Clin. exp. Immunol. (1991) 85, 418-423
ADONIS
0009910491002531
CD5 + B cells and naturally occurring autoantibodies in cancer patients R. STEIN, I. P. WITZ*, J. OVADIAt, D. M. GOLDENBERG & I. YRON* Garden State Cancer Center and Center for Molecular Medicine and Immunology, Newark, NJ, USA, *Department of Microbiology and the Moise and Frida Eskenasy Institute for Cancer Research, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, and tDepartment of Obstetrics and Gynaecology, Beilinson Medical Centre, Sackler School of Medicine, Petah Tikva, Israel
(Accepted for publication 25 March 1991)
SUMMARY We have determined the percentage of CD5+ B lymphocytes in the peripheral blood of cancer patients and healthy controls, using antibodies directed at the CD5 and CDl9 (pan-B) markers. The frequencies of CD5+ B cells, expressed as a percentage of total B cells, ranged from 14-3 to 57.50/) in the controls and from 14 8 to 82-8% in the patient population. One-third of the cancer patients had frequencies > 2 s.d. above the mean of the control population. The CD5 + B cell fraction expressed as a percentage of total lymphocytes was also significantly elevated in this group of cancer patients. These results suggest that the CD5+ B cell compartment may be affected by the malignant process or by the therapy modality employed. The plasma levels of several naturally occurring autoantibodies, the products of the CD5+ B cells, were also assessed in cancer patients and controls. No significant differences were observed when reactivity to several autoantigens was measured. These included nuclear components and phospholipids.
Keywords CD5+ B cells autoantibodies
cancer
INTRODUCTION Autoimmunity and the development of non-lymphoid malignancies may be linked by several independent mechanisms. Immune responses against autoantigens may follow the induction of immunity against tumour-associated epitopes if antibodies induced by tumour-associated epitopes cross-react with epitopes expressed on normal cells, or if the expression of the tumour-specific epitopes is in close enough proximity with autoepitopes to cause the termination of tolerance to the autoepitopes. We have supported the occurrence of such a mechanism in a previous study (Witz et al., 1988) which demonstrated that mice specifically immunized against the polyoma virus TSTA also formed autoantibodies against T cells. A second mechanism by which malignancy and autoimmunity may be connected involves the CD5+ B cell subset and the products of these cells. Although in humans cells with this phenotype were first noted as the cell involved in chronic lymphocytic leukaemia, B cells expressing the CD5 T cell marker (CD5+ B cells) have been shown during the last decade to be normal constituents of the immune system of healthy
individuals (reviews by Herzenberg et al., 1986; Casali & Notkins, 1989; Kipps, 1989). In many cases, the immunoglobulin products of CD5 + B cells consist of autoantibodies which have been referred to as naturally occurring autoantibodies (NOA) (Dighiero, Guilbert & Avrameas, 1982; Casali et al., 1989; Kipps, 1989). NOA are typically present in apparently healthy humans and animals that have not been intentionally immunized (reviews by Casali et al., 1989; Avrameas et al., 1987). Among the autoantigens recognized by NOA are the Fc portion of IgG, ssDNA, cryptic epitopes on erythrocytes, several internal cellular proteins, and membrane constituents of nucleated cells, including lymphocytes. A striking characteristic of many NOA is their polyreactivity (Casali et al., 1989). Thus, unlike conventional antibodies, NOA may react with multiple unrelated antigens, such as non-self antigens, in addition to autoantigens. In vitro studies show that NOA may react with various tumour cells (review by Witz & Agassy-Cahalon, 1987). The tumour reactivity is most probably a manifestation of the polyreactivity of NOA rather than a specific reactivity against tumour-associated antigens. Since NOA are able to react with tumour cells and certain normal cells, such as lymphocytes, or cytokine products of normal cells, these antibodies may, under certain circumstances, influence the progression and proliferation of tumour cells. To date, the evidence to support this
Correspondence: Rhona Stein, PhD, Garden State Cancer Center and Center for Molecular Medicine and Immunology, One Bruce Street, Newark, NJ 07103, USA.
418
CD5 B cells in cancer possibility has come exclusively from murine systems, where it was shown that NOA control the development of certain tumours. NOA have been shown to contribute towards heightened resistance against tumour growth (Chow & Chan, 1987; Chow & Bennet, 1989; Tough & Chow, 1988), and to cause a change (either retardation or enhancement) in tumour development when administered to tumour-bearing mice (AgassyCahalon et al., 1988; Gonen et al., 1988). In contrast to animal models, there are essentially no data available to assign a biological role for NOA or NOAproducing CD5+ B cells in the development or progression of cancer in humans. As a first step toward addressing this question, we compared the levels of the CD5 + B cell subset and the levels of several NOA between cancer patients and a healthy control population, and described quantitative alterations which occur in the patients. SUBJECTS AND METHODS Patients and controls Two independent patient and control groups were studied. The patients belonging to the first group are those referred to the Center for Molecular Medicine and Immunology (CMMI) in Newark, NJ, either for radioimmunodetection or for radioimmunotherapy (Goldenberg, 1988). These patients are characterized by tumours that produce antigens for which the CMMI has developed monoclonal antibodies, e.g. carcinoembryonic antigen, alpha-fetoprotein, human chorionic gonadotropin, prostatic acid phosphatase, colon-specific antigen-p (Goldenberg, 1989). They represent an in-house group of readily accessible, carefully studied patients. The patient population consisted of 26 women (age range 3-79 years, mean 56-2 + 15-5) and 27 men (age range 24-81 years, mean 60 6+ 14.6). Controls for this patient population were CMMI employees: 11 women (age range 20-65 years, mean 39-2 + 14-2) and 14 men (age range 21-68 years, mean 35-8 + 14-9). This group of patients and controls is referred to hereafter as the NJ group. The diagnosis of the primary tumour in 51 out of the 53 NJ cancer patients studied is as follows: 23 colon, three rectal, two caecum, three liver, two pancreas, four breast, four lung, two prostate and one of each stomach, duodenum, sigmoid/colon, testes, thyroid, lymphoma, and teratoma. One patient had both breast and pancreatic primaries. The diagnosis of two primary tumours was unknown. All patients were treated before being tested in this study by surgery, chemotherapy and/or irradiation. The second patient and control group has been described previously (Yron et al., 1986) and is referred to as the Tel Aviv group. This group included 26 women with stage I endometrial carcinoma (EC), prior to treatment; 32 women with hyperplasia of the endometrium, including women with adenomatous and atypical hyperplasia, the pre-EC stages. All women included in this group were over 45 years of age; and 29 healthy women, over 45 years of age, either with no history of major gynaecological problems or with post-menopausal bleeding, and diagnosed as disease free. Collection of blood and plasma samples Whole blood from patients and controls was collected into EDTA vacutainer tubes. After centrifugation at 300 g for 10 min, the plasma, buffy coat, and packed erythrocytes were
