Prognostic information of gastric carcinoma using Goseki system in relation to nuclear organizer region ( AgNOR ) and proliferating cell nuclear antigen ( PCNA ) expression

Objective: The present study was undertaken to find out the relationship of Goseki grading system (I-IV) with existing classification of WHO, Lauren and tumour differentiation of gastric carcinoma and its prognostic information in relation to AgNOR & PCNA expression. Materials and methods: To assess the reproducibility and usefulness of Goseki grading system thirty five gastric carcinoma were selected from January 2007 to July 2009 in the department of Pathology, Burdwan Medical College in West Bengal, India and analyzed in relation to existing grading system by chi-squared testing. Mean AgNOR count & mean PCNA leveling index quantities of different tissue sections were assessed according to different classification system of gastric carcinoma and interobserver variations of all data were evaluated from Spearman Rank-order Correlation Coefficient (rs ). Results: Highly significant predictable correlation of Goseki grading system for existing classification of gastric carcinoma was obtained statistically. Increasing values of mean AgNOR count and PCNA leveling index (2.35% & 15.14%, 2.91% & 21.32%, 3.08 %& 24.76% and 3.2% & 25.12 respectively) were observed from Grade I – IV of Goseki grade. Mucin rich (3.05% & 23.22%) and tubule poor (3.14% & 24.76%) tumours higher values than mucin poor (2.71% & 19.95%) and tubule rich (2.63% & 18.23%) tumors. No significant correlations were observed in other grading system. Conclusion: Following Goseki grading system increasing expression of proliferating marker in mucin rich than mucin poor tumours and tubules poor than tubules rich tumors indicate poor prognosis and tumour behavior. Simple system may help to select patients for adjuvant therapy.


Introduction
Gastric carcinoma is highly variable lesion.Many different classification systems have been proposed for the morphological, histological classification, grading and staging system to find out tumour prognosis to patient survival.None of them are satisfactory.The surgical status of the specimen resection is highly variable due to palliative as well as curative operations.Thus the pathologist has an important role in determining the cancer progression and depends on thorough tissue sampling as well as meticulous microscopic examination.In 1992 Goseki 1 proposed a classification system of carcinoma of stomach based on tubular differentiation and intracytoplasmic mucin production and was thought superior to other classification system, such as those of World Health Organization (WHO), Lauren system and tumour differentiation.According to Goseki four grades depending on tubule formation and intracellular mucin production by the tumour cells categorized as Grade I tubules well formed and mucin poor; II tubules well mucin rich; III tubules poor differentiated mucin poor; IV tubules poor mucin rich.Significant reliability with interobserver agreement with this grading system was observed by different workers 2, 3 and was proposed the only system with prognostic value in additional to TNM staging 4 .
The clinical significance of these classifications was limited, only the Lauren and perhaps the Goseki classifications were proposed to have prognostic assessments 5 .However, the Lauren classification has been the most successful system, as it defines two distinct histological entities, which clearly exhibit different clinical and epidemiological characteristics, even in advanced gastric cancers 6 .
The prognosis of gastric carcinoma is related to depth of tumour invasion, size, histological type, DNA ploidy, cell proliferation, tumour grading & staging.The cell proliferation indices like AgNOR (Nucleolar organizer regions) counting and PCNA LI (Proliferate cell nuclear antigen labeling index) have been found to be a prognostic marker along with histological classification and tumour grade of different neoplastic lesions.Various study showed increased numbers of AgNORs count in nuclei from normal mucosa to gastric carcinoma and significantly higher in intestinal type than diffuse type gastric carcinomas in Lauren type 7. PCNA LI and AgNOR expressions were significantly increased from normal to dysplasia to gastric adenocarcinoma 8 .Another study did not found any evidence for prognostic value with PCNA in gastric carcinoma 9 .
Nucleolar organizer regions (NORs) are the DNA loops which are transcribed to rRNA and ultimately to protein namely C 23 and B 23 10 are responsible for silver staining of nucleoli and are easily visible under light microscope by staining with freshly prepared silver colloidal developer solution.AgNOR dots are black dots within the nucleus usually increased in number of rapidly proliferating cells and acts as a proliferate activity of tissues, particularly in neoplastic cells.AgNOR quantity was measured by counting number of dots per nucleus in 100 tumour cells and average was taken as mean AgNOR count.
PCNA is a nonhistone nuclear protein (36 KD) has a role in initiation of cell proliferation 11 and the level of PCNA increased in the nucleus during late G1 phase immediately prior to the onset of DNA synthesis and maximum in S phase of cell cycle.Immunolocalization of PCNA in tumour used as proliferate activity of tumour and acts as proliferative tumour marker.The progressive growth of the tumour is determined by excess of cell proliferation over cell death and major fractions of transformed cells are in proliferative pool.
Two recent studies found that the Goseki histological classification was predictive of survival in patients with gastric cancer 3,12 , although two other studies did not get any predictive value 13,14 and thus the prognostic value of the system remains controversial 15 .
Purpose of the present study was to find out a comparison of Goseki's grading system for existing classification systems of gastric adenocarcinoma along with expression of mean AgNOR count and PCNA labeling index (PCNA LI) and an assessment of their reproducibility.

