Lymphocyte subsets in tumour of patients with undifferentiated nasopharyngeal carcinoma: presence of lymphocytes with the phenotype of activated T cells.

We have analyzed lymphocytes infiltrating nasopharyngeal carcinomas, using a combination of immunoperoxidase staining of frozen and paraffin-embedded sections, and immunofluorescence on lymphocyte suspensions recovered from teased tumours. A panel of monoclonal antibodies was used to define lymphocytic subsets on frozen sections of 14 different tumours. The vast majority of peri- and intra-tumoral lymphocytes were stained by OKT3 antibody. In 8 sections, T4 positive cells were largely predominant, while T8 positive cells were the majority in three sections. Twenty-nine paraffin-embedded sections from other NPC patients stained with HNK-1 antibody showed a variable percentage of positive cells reaching 6 to 15% in nine patients. Most HNK-1 positive cells had the morphology of large granular lymphocytes typical of natural killer cells. Double staining experiments on lymphocytes isolated from 7 tumours revealed a constant presence of T3 positive, HLA-DR positive lymphocytes (from 6 to 29% of mononuclear cells), and of lymphocytes coexpressing the T3 and the Tac (IL-2 receptor) antigens (from 5 to 12% of mononuclear cells). Lymphocytes with a phenotype of activated T-cells are thus constantly found in NPC tumours.

Undifferentiated carcinoma of the nasopharyngeal type Material and methods (UCNT) is an intriguing tumour. Although rare in most parts of the world, it occurs with a strikingly high incidence Patients and tissue samples in South East Asia, and is frequent in North Africa and Fresh tumour biopsy specimens were obtained at the time arctic regions (De The, 1980). Serological and DNA hybridi-of diagnosis from 14 untreated patients with UCNT, most zation studies have shown its association with the Epsteinof them originating from North Africa. Diagnosis of Barr virus (EBV): A close relationship exists between the UCNT was performed by the same pathologist (C.M.) and humoral immune response to EBV antigens and the clinical based upon histological criteria previously described course of the disease (De The et al., 1975; (Shanmugaratnam et al., 1979). Slides were frozen for 1973). Serum IgA antibodies to the viral capsid antigen immunoperoxidase staining. When allowed by the size of the (VCA) and to early antigens (EA), a typical feature in biops rt was songen ased by mphocyte nasopharyngeal carcinoma (NPC) (Henle et al., 1976), can biopsy, a part was also gently teased for lymphocyte be used for early diagnosis of the tumour and for detecting recovery on Ficoll cushion in seven cases. In addition individuals at high risk . EBV-specific paraffin-embedded specimens from 29 earlier-diagn od cell-mediated immunity has been less investigated but UCNT patients were retrospectively collected for immuno antibody-dependent cell-mediated cytotoxicity (ADCC) has peroxidase staining with monoclonal antibody HNK 1 been clearly shown to be related to the course and prognosis Antibodies of the disease (Pearson et al., 1978).
NPC cells express EBV-associated nuclear antigen EBNA Conjugated antisera were from Dako Laboratories, and contain the EBV genome (Wolf et al., 1973;Zur Hausen Denmark. Monoclonal antibodies used included OKT3 to et al., 1970). However, because no EBV receptor has been the T3 pan-T-cell antigen, OKT4 to the helper/inducer T-cell demonstrated at the surface of any human epithelial cell so subset, OKT8 to the suppressor/cytotoxic subpopulation, far, the mechanism of EBV penetration into nasopharyngeal OKTIO that detects activated T-cells and haematopoietic cells remains a matter of debate (discussed in Rickinson, progenitors (Hercend et al., 1981;Reinherz et al., 1979), all 1984& Wolf, 1984. Furthermore, the role of the virus in the from Ortho Diagnostics. The Ll/l/12 hybridoma producing carcinogenic process is essentially still not understood.
an anti-HLA-DR monoclonal antibody (Kalil et al., 1982) Typical UCNT histopathologic features consist of a nexus was kindly supplied by George Khalil (Hopital St. Louis, of large epithelial tumour cells and numerous lymphocytes Paris). The HNK-1-producing hybridoma (Abo et al., 1981) ( Shanmugaratnam et al., 1979). The lymphoid cells within was purchased from the American Type Culture Collection the tumour are cytologically normal, do not contain EBV (Rockville, MD). Anti-Tac monoclonal antibody, that and have been shown to belong largely to the T-lineage detects the IL-2 receptor (Uchiyama et al., 1981;Wakasugi (Gallili et al., 1980). Whether these lymphocytes are et al., 1985) was kindly provided by Dr. T. Waldmann. They remnants of the cells present in the normal nasopharyngeal were used as culture supernatants or as purified Ig diluted mucosa or rather indicators of a local immune reaction to appropriately. the malignant cells might be a keypoint in the understanding of this nexus of epithelial and Iymphocytic cells.
