Peripheral t and putative natural killer cell lymphomas commonly coexpress CD95 and CD95 ligand

Peripheral T and Putative Natural Killer Cell Lymphomas Commonly Coexpress CD95 and CD95 Ligand CHI-SING NG, FRCPATH,STEPHENT.H. LO, MSc, AND JOHN K.C. CHAN, FRCPATH The CD95 (Fas)/CD95 ligand (CD95L) system is an important mechanism triggering apoptosis, and CD95L expression has recently been implicated for immune evasion and aggressive behavior in malignancies. This study aimed to investigate CD95 and CD95L expression in lymphomas and the possible relationship with tumor cell apoptosis, with emphasis on the natural killer (NK) cell lymphomas, which are highly aggressive neoplasms and frequently exhibit tumor cell apoptosis/necrosis. Frozen sections of 82 cases of lymphomas obtained from Queen Elizabeth Hospital and Caritas Medical Center, Hong Koug, were immunostained with polyclonal anti-CD95 and anti-CD95L antibodies. The NK-cell lymphomas were also studied for apoptosis by in situ end labeling (ISEL) method, and zonal tumor cell death was evaluated semiquantitatively. The cases studied included 27 NK-, 22 T-, and 33 B-cell lymphomas. CD95 was expressed in 25 (93%) NK-, 11 (50%) T-, and 14 (42%) B-cell lymphomas. CD95L was expressed in 19 (70%) NK-, 15 (68%) T-, and 3 (9%) B-cell lymphomas. There was significant difference in the frequency of CD95 expression between B- and NK- (P < .001), and between 37- and NK-cell lymphomas (P < .05), and in CD95L expression between B-

and T- (P < .01) or NK-cell (P < .01) lymphomas. Zonal tumor cell death was present in 21 (78%) NK-cell lymphomas and 1 (4.5%) T-cell lymphoma and showed no correlation with CD95 or CD95L expression. ISEL analysis showed apoptosis predominantly in the viable areas in only 5 (24%) NK-cell lymphomas. In conclusion, CD95L is frequently expressed in NK- and T-cell lymphomas, but rarely in B-cell lymphomas. Zonal tumor cell death is not correlated with CD95 or CD95L expression and thus the CD95/CD95L system probably does not contribute significantly to this phenomenon. We postulate that the frequent expression of CD95L by NK- and T-cell lymphomas may mediate local or systemic tissue damage and immune evasion, and may contribute to the clinical aggressiveness of these tumors. HuM PATrIOL30:48-53. Copyright © 1999 byW.B. Saunders Company Key words: CD95, CD95 ligand, lymphomas, apoptosis. Abbreviations: NK, natural killer; EBV, Epstein-Barr virus; ISEL, in situ end labelling; TBS, Tris-buffered saline; NBT/BCIP, nitroblue tetra zolium/bromochloroindoxyl phosphate; PBS, phosphatebuffered saline; TdT, terminal deoxynucleotidyl transferase.

The CD95 (Fas) molecule belongs to the tumor necrosis factor receptor superfamily which on interaction with the CD95 ligand (CD95L, also known as Fas ligand) expressed on cytolytic effector cells (activated T cells and natural killer [NK] cells) results in target cell apoptosis medicated by complex intracellular signals. 1,2 The CD95/CD95L-mediated target cell apoptosis takes part in immunologic regulation, autoimmunity, immunologic privilege (tolerance), graft rejection, and imm u n e evasion by malignant neoplasms. 39 The latter, a recent observation, is of particular interest because it may at least partly explain the aggressiveness of some tumor types. Carriers of C~95 gene mutation may also develop the atypical lymphoproliferative syndrome and lymphoma. This may be related to defects in regulation of the i m m u n e response attributed to the CD95/CD95L system.10,11 CD95 is expressed in many different types of normal tissues (including T cells and NK cells) and neoplasms. 12q6 Various lymphoma types such as f o l l i c u -

