Estimation of Silver-Binding Nucleolar Organizer Regions (AgNORs) in Lymphoproliferative Disorders of Gastrointestinal Tract

Path. Res. Pract. 190,350-361 (1994)

Estimation of Silver-Binding Nucleolar Organizer Regions (AgNORs) in Lymphoproliferative Disorders of Gastrointestinal Tract H. Takahashi, S. Fujita and H. Okabe Department of Oral Pathology, Nagasaki University School of Dentistry, Nagasaki, Japan

N.Tsuda Pathology Division of Central Diagnostic Laboratory, Nagasaki University Hospital, Nagasaki, Japan

F. Tezuka The Research Institute for Tuberculosis and Cancer, Tohoku University, Sendai, Japan

SUMMARY

We quantified nucleolar organizer regions demonstrable by silver staining technique (AgNORs) in six cases of reactive lymphoid hyperplasia (RLH), and in 14 low-grade B-cell lymphomas (mucosa-associated lymphoid tissue (MALT) 5, centrocytic 3, centroblasticcentrocytic 6) and 49 high-grade B-celllymphomas (centro blastic 44, immunoblastic 3, Burkitt's one, large cell anaplastic one). The pooled mean AgNOR number in low-grade B-celllymphomas was significantly higher than that in RLH, and significantly lower than that in high-grade B-cell lymphomas. There was a statistically significant difference between RLH versus centroblastic-centrocytic lymphoma in pooled mean AgNOR number and in RLH versus centrocytic lymphoma but not in RLH versus MALT lymphoma. The AgNOR numbers in the cellular components of RLH were also analyzed. The follicle center centroblasts of RLH exhibited a significantly higher pooled mean AgNOR number than other kinds of lymphoid cells in RLH. Furthermore, AgNOR numbers of gastrointestinal lymphomas were compared with those of cellular elements in RLH. There was a statistically significant increased AgNOR number of centroblastic lymphoma when compared with neoplastic centroblasts in centroblastic-centrocytic lymphoma and follicular centroblasts in RLH. By contrast, there was no significant difference in AgNOR numbers between MALT lymphomas and interfollicular lymphocytes, mantle zone lymphocytes or follicular centrocytes in RLH, respectively. This study demonstrated a similar increase when non-Hodgkin's B-celllymphomas in gastrointestinal tract were compared with histogenetically related cellular elements of RLH. AgNOR counting might be useful adjunct in the classification and grading of lymphoproliferative disorders in gastrointestinal tract.

Introduction

The gastrointestinal tract is the commonest primary site of extranodallymphomas 13. As reported in the literature, most gastric B-celllymphomas are large cell ("histiocytic") 0344-0338/94/0190-0350$3.50/0

type 15 ,24,28, and many of them possess nuclei resembling centro blasts according to the Kiel classification27,35. The association between reactive lymphoid hyperplasia (RLH) and primary B-celllymphoma of the gastrointestinal tract is well recognized25 ,28; nevertheless, there is © 1994 by Gustav Fischer Verlag, Stuttgart

AgNORs in Gastrointestinal Lymphoproliferative Disorders· 351

controversy about the differential diagnosis between the two conditions. Recent authors have questioned the validity of distinguishing between RLH and primary gastrointestinal lymphoma and it has become apparent that traditional morphological criteria, particularly the presence of germinal centers, are unreliable 38 . In a previous study of primary gastric lymphoma, which included cases of "pseudolymphoma", Isaacson et al.1 8 demonstrated light chain restriction in all cases. In the course of their study these authors characterized the B-cell mucosaassociated lymphoid tissue (MALT) lymphomas of the stomach as indolent tumors that usually remain localized for a long time before dissemination occurs l8 , 19. In low-grade B-celllymphoma of the stomach, MALT lymphoma has been reported as the most frequent histologic subtype4 . Some cases of gastric lymphoma previously reported as immunocytoma, centroblastic-centrocytic, or centrocytic lymphoma may represent examples of this tumor 35 • Furthermore, unusual histologic appearance of MALT lymphoma may lead to misdiagnosis, usually as a or so-called benign lesion, pseudolymphoma, RLH3,23,30,38. Silver-stained nucleolar organizer regions (AgNORs) contain loops of DNA (rDNA) encoded for ribosomal RNA (rRNA) production and associated with nucleoli. AgNORs estimation has been shown to aid in the distinction between certain benign and malignant tumors, and between low- and high-grade tumors 2,5, 7,11 and may help as an indication of cell proliferation6,14,16. In a few previous studies of node-based non-Hodgkin's lymphomas and reactive follicular hyperplasia, a significantly increased mean AgNOR number was found in high-grade lymphomas when compared with low-grade lymphomas5 , whereas mean AgNOR number of reactive follicular hyperplasia was greater than that in follicular lymphoma9 • It is not yet clear, on the other hand, to what extent such an AgNOR enumeration is of value for the discrimination among RLH, low-grade B-cell lymphoma or high-grade B-celilymphoma of gastrointestinal tracts. In this study using a silver staining technique, nucleolar organizer region-associated proteins have been studied in paraffin sections of 63 cases of non-Hodgkin's lymphoma and 6 cases of RLH in gastrointestinal tracts. The main aims of this study were: firstly, to assess the reliability of AgNOR staining method in distinguishing among RLH, low-grade B-celilymphoma and high-grade B-celilymphorna; secondly, to establish whether the mean number of AgNORs can be used as a marker of classification between MALT lymphoma and RLH; and thirdly, to determine if there is any relationship between the mean AgNOR count in lymphoma cells and that in the histogenetically related cellular components of RLH. Material and Methods Tissues Histology. Sixty-three operatively resected primary gastrointestinal non-Hodgkin's lymphomas were gathered from the files of the Pathology Division of Central Diagnostic Laboratory,

