counting

In the gastric mucosa the epithelium of the neck region of the glands is the proliferating compartment. Cell reduplication takes part mainly in the bottom of the pit and the cells produced gradually migrate to the surface. Consequently, PC10 staining is expectedly more intensive near the bottom than on the luminal surface of the gastric pit. In our study, all immunostained sections were examined by one observer using a X40 objective with an eyepiece graticule. A minimum of ten different high power fields (X400) were examined in each section and, in an attempt for a most objective assessment of the labelling index, only the nuclei of axial pit sections were included in the count (Fig. 1). The total number of nuclei in the pit and the labelled nuclei were noted and the labelling index was expressed as the percentage of labelled epithelial cells per total number of epithelial cells counted. The way of counting described above is a modification of the one used by Sarraf et al¹¹.

Statistical analysis

Fisher's exact test was used for the comparison of H.P. infection between patients with highly differentiated and poorly differentiated carcinomas. Using the Colmogorov-Smirnov test, we noticed that frequency distribution of PCNA in each group of patients was close to normal, so the student's t-test was selected as the suitable test for quantitative comparison of mean values of PCNA expression. Finally, the results for PCNA were expressed as mean value ± SD, while for the titres of serum antibodies we gave the range and the median values.

RESULTS

IgG antibodies against H.P. and histology detection of H.P.

All patients of group B were characterised by an unequivocal existence of H.P. in biopsy specimens. Nineteen of them were seropositive against H.P. and only one had a titre indicating doubtful infection (Table 1). Twelve patients in group C were seronegative, 4 were seropositive and the remaining 4 had doubtful infection, while in none of the latter was there a histologically obvious H.P. infection. If we take the patients of groups B and C into account and considering that patients with doubtful titres were neither seropositive nor seronegative, the serology test for H.P. detection has a specificity of 73.3% while its sensitivity is approximately 100%.

Unlike these results, in group A patients there was no such concordance between pathology detection of H.P. and seropositivity. While 34 of these patients were seropositive, it was feasible to reveal the bacteria in only 18 of them during histological examination (Table 2). All neoplasms were divided into highly and poorly differentiated. Twenty three were found to be highly and 17 poorly differentiated. Twenty two out of 23 (95.6%) patients with low grade cancers and 13 out of 17 (76.5%) with high grade tumors harboured bacteria. This difference was not statistically significant (p>0.1; Fisher's exact test).

Immunohistochemical staining with PC10 Ab

PCNA immunolabelled nuclei were easily identified and PCNA positivity was greater near the bottom of the pits in comparison with the surface. Using cell counts, mean PCNA immunostaining index was found to be more high in group A, less high in group C and with an intermediate value in group B. Results are shown in Tables 2 and 3. In detail, the mean labelling index for PCNA staining in groups A, B and C was 23.2%±9, 11.1%±7.3 and 5.2%±4.1 respectively. We noticed that the labelling indexes were quite similar between the two different pathology specimens of each patient in groups B and C. In group A, however, there was a big difference between the two specimens of each patient; specimens coming from the mucosa adjacent to the tumors evidently demonstrated higher labelling indexes. This probably has to do with the severely dysplastic epithelium observed on the mucosa adjacent to the tumor which shows higher PCNA expression (Fig. 3).

DISCUSSION

Most patients colonised with H.P. elicit a measurable systemic antibody response. The immunoglobulin classes and subclasses of these circulating anti-H.P. Abs are consistent with a prolonged chronic infection; IgG and IgA Abs are predominating whereas IgM Abs are rarely seen^13,14. In a well documented volunteer study, where ingestion of H.P. eventually led to chronic gastritis, IgG seroconversion occurred between the 22nd and 23rd day post infection¹⁵. This specific IgG Ab remains constant throughout the infection and the antibody levels slowly decrease after treatment and concominant clearance of the organism with patients attaining normal levels no sooner than after 6 months^16,17. The measurement of IgG antibodies is an easy, commonly used and accurate method for identification of H.P. infected patients. A specificity of 73.3% and a sensitivity of approximately 100%, as found in the ELISA test of this study, are in accordance with other reports^16,18.