Material and methods
The study was done in the year of January 2007 to July 2009 from the tissues received in the Department of pathology; Burdwan Medical College and Hospital, West Bengal, India.35 cases of tissue paraffin blocks of primary gastric adenocarcinoma were included in this study after taking permission of ethical committee.The tissue sections of 5 micron size were taken on microscopic glass slides precoated with 3% gelatin from routinely processed paraffin embedded blocks.The sections were then stained with routine haematoxxylin & eosin stain and alcian blue PAS pH 2.5 to detect intracytoplasmic mucin production for classification of Goseki grade.
The tumors were categorized according to WHO classification into tubular, papillary, mucinous, signet ring cell type, adenosquamous, and small cell and undifferentiated type determined by prominent component of the tumour.According to Lauren as intestinal and diffuse type and few tumors not categorized were designated as mixed (tumors with equal quantity of intestinal and diffuse type).Tumors were also categorized as well, moderate and poor according to differentiation.In the Goseki system four categories were graded into Grade I tubules well differentiated and mucin poor; II tubules well mucin rich; III tubules poor differentiated mucin poor; IV tubules poor mucin rich gastric carcinoma.The tumors not categorized in any systems were excluded in this study.

AgNOR staining and counting
The deparaffinised tissue sections were first thoroughly washed with double distilled water for 15-20 minutes.Working solution was freshly prepares mixture of one volume of 2% gelatin in 1% formic acid solution and two volume of 50% aqueous silver nitrate solution in a dark room condition.Working solution was poured over the sections kept in a dark place for 45 minutes at 37 0 C.After washing with double distilled water all the sections were mounted in D.P.X.AgNOR dots were counted within the nucleus by Crocker's method 16 both single dot and in clusters (counted as single dot) under oil immersion.Total dots were counted in 100 tumour cells and average was taken as mean AgNOR count.

PCNA immunostaining
Immunohistochemistry were done in all tissue sections by standard procedure by using anti PCNA antibodies (PC 10 DAKOPATTS) at a dilution of 1 in 50 using ABC technique with AEC as chromogen and mounted in glycerin jelly.High grade breast carcinoma was taken as positive control.The PCNA positive cells showed reddish colour nucleus with AEC chromogen.The percentage of cells positive for PCNA among 500 tumour cells nuclei were counted under oil immersion as Proliferating Cell Nuclear Antigen Lebiling Index (PCNA LI).
Both AgNOR count and PCNA LI of all tissue sections were performed by two different observers to avoid inter-observer variation.

Statistical evaluation
The relationship of Goseki's grade to the other classifications and grading system was performed from the results of contingency table X 2 statistical test (Chisquared testing).The results of AgNOR counts and PCNA LI of two different observers were evaluated from Spearman Rank-order Correlation Coefficient (r s ).Multivariate analyses of different Goseki grades were analyzed by 'F' test.

Results and analysis:
The results of this study consists of 35 primary adenocarcinoma of stomach whose ages ranged from 33 to 79 years of ages with 23 male and 12 female patients.The AgNOR dots were ranged from 1-5 brownish black dots per tumour cell nucleus and mean AgNOR count in different tumors were 1.2-5.1%.The PCNA immunostaining showed nuclear positivity.Few foci in some tumors also showed mild cytoplasmic positivity and mean PCNA LI were 3-42.6%.The frequency of observations in the four Goseki grade for other histological classification and differentiation were tabulated in Table I.According to Goseki's grading 6(17.14%), 13 (37.14%),11(31.42%)and 5(14.30%)tumors were found in grade I, II, III, and IV respectively.Good tubular differentiation of Goseki grade I and II were reflected in the relationship of WHO classification, intestinal type of Lauren classification and histological differentiation.84% of WHO tumour of tubular differentiation were fell in Goseki grade I and grade II of tubule rich category.Similarly 94.73% of intestinal type of Lauren and well and moderately differentiated tumors were fallen in same category.Goseki's tubule poor grade III and IV showed 75% of poorly differentiated tumors.However 14.28% of WHO tubular tumors and 2.5% of intestinal tumour of Lauren type were found in tubular poor of Goseki's grade III and grade IV.In mucin content tumour all 4 of WHO type were in Goseki II and III grade and one signet ring cell type was categorized in grade IV Goseki system.High significant correlation obtained on Chi-squared testing of Goseki's grading system with WHO system, Lauren system and tumour differentiation and Degree of freedom & probability were 56.7(18) &0.000, 23.8(6) &0.001 and 32.0(6) &0.000 respectively.

WHO classification
The maximum AgNOR dots were found in mucinous (3.43%) and small cell type (4.11%).The mean PCNA LI was maximum in small cell type (29.13%) and mucinous type (27.97%).The tubular type of tumors (40%) did not show any significant result.No correlations were found among different tumors of this classification.