separated. Erythrocytes remaining in the buffy
419 coat fraction
were lysed with 20 volumes of lysing solution (155 mm NH4CI, 10 mm KHCO3, 0-1 1 mm EDTA). The white blood cells (WBC) were washed with phosphate-buffered saline (PBS) and then
labelled with fluorescent conjugates of monoclonal antibodies directed against the antigens under study. Flow cytometry of lymphocytes For labelling, WBC from 5 ml of blood were resuspended in 1 5 ml of PBS with 0-1% NaN3 and 1% horse serum. Routinely, five tests were run on each patient's specimen as follows: (i) no labelled antibody (PBS only); (ii) isotype-matched irrelevant mouse immunoglobulins labelled with PE and FITC; (iii) PEconjugated anti-Leu 1 (CD5); (iv) FITC-conjugated anti-Leu 12 (CD19); and (v) PE-anti-Leu 1 and FITC-anti-Leu 12 Washed WBC (60 ,ul) were added to the tubes containing the antibodies, incubated for 30 min in an ice-water bath, washed twice with PBS containing NaN3 and horse serum, and then analysed by flow cytometry using a FACScan flow cytometer (Becton Dickinson Immunocytometry Systems, Mountain
View, CA). Using lymphocyte gating to exclude monocytes and granulocytes, 10 000 cells were analysed for fluorescence. By quadrant analysis on a correlated two-parameter display (PE-anti-Leu I versus FITC-anti-Leu 12), cells in each of the following classifications were quantified: (i) unlabelled, null cells; (ii) cells displaying red fluorescence only, T cells; (iii) cells displaying green fluorescence only, CD5- B cells; and (iv) cells displaying both red and green fluorescence, CD5+ B cells. Determination of circulating autoantibody levels Levels of autoantibodies directed against the following antigens were assessed: (i) nuclear components (Sm; snRNP; Sjogren's syndrome-related antigens (SS-B/La); dsDNA; DNP); (ii) M-2 mitochondrial antigen (pyruvate dehydrogenase (PDH); and
(iii) phospholipids (cardiolipin; phosphatidyl-L-serine). Detection was performed using ELISA kits purchased from BioHyTech (Ramat Gan, Israel), and performed according to the manufacturer's instructions as described elesewhere (Firer et
al., 1988, Zurgil et al., 1988). Briefly, the antigens were each coated optimally to wells of Nunc (Roskilde, Denmark) polystyrene microtitre plates. Serum samples were diluted 1/200 in assay diluent, added to the wells and incubated for 30 min at 37°C. After washing the plates with PBS-Tween 20 wash buffer, (phospholipid buffer did not contain Tween), and a trivalent conjugate containing a mixture of alkaline-phosphatase-conjugated goat anti-human IgG, IgM and IgA (each optimally diluted for the particular antigen; dilution range 1/1000-1/ 3000), was added to the wells and the plates were incubated for a further 30 min at 37°C. The plates were washed again and diethanolamine substrate buffer containing p-nitrophenylphosphate and MgCl2 was added. After 60 min at 37°C, the optical density in each well at 405 nm was read using a Titertek ELISA reader. Standards provided in BioHyTech kits or known positive and negative sera, as well as a reagent blank, were included in each plate. Where autoantibody isotypes were assessed (as was the case in the cardiolipin assays), assays were performed as described above, except that plates were incubated separately with antiIgG or anti-IgM conjugates.
420
R. Stein et al. loor-
0 0
801-
0 0o
g8-08 o8o
601-
0
0
0-
08o
cl
1*t
40 F 0
40S.
00 060
20
F
000
*l00
00
to
00
Cancer
0
Cancer
Control~~~~~~~~~~~~~~~~~
Control
patients
Fig. 1. CD5+ B cells as a percentage of B cells (0, cancer patients; 0, healthy control individuals). Mean values are indicated by horizontal bars for each population.
Optical densities
converted to arbitrary autoantibody curve of a high-titred included in each plate. Cut-off values for provided by BioHyTech, and represent the were
U/ml by comparison with the dilution
positive serum positivity were mean + 3 s.d. of antibody levels (U/ml) obtained from 210 blood bank control sera. Autoantibody activities in these sera showed a normal distribution (data not shown). All experimental sera with autoantibody levels higher than the cut-off value were considered to be positive. Statistical analysis One-way analysis of variance, the Mann-Whitney U-test, and x2 analysis were used to assess the significance of the results of the autoantibody determinations. Student's t-test was used to assess significance of the results in the lymphocyte subset analyses. RESULTS
CDS+ B cells in the peripheral blood of cancer patients To investigate whether CD5 + B cells are altered in cancer patients, we surveyed the levels of lymphocyte subpopulations in the peripheral blood of cancer patients and healthy controls. In these experiments, peripheral blood cells of the NJ groups were analysed. The variability in the percentage values of these lymphocyte subpopulations between individual cancer patients somewhat greater than within the control group. However, the mean values of T, B, null and CD5+ B cells from cancer patients and controls were not statistically different. B cell range was 0 9-32-5% (mean 8-4+7-6%) and 2-8-18-2% (mean 9.4+4.5%) in the cancer patient and control populations, respectively. The percentages of CD5+ B (CD5+, CDl 9+) cells (calculated as a percentage of total circulating lymphocytes) were 0-3-26-0% (mean 4-1+49%) and 0-7-6-9% (mean
was
3-0 + 1 6%) in the peripheral blood of cancer patients and of normal controls, respectively. In addition to the above calculation of CD5+ B cells as a percent of total lymphocytes, the percentage of B cells that express the CD5 molecule was also determined as a per cent of the total B cell population. Figure I is a scattergram of individual values of the cancer and the control groups. The mean value + s.d. of the CD5 + B cells (expressed as a percentage of total B cells) of the control group is 33 5 + 12 2, and that of the cancer patient group is 452 + 18 2. The difference in these values is statistically significant (P < 0-0 1). The ranges are 14-357 5% for the control group and 14-8-82-8% for patient population. In order to assess further the cancer patients who express elevated CD5+ B cell levels, we divided the patient group into two populations: those of the tested cancer patients with higher than normal levels of CD5+ B cells in their peripheral blood were separated from the remainder of the patient population. Higher than normal levels of CD5+ B cells (as percentage of total B cells) were defined as values greater than the mean+2 s.d. of the normal group. According to the results presented above, this value is 57 9%. Employing this setting, it was determined that none of the 25 control subjects had an 'above normal' value of CD5+ B cells. However, 30% of the cancer patients (16 out of 53) showed 'above normal' values of CD5+ B cells (as percentage of total B cells) in peripheral blood. Table 1 lists