lar, diffuse B-cell, T-cell, and Hodgkin's lymphomas have been reported to show variable frequencies of CD95 expression, 13-16 but the rare aggressive NK-cell lymphomas have not been systematically studied. 1725 CD95L shows much more restricted expression in normal tissues, and is expressed by activated cytotoxic T cells and NK cells. 7 Recently, high levels of soluble serum CD95L are detected in NK-cell leukemia and lymphoma, and CD95L has been postulated to be responsible for the frequent systemic tissue damage and aggressive clinical behavior of these neoplasms26,27 CD95L has been shown in T- and NK-large granular lymphocytic l e u k e m i a s y nasal putative NK-cell lymphoma, 2s and multiple myeloma, s However, systematic studies on C~95L expression in lymphomas are not available, although a small study on eight nasal lymphomas (two T/NK-cell and six NK-cell lymphomas) was recently published2 s Apoptosis is c o m m o n in many lymphomas and leukemias, 13a4 including NK-cell lymphomas. 24Although massive tumor cell death in NK-cell lymphomas may be attributable to angioinvasion, vascular damage, cytolytic effector molecules, and EpsteinBarr virus (EBV)-mediated cytokines and chemokines, 23-29the role of the CD95/CD95L-based mechanism remains to be explored in these lymphomas. This study aimed to investigate CD95 and C~95L expression in a large series of various types of lymphomas, including the rare NK-cell lymphomas and the possible role of the CD95/CD95L-based mechanism in tumor cell apoptosis with emphasis on the NK-cell lymphomas.

From Department of Pathology, Caritas Medical Center and Queen Elizabeth Hospital, Hong Kong. Accepted for publication August 13, 1998. Presented in part at the 86th Annual Meeting, US-Canadian Academyof Pathology,Florida,USA,March 1997. Address correspondence and reprint requests to Chi-Sing Ng, FRCPath, Department of Pathology, Caritas Medical Center, Shamshuipo, Kowloon,Hong Kong. Copyright© 1999by W.B.SaundersCompany 0046-8177/99/3001-0009510.00/0 48

CD95 LIGANDAND AGGRESSIVELYMPHOMAS(Ng et al) MATERIALS AND METHODS

SEMIQUANTITATIVE EVALUATION

Tissue Samples

Percentage of CD95 or CD95L-Positive Tumor Cells

The tissue samples were obtained from the archives of the Queen Elizabeth Hospital and Caritas Medical Center, Hong Kong. All samples were newly diagnosed cases without prior therapy. The cases selected were not consecutive, but included a significant number of NK-cell lymphomas. The tissue samples were partly fixed in 10% neutral-buffered formalin and partly snap frozen in liquid nitrogen. The fixed tissues were processed for paraffin sections as usual. The immunohistochemical and in situ end labeling (ISEL) studies were performed on frozen sections. Cases studied included: 27 NK-cell lymphomas (25 nasal, 2 tonsillar), 22 T-cell lymphomas, and 33 B-cell lymphomas.

The percentage of t u m o r cells expressing CD95 or CD95L was evaluated in tumor-bearing areas by counting the n u m b e r of t u m o r cells showing granular cell surface staining in 20 high power fields (×400) and obtaining the average figure. T u m o r cells were distinguished from infiltrating reactive lymphocytes by their larger size and nuclear atypia. Identification of t u m o r cells in NK-cell lymphomas was also facilitated by comparison with a corresponding frozen section immunostained for CD56. A case was arbitrarily considered positive if 10% or more of the t u m o r cells were stained.

Immunohistochemistry

Percentage of In Situ Apoptotic Cells and Area of Zonal Tumor Cell Death in Tumor Involved Tissues