Nagasaki University Hospital and its affiliated hospitals between 1962 and 1984. Surgical specimens were routinely fixed in formalin and processed in paraffin. Hematoxylin and eosin, Giemsa and periodic acid-Schiff stained paraffin sections of gastrointestinal resected tissues were examined, and classified according to the updated Kiel system 22 by one (H. T.) of the authors. These comprised the following Kiel subtypes of nonHodgkin's lymphomas: MALT (five), centrocytic (three), centroblastic-centrocytic (six), centroblastic (44), immunoblastic (three), Burkitt (one), large cell anaplastic (one) (Table 1). In addition, six cases of RLH were also examined. The criteria applied in the diagnosis of RLH were fundamentally those described by Faris and Saltzstein (1964)12.

Table 1. Results of classification in gastrointestinallymphoproliferative disorders Histologic type

No. of cases

Reactive lymphoid hyperplasia Non-Hodgkin's B-celllymphoma Low-grade B-celllymphoma Mucosa-associated lymphoid tissue lymphoma Centrocytic lymphoma Centroblastic-centrocytic lymphoma High-grade B-celllymphoma Centro blastic lymphoma Immunoblastic lymphoma Burkitt's lymphoma Large cell anaplastic lymphoma

6 63

Total

69

14

5

3 49

6

44 3 1

1

Histologic subtypes of gastrointestinal lymphomas according to the updated Kiel classification.

Immunohistochemistry. Immunoperoxidase stammg was performed on paraffin tissue sections in 69 cases of RLH and non-Hodgkin's lymphomas, and thereafter the immunological phenotype of all cases was determined. Paraffin sections cut at 3 [tm were stained by the use of the avidin-biotin complex technique as described previously33. Table 2 lists the panel of primary antibodies used, all of which have been shown to be reactive in formalin-fixed, paraffin-embedded tissues. The antibodies chosen included those reported to detect markers of B-cell (immunoglobulins, L26, KiB3, LNl, LN2) and T-cell (UCHLl, MTl, DFT1), as well as the HLA-DR antigen marker LN3, BerID (activated lymphoid cells) and Leu-7 (natural killer cells). Antibodies detecting cathepsin D, lysozyme and alpha-I-antitrypsin were used as markers of histiocytic origin, in addition to KPI and LN5. Antibody to epithelial membrane antigen (EMA) was used to detect tumors of epithelial origin. In all cases of non-Hodgkin's lymphomas, the specimens were characterized by means of the demonstration of monotypic immunoglobulin and B-cell markers (L26, KiB3, LNI and/or LN2), and were all of established B-cell lineage. Conversely, stains with T-cell marker (UCHLl) and histiOCytic markers (cathepsin D, lysozyme, alpha-I-antitrypsin, KPl and LN5) were totally negative. Neoplastic cells in only one of 63 non-Hodgkin's lymphomas showed BerH2 reactivity, and this case was diagnosed as large cell anaplastic lymphoma. Lymphoid cells in six of 69 cases showed a polytypic immunoglobulin pattern and these cases belonged to RLH.

352 . H. Takahashi et a!. Table 2. Details of the antibodies used in this study Antibody

Isotype

Predominant reactivity

Working dilution

Source

B-cell markers IgG IgM IgA IgD Kappa Lambda L 26 KiB3 LN1 LN2

Polyclonal Polyclonal Polyclonal Polyclonal Polyclonal Polyclonal IgG 2a IgG j IgM IgG l

y heavy chain

1:2000 1:4000 1:4000 1:4000 1:4000 1:4000 1:50 1 :5000 prediluted prediluted

Dako Ltd., Copenhagen, Denmark Dako Ltd. Dako Ltd. Dako Ltd. Dako Ltd. Dako Ltd. Dako Ltd. Patho!' Inst. Univ. Kiel, Germany Techniclone Int., CA, USA Techniclone Int.

IgG 2a IgG j IgG j

T cells, epithelioid cells T cells, myeloid cells, macrophages T cells, myeloid cells, macrophages

1 :100 1:10 1:100

Dako Ltd. Dako Ltd. Dako Ltd.

T-cell markers UCHLl MT1

DFTl

Histiocytic markers Cathepsin D Polyclonal Polyclonal Lysozyme

!t heavy chain CL heavy chain b heavy chain Kappa light chain Lambda light chain B cells B cells Follicular center cells Follicular center cells, mantle zone cells, interdigitating histiocytes, T cells

Histiocytes, macrophages 1:150 Myeloid cells, monocytes/macrophages 1:300

Novocastra Lab. Ltd., Newcastle, UK. Dako Ltd.

CLrantitrypsin

Polyclonal

Monocytes/macrophages, some epithe- 1: 1500 lial cells

Dako Ltd.

KP1 LN5

IgG j IgM

Macrophages 1:100 Monocytes/macrophages, mantle zone prediluted B cells, gastric chief cells, interdigitating reticulum cells

Dako Ltd. Techniclone Int.

Miscellaneous LCA

IgG j

1:200

Dako Ltd.