In the present study, we distinguished infected from uninfected patients based on either seropositivity or histologic findings. As seen in the results of groups B and C, the accuracy of these two methods is quite similar. In group A, however, we noticed that pathology examination tends to underestimate the number of H.P. infected patients. Perhaps, this is due to prolonged exhibition of the surgical specimens in atmospherical air; a factor causing death of the bacteria. The cause of death is the atmosphere temperature as well as the concentration of O₂ and CO₂ in the atmosphere which is not at all suitable for H.P. growth¹⁹. In addition, all surgical specimens are washed out of blood and mucus before their formalin fixation. So, most of the bacteria may be carried away by washing. These disadvantages render the pathology observation an unreliable method of H.P. identification in surgical specimens. Nevertheless, in biopsy specimens, these disadvantages do not exist and thus pathology examination seems to be a quite reliable method for H.P. identification in this sort of material. As a sequel of the above, we consider all the seropositive patients plus the seronegative ones who may paradoxically harbour bacteria in pathology examination as H.P. infected. Thirty-four out of 40 patients (85%) in group A were seropositive, while in only 18 of all these patients (45%) bacteria could be histologically detected. If we take both methods into account, a total of 35 cancer patients (87.5%) were H.P. infected; a proportion in keeping with other studies^8,20.

PCNA is a proliferation-associated, nuclear protein which is mainly expressed during the S phase of the cell cycle and appears to play a role in DNA replication, possibly as co-factor of DNA polymerase-δ. Although several markers such as 5-bromo-2-deoxyuridine and tritiated thymidine, have been used to assess proliferation in human tissues, PCNA (and Ki-67) have the advantages of being reproducible, not interventional and also that they can be performed in routinely processed tissues^11,21 on condition that uniform standards of fixation length are maintained. Convincing data in the literature indicate that PCNA can be used as a useful tool for the estimation of cell proleferation in gastro-intestinal epithelia¹¹.

In this study, the immunohistochemical cellular staining for PCNA was characterized by a consistent topographical distribution, being mainly localized in gastric pits' deep epithelia, which are known to be the actively proliferating parts of the gastric pits. The relative mean labelling index in group B was 11.1%±7.3 and in group C it was only 5.2%±4.1. This difference, which is statistically significant (p<0.01) indicates that H.P. infection forces gastric epithelium to a higher proliferation fraction. A similar difference was noticed between patients in groups A and B with mean labelling indexes 23.2%±9 and 11.1%±7.3 respectively (p<0.001). This higher immunohistochemical PCNA staining in group A is most possibly due to the gastric mucosal alterations occurring in tumor neighbouring mucosa. The mucosa with such changes demonstrates high PCNA positivity; this finding is at least indicative of a high proliferation fraction. Consequently, the hyperproliferation of the non malignant gastric epithelium in group A could be considered as a result of H.P. infection as well as of any other factor causing tumor adjacent, severely dysplastic epithelial changes.

H.P. infection has been implicated in several studies as a causative factor for gastric mucosal changes that predispose to gastric cancer^22,26. As a result of this study, we can add that H.P. infection provokes a high proliferation fraction of the gastric mucosa. Recently, this hyperproliferation has been supported by other authors too^27,28. The high proliferation fraction of course, on its own, cannot be incriminated as a precancerous condition, but the more intensive the proliferation, the more vulnerable the epithelium becomes to several mutagen agents. The latter may be the process via which H.P. potentially promotes gastric carcinogenesis.

As Correa and Ruiz report, agents involved in the aetiology of cancer are distinguished as either initiators or promoters²³. H.P. could be classified in the group of promoters, being a factor which increases the risk for gastric cancer development on condition that gastric mucosa is exposed to an initiator.

An epidemiological association between H.P. and gastric cancer has been established. It is of interest that a constant decrease in gastric cancer incidence has been established during the last few decades in West Europe, USA and other developed countries. It is well known that people in these countries extensively use antibiotics-especially during their childhood- for upper respiratory infections or other reasons. Considering the fact that H.P. is sensitive to most of these antibiotics, the decrease of gastric cancer incidence in these countries could reflect the eradication of H.P. in the relative populations.

In conclusion, we can speculate that hyperproliferation of the gastric epithelium might be the mechanism via which H.P. promotes gastric carcinogenesis at least in those cases where several mutagens are also present and ready to act.

References