Lauren system
In

Tumour differentiation
Mean AgNOR count were 2.43%, 3.15% and 2.75% in well, moderately and poorly differentiate tumours respectively.Similarly the mean PCNA LI were 13.85%, 24.95% and 20.46% in well, moderate and poorly differentiated tumors.The moderately differentiated tumors (48.57%) showed higher values than remaining types.Tubule rich (Goseki grade I&II) & tubule poor (Goseki grade III&IV) tumors and mucin rich (Goseki grade II&IV) & mucin poor (Goseki grade I&III) of different gastric carcinoma classification systems were tabulated.Mean AgNOR count 2.88% and PCNA LI 20.46% of poorly differentiated carcinomas showed higher values than well and moderately differentiated carcinomas.Mucin rich tumors of different differentiation falling in Goseki II and IV showed higher mean AgNOR (3.16%) and PCNA LI(25.91 %) than mucin poor of Goseki grade I and III ( AgNOR 3.2% and 25.82%).Tubule poor tumors of Goseki's III and IV grade showed higher mean AgNOR and PCNA LI (3.81% and 31.97%)than tubule rich type of grade I and II (2.86% and 22.81%).

Discussion:
In spite of different classifications of gastric carcinoma system none of them showed any definitive prognosis and treatment prediction.New classification system with prognostic value still trying by different workers.WHO group was recognized as descriptive classification and not recognized as much prognostic out coma.In this system 63.15% of the tumors fell into tubular category and 23.33% tumors were undifferentiated (Table I).The present study did not show significant values of mean AgNOR count or PCNA LI (Table II) in any histological subtypes which helped in patient for good clinical management or survival.
Lauren classification system according to Jarvi et al 17 showed prognostic and useful in epidemiological studies.Prakash I et al 18 found higher values of mean AgNOR and PCNA LI in intestinal type than diffuse type of adenocarcinoma of Lauren system.Oshima et al 7 also found same result considering AgNOR.But opposite results were obtained in the present study (Table II).In intestinal type 66.66% and diffuse type 33.33% were mucin-rich tumors and mean AgNOR count were higher in diffuse type (  19 observer in a study of five year survival of patients with mucus rich (Goseki II and IV) T3 tumours was significantly worse than that of patients with mucus poor (Goseki I and III) T3 tumours (18% v 53%, p < 0.003).Goseki grading identifies subgroups of patients with a poorer prognosis than is predicted by TNM staging alone.Accordingly tubule poor tumors (Goseki grade I and II) expressed higher mean AgNOR and PCNA LI (3.14 % and 24.94%) than tubule rich (Goseki III and IV) type (3.14 and 24.94).Thus higher proliferative activity markers AgNOR and PCNA in Goseki's mucin rich tumors and tubule poor tumors indicating tumour progression.Goseki in 1992 proposed four grading system depending on tubular differentiation and degree of intracytoplasmic mucin production by the tumour cells.Along with other authors 19,20 Goseki suggested that mucin production is most important than tubule formation for assessment of prognosis.Goseki and his coworkers suggested from their 200 sample series that grade I tumors (tubular rich and mucin poor) had higher frequency of haematogenous metastasis, while type IV tumors (mucin rich and tubule poor) spread directly into the lymph nodes and peritoneum.In one study by Dixon et al 2 showed 55% patients with mucin-rich carcinomas died within 5 years than mucin poor patients.
In the present study, correlations of intra cytoplasmic mucin production and tubule formation by tumors falling in different histological classifications gastric adenocarcinoma (WHO, Lauren, tumour differentiation) with Goseki's grading system along with proliferating markers (AgNOR and PCNA ) helps to determine the efficacy of Goseki's grading system.Mean AgNOR count and PCNA LI expression of different histological classification systems in WHO, Lauren and tumor differentiation did not show any significant correlations, probably due to tumour heterogenecity and some form of deregulated expression of cell proliferative markers from tumour to tumour, from superficial to deeper layers of advanced gastric carcinoma and different tissue sections and microscopic fields.Increasing quantifications of mean AgNORs and PCNA LI of gastric carcinoma according to Goseki grading system along with the mucin rich and mucin poor tumors together with tubule rich & poor tumors may reflect the tumour behavior and prognosis in addition to tumour staging.The present study is a cost effective simple laboratory procedure may help to select patients for adjuvant therapy in rural hospital and needs further study with larger sample size.______________

Figure I :
Figure I: Photomicrograph showing multiple AgNOR dots within nucleus of tubules of Gastric Carcinoma (x1000)

Table I :
Relationship of Goseki grade for other classification of gastric carcinoma

Table II :
Spearman Rank-Order Correlation coefficient (r s ) of AgNOR & PCNA LI values expressed in different grades of gastric carcinoma *High degree of similarity.** Moderate degree of similarity.Obser-Observer.