Ber-H2

IgG j

1:20

Dako Ltd.

LN3 EMA

IgG 2a IgG2a

prediluted 1:100

Techniclone Int. Dako Ltd.

Leu7

IgM

1:80

Becton-Dickinson, CA, USA

Vim en tin LN6

IgG j IgM

Lymphocytes, myeloid cells, monocytes/macrophages Hodgkin cells, Sternberg-Reed cells, activated T and B cells Cells expressing HLNDR Epitheli
1:10 prediluted

Dako Ltd. Techniclone Int.

AgNOR Staining Method Sections were cut at 3 !tm from routinely processed paraffin blocks, dewaxed in xylene, and hydrated through graded methanol into deionized distilled water. Thereafter a one-step silver nitrate staining was performed. The staining solution was freshly prepared by dissolving 2 % gelatin in deionized distilled water adding formic acid to a final concentration of 1 %. This solution was mixed with 2 volumes of 50 % aqueous silver nitrate to obtain the final working solution. The paraffin sections were immersed in an AgNOR staining solution for 40 min under darkroom conditions at room temperature. Gold-enhancement was performed by covering the sections with 0.2 % aqueous gold chloride for 10 min, followed by 1: 5-diluted photographic fixing solution for 5 min in oxder to prevent discoloration of the AgNOR dots. The sections were washed with deionized distilled water prior to counterstaining with Mayer's hematoxylin, dehy-

drated through alcohols to xylene and mounted with synthetic medium. All of the sections for examination were stained at the same time and with the same solution.

Counting Procedure and Statistical Analysis Counting was done by using a X 100 oil immersion objective to a total magnification of X 1000 by one and the same person (H. T.) of the authors. In RLH of gastrointestinal tracts, AgNORs were counted in 200 cells of follicular centro blasts, follicular ceptrocytes, mantle zone lymphocytes, and interfollicular lymphocytes, respectively. In every case of gastrointestinallymphomas, AgNOR dots were counted in at least 200 nuclei. The nuclei for the study were selected randomly, and a counting grid was employed to ensure that there was no duplication of counting. All separate silver stained dots in the nucleoli and nucleoplasm were

AgNORs in Gastrointestinal Lymphoproliferative Disorders . 353 counted as AgNO Rs and polycyclic structures were counted as ones. Focussing was adjusted for each nucleus to include all the AgNORs in all possible foca l planes in the counts. The comparison was made among follicular centro blasts in RLH, neoplastic centro blasts in centroblastic-centrocytic lymphoma, and lymphoma cells in centroblastic lymphoma; among follicular centrocytes in RLH, neoplastic cells of MALT lymphoma, and neoplastic centrocytes in centroblastic-centrocytic lymphoma and centrocytic lymphoma; among mantle zone lymphocytes and interfollicular lymphocytes in RLH, and neoplastic cells of MALT lymphoma; and among lymphoprolifera tive disorders of gastrointestinal tracts, i.e., RLH, low-grade B-cell lymphoma, and high-grade B-celilymphoma, in order to examine the correlation of AgNORs with hyperplastic and neoplastic changes of lymphoid cells. The mean number of AgNORs per nucleus was calculated and all the data were statistically assessed. The mean values for all groups were compared using the Student's t-test.

Results

A. AgNOR Counts AgNORs were seen as small, well defined black silverstained intranuclear structures, sometimes in aggregates

within the nucleolus, in all of the gastrointestinal tracts examined in this study. They exhibited somewhat varying size and configuration, and occurred in all kinds of cells investigated. There appeared to be a somewhat greater variation in size and configuration of the AgNORs in malignant than in reactive lymphoid cells. The pooled mean number of AgNORs in RLH, low-grade B-cell lymphoma and high-grade B-celilymphoma are summarized in Table 3 and a scattergram (Fig. 1). Reactive Lymphoid Hyperplasia (RLH)

In six cases of RLH, the pooled mean number of AgNORs ranged between 2.16 and 2.38 (mean, 2.28 ± 0.09) (Table 3). The pooled mean number of AgNORs were determined from the following cell types; mantle zone lymphocytes (mean, 1.33 ± 0.18) (Fig. 2), interfollicular lymphocytes (mean, 1.34 ± 0.19) (Fig. 3), follicular centrocytes (mean, 3.28 ± 0.21) (Fig. 4), and follicular centroblasts (mean, 4.54 ± 0.27) (Fig. 4). Analysis of pooled means showed a significant difference between the AgNOR numbers derived from follicular centro blasts and follicular centrocytes (p < 0.01). There

Table 3. Comparison of pooled mean number of AgNORs per nucleus in gastrointestinal lymphoproliferative disorders using Student's t-test Histologic and cell types Reactive lymphoid hyperplasia (I) Mantle zone lymphocyte (II) Interfollicu lar lymphocyte (III) Follicular centrocyte (IV)

No. of cases

Pooled mean Range (mean ± SED)

± 0.09 ± 0.18 ± 0.19 ± 0.21

6 6 6 6

2.28 1.33 1.34 3.28

6

4.54 ± 0.27 4.21- 5.01

14 5

3.81 ± 1.60 1.12- 6.14 2.09 ± 1.3 7 1.12- 4.11

Centrocytic lymphoma (VIII)

3

4.23 ± 0.71 3.62- 5.01

Centroblastic-centrocytic lymphoma (IX)

6

5.03 ± 0.68 4.59- 6.14

6 6 49

6.04 ± 0.93 5.89- 8.05 3.44 ± 0.44 3.11- 4.21 8.24 ± 1.03 5.03-11.34

44

8.19 ± 0.76 6.51-10.09

3 1 1

8.22 ± 2.91 5.03-10.75 11.34 7.53

Follicular centroblast (V) Low-grade B-celllymphoma (VI) Mucosa-associated lymphoid tissue lymphoma (VII)

Centroblast in centroblastic-centrocytic lymphoma (X) Centrocyte in centroblastic-centrocytic lymphoma (XI) High-grade B-celllymphoma (XII) Centroblastic lymphoma (XIII) Immunoblastic lymphoma (XIV) Burkitt's lymphoma (XV) Large cell anaplastic lymphoma (XVI) NS, Not significantly altered.