Polyclonal rabbit anti-FAS (C-20, CD95) and anti-FASL (C-20, CD95L) antibodies were purchased from Santa Cruz Biotechnology Inc., California and both were used at a dilution of hl00. According to the manufacturer's specifications, anti-FAS detects an epitope corresponding to amino acids 316 to 335 mapping at the carboxy terminus of the FAS precursor of human origin. Anti-FASL detects an epitope corresponding to amino acids 260 to 279 mapping at the carboxy terminus of FAS-L of human origin. Specifications of both antibodies have been validated by western blotting by the manufacturer. Briefly, air dried 4-micron thick frozen sections were fixed in acetone for 10 minutes at room temperature. After removing excessive fixative with three changes of buffer wash (0.05M Tris buffered saline [TBS], pH 7.6), the sections were incubated with 10% normal rabbit serum for 10 minutes. They were then incubated with the primary antibodies at room temperature for 45 minutes. Signal detection was achieved using a Universal LSAB-2 kit 14674, (Dako A/S, Copenhagen, Denmark) with alkaline phosphatase conjugate, followed by color development with nitroblue tetrazolium/ bromochloroindoxyl phosphate (NBT/BCIP). The nuclei were counterstained with chloroform extracted 2% methyl green solution or neutral red. Positive result was shown as granular cell surface staining in the lymphoma cells.

The percentage of in situ apoptotic cells detected by the positive ISEL signal in tumor-involved tissues was e n u m e r a t e d in 20 high power fields (X400) and the average figure obtained. The percentage area of zonal t u m o r cell death in tumor-involved tissues was evaluated in all available materials (frozen and paraffin sections). Zonal t u m o r cell death was considered significant when it involved more than 10% area of tumor-involved tissue.

Statistical Analysis The differences in CD95 and C~95L expression a m o n g B-, NK-, and T-cell lymphomas were analysed by the nonparametric ANOVA test using Ranks Sum analysis. Correlation a m o n g the parameters ISEL-apoptotic index, zonal t u m o r cell death, CD95, and C~95L expression in various groups of lymphomas was also studied.

RESULTS NK-Cell Lymphomas (n = 27) Apoptosis, morphologically characterized by cell shrinkage with deeply eosinophilic cytoplasm and pyknotic or karyorrhectic nuclei, was constantly present. It was interspersed t h r o u g h o u t or in the form of small ill-defined foci within viable t u m o r areas or large circumscribed areas of zonal t u m o r cell death (Fig 1). Significant zonal t u m o r cell death was present in 21 cases (74%). ISEL analysis showed apoptosis in 5 (24%) of 24 cases studied; in these cases, ISEL-positive signals were predominantly detected within viable t u m o r areas and the percentage of positive cells ranged from 5% to 20%. ISEL positive signal was detected only in rare cells within the areas of zonal t u m o r cell death, despite the presence of many morphological apoptotic cells in these areas (Fig 2). The l y m p h o m a cells expressed CD95 in 25 cases (93%), with 15% to 95% (mean, 72%) of cells being positive. CD95L was expressed in 19 cases (70%), with the percentage of positive cells ranging from 15% to 95% (mean, 72%). All 19 CD95L-positive cases also expressed CD95. (Fig 1; Table 1)

In Situ Apoptosis Detection Detection of in situ apoptosis on frozen sections was achieved by the ISEL technique, using the In Situ Cell Death Kit, AP (Boehringer Mannheim GmbH, Germany). Briefly, 4-micron thick frozen sections were fixed in 4% phosphatebuffered paraformaldehyde at 4°C for 15 minutes. They were washed three times in pH 7.4 phosphate-buffered saline (PBS) in 5 minutes. The slides were permeabilized with 0.1% Triton X-100 in pH 7.4 PBS for 2 minutes. They were then incubated in a premixed ISEL reaction mixture containing terminal deoxynucleotidyl transferase (TdT) and fluorescein-dUTP for 4 hours at 37°C. TdT catalysed the addition of fluoresceindUTP at 3'-OH groups in single- and double-stranded DNA breaks occurring at early stages of apoptosis. The slides were then washed with pH 7.4 TBS, followed by inclubation with alkaline phosphatase conjugated mouse antifluorescein antibody at room temperature for 30 minutes. Color development was achieved by NBT-BCIR and nuclear fast red was used as counterstain. Positive ISEL reaction manifested as dark blue nuclear staining. Positive control using DNase 1 pretreated slides and negative control with omission of TdT from the ISEL reaction mixture were included in each batch.