Section immunoperoxidase staining In order to characterize the phenotype of these infiltrating Inietmuoprxds sanngws efred n lymphocytes, we have stained frozen and paraffin-embedded frzndisectiomnsoitealronoclona saintibodies pexorept HnK sections of NPC as well as lymphocytes isolated from teased foe etoswt l oolnlatbde xetHK tumours, using a panel of monoclonal antibodies.
tumour specimens fixed in acetone, stored at -70°C then Received 17 June 1986; and in revised form, 30 September 1986. rehydrated in PBS. Paraffin sections were treated with xylene, rehydrated with ethanol and water and immersed in V " .. Tris HCI buffer for 10min. Endogenous peroxidase was _ blocked by incubation with 0.3% H20. Non-specific reactions were prevented by incubation with normal rabbit serum. Sections were then incubated for 30min with staining antibody (diluted 1:100 for OKT antibodies or 1:10 for anti-HLA-DR antibody) washed in cold PBS, then incubated for 30min with a peroxidase-conjugated rabbit anti-mouse Ig antiserum (diluted 1:100). The reactions were revealed with -H202 and 3-amino-9-ethyl-carbazol or diaminobenzidine. Finally slides were counterstained with haematoxylin for : cytological examination. Detection of Ig-bearing cells was carried out by the same method using a rabbit anti-human Ig antiserum (diluted 1:100) followed by a peroxidaseconjugated swine anti-rabbit Ig antiserum (diluted 1:100) as the developing reagent. The estimation of the number of positive cells could only be semi-quantitative on frozen Figure  Mononuclear cells from dissociated tumours were isolated by Ficoll centrifugation. Indirect immunofluorescence assays were performed according to standard protocols: First step antibody staining was revealed after washing in PBS with a fluorescein or rhodamine-conjugated goat anti-mouse Ig antiserum. Double staining assays were carried out with IgG 1 in the first step revealed by fluorescein-conjugated goat anti-mouse IgG antiserum, and anti-HLA-DR monoclonal antibody, of IgM isotype, in the second step, developed with rhodamine conjugated, goat anti-mouse IgM antiserum. located, appeared to surround the tumour masses (Figure 3). In only 3 patients were OKT8 + cells more numerous. In the last three patients, OKT4+ and OKT8 + cells were present To estimate the percentage of natural killer cells present in in approximately equal numbers. In six speciments tested the lymphocytic infiltrates, 29 paraffin-embedded additional with the OKT1O monoclonal antibody, a varying number of specimens were stained with the HNK-1 monoclonal antipositive cells were constantly detected with reactivity ranging body (Table II). In all patients the percentage of HNK-1 + from + to + ++. HLA-DR positive lymphocytes were cells was below 15% of total lymphocytes. Approximately present in all specimens from + up to + + ++. In some two-thirds of the sections showed no or very few scattered specimens clear staining of tumour epithelial cells by the stained cells. When more numerous, HNK-1 positive cells anti-HLA-DR antibody was apparent (results not shown).
were located predominantly within the tumour masses. Most  (Figure 4) typical of human frozen sections (grading explained in the text) NK cells (Saksela et al., 1979). No study on membrane Tassociated antigens could be performed on these paraffin- In each case tested with both OKT4 and OKT8 antibodies, 11-15 4 the DR-positive T-cell population and the Tac-positive T-cell population included both T4 and T8 subsets in roughly equal proportion (data not shown).