2.161.161.093.01-

Statistical significance

2.38 1.62 1.62 3.56

II-III NS II-IV p < 0.01 III-IV p < 0.01 II-V p < 0.01 III-V p < 0.01 IV-V p < 0.01 I-VI p < 0.01 I-VII NS II-VII NS III-VII NS IV-VII NS V- VII P < 0.01 I-VIII p < 0.05 II-VIII P < 0.01 III-VIII P < 0.01 IV-VIII NS V-VIII NS VII-VIII . P < 0.05 I-IX p < 0.01 VII-IX P < 0.05 VIII-IX NS V-VIII P < 0.01 IV-XI NS I-XII p < 0.01 VI-XII p < 0.01 IX-XIII p < om X-XIII p < 0.01

354 . H. Takahashi et al. 12 0

11

0 0

10

t

9 VI :J OJ

U

8

:J C Co

OJ

a.

7

VI

n:

0

z\lD

0

0

6

<{

....0 Co

OJ

0 0

5

.0

"

c c

III OJ

0

COO

E

00

4

-0

::;E

3

o

2

00

o

2

o~------------------------------------Reactive Low-grade lymphoid hyperplasia

High-grade

8-cell lymphoma

Fig. 1. Scattergram showing the pooled mean AgNOR numbers per nucleus in reactive lymphoid hyperplasia, low-grade B-cell lymphoma and high-grade B-cell lymphoma of gastrointestinal tracts. Horizontal bars indicate mean in each ,group.

was also a significant difference between pooled mean AgNOR numbers from follicular centrocytes and mantle zone and interfollicular lymphocytes (p < 0.01), but not between the latter two cell types.

Fig. 2. Mantle zone lymphocytes in reactive lymphoid hyperplasia. The nuclei contain one to three AgNOR dots. Silver colloid staining, original magnification x 500. Fig. 3. Interfollicular lymphocytes in reactive lymphoid hyperplasia. Almost all nuclei contain one to two AgNOR dots. Silver colloid staining, original magnification x 500.

AgNORs in Gastrointestinal Lymphoproliferative Disorders· 355

Fig. 4. Follicular centro blasts and centrocytes in reactive lymphoid hyperplasia. Centroblasts containing three to nine AgNOR dots in nuclei and centrocytes showing two to six dots per nucleus. Silver colloid staining, original magnification x 500.

Fig. 5 . MALT lymphoma. Most neoplastic cells contain one to four AgNOR dots. Silver colloid staining, original magnification x 500.

"

Non-Hodgkin's B-cell Lymphoma Low-grade B-celllymphoma. In 14 cases of low-grade B-ceillymphomas, the pooled mean number of AgNORs was 3.81 ± 1.60. The group of MALT iymphoma contained the lowest mean number of AgNORs (mean, 2.09 ± 1.37) among three types of gastrointestinal lowgrade B-cell lymphomas investigated (Fig. 5). As can be seen on the scattergrams (Figs. 1 and 6), the AgNOR count in neoplastic cells of MALT lymphoma was in the range of lymphoid cells of RLH. The pooled mean number of AgNORs in centrocytic lymphoma (mean, 4.23 ± 0.71) (Fig. 7) was found to be significantly higher than that in MALT lymphoma (p < 0.05). The highest value in lowgrade B-ceillymphoma was found in centroblastic-centrocytic lymphoma (mean, 5.03 ± 0.68) (Fig. 8). The pooled mean number of AgNORs in this lymphoma was significantly higher than that in MALT lymphoma (p < 0.05). However, there was no significant difference when compared with centrocytic lymphoma.

High-grade B-celllymphoma. The pooled mean number of AgNOR dots in high-grade B-ceillymphoma.consisting of 49 cases was 8.24 ± 1.03. All four high-grade malignancy varieties of non-Hodgkin's lymphoma, i.e. centroblastic lymphoma (mean, 8.19 ± 0.76) (Fig. 9), Burkitt's lymphoma (mean, 11.34) (Fig. 10), immunoblastic lymphoma (mean, 8.22 ± 2.91) (Fig. 11), and large cell anaplastic lymphoma (mean, 7.53) (Fig. 12) showed an extremely high level in their counts. B. Comparison of the AgNORs among RLH, Low-grade B-cell Lymphoma and High-grade B-cell Lymphoma The results of AgNOR numbers in gastrointestinal lymphomas in comparison with that in the cells constituting the RLH were summarized in Table 3 and Figs. 1 and 2. When treating all high-grade B-celilymphomas as one group, the pooled mean number of AgNORs was signifi-

356 . H. Takahashi et al. 12

0 0

11 III

0

10

t

::J

..'!! u

::J

9

C L

Ql

DIII

0:: 0 Z

tlll

7

«

6

L

5

'0 ill

-"