49

HUMAN PATHOLOGY

Volume 30, No. 1 (January 1999)

FIGURE 1. NK-cell lymphoma, (A) An area of zonal tumor cell death is present in lower half and viable tumor tissue in upper half of field. Note morphological apoptotic cells (arrow) in the zonal tumor cell death area (H&E, original magnification x200). (B) Small foci of apoptotic tumor ceils (arrow) are present in the viable tumor involved areas (H&E, original magnification ×400). (C) CD95 staining of lymphoma cells (arrow shows unstained cell) (Immunoalkaline phosphatase, original magnification x400). (D) CD95L staining of lymphoma cells (arrow shows unstained cell) (Immunoalkaline phosphatase, original magnification x400), T-Cell L y m p h o m a s (n = 22)

fled showed an immunophenotype of CD2+ CD3+ CD8+ CD56+, expressed perforin, granzyme B and TIA1, and showed extensive zonal tumor cell death. All other postthymic T-cell lymphomas were CD8-negatire.

The 22 T-cell lymphomas included lymphoblastic (n = 4), lymphoepithelioid (n = 4), anaplastic large cell (n = 6), and peripheral T-cell lymphoma unspecified (n = 8). One peripheral T-cell lymphoma unspeci50

CD95 LIGAND AND AGGRESSIVE LYMPHOMAS (Ng et al)

FIGURE 2. NK-cell lymphoma. (A) Only rare apoptic cells in an area of zonal tumor ceil death show ISEL-positive signal (arrow). Many apoptic cells do not show ISEL-positive signal (arrow head) (ISEL, original magnification ×200). (B) ISEL-positive signals in an area of viable tumor tissue (ISEL, original magnification ×200).

Among the postthymic T-cell lymphomas, 10 cases (56%) expressed CD95 with 15% to 95% (mean, 58.5%) of cells being positive. CD95L was expressed in 12 cases (67%), with the percentage of positive cells ranging from 15% to 95% (mean, 61%). Eight of the 12 CD95L-positive cases also expressed CD95. Among the four cases of T-lymphoblastic lymphomas, three (75%) expressed CD95L and one of these positive cases also expressed C~95 (Table 1).

TABLE 1.

B-Cell L y m p h o m a s (n : 33)

The 33 B-cell lymphomas included 8 small cell lymphomas (2 small lymphocytic, 1 low-grade B-cell mucosa-associated lymphoid tissue type, 5 mantle cell), 4 follicular lymphomas, 19 diffuse large B-cell lymphomas, and 2 Burkitt's lymphomas. CD95 expression was seen in three mantle cell, four follicular, and seven diffuse large B-cell lymphomas. The percentage of positive cells ranged from 15% to 90% (mean, 68%). One mantle cell and two diffuse large B-cell lymphomas expressed C~95L, with the percentage of positive cells ranging from 15% to 90% (mean, 42%). One mantle cell and one diffuse large B-cell lymphoma coexpressed CD95 and CD95L (Table 1). The intensity of staining for CD95 and CD95L in B-cell lymphomas are generally weaker compared with that in NK- and T-cell lympho-

CD95 a n d CD95 Ligand Expression in Malignant Lymphomas T-Lymphomas B-Lymphomas PTL (n = 33) (n = 18)

CD95 Mean* (all cases) Median* (positive cases) No. of Positive Cases (%) CD95L Mean* (all cases) Median* (positive cases) No. of Positive Cases (%) CD95/CD95L coexpression No. of cases (%)

LBL (n = 4)

NKLymphomas (n = 27)

mas.

28.5%

37.0%

31.0%

68.5%

Statistical Analysis

68.0%

58.5%

24.0%

72.0%

14 (42%)

10 (56%)

There was significant difference in CD95 expression between B - a n d NK-cell lymphomas ( P < .001), and between T- and NK-cell lymphomas (P < .05). The expression of CD95L was also significantly different between B- and T-cell lymphomas ( P < .01), and between B- and NK-cell lymphomas ( P < .01) (Table 1). Within the different lymphoma groups, there was significant correlation between CD95 and CD95L in B-cell lymphomas ( P < .001), T-cell lymphomas ( P < .001) and in NK-cell lymphomas ( P < .01). There was no correlation between CD95 or C~95L expression with zonal tumor cell death or ISEL-apoptotic index. Among NK-cell lymphomas, there was no significant correlation between zonal tumor cell death and ISEL-apoptotic index.