the characteristics of the patients with elevated CD5+ B cells (as percentage of total B cells). This population represents a heterogeneous group in terms of tumour type, stage, age, sex, and prior treatment. The fractions of T, B, null and CD5+ B cells (as percentage of total lymphocytes) in the peripheral blood of these two patient populations are compared in Table 2. The results show that neither division of cancer patients differed significantly from the control population in levels of T, B, or null cells. However, the group of patients with elevated values of CD5+ B cells, expressed as a percentage of total B cells, also had a significantly increased proportion of CD5 + B cells expressed as percentage of total lymphocytes (P< 005). This is in contrast to the cancer group as a whole, as described above. Since the mean age of the control group was lower than that of the cancer patient group, we examined whether advanced age correlated positively with higher levels of CD5+ B cells. The levels of CD5+ B cells (as percentage of total B cells) of individuals in the control and in the cancer group as a function of their age were calculated and no positive correlation between these parameters could be detected (the correlation coefficients were -0065 and -0265 for the control and cancer patient populations, respectively). Furthermore, the selected subpopulation of cancer patients with higher than normal levels of CD5 + B cells did not belong, as a group, to the older population. Autoantibody levels in cancer patients The NJ group. Since CD5+ B cells have been shown to be autoantibody producers, we measured the levels of autoantibodies in the plasma of patients with a high frequency as well as in patients with a low frequency of CD5+ B cells. The healthy individuals mentioned above served as controls. The levels of antibodies against three categories of antigens, nuclear components, mitochondrial PDH and cardiolipin, are shown in Table 3. No difference in reactivity with these antigens was
CD5+ B cells in cancer
421
Table 1. Characteristics of patients with high counts of CD5+ B cells
Patient no. 799 967 1014 1058 1075 1077 1080 1101 1103 1116 1124 1134 1135 1136 1146 1148
Stage at diagnosis
Cancer diagnosis
Testicular Sigmoid colon Breast/pancreatic Lung Rectal Non-Hodgkin's lymphoma Breast Colon Prostate Stomach Colon Hepatoblastoma Colon Hepatocellular Rectal Colon
Stage at study
Age/ sex
Prior treatment
CD5 + B cells (0%, of total)
D D D D D III D D NED D
32/M 70/M 67/F 41/F 72/M 49/M 55/F 53/M 76/M 51/M 42/M 3/F 75/M 64/F 51/M 68/F
Surgery, chemotherapy Surgery, chemotherapy Surgery Surgery, XRT, chemotherapy Surgery Chemotherapy Surgery, XRT, chemotherapy Surgery, XRT, chemotherapy Surgery Surgery, XRT, chemotherapy Surgery, XRT, chemotherapy Surgery, chemotherapy Surgery Surgery, chemotherapy Surgery, XRT, chemotherapy Surgery, XRT, chemotherapy
65 9 67-2 67-3 80-0 62-0 62 5 61-7 82-8 70 5 63 7 58-1 73 9 66-3 71 5 64.3 64-6
III
Dukes C-I II III Dukes A III II Dukes C-II D2 T2, N2, M l Dukes C-II I Dukes Bl I 0-I Dukes D
D NED D L D D
Stages: D, distant disease; NED, no evidence of disease; L, local; Breast, UICC classification with lymph node criteria; Colon, standard Dukes classification, with C-I, involvement of 1-4 lymph nodes and C-II,5 or more lymph nodes; Lymphoma, Ann Arbor classification; all others according to standard clinical staging. XRT, external radiotherapy Table 2. Lymphocyte subpopulations in peripheral blood of cancer patients (NJ group) with high or normal counts of CD5+ B cells
Normal CD5'
(n=37)* Lymphocyte subpopulation T cells B cells Null cells CD5+ B cells
High CD51 (n= 16)t
Mean + s.d.
Range
Mean + s.d.
Range
72-5 + 12 6
25 7-92-9 0 9-32-3 6-2-57 8 03-12-3
65 2+ 14 6 9 5+9 2 24-1 + 14 3 6-8+7-2
36-1-86-3 1-4-32-5 76-60 8 09-260
8-0+±70 19-2+ 11-0 31+3-2
group of age-matched healthy women (data not shown). Similarly, the proportion of autoantibody-positive individuals was similar in the three groups. It should be noted that the anti-DNP autoantibody levels of the EC and hyperplasia patients of the Tel Aviv group were higher than those of the NJ patients. However, this difference was not statistically significant. Autoantibodies to DNP appear
in high titres in patients with autoimmune rheumatic disease,
Cancer patients with control frequency of peripheral CD5+ B cells. t Cancer patients with frequency of peripheral CD5 + B cells higher than the mean + 2 s.d. above control level. *
observed between the three groups. Similarly, the proportion of autoantibody-positive individuals in the three groups was similar. The levels of the mitochondrial autoantibodies, antiPDH, were found to be somewhat (not significantly) higher in the NJ patients. High titres of such autoantibodies characterize autoimmune primary biliary cirrhosis (Mackay & Gershwin, 1989). Thus, the slightly higher levels of such antibodies in certain NJ patients may suggest hepatic involvement in their disease, although no direct liver function tests were performed. The Tel Aviv group. Based on the observation that autoantibody production can be enhanced by oestrogens (Ahmed et al., 1989), a second group of patients was also included in this study. Autoantibody activity was measured in the plasma of patients with EC or hyperplasia of the endometrium, situations which are highly correlated with longer term exposure to oestrogens unopposed by progesterone (Henderson et al., 1982). No difference in autoantibody levels directed against any of the antigens tested could be observed between the patients and the
most of whom are women (Tan, 1989). It is possible,
therefore,
that the production of such antibodies is enhanced in oestrogenrelated diseases.
DISCUSSION B cells expressing the CD5 T cell marker and the naturally occurring autoantibody products of these cells have been identified in mice and humans only within the last decade. Evidence obtained in several laboratories, including our own, has shown a relationship between NOA and tumourigenesis in several murine systems (Witz & Agassy-Cahalon, 1987). Using flow cytometry, we have surveyed the levels of CD5+ B cells in the peripheral blood of cancer patients and healthy controls. One-third of the cancer patients had frequencies more than 2 s.d. above the mean of the control populations when expressed as a percentage of B cells. The CD5 + B cell fraction expressed as a percentage of total lymphocytes was also significantly elevated in this group of cancer patients. This increase may be of biological significance, especially in view of the fact that in this group of patients the CD5+ B cell lymphocyte subpopulation was the only one to be quantitatively altered when compared with normal controls. No significant deviation from control
values was detected in the other parameters that were measured. To our knowledge, this is the first report of this observation. The significance of this finding is unclear at present. Thus far, we have not found any correlation between a high CD5 + B cell count and clinical parameters, such as the type of the
422
R. Stein et al. Table 3. Levels of autoantibodies in the NJ group Normal CD5+t
High CD5+t
Healthy controls
Cut-off value*
Mean + s.d.
H/T
Mean + s.d.
H/T
Mean + s.d.