E ::J c c

'" ill

2

0 0

8

4

Q§) 0

V

0

2

~ 0

cb

cc

MZL

• IFL

0

0

0

---00

~

3

_0_

0

-e-

cP 0

CB-CC

MALT

CC

CB

Fig. 6. Scattergram representing the pooled mean AgNOR numbers per nucleus in constituent cells of lymphogastrointestinal proliferative disorders. Horizontal bars indicate mean in each cell type or in each histological type. cb = centroblast, cc = centrocyte, MZL = mantle zone lymphocyte, IFL = interfollicular lymphocyte, CB-CC = centroblastic-centrocytic lymphoma, CC = centrocytic lymphoma, MALT = lymmucosa-associated phoid tissue lymphoma, CB = centroblastic lymphoma, Others = incluce immunoblastic, Burkitt's and large cell anaplastic lymphomas.

Others

~~~-

Reactive lymphoid hyperplasia

High-grade

Low-grade

Fig. 7. Centrocytic lymphoma. The cleaved medium-sized nuclei contain one to seven AgNO R dots. Silver colloid staining, original magnification x 500.

B-celi lymphoma

Fig. 8. Centroblastic-centrocytic lymphoma. The nuclei of neoplastic centro blasts and centrocytes present the highest number of AgNORs among low-grade B-celllymphomas investigated. Silver colloid staining, original magnification x 500.

AgNORs in Gastrointestinal Lymphoproliferative Disorders . 357

Fig. 9. Centro blastic lymphoma. All cells contain up to 10 small and grouped AgNOR dots. Silver colloid staining, original magnification x 500.

Fig. 10. Burkitt's lymphoma. All cells contain numerous grouped and scattered AgNORs within the nucleus. Silver colloid staining, original magnification x 500.

candy higher than that in low-grade B-cell lymphomas (p < 0.01). Low-grade B-cell lymphomas exhibited a significantly higher pooled mean AgNOR number than the RLH (p < 0.01). In RLH, the highest value of pooled mean AgNOR numbers was identified in follicular centro blasts. Among low-grade B-cell lymphomas, the highest value of AgNORs was found in centroblastic element of centroblastic-centrocytic lymphoma. The pooled mean number of AgNORs in neoplastic centro blasts of centroblasticcentrocytic lymphoma was significantly lower than that in centroblasticlymphoma (p < 0.01) and higherthan that in follicular centro blasts of RLH (p < 0.01). The mean number of AgNORs per nucleus in centrocytic lymphoma was lower than that in follicular centroblasts of RLH and higher than those in follicular centrocytes of RLH, mantle zone lymphocytes of RLH and neoplastic centrocytes of centroblastic-centrocytic lymphoma. But only the difference between centrocytic lymphoma and reactive mantle zone lymphocytes reached a statistically significant level (p < 0.01). In neoplastic cells of MALT lymphomas, there was a slight increase of pooled mean number.of AgNORs

when compared with mantle zone lymphocytes and interfollicular lymphocytes in RLH. However, the difference did not reach a statistically significant level. Moreover, the neoplastic cells of MALT lymphomas showed a nonsignificantly lower pooled mean AgNOR number than that of the follicular centrocytes in RLH. In comparison with follicular centroblast in RLH, a significant decrease (p < 0.01) in pooled mean AgNOR numbers was found for the MALT lymphomas. Discussion The characterization and quantitation of nucleolar organizer regions (NORs) by a simple silver staining on formalin-fixed, paraffin-embedded sections have added another approach to understanding biological features of a variety of human disorders. The argyrophilic method is based on reaction of silver ions with carboxyl groups and sulphydryl groups of phosphoproteins bound to NORs (AgNORs)21 ,26. It had been suggested that the number of AgNORs may reflect processes such as hyperplastic and

358 . H. Takahashi et al.

Fig. 11. Immunoblastic lymphoma. Most of these nuclei contain multiple dots. Fusion of these dots is observed in some cells but in most cells can be resolved by adjusting the focus. Silver colloid staining, original magnification X 500.

Fig. 12. Large cell anaplastic lymphoma. Giant cells contain up to 12 grouped and scattered AgNORs within the nucleus. Silver colloid staining, original magnification X 500.

neoplastic tissues, and there is increasing evidence to suggest that NOR numbers are a useful histological marker of cell proliferations, 7, 10, 11, 16,34. As far as the lymphoid tissue is concerned, Crocker and Nar (1987)5 have used AgNOR counting as a diagnostic discriminant for malignancy since it was found that AgNORs may be increased in nodal lymphomas compared with their benign counterparts. They have also indicated that the number of AgNORs is increased in high-grade lymphom;ls compared with low-grade lymphomas. To our knowledge, however, AgNOR estimation in reactive lymphoid hyperplasia (RLH) and B-celllymphoma of gastrointestinal tracts has not been reported in the pertinent literature before. Especially, AgNOR numbers of mucosa-associated lymphoid tissue (MALT) lymphoma and large cell anaplastic lymphoma, which were established on the basis of their special histopathologic features 22 , were firstly discussed in the present study. Gastrointestinal malignant lymphoma holds an important position among. extranodallymphomas 13 • There are many clinicopathological studies on gastrointestinal nonHodgkin's lymphomas, however, ther.e are only a few