1 (25%)

25 (93%)

4.0%

46.0%

39.0%

45.0%

42.0%

61.0%

53.0%

72.0%

3 (9%)

12 (67%)

3 (75%)

19 (70%)

2 (6%)

8 (44%)

1 (25%)

19 (70%)

Abbreviations: PTL, postthymic T-cell lymphomas; LBL, Tlymphoblastic lymphomas. *Percentage positively stained lymphoma cells,

51

HUMAN PATHOLOGY

Volume 30, No. 1 (January 1999)

DISCUSSION

CD95L expression in these aggressive tumors or use of neutralizing antibodies to CD95L. The frequent occurrence of apoptosis and zonal tumor cell death in NK-cell lymphomas is of special interest, 24 because this is much less common in CD95Lpositive solid tumors, T-cell lymphomas, and multiple myelorna. It has been proposed that CD95L-positive solid tumors develop resistance to CD95/CD95L induced apoptosis by acquiring an intracellular mechanism or by partial or total loss of CD95 expression. ~-7 Because the CD95L-positive NK- and T-cell lymphomas commonly coexpress CD95, it is likely that these tumors have developed intracellular mechanisms that make them resistant to CD95/CD95I_~mediated cytolysis. This study also does not show any correlation between zonal tumor cell death and CD95 or CD95L expression. The CD95/CD95L-based mechanism, therefore, is probably not a major contributing factor to tumor cell death in NK-cell lymphomas. Our previous study suggests that the perforin-based cytolytic mechanism, perhaps in conjunction with tumor angioinvasion and EBV-related pathways (tissue necrosis and vascular damage mediated by the chemokines Mig and IP-10) are the major contributing factors to zonal tumor cell death in NK-cell lymphomas. 24-29 One other finding is the lack of correlation between ISEL-apoptotic index and the percentage area of zonal tumor cell death in NK-cell lymphomas. Apoptotic cells with ISEL-positive signals are shown predominantly in areas with viable tumor cells, but only rarely in the areas of zonal tumor cell death, and Ohshima et al have recently reported similar findings. 2s This may be because ISEL is a method for detecting early apoptosis, and therefore fails to detect the more advanced apoptotic cells in areas of zonal tumor cell death. In conclusion, NK- and T-cell lymphomas frequently express CD95L, which may have facilitated local tissue invasion, distant metastasis, systemic tissue damage, and immune evasion, contributing to the marked aggressive behavior of these tumors.

T-cell lyrnphomas account for 10% to 30% of all non-Hodgkin's lymphomas and are aggressive neoplasms with frequent extranodal involvement, highstage disease and poor clinical outcome, s° NK-cell lymphomas (or so-called NK/T-cell lymphomas) are even less common, but appear to be more prevalent among Asians. They are characterized by large granular lymphocyte morphology, surface CD3- CD56 + phenotype, lack of T-cell receptor g e n e rearrangement, frequent expression of cytolytic effector molecules, strong association with EBV, presentation most commonly in the upper aerodigestive tract and other extranodal sites, and highly aggressive clinical behavior. 17-25,31 This study shows that CD95 and CD95L are frequently expressed in NK-cell and T-cell lymphomas, whereas CD95 is expressed less commonly and CD95L rarely in B-cell lymphomas. CD95 expression has been reported in follicular, diffuse B-cell, T-cell, and Hodgkin's lymphomas, 13-16with a reported frequency of approximately 60% among B-cell or T-cell lymphomas. 13,14,16 Our results show a comparable frequency of CD95 expression in T-cell lymphomas (50%) but a slightly lower percentage in B-cell lymphomas (42%). This study also shows frequent CD95L expression in NK(70%) and T-ceU lymphomas (68%). A recent study of eight cases of nasal T- or NK-ce!l lymphomas (two T/NK-cell and six NK-cell lymphomas) similarly shows a high frequency of CD95 and C~95L expression. 2s These results are not surprising because NK and T cells normally express CD9512,16and CD95L on activation. 7 The demonstration of high levels of soluble serum CD95L in NK-cell leukemia and lymphoma in recent studies raises the possibility that systemic release of CD95L from these tumors may be responsible for systemic tissue damage, ~6,e7mediated through the binding of circulating CD95L to CD95 thus triggering apoptosis of the CD95-bearing tissues. I2-16 In NK-cell lymphomas, the process may be exaggerated by frequent massive tumor cell death, 24 resulting in release of large quantities of CD95L into the circulation. The expression of CD95L in NK- and T-cell lymphomas may also lead to local damage of the involved CD95-positive tissues thus facilitating local tissue invasion. 7 The CD95/ CD95I_~mediated damage of blood vessels and lymphatic vessels (endothelium and tunica media being CD95-positive a6) may also facilitate vascular invasion and subsequent dissemination of the CD95L-positive lymphomas. Recently, CD95L has been implicated in the immunologic evasion by a variety of CD95L-positive neoplasms including melanoma, carcinomas of colon, liver and lung, and multiple myeloma. It is postulated that CD95L expressed on the tumor cells interact with the CD95 antigen on the tumor-infiltrating cytotoxic T cells, resulting in killing of the T-cells. 5-9 A similar phenomenon can occur in the NK- and T-cell lymphomas by virtue of the CD95L expression. All the above factors probably contribute to the aggressive behavior of these tumors. This observation may potentially lead to new therapeutic interventions such as modulating