H/T
310 280 250 275 250
152+74 246+ 307 189+326 191 +266 230+85
1/40 7/40 5/40 3/40 4/40
146+80 298 + 392 181 +351 268+423 258+ 144
0/16 4/16 1/14 3/16 4/16
158+87 224+ 274 98+57 150+79 193+68
0/19 2/19 0/16 1/19 1/18
Cardiolipin§ IgG 320 IgM 320
166±58 188+ 102
1/40 3/40
151 +47 151 +74
0/16 1/16
132+21 157+57
0/19 1/19
Antigen DNP RNP Sm SSb PDH
*
Cut-off values represent the *Means (U/ml +3 s.d.) of 210 blood bank control sera.
t Plasma from cancer patients with control frequency of peripheral CD5 + B cells. Values are arbitrary units (U/ml), calculated in reference to a known high titered positive serum, of all individuals included in the group. T Plasma from cancer patients with frequency of peripheral CD5 + B cells higher than the mean+ 2 s.d. above control level. § Cardiolipin autoantibody isotypes were assessed separately for IgG and IgM. H/T (high/total), the ratio between the number of sera in which the levels of antibodies were above cut-off values and the total number of individuals included in the group.
primary cancer, stage of disease, or previous therapy. However, one may make several assumptions which could serve as a working hypothesis for future studies. One of these assumptions is that chemotherapy or external radiotherapy in themselves were not responsible for the elevated levels of CD5 + B cells where detected. This assumption stems from the fact that four out of the 16 patients in whom high levels of such cells were detected had only surgery as prior treatment (Table 1). Obviously, in future studies one should attempt to analyse patients with more uniform diagnosis and prior to any form of therapy, and should include an analysis of CD5+ B cells in secondary lymphoid tissues, such as draining lymph nodes, of untreated patients since the bloodstream is only a transit site. One interesting observation was that autoantibodies were not increased even in those NJ cancer patients exhibiting a higher than normal level of CD5+ B cells. It is possible that autoantibodies other than those tested in our study may have been altered concomitantly with the increase in the levels of CD5+ B cells. For example, the CD5+ B cells in those patients whose numbers are increased could be recognizing glycolipid tumour-associated antigens. Patients with EC or with preneoplastic stages of the disease (the Tel Aviv group) were included in the NOA study since these conditions are known to be highly related to long-term exposure to oestrogens unopposed by progesterone (Henderson et al., 1982), and it has been observed that oestrogens can induce increased production of NOA in normal mice by augmenting the activity of CD5 + B cells (Ahmed et al., 1989). However, in this group of patients we were unable to observe a change in NOA levels. CD5+ B cells may also have other functions in addition to autoantibody production. MacKenzie, Paglieroni & Warner (1987) reported that such cells exhibit immunoregulatory functions and may be responsible for the immune suppression occurring in patients with multiple myeloma. Although we have not detected a gross immunodeficiency with respect to percent-
age ofcirculating T, B or null cells in the cancer patients with the elevated CD5+ B cell counts, it is possible that a more elaborate immunological profile of these patients would have detected modulated immune reactivities. In fact, the significant increase in the percentage of circulating CD5+ B cells in certain cancer patients has to be at the expense of other cell types, since there was no change in the total B cells. These results suggest that the CD5+ B cell subset may be affected by the malignant process or by the therapy modality employed, and suggest that further studies are warranted.
ACKNOWLEDGMENTS We thank Susan Chen, Glen Kelly, and Lea Shohat for their excellent technical assistance. We also thank Dr Benjamin Fisch, Dr Yaron Lidor, Dr Haim Pinkas, Dr Joseph Pardo, Dr Linda Hanel, and Dr Ian Katz, from the Department of Obstetrics and Gynaecology, Berlinson Medical Centre, Petah Tikva, Israel, for their help in collecting blood samples for the Tel Aviv study group. This work was supported in part by the Fainbarg Family Fund, Orange County, CA, the Mary A. Pikovski Fund for Cancer Research, Jerusalem, and USPHS grant CA39841 from the NIH. I.P.W. is the incumbent of the David Furman Chair in Cancer Immunology.
REFERENCES AGASSY-CAHALON, L., YAAKUBOWICZ, M., WITZ, I.P. & SMORODINSKY, N.I. (1988) The immune system during the precancer period. Naturally occurring tumor-reactive monoclonal antibodies and urethan carcinogenesis. Immunol. Lett. 18, 181. AHMED, S.A., DAUPHINEE, M.J., MONTOYA, A.I. & TALAL, N. (1989) Estrogen induces normal murine CD5 B cells to produce autoantibodies. J. Immunol. 142, 2647. AVRAMEAS, S., GUILBERT, B., MAHARA, W., MATSIOTA, P. & TERNYNCK, T. (1987) Recognition of self and non-self constituents by polyspecific autoreceptors. Int. Retz. Immunol. 3, 1. CASALI, P. & NOTKINS, A.L. (1989) CD5 B lymphocytes polyreactive antibodies and the human B-cell repertoire. Immunol. Today, 10, 364.
CD5+ B cells in cancer CHOW, D.A. & BENNETT, R.D. (1989) Low natural antibody and low invivo tumor resistance in Xid-bearing B-cell deficient mice. J. Immunol. 142, 3702. CHOW, D.A. & CHAN, J. (1987) Tumor progression in vitro: the paradoxical natural antibody and complement-selected phenotype. Nat. Immun. Cell Growth Regul. 6, 189. DIGHIERO, G., GUILBERT, B. & AVRAMEAS, S. (1982) Naturally occurring antibodies against nine common antigens in human sera. II. High incidence of monoclonal Ig exhibiting antibody activity against actin and tubulin and sharing antibody specificities with natural antibodies. J. Immunol. 128, 2788. FIRER, M.A., SHOENFELD, Y., ISENBERG, D.A. & SLOR, H. (1988) The use of ELISA for the measurement of autoantibodies to nuclear antigens: comparison with other methodologies. Isr. J. med. Sci. 24, 747. GOLDENBERG, D.M. (1988) Targeting of cancer with radiolabeled antibodies prospects for imaging and therapy. Arch. Pathol. Lab. Med. 112, 580. GOLDENBERG, D.M. (1989) Imaging and therapy of cancer with radiolabeled monoclonal antibodies. In Immunity to Cancer II(ed. by M. S. Mitchell) p. 413. Alan R. Liss, New York. GONEN, B., RAN, M., WRESCHNER, D., CAHALON, L., ANAVI, R., SHLOMO, Y. & WITZ, I.P. (1988) Some immunological properties of high and low tumorigenic cellular variants of c-H-ras transformed 3T3 cels. Nat. Immun. Cell Growth Regul. 7, 144. HENDERSON, B.E., Ross, R.K., PIKE, M.C. & CASAGRANDE, J.T. (1982) Endogenous hormones as a major factor in human cancer. Cancer Res. 42, 3232. HERZENBERG, L.A., STALL, A.M., LALOR, P.A., SIDMAN, C., MOORE,
423
W.A. & PARKS, D.R. (1986) The Ly-l B cell lineage. Immunol. Rev. 93,81. Kipps, T.J. (1989) the CD5 B cell. Adv. Immunol. 47, 117. MACKAY, I.R. & GERSHWIN, M.E. (1989) Molecular basis of mitochondrial autoreactivity in primary biliary cirrhosis. Immunol. Today, 10, 315. MACKENZIE, M.R., PAGLIERONI, T.G. & WARNER, N.L. (1987) Multiple Myeloma: an immunologic profile. IV. The EA rosette-forming cell is a Leu-l positive immunoregulatory B cell. J. Immunol. 139, 24. TAN, E.M. (1989) Antinuclear antibodies: diagnostic markers for autoimmune diseases and probes for cell biology. Adv. Immunol. 44,93. TOUGH, D.F. & CHOW, D.A. (1988) Tumorigenicity of murine lymphomas selected through fluorescence-detected natural antibody binding. Cancer Res. 48, 270. WITZ, I.P. & AGASSY-CAHALON, L. (1987) Do naturally occurring antibodies play a role in the progression and proliferation of tumor cells? Int. Rev. Immunol. 3, 133. WITZ, I.P., CAHALON, L., SMORODINSKY, N.I. & RAN, M. (1988) The bearing of certain tumors may result in autoimmunity. Isr. J. med. Sci. 24, 744. YRON, I., SCHICKLER, M., FISCH, B., PINKAS, H., OVADIA, J. & WITZ, I.P. (1986) The immune system during the precancer and the early cancer period. IL-2 production by PBL from postmenopausal women with and without endometrial carcinoma. Int. J. Cancer, 38, 331. ZURGIL, N., SHAFRIR, S., SLOR, H. & SHOENFELD, Y. (1988) Identification of anti-Sm and anti-RNP autoantibodies by immunoblotting and immunoassay. Isr. J. med. Sci. 24, 751.