which stressed a close relationship between gastrointestinal non-Hodgkin's lymphoma and benign lymphoid lesions (coeliac disease, RLH or pseudolymphoma)3,17,2S,36. Considering all the immunohistochemical data reported and the follow up of these diagnosed patients, pathologists and clinicians have divided gastrointestinallymphoproliferative disorders into three categories, RLH, low-grade B-cell lymphoma and high-grade B-cell lymphoma. This categorization of gastrointestinallymphoproliferative disorders is not straightforward as a grey area exists between these categories, which in some cases can be resolved by immunohistochemistry28. Newer techniques and approaches are required to achieve satisfying results for classifying gastrointestinal lymphoid lesions. We therefore stained a variety of human gastrointestinallymphoproliferative disorders with the AgNOR technique to see if 'some conclusive answer could be obtained. The overall mean nuclear AgNOR count for high-grade B-celllymphomas was 8.24, and for low-grade B-celllymphomas it was 3.81. In agreement with previously published results in nodal lymphomaS obvious distinction could be made

AgNORs in Gastrointestinal Lymphoproliferative Disorders . 359

between low-grade and high-grade B-celllymphomas on the basis of NOR numbers. Moreover, since significantly different pooled mean AgNOR numbers were observed among RLH, low-grade B-celilymphoma and high-grade B-celllymphoma of gastrointestinal tract, it appears that quantificabon of AgNORs might also be of importance in the grading of these three groups. As referred to the histological differentiation between RLH and MALT lymphoma is occasionally problematic. In order to assist the separation of MALT lymphoma from RLH, a variety of morphological and immunohistochemical methods have been reported 28 ,38. Low-grade B-cell lymphomas arising from gastrointestinal-associated lymphoid tissue have recently been reported to show the features of MALT lymphoma 19. Gastrointestinallymphomas previously diagnosed as lymphoplasmacytoid-lymphoplasmacytic, centrocytic, or centroblastic-centrocytic lymphoma, and many cases of gastric "pseudolymphoma" may represent examples of this tumor 30,35 . In a previous study of primary gastric lymphoma, which included cases of "pseudolymphoma", Isaacson et al.l 8 demonstrated light chain restriction in all cases. Although the histogenesis of these MALT lymphomas is not fully understood, ample evidence favors' derivation from a B-cell subcohort native to mucosal linings. In addition, a relationship has been proposed to the B-cells of the marginal zone and/or of lymph follicles I9 ,29. In the present study, the AgNOR numbers differed among low-grade B-celllymphomas, and MALT lymphomas had the lowest mean values of AgNORs. As described in "Material and Methods" of this study, five MALT lymphomas showed monotypic immunoreactivity for immunoglobulin of neoplastic cells. Although the AgNOR number in one of five MALT lymphomas was quite similar to that of centrocytic lymphoma, most MALT lymphomas had lower numbers than other low-grade B-celilymphomas. Interestingly, the present study demonstt ated that the pooled mean AgNOR nllmbers per nucleus in histopathologically and immunohistochemically verified MALT lymphoma were not significantly lower than those in the overall lymphoid cells of RLH. This low value of MALT lymphoma is in agreement with clinical follow-up studies revealing that some MALT lymphomas grow very slowly or may be stationary for several years 4,35 . Recently, Cronin et a\.9 and Weiss et alY have reported that reactive follicular hyperplasia shows greater proliferative activity than follicular lymphoma. Most follicular lymphomas are low-grade lymphomas, possibly low cell proliferation, whereas hyperplastic follicles are highly active units, as can be seen in any reactive lymph node immunostained with monoclonal antibody Ki67. This raises the possibility that increased numbers of AgNORs may reflect cell proliferation rather than biological malignancy. Large cell anaplastic lymphoma is an uncommon and hitherto poorly recognized form of high-grade nonHodgkin's lymphoma which may cause considerable diagnostic difficulty. The cytological appearance and pattern of nodal infiltration of these tumors can mimic malignant melanoma and carcinomi20,3 1,32. The existence of this entity was recognized using the monoclonal antibody Ki-l