REFERENCES 1. Goldstein P: Fas-based T-cell-mediated cytotoxicity, in Tschopp J, Goldstein P (eds): Current Topics in Microbiology and Immunology. Vol. 198, Berlin, Springer Verlag, 1995, pp 25-37 2. Liu CC, Young LHY, YoungJD-E: Lymphocyte-mediated cytolysis and disease. N EngJ Med 335:1651-1659, 1996 3. Lynch DH, Ramsdell F, Aldersou MR: Fas and FasL in the homeostatic regulation of immune responses. Immunol Today 16:569574, 1995 4. Griffith TS, Ferguson TA: The role of FasL-induced apoptosis in immune privilege. Immunol Today 18:240-244, 1997 5. Hahne M, Rimoldi D, Schr6ter M, et al: Melanoma cell expression of Fas (APO-1/CD95) ligand: Implications for tumor immune escape. Science 274:1363-1366, 1996 6. Strand S, Hofmann WJ, Hug H, et al: Lymphocyte apoptosis induced by CD95 (APO-1/Fas) ligand expressing tumor cells--A mechanism of immune evasion? Nat Med 2:1361-1366, I996 7. Nagata S: Fas ligand and immune evasion. Nat Med 2:13061307, 1996 8. Villunger A, Egle A, Marschitz I, et al: Constitutive expression of Fas (Apo-1/CD95) Ligand on Multiple Myeloma cells: A potential mechanism of tumor-induced suppression of immune surveillance. Blood 90:12-20, 1997