ADONIS
0009910491002531
CD5 + B cells and naturally occurring autoantibodies in cancer patients R. STEIN, I. P. WITZ*, J. OVADIAt, D. M. GOLDENBERG & I. YRON* Garden State Cancer Center and Center for Molecular Medicine and Immunology, Newark, NJ, USA, *Department of Microbiology and the Moise and Frida Eskenasy Institute for Cancer Research, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, and tDepartment of Obstetrics and Gynaecology, Beilinson Medical Centre, Sackler School of Medicine, Petah Tikva, Israel
(Accepted for publication 25 March 1991)
SUMMARY We have determined the percentage of CD5+ B lymphocytes in the peripheral blood of cancer patients and healthy controls, using antibodies directed at the CD5 and CDl9 (pan-B) markers. The frequencies of CD5+ B cells, expressed as a percentage of total B cells, ranged from 14-3 to 57.50/) in the controls and from 14 8 to 82-8% in the patient population. One-third of the cancer patients had frequencies > 2 s.d. above the mean of the control population. The CD5 + B cell fraction expressed as a percentage of total lymphocytes was also significantly elevated in this group of cancer patients. These results suggest that the CD5+ B cell compartment may be affected by the malignant process or by the therapy modality employed. The plasma levels of several naturally occurring autoantibodies, the products of the CD5+ B cells, were also assessed in cancer patients and controls. No significant differences were observed when reactivity to several autoantigens was measured. These included nuclear components and phospholipids.
Keywords CD5+ B cells autoantibodies
cancer
INTRODUCTION Autoimmunity and the development of non-lymphoid malignancies may be linked by several independent mechanisms. Immune responses against autoantigens may follow the induction of immunity against tumour-associated epitopes if antibodies induced by tumour-associated epitopes cross-react with epitopes expressed on normal cells, or if the expression of the tumour-specific epitopes is in close enough proximity with autoepitopes to cause the termination of tolerance to the autoepitopes. We have supported the occurrence of such a mechanism in a previous study (Witz et al., 1988) which demonstrated that mice specifically immunized against the polyoma virus TSTA also formed autoantibodies against T cells. A second mechanism by which malignancy and autoimmunity may be connected involves the CD5+ B cell subset and the products of these cells. Although in humans cells with this phenotype were first noted as the cell involved in chronic lymphocytic leukaemia, B cells expressing the CD5 T cell marker (CD5+ B cells) have been shown during the last decade to be normal constituents of the immune system of healthy
individuals (reviews by Herzenberg et al., 1986; Casali & Notkins, 1989; Kipps, 1989). In many cases, the immunoglobulin products of CD5 + B cells consist of autoantibodies which have been referred to as naturally occurring autoantibodies (NOA) (Dighiero, Guilbert & Avrameas, 1982; Casali et al., 1989; Kipps, 1989). NOA are typically present in apparently healthy humans and animals that have not been intentionally immunized (reviews by Casali et al., 1989; Avrameas et al., 1987). Among the autoantigens recognized by NOA are the Fc portion of IgG, ssDNA, cryptic epitopes on erythrocytes, several internal cellular proteins, and membrane constituents of nucleated cells, including lymphocytes. A striking characteristic of many NOA is their polyreactivity (Casali et al., 1989). Thus, unlike conventional antibodies, NOA may react with multiple unrelated antigens, such as non-self antigens, in addition to autoantigens. In vitro studies show that NOA may react with various tumour cells (review by Witz & Agassy-Cahalon, 1987). The tumour reactivity is most probably a manifestation of the polyreactivity of NOA rather than a specific reactivity against tumour-associated antigens. Since NOA are able to react with tumour cells and certain normal cells, such as lymphocytes, or cytokine products of normal cells, these antibodies may, under certain circumstances, influence the progression and proliferation of tumour cells. To date, the evidence to support this
Correspondence: Rhona Stein, PhD, Garden State Cancer Center and Center for Molecular Medicine and Immunology, One Bruce Street, Newark, NJ 07103, USA.