and/or the Ki-l equivalent antigen BerH2. The majority of large cell anaplastic lymphomas are reported to consist of activated T cells 1,32. Because of this rarity, their proliferative activity has not been reported. In the present study, large cell anaplastic lymphoma with immunoreaction for BerH2 and B-cell marker was analyzed by AgNOR technique. The high AgNOR counts in this tumor, as demonstrated in our study, reflect high nuclear proliferative activity and provided further evidence of the highgrade malignant potential of large cell anaplastic lymphoma. A relationship between AgNOR numbers in neoplastic cells of B-cell lymphomas and those in reactive counterparts of RLH was analyzed. Centroblastic lymphomas are usually considered to develop from follicular centroblasts in RLH, or from centroblastic-centrocytic lymphoma undergoing high-grade malignant transformation. Hence the data for the centroblastic lymphomas was compared with that for follicular centroblasts in RLH and with that for centroblastic-centrocytic lymphomas. In centro blastic lymphomas and centro blasts of centroblastic-centrocytic lymphomas, in general, the mean AgNOR numbers were higher than in their corresponding non-tumorous cell types. It is now obvious that centrocytic lymphomas are not derived from the follicular centrocytes, but rather from the mantle zone lymphocytes 24 . Hence we used AgNOR numbers for the reactive mantle zone lymphocytes and neoplastic centrocytes for statistical evaluation. The corresponding value of 4.23 in centorcytic lymphoma is significantly higher than that of 1.33 in mantle zone lymphocytes of RLH. Although there are still some uncertainties as to the histogenesis of the MALT lymphomas, a relationship has been proposed to the B-cells of marginal zone and/or germinal centers 19 , 29. Therefore, it seems adequate for the present purpose to compare the data for MALT lymphomas with those for mantle zone lymphocytes, interfollicular lymphocytes and follicular centrocytes in RLH. The pooled mean AgNOR numbers of 2.09 per nucleus in MALT lymphomas was higher than those of 1.33 for mantle zone lymphocytes and 1.34 in interfollicular lymphocytes of RLH. On the other hand, AgNOR numbers in MALT lymphomas were lower than those in follicular centrocytes of RLH. However, no apparent difference was observed between MALT lymphoma and three kinds of cellular elements of RLH. The observations in this study suggest that MALT lymphoma was not significantly different from the AgNOR reading of non-tumorous counterpart. By contrast, we found a statistically significant increase in the number of AgNORs in centrocytic lymphomas and centro blastic lymphomas compared with histogenetically related reactive mantle zone lymphocytes and follicular centroblasts, respectively. Our findings indicated that AgNOR numbers in RLH progressively increased from mantle zone and interfollicular lymphocytes to follicular centrocytes and follicular centroblasts. Furthermore, the AgNOR numbers in nonHodgkin's B-cell lymphoma exhibited progressive increases from MALT lymphoma to centrocytic lymphoma or centroblastic-centrocytic lymphoma, and highgrade malignancies. Our results are in disagreement with a

360 . H. Takahashi et al.

previous study by Crocker and Nar (1987)5, in which no significant differences were found in AgNOR numbers between mantle zone and interfollicular lymphocytes and follicular centrocytes of reactive lymphadenitis. Furthermore, in node-based low-grade lymphoma, they described a considerable overlap in AgNOR numbers among immunocytoma, centrocytic lymphoma and centroblastic-centrocytic lymphoma. This discrepancy between our results and data of Crocker and Nar (1987)5 may be due to the different organ, or different method of AgNOR counting, since in our study we took into consideration all of the separable AgNOR in the nucleus and therefore counted a higher number of AgNOR in follicular centrocytes and follicular centro blasts of RLH and neoplastic cells of non-Hodgkin's lymphomas.

References 1 Agnarsson BA, Kadin ME (1988) Ki-1 positive large cell lymphoma: A morphologic and immunologic study of 19 cases. Am J Surg Pathol12: 264-274 2 Boquist LLV (1990) Nucleolar organizer regions in normal, hyperplastic and neoplastic parathyroid glands. Virchows Arch A (Pathol Anat) 417: 237-241 3 Brooks J], Enterline HT (1983) Gastric pseudolymphoma. Its three subtypes and relation to lymphoma. Cancer 51: 476--486 4 Cogliatti SB, Schmid U, Schumacher U, Eckert F, Hansmann ML, Hedderich J, Takahashi H, Lennert K (1991) Primary B-cell gastric lymphoma: A clinicopathological study of 145 patients. Gastroenterology 101: 1159-1170 5 Crocker J, Nar P (1987) Nucleolar organizer regions in lymphomas. J Pathol151: 111-118 6 Crocker J, Macartney JC, Smith PJ (1988) Correlation between DNA flow cytometric and nucleolar organizer region data in non-Hodgkin's lymphoma. J Pathol154: 151-156 7 Crocker J, McGovern J (1988) Nucleolar organizer regions in normal, cirrhotic, and carcinomatous live,s. J Clin Pathol 41: 1044-1048 8 Crocker J, Boldy DAR, Egan MJ (1989) How should we count AgNORs? Proposal for a standardized approach. J Pathol 158: 185-188 9 Cronin K, Loftus BM, Dervan PA (1989) Are AgNORs useful in distinguishing follicular hyperplasia from follicular lymphoma? J Clin Pathol 42: 1267-1268 10 Dervan PA, Gilmartin LG, Loftus BM, Carney DN (1989) Breast carcinoma kinetics: argyrophilic nucleolar organizer region counts correlate with Ki67 scores. Am J Clin Pathol 92: 401--407 11 Egan MJ, Crocker J (1988) Nucleolar organizer regions in cutaneous tumours. J Pathol154: 247-253 12 Faris TD, Saltzstein SL (1964) Gastric lymphoid hyperplasia: A lesion confused with lymphosarcoma. Cancer 17: 207-212 13 Freeman C, Berg JW, Cutler SJ (1972) Occurrence and prognosis of extranodallymphomas. Cancer 29: 252-260 14 Giri DD, Nottingham JF, Lawry J, Dundas SAC, Underwood JCE (1989) Silver-binding nucleolar organizer regions (Ag-NORs) in benign and malignant breast lesions: correlations with ploidy and growth phase by DNA flow cytometry. J Pathol 157: 307-313 15 Grody WW, Weiss LM, Warnke RA, Magidson JG, Hu E, Lewin KJ (1985) Gastr.9intestinal lymphomas, immunohistochemical studies on the cell of origin. Am J Surg Pathol 9: 328-337