No evident relationship could be found on this small population of patients when these data were analyzed with respect to the stage of the disease, lymphoproliferative *~~r esponses to PHA or ConA mitogens (unpublished data), titres of antibodies to EBY-associated antigens or HLA phenotypes of the patients (Herait et al., 1983; and data not > Discussion We have undertaken a phenotypic analysis of lymphocytes present in UCNT by a combination of immunohistological staining of frozen and paraffin sections of tumour biopsies -~and of immunofluorescence assays on lymphocytic cells isolated from tumour specimens. Most of the lymphoid infiltrate was found to comprise lymphocytes expressing the T3 antigen characteristic of the T-cell lineage. Among these T-lymphocytes, the distribution  T4 positive cells in most patients, T8 positive lymphocytes tended to be predominantly located around tumour cell masses in some patients. The T4/T8 ratio was not strongly correlated with any clinical or biological feature. Because this phenotypically-defined T8 positive subset is known to include the functional population of cytotoxic cells, it is tempting to speculate that the T8 expressing lymphocytes found in close relationship with the malignant cells might play a role in the immune reaction to the tumour. In this regard, it was striking to observe that a relatively highalthough varying from one tumour to anotherproportion of T3 positive cells also expressed HLA-DR molecules as detected in double staining assays. In fact 54% (range 12%-90%) of T-cells expressed HLA-DR antigens. This was in agreement with the observation of more numerous HLA-DR positive than sIg bearing cells in lymphocytes isolated from the tumour. The TIO antigen which is also found on activated T-cells (Hercend et al., 1981) was also detected in all tumours tested for this marker.
Whether the presence of activated T-cell associated antigens was restricted to the subpopulation expressing the T8 antigen or was rather shared by T4 positive cells as well as suggested by the preliminary results will be a matter for further study.
The reactivity of intra-UCNT lymphocytes was also studied with a monoclonal antibody directed to the receptor for interleukin-2 (IL-2) (Tac antigen) (Uchiyama et al., 1981), which is only expressed on activated T-cells. Again, in all patients studied we found that a strikingly high number of T-cells (17-28%) although less numerous that T3-HLA-DR positive cells expressed the IL-2 receptor. In addition we have performed bulk cultures of these T-lymphocytes which are easily maintained in IL-2 containing medium. Clones derived from these populations will be necessary to analyze their potentially specific immune functions.
The mechanism of T-cell activation remain to be investigated. Thomas et al. (1984) reported the expression of HLA-DR antigens on NPC epithelial cells by immunoperoxidase labeling. We confirmed this finding in the course of this study. Furthermore we have recently found very high MHCclass II antigen expression on malignant EBNA-positive epithelial cells from nude mice-grown NPC tumours. These malignant NPC cells were shown to produce a monokine able to activate T-cell responses in the presence of mitogens, and probably similar to monocyte-derived interleukin-I (to be reported elsewhere). NPC cells could thus share several characteristics with accessory cells able to induce T-cell activation.
EBNA is the only serologically-defined EBV-associated antigen constantly detected in malignant NPC cells. In infectious mononucleosis (IM), a benign disease caused by a polyclonal activation of B-lymphocytes stimulated by EBV, a cytotoxic immune response is based upon NK and T-cells (Lipinski et al., 1979). That NK cells, as recognized by the HNK-1 monoclonal antibody, are present in UCNT tumours, might be of relevance for a potential anti-tumour activity in situ. In this context, EBV-specific T-cells could also play a role (Rickinson et al., 1980), despite the depression of anti-EBV T-cell-mediated immunity observed elsewhere (Moss et al., 1983) in the peripheral blood of NPC patients using the regression of EBV transformation assay. The target structure of EBV-specific cytotoxic T-cells is the LYDMA antigen that has not been serologically defined but whose location in the EBV genome has recently been suggested (Hennesy et al., 1984). It will be interesting to find out whether UCNT cells can be killed by LYDMA-directed T-cells, and whether lymphocytes isolated from UCNT can give rise to LYDMA-specific cytotoxic T-cell clones.
In conclusion, we have reported that T-cells infiltrating UCNT demonstrate interesting phenotypic features: (i) large variation in the T4/T8 ratio, with usually an excess of T4 positive cells, but with a few striking exceptions. (ii) detection of HNK-1 positive lymphocytes, often with the morphology of LGL, sometimes in high numbers (6 to 15% of the lymphocytes in 9/29 patients), suggesting a role for NK cells in the local defence against UCNT: (iii) constant presence of T-cells expressing HLA-DR antigens, the IL-2 receptor and the TIO antigen, and therefore with the phenotype of activated T-cells.
These data suggest the existence of local immunological reactions in UCNT, involving both T-and NK cells. Further studies will be needed to elucidate their precise nature and their possible role in the control of the disease. Whether the immunological features described in this report can be related with the clinical grades and/or the prognosis of UCNT will be investigated on a large population of patients.