418
CD5 B cells in cancer possibility has come exclusively from murine systems, where it was shown that NOA control the development of certain tumours. NOA have been shown to contribute towards heightened resistance against tumour growth (Chow & Chan, 1987; Chow & Bennet, 1989; Tough & Chow, 1988), and to cause a change (either retardation or enhancement) in tumour development when administered to tumour-bearing mice (AgassyCahalon et al., 1988; Gonen et al., 1988). In contrast to animal models, there are essentially no data available to assign a biological role for NOA or NOAproducing CD5+ B cells in the development or progression of cancer in humans. As a first step toward addressing this question, we compared the levels of the CD5 + B cell subset and the levels of several NOA between cancer patients and a healthy control population, and described quantitative alterations which occur in the patients. SUBJECTS AND METHODS Patients and controls Two independent patient and control groups were studied. The patients belonging to the first group are those referred to the Center for Molecular Medicine and Immunology (CMMI) in Newark, NJ, either for radioimmunodetection or for radioimmunotherapy (Goldenberg, 1988). These patients are characterized by tumours that produce antigens for which the CMMI has developed monoclonal antibodies, e.g. carcinoembryonic antigen, alpha-fetoprotein, human chorionic gonadotropin, prostatic acid phosphatase, colon-specific antigen-p (Goldenberg, 1989). They represent an in-house group of readily accessible, carefully studied patients. The patient population consisted of 26 women (age range 3-79 years, mean 56-2 + 15-5) and 27 men (age range 24-81 years, mean 60 6+ 14.6). Controls for this patient population were CMMI employees: 11 women (age range 20-65 years, mean 39-2 + 14-2) and 14 men (age range 21-68 years, mean 35-8 + 14-9). This group of patients and controls is referred to hereafter as the NJ group. The diagnosis of the primary tumour in 51 out of the 53 NJ cancer patients studied is as follows: 23 colon, three rectal, two caecum, three liver, two pancreas, four breast, four lung, two prostate and one of each stomach, duodenum, sigmoid/colon, testes, thyroid, lymphoma, and teratoma. One patient had both breast and pancreatic primaries. The diagnosis of two primary tumours was unknown. All patients were treated before being tested in this study by surgery, chemotherapy and/or irradiation. The second patient and control group has been described previously (Yron et al., 1986) and is referred to as the Tel Aviv group. This group included 26 women with stage I endometrial carcinoma (EC), prior to treatment; 32 women with hyperplasia of the endometrium, including women with adenomatous and atypical hyperplasia, the pre-EC stages. All women included in this group were over 45 years of age; and 29 healthy women, over 45 years of age, either with no history of major gynaecological problems or with post-menopausal bleeding, and diagnosed as disease free. Collection of blood and plasma samples Whole blood from patients and controls was collected into EDTA vacutainer tubes. After centrifugation at 300 g for 10 min, the plasma, buffy coat, and packed erythrocytes were
separated. Erythrocytes remaining in the buffy
419 coat fraction
were lysed with 20 volumes of lysing solution (155 mm NH4CI, 10 mm KHCO3, 0-1 1 mm EDTA). The white blood cells (WBC) were washed with phosphate-buffered saline (PBS) and then
labelled with fluorescent conjugates of monoclonal antibodies directed against the antigens under study. Flow cytometry of lymphocytes For labelling, WBC from 5 ml of blood were resuspended in 1 5 ml of PBS with 0-1% NaN3 and 1% horse serum. Routinely, five tests were run on each patient's specimen as follows: (i) no labelled antibody (PBS only); (ii) isotype-matched irrelevant mouse immunoglobulins labelled with PE and FITC; (iii) PEconjugated anti-Leu 1 (CD5); (iv) FITC-conjugated anti-Leu 12 (CD19); and (v) PE-anti-Leu 1 and FITC-anti-Leu 12 Washed WBC (60 ,ul) were added to the tubes containing the antibodies, incubated for 30 min in an ice-water bath, washed twice with PBS containing NaN3 and horse serum, and then analysed by flow cytometry using a FACScan flow cytometer (Becton Dickinson Immunocytometry Systems, Mountain
View, CA). Using lymphocyte gating to exclude monocytes and granulocytes, 10 000 cells were analysed for fluorescence. By quadrant analysis on a correlated two-parameter display (PE-anti-Leu I versus FITC-anti-Leu 12), cells in each of the following classifications were quantified: (i) unlabelled, null cells; (ii) cells displaying red fluorescence only, T cells; (iii) cells displaying green fluorescence only, CD5- B cells; and (iv) cells displaying both red and green fluorescence, CD5+ B cells. Determination of circulating autoantibody levels Levels of autoantibodies directed against the following antigens were assessed: (i) nuclear components (Sm; snRNP; Sjogren's syndrome-related antigens (SS-B/La); dsDNA; DNP); (ii) M-2 mitochondrial antigen (pyruvate dehydrogenase (PDH); and
(iii) phospholipids (cardiolipin; phosphatidyl-L-serine). Detection was performed using ELISA kits purchased from BioHyTech (Ramat Gan, Israel), and performed according to the manufacturer's instructions as described elesewhere (Firer et
al., 1988, Zurgil et al., 1988). Briefly, the antigens were each coated optimally to wells of Nunc (Roskilde, Denmark) polystyrene microtitre plates. Serum samples were diluted 1/200 in assay diluent, added to the wells and incubated for 30 min at 37°C. After washing the plates with PBS-Tween 20 wash buffer, (phospholipid buffer did not contain Tween), and a trivalent conjugate containing a mixture of alkaline-phosphatase-conjugated goat anti-human IgG, IgM and IgA (each optimally diluted for the particular antigen; dilution range 1/1000-1/ 3000), was added to the wells and the plates were incubated for a further 30 min at 37°C. The plates were washed again and diethanolamine substrate buffer containing p-nitrophenylphosphate and MgCl2 was added. After 60 min at 37°C, the optical density in each well at 405 nm was read using a Titertek ELISA reader. Standards provided in BioHyTech kits or known positive and negative sera, as well as a reagent blank, were included in each plate. Where autoantibody isotypes were assessed (as was the case in the cardiolipin assays), assays were performed as described above, except that plates were incubated separately with antiIgG or anti-IgM conjugates.
420
R. Stein et al. loor-
0 0
801-
0 0o
g8-08 o8o
601-
0
0
0-
08o
cl
1*t
40 F 0
40S.
00 060
20
F
000
*l00
00
to
00
Cancer
0
Cancer
Control~~~~~~~~~~~~~~~~~
Control
patients
Fig. 1. CD5+ B cells as a percentage of B cells (0, cancer patients; 0, healthy control individuals). Mean values are indicated by horizontal bars for each population.