16 Hall PA, Crocker J, Watts A, Stansfeld AG (1988) A comparison of nucleolar organizer region staining and Ki-67 immunostaining in non-Hodgkin's lymphoma. Histopathology 12: 373-381 17 Isaacson P (1981) Primary gastrointestinal lymphoma. Virchows Arch A (Pathol Anat) 391: 1-8 18 Isaacson PG, Spencer J, Finn T (1986) Primary B-cell gastric lymphoma. Hum Pathol 17: 72-82 19 Isaacson PG, Spencer J (1987) Malignant lymphoma of mucosa-associated lymphoid tissue. Histopathology 11: 445--462 20 Kadin ME, Sako D, Berliner N, Franklin W, Woda B, Borowitz M, Ireland K, Schweid A, Herzog P, Lange B, Dorfman R (1986) Childhood Ki1lymphoma presenting with skin lesions and peripheral lymphadenopathy. Blood 68: 1042-1049 21 Kling H, Lepore L, Krone W, Olert J, Sawatzki G (1980) Chemical studies on the silver staining of nucleoli. Experientia 36: 249-251 22 Lennert K, Feller AC (1990) Histopathologie der NonHodgkin-Lymphome (nach der aktualisierten Kiel-Klassifikation). Springer Verlag-Berlin-New York 23 Lerman ST, Ziv Y, Rubin M, More C, Dintsman M, Kadish U (1985) Gastric lymphoma versus pseudolymphoma: the importance of immunological differentiation. Am J Gastroenterol 80: 763-766 24 Lewin KJ, Ranchod M, Dorfman RF (1978) Lymphomas of the gastrointestinal tract, a study of 117 cases presenting with gastrointestinal disease. Cancer 42: 693-707 25 Mohri N (1987) Primary gastric non-Hodgkin's lymphomas in Japan. Virchows Arch A (Pathol Anat) 411: 459--466 26 Pfeifle J, Boller K, Anderer FA (1986) Phosphoprotein pp 135 is an essential component of the nucleolus organizer regions (NOR). Exp Cell Res 162: 11-22 27 Radaszkiewicz T, Dragosics B (1980) Primary lymphomas of the gastrointestinal tract: A clinicopathologic study of 60 cases. Path Res Pract 169: 353-365 28 Saraga P, Hurlimann J, Ozzello L (1981) Lymphomas and pseudo lymphomas of the alimentary tract, an immunohistochemical study with clinicopathologic correlation. Hum Pathol 12: 713-722 29 Spencer J, Finn T, Pulford KAF, Mason DY, Isaacson PG (1985) The human gut contains a novel population of B lymphocytes which resemble marginal zone cells. Clin Exp Immunol 62: 607-612 30 Spencer J, Smith WJ, Diss TC, Isaacson PG (1989) Primary B cell gastric lymphoma and 'pseudolymphoma': A genotypic analysis. Am J Pathol135: 557-564 31 Stein H, Mason DY, Gerdes J, O'Connor N, Wainscoat J, Pallesen G, Gatter KC, Falini B, Delsol G, Lemke H, Schwarting R, Lennert K (1985) The expression of the Hodgkin's disease associated antigen Ki1 in reactive and neoplastic lymphoid tissue: Evidence that Reed-Sternberg cells and histiocytic malignancies are derived from activated lymphoid cells. Blood 66: 848-858 32 Suchi T, Lennert K, Tu L-Y, Kikuchi M, Sato E, Stansfeld AG, Feller AC (1987) Histopathology and immunohistochemistry of peripheral T cell lymphomas: a proposal for their classification. J Clin Pathol 40: 995-1015 33 Takahashi H, Fujita S, Okabe H, Tsuda N, Tezuka F (1993) Immunophenotypic analysis of extranodal non-Hodgkin's lymphomas in the oral cavity. Path Res Pract 189: 300-311 34 Trere D, Farabegori F, Cancellieri A, Ceccarelli C, Eusebi V, Derenzini M (1991) AgNOR area in interphase nuclei of human tumours correlates with the proliferative activity evaluated by bromodeoxyuridine labelling and Ki-67 immunostaining. J Patho1165: 53-59 35 Van KriekenJHJM, Otter R, HermansJ, Van Groningen K, Spaander pJ, Van de Sandt, Keuning JF, Kluin PhM (1989)

AgNORs in Gastrointestinal Lymphoproliferative Disorders . 361 Malignant lymphoma of the gastrointestinal tract and mesentery. A clinico-pathologic study of the significance of histologic classification. 135: 281-289 36 Vanden Heule B, Van Kerkem C, Heimann R (1979) Benign and malignant lymphoid lesions of the stomach. A histological reappraisal in the light of the Kiel classification of non-Hodgkin's lymphomas. Histopathology 3: 309-320 37 Weiss LM, Strickler jG, Medeiros Lj, Gerdes j, Stein H,

Warnke RA (1987) Proliferative rates of non-Hodgkin's lymphomas as assessed by Ki-67 antibody. Hum Pathol 18: 1155-1159 38 Zukerberg LR, Ferry jA, Southern jF, Harris NL (1990) Lymphoid infiltrates of the stomach. Evaluation of histologic criteria for the diagnosis of low-grade gastric lymphoma on endoscopic biopsy specimens. Am j Surg Pathol 14: 1087-1099

Received july 13, 1993 . Accepted in revised form October 8, 1993

Key words: Non-Hodgkin's B cell lymphoma - Reactive lymphoid hyperplasia - Nucleolar organizer regions Gastrointestinal tract Hiroshi Takahashi, D. D. S., Dr. med. Sc., Associate Professor, Department of Oral Pathology, Nagasaki University School of Dentistry, 1-7-1 Sakamoto, Nagasaki 852, japan