52

CD95 LIGAND AND AGGRESSIVE LYMPHOMAS (Ng et al) study of 49 cases of an uncommon aggressive neoplasm. Blood 89:4501-4513, 1997 21. EmileJF, Boulland ML, Haioun C, et al: CD5- CD56 + T-cell receptor silent peripheral T-cell lymphomas are natural killer cell lymphomas. Blood 4:1466-1473, 1996 22. Mori N, Yatabe Y, Oka K, et al: Expression of perforin in nasal lymphoma. Additional evidence of its natural killer cell derivation. AmJ Patho1149:699-705, 1996 23. Jaffe ES, Chan JKC, Su IJ, et al: Report of the workshop on nasal and related extranodal angiocentric T/natural killer cell lymphomas. Definitions, differential diagnosis and epidemiology. Am J Surg Patho120:103-111, 1996 24. Ng CS, Lo STH, Chan JKC, et al: CD56 + putative natural killer cell lyanphomas: Production of cytolytic effectors and related proteins mediating tumor cell apoptosis? HUM PATHOL 28:1276-1282, 1997 25. Ho FCS, Choy D, Loke SL: Polymorphic reticulosis and conventional lyrnphomas of the nose and upper aerodigestive tract: A clinicopathologic study of 70 cases, and immunophenotypic studies of 16 cases. HUM PATHOL21 :1041-1050, 1990 26. Sato K, Kimura F, Nakamura Y, et al: An aggressive nasal lymphoma accompanied by high levels of soluble Fas ligand. Br J Haemato194:379-382, 1996 27. Tanaka M, Suda T, Haze K, et al: Fas ligand in human serum. Nat Med 2:317-322, 1996 28. Ohshima K, SuzumiyzJ, Shimazaki K, et al: Nasal T/NK cell lymphomas commonly express perforin and Fas ligand: Important mediators of tissue damage. Histopathology 31:444-450, 1997 29. Teruya-Feldstein J, Jaffe ES, Burd PR, et al: The role of Mig, the monkine induced by interferin-~ and IP-10, the interferon-'yinducible protein-10, in tissue necrosis and vascuIar damage associated with Epstein-Barr virus-positive lymphoproliferative disease. Blood 90:4099-4105, 1997 30. The Non-Hodgkin's Lymphoma Classification Project: A clinical evaluation of the International Lymphoma Study Group classification of non-Hodgkin's lymphoma. Blood 89:3909-3918, 1997 31. Cheung MMC, Chan JKC, Lau WH, et aI: Primary nonHodgkin's lymphoma of the nose and nasopharynx: Clinical features, tumor immunophenotype and treatment outcome in 113 patients. J Clin Onco116:70-77, 1998

9. Niehans GA, Brunner T, Frizelle SP, et al: Human lung carcinomas express Fas ligand. Cancer Res 57:1007-.1012, 1997 10. Sneller MC, Wang J, Dale JK, et al: Clinical, immunologic, and genetic features of an autoimmune lymphoprolferative syndrome associated with abnormal lymphocyte apoptosis. Blood 89:1341-1348, 1997 11. Fisher A, Cavazzana-Calvo M, De Saint Basile G, et al: Naturally occurring primary deficiencies of the immune system. Annu Rev Immuno115:93-124, 1997 12. Robertson MJ, Ritz J: AA7 Cluster report: CD95, in Schlossman SF, Boumsell L, Gilks W, et al (eds): Leucocyte Typing V. White Cell Differentiation Antigens. Proceedings of the Fifth International Workshop and Conference. Vol. 1 Boston, MA. Oxford University Press, 1995, pp 1142-1143 13. Kondo E, Yoshino T, Yamadori I, et al: Expression of bcl-2 protein and Fas antigen in non-Hodgkin's lymphomas. Am J Pathol 145:330-337, 1994 14. Aftabuddin M, Yamadori I, Yoshino T, et al: Correlation between the number of apoptotic ceils and expression of the apoptosis-related antigens Fas, LeY and bcl-2 protein in nonHodgkin's lymphomas. PathoI Int 45:422-429, 1995 15. Carbuccia LX, Parc P, HassounJ, et al: Frequent expression of Fas/APO-1 in Hodgkin's disease and anaplastic lymphomas. Histopathology 27:235-241, 1995 16. Leith/iuser F, DheinJ, Mechtersheimer G, et al: Constitutive and induced expression ofAPO-1, a new member of the nerve growth factor/tumor necrosis factor receptors superfamily, in normal and neoplastic cells. Lab Invest 69:415-429, 1993 17. Ng CS, Chan JKC, Lo STH: Expression of natural killer cell markers in non-Hodgkin's lymphomas. HuM PATHOL 18:1257-1262, 1987 18. Wong KF, Chan JKC, Ng CS, et al: CD56 (NKH1)-positive hematolymphoid malignancies: An aggressive neoplasm featuring frequent cutaneous/mucosal involvement, cytoplasmic azuzophilic granules, and angiocentricity. HuM PATHOL23:798-804, 1992 19. Nakumura S, Suchi T, Koshikawa T, et al: Clinicopathologic study of CD56 (NCAM) positive angiocentric lymphoma occuriring in sites other than the upper and lower respiratory tract. Am J Surg Pathol 19:284-296, 1995 20. Chan JKC, Sin VC, Wong KF, et al: Non-nasal lymphoma expressing the natural killer cell marker CD56: A clinicopathologic

53