Optical densities
converted to arbitrary autoantibody curve of a high-titred included in each plate. Cut-off values for provided by BioHyTech, and represent the were
U/ml by comparison with the dilution
positive serum positivity were mean + 3 s.d. of antibody levels (U/ml) obtained from 210 blood bank control sera. Autoantibody activities in these sera showed a normal distribution (data not shown). All experimental sera with autoantibody levels higher than the cut-off value were considered to be positive. Statistical analysis One-way analysis of variance, the Mann-Whitney U-test, and x2 analysis were used to assess the significance of the results of the autoantibody determinations. Student's t-test was used to assess significance of the results in the lymphocyte subset analyses. RESULTS
CDS+ B cells in the peripheral blood of cancer patients To investigate whether CD5 + B cells are altered in cancer patients, we surveyed the levels of lymphocyte subpopulations in the peripheral blood of cancer patients and healthy controls. In these experiments, peripheral blood cells of the NJ groups were analysed. The variability in the percentage values of these lymphocyte subpopulations between individual cancer patients somewhat greater than within the control group. However, the mean values of T, B, null and CD5+ B cells from cancer patients and controls were not statistically different. B cell range was 0 9-32-5% (mean 8-4+7-6%) and 2-8-18-2% (mean 9.4+4.5%) in the cancer patient and control populations, respectively. The percentages of CD5+ B (CD5+, CDl 9+) cells (calculated as a percentage of total circulating lymphocytes) were 0-3-26-0% (mean 4-1+49%) and 0-7-6-9% (mean
was
3-0 + 1 6%) in the peripheral blood of cancer patients and of normal controls, respectively. In addition to the above calculation of CD5+ B cells as a percent of total lymphocytes, the percentage of B cells that express the CD5 molecule was also determined as a per cent of the total B cell population. Figure I is a scattergram of individual values of the cancer and the control groups. The mean value + s.d. of the CD5 + B cells (expressed as a percentage of total B cells) of the control group is 33 5 + 12 2, and that of the cancer patient group is 452 + 18 2. The difference in these values is statistically significant (P < 0-0 1). The ranges are 14-357 5% for the control group and 14-8-82-8% for patient population. In order to assess further the cancer patients who express elevated CD5+ B cell levels, we divided the patient group into two populations: those of the tested cancer patients with higher than normal levels of CD5+ B cells in their peripheral blood were separated from the remainder of the patient population. Higher than normal levels of CD5+ B cells (as percentage of total B cells) were defined as values greater than the mean+2 s.d. of the normal group. According to the results presented above, this value is 57 9%. Employing this setting, it was determined that none of the 25 control subjects had an 'above normal' value of CD5+ B cells. However, 30% of the cancer patients (16 out of 53) showed 'above normal' values of CD5+ B cells (as percentage of total B cells) in peripheral blood. Table 1 lists the characteristics of the patients with elevated CD5+ B cells (as percentage of total B cells). This population represents a heterogeneous group in terms of tumour type, stage, age, sex, and prior treatment. The fractions of T, B, null and CD5+ B cells (as percentage of total lymphocytes) in the peripheral blood of these two patient populations are compared in Table 2. The results show that neither division of cancer patients differed significantly from the control population in levels of T, B, or null cells. However, the group of patients with elevated values of CD5+ B cells, expressed as a percentage of total B cells, also had a significantly increased proportion of CD5 + B cells expressed as percentage of total lymphocytes (P< 005). This is in contrast to the cancer group as a whole, as described above. Since the mean age of the control group was lower than that of the cancer patient group, we examined whether advanced age correlated positively with higher levels of CD5+ B cells. The levels of CD5+ B cells (as percentage of total B cells) of individuals in the control and in the cancer group as a function of their age were calculated and no positive correlation between these parameters could be detected (the correlation coefficients were -0065 and -0265 for the control and cancer patient populations, respectively). Furthermore, the selected subpopulation of cancer patients with higher than normal levels of CD5 + B cells did not belong, as a group, to the older population. Autoantibody levels in cancer patients The NJ group. Since CD5+ B cells have been shown to be autoantibody producers, we measured the levels of autoantibodies in the plasma of patients with a high frequency as well as in patients with a low frequency of CD5+ B cells. The healthy individuals mentioned above served as controls. The levels of antibodies against three categories of antigens, nuclear components, mitochondrial PDH and cardiolipin, are shown in Table 3. No difference in reactivity with these antigens was
CD5+ B cells in cancer
421
Table 1. Characteristics of patients with high counts of CD5+ B cells
Patient no. 799 967 1014 1058 1075 1077 1080 1101 1103 1116 1124 1134 1135 1136 1146 1148
Stage at diagnosis
Cancer diagnosis
Testicular Sigmoid colon Breast/pancreatic Lung Rectal Non-Hodgkin's lymphoma Breast Colon Prostate Stomach Colon Hepatoblastoma Colon Hepatocellular Rectal Colon
Stage at study
Age/ sex
Prior treatment
CD5 + B cells (0%, of total)
D D D D D III D D NED D
32/M 70/M 67/F 41/F 72/M 49/M 55/F 53/M 76/M 51/M 42/M 3/F 75/M 64/F 51/M 68/F
Surgery, chemotherapy Surgery, chemotherapy Surgery Surgery, XRT, chemotherapy Surgery Chemotherapy Surgery, XRT, chemotherapy Surgery, XRT, chemotherapy Surgery Surgery, XRT, chemotherapy Surgery, XRT, chemotherapy Surgery, chemotherapy Surgery Surgery, chemotherapy Surgery, XRT, chemotherapy Surgery, XRT, chemotherapy
65 9 67-2 67-3 80-0 62-0 62 5 61-7 82-8 70 5 63 7 58-1 73 9 66-3 71 5 64.3 64-6
III
Dukes C-I II III Dukes A III II Dukes C-II D2 T2, N2, M l Dukes C-II I Dukes Bl I 0-I Dukes D
D NED D L D D
Stages: D, distant disease; NED, no evidence of disease; L, local; Breast, UICC classification with lymph node criteria; Colon, standard Dukes classification, with C-I, involvement of 1-4 lymph nodes and C-II,5 or more lymph nodes; Lymphoma, Ann Arbor classification; all others according to standard clinical staging. XRT, external radiotherapy Table 2. Lymphocyte subpopulations in peripheral blood of cancer patients (NJ group) with high or normal counts of CD5+ B cells
Normal CD5'
(n=37)* Lymphocyte subpopulation T cells B cells Null cells CD5+ B cells
High CD51 (n= 16)t
Mean + s.d.
Range
Mean + s.d.
Range
72-5 + 12 6
25 7-92-9 0 9-32-3 6-2-57 8 03-12-3
65 2+ 14 6 9 5+9 2 24-1 + 14 3 6-8+7-2
36-1-86-3 1-4-32-5 76-60 8 09-260
8-0+±70 19-2+ 11-0 31+3-2
group of age-matched healthy women (data not shown). Similarly, the proportion of autoantibody-positive individuals was similar in the three groups. It should be noted that the anti-DNP autoantibody levels of the EC and hyperplasia patients of the Tel Aviv group were higher than those of the NJ patients. However, this difference was not statistically significant. Autoantibodies to DNP appear
in high titres in patients with autoimmune rheumatic disease,
Cancer patients with control frequency of peripheral CD5+ B cells. t Cancer patients with frequency of peripheral CD5 + B cells higher than the mean + 2 s.d. above control level. *
observed between the three groups. Similarly, the proportion of autoantibody-positive individuals in the three groups was similar. The levels of the mitochondrial autoantibodies, antiPDH, were found to be somewhat (not significantly) higher in the NJ patients. High titres of such autoantibodies characterize autoimmune primary biliary cirrhosis (Mackay & Gershwin, 1989). Thus, the slightly higher levels of such antibodies in certain NJ patients may suggest hepatic involvement in their disease, although no direct liver function tests were performed. The Tel Aviv group. Based on the observation that autoantibody production can be enhanced by oestrogens (Ahmed et al., 1989), a second group of patients was also included in this study. Autoantibody activity was measured in the plasma of patients with EC or hyperplasia of the endometrium, situations which are highly correlated with longer term exposure to oestrogens unopposed by progesterone (Henderson et al., 1982). No difference in autoantibody levels directed against any of the antigens tested could be observed between the patients and the
most of whom are women (Tan, 1989). It is possible,
therefore,
that the production of such antibodies is enhanced in oestrogenrelated diseases.
DISCUSSION B cells expressing the CD5 T cell marker and the naturally occurring autoantibody products of these cells have been identified in mice and humans only within the last decade. Evidence obtained in several laboratories, including our own, has shown a relationship between NOA and tumourigenesis in several murine systems (Witz & Agassy-Cahalon, 1987). Using flow cytometry, we have surveyed the levels of CD5+ B cells in the peripheral blood of cancer patients and healthy controls. One-third of the cancer patients had frequencies more than 2 s.d. above the mean of the control populations when expressed as a percentage of B cells. The CD5 + B cell fraction expressed as a percentage of total lymphocytes was also significantly elevated in this group of cancer patients. This increase may be of biological significance, especially in view of the fact that in this group of patients the CD5+ B cell lymphocyte subpopulation was the only one to be quantitatively altered when compared with normal controls. No significant deviation from control
values was detected in the other parameters that were measured. To our knowledge, this is the first report of this observation. The significance of this finding is unclear at present. Thus far, we have not found any correlation between a high CD5 + B cell count and clinical parameters, such as the type of the
422
R. Stein et al. Table 3. Levels of autoantibodies in the NJ group Normal CD5+t
High CD5+t
Healthy controls
Cut-off value*
Mean + s.d.
H/T
Mean + s.d.
H/T
Mean + s.d.
H/T
310 280 250 275 250
152+74 246+ 307 189+326 191 +266 230+85
1/40 7/40 5/40 3/40 4/40
146+80 298 + 392 181 +351 268+423 258+ 144
0/16 4/16 1/14 3/16 4/16
158+87 224+ 274 98+57 150+79 193+68
0/19 2/19 0/16 1/19 1/18
Cardiolipin§ IgG 320 IgM 320
166±58 188+ 102
1/40 3/40
151 +47 151 +74
0/16 1/16
132+21 157+57
0/19 1/19
Antigen DNP RNP Sm SSb PDH
*
Cut-off values represent the *Means (U/ml +3 s.d.) of 210 blood bank control sera.
t Plasma from cancer patients with control frequency of peripheral CD5 + B cells. Values are arbitrary units (U/ml), calculated in reference to a known high titered positive serum, of all individuals included in the group. T Plasma from cancer patients with frequency of peripheral CD5 + B cells higher than the mean+ 2 s.d. above control level. § Cardiolipin autoantibody isotypes were assessed separately for IgG and IgM. H/T (high/total), the ratio between the number of sera in which the levels of antibodies were above cut-off values and the total number of individuals included in the group.
primary cancer, stage of disease, or previous therapy. However, one may make several assumptions which could serve as a working hypothesis for future studies. One of these assumptions is that chemotherapy or external radiotherapy in themselves were not responsible for the elevated levels of CD5 + B cells where detected. This assumption stems from the fact that four out of the 16 patients in whom high levels of such cells were detected had only surgery as prior treatment (Table 1). Obviously, in future studies one should attempt to analyse patients with more uniform diagnosis and prior to any form of therapy, and should include an analysis of CD5+ B cells in secondary lymphoid tissues, such as draining lymph nodes, of untreated patients since the bloodstream is only a transit site. One interesting observation was that autoantibodies were not increased even in those NJ cancer patients exhibiting a higher than normal level of CD5+ B cells. It is possible that autoantibodies other than those tested in our study may have been altered concomitantly with the increase in the levels of CD5+ B cells. For example, the CD5+ B cells in those patients whose numbers are increased could be recognizing glycolipid tumour-associated antigens. Patients with EC or with preneoplastic stages of the disease (the Tel Aviv group) were included in the NOA study since these conditions are known to be highly related to long-term exposure to oestrogens unopposed by progesterone (Henderson et al., 1982), and it has been observed that oestrogens can induce increased production of NOA in normal mice by augmenting the activity of CD5 + B cells (Ahmed et al., 1989). However, in this group of patients we were unable to observe a change in NOA levels. CD5+ B cells may also have other functions in addition to autoantibody production. MacKenzie, Paglieroni & Warner (1987) reported that such cells exhibit immunoregulatory functions and may be responsible for the immune suppression occurring in patients with multiple myeloma. Although we have not detected a gross immunodeficiency with respect to percent-
age ofcirculating T, B or null cells in the cancer patients with the elevated CD5+ B cell counts, it is possible that a more elaborate immunological profile of these patients would have detected modulated immune reactivities. In fact, the significant increase in the percentage of circulating CD5+ B cells in certain cancer patients has to be at the expense of other cell types, since there was no change in the total B cells. These results suggest that the CD5+ B cell subset may be affected by the malignant process or by the therapy modality employed, and suggest that further studies are warranted.
ACKNOWLEDGMENTS We thank Susan Chen, Glen Kelly, and Lea Shohat for their excellent technical assistance. We also thank Dr Benjamin Fisch, Dr Yaron Lidor, Dr Haim Pinkas, Dr Joseph Pardo, Dr Linda Hanel, and Dr Ian Katz, from the Department of Obstetrics and Gynaecology, Berlinson Medical Centre, Petah Tikva, Israel, for their help in collecting blood samples for the Tel Aviv study group. This work was supported in part by the Fainbarg Family Fund, Orange County, CA, the Mary A. Pikovski Fund for Cancer Research, Jerusalem, and USPHS grant CA39841 from the NIH. I.P.W. is the incumbent of the David Furman Chair in Cancer Immunology.
REFERENCES AGASSY-CAHALON, L., YAAKUBOWICZ, M., WITZ, I.P. & SMORODINSKY, N.I. (1988) The immune system during the precancer period. Naturally occurring tumor-reactive monoclonal antibodies and urethan carcinogenesis. Immunol. Lett. 18, 181. AHMED, S.A., DAUPHINEE, M.J., MONTOYA, A.I. & TALAL, N. (1989) Estrogen induces normal murine CD5 B cells to produce autoantibodies. J. Immunol. 142, 2647. AVRAMEAS, S., GUILBERT, B., MAHARA, W., MATSIOTA, P. & TERNYNCK, T. (1987) Recognition of self and non-self constituents by polyspecific autoreceptors. Int. Retz. Immunol. 3, 1. CASALI, P. & NOTKINS, A.L. (1989) CD5 B lymphocytes polyreactive antibodies and the human B-cell repertoire. Immunol. Today, 10, 364.
CD5+ B cells in cancer CHOW, D.A. & BENNETT, R.D. (1989) Low natural antibody and low invivo tumor resistance in Xid-bearing B-cell deficient mice. J. Immunol. 142, 3702. CHOW, D.A. & CHAN, J. (1987) Tumor progression in vitro: the paradoxical natural antibody and complement-selected phenotype. Nat. Immun. Cell Growth Regul. 6, 189. DIGHIERO, G., GUILBERT, B. & AVRAMEAS, S. (1982) Naturally occurring antibodies against nine common antigens in human sera. II. High incidence of monoclonal Ig exhibiting antibody activity against actin and tubulin and sharing antibody specificities with natural antibodies. J. Immunol. 128, 2788. FIRER, M.A., SHOENFELD, Y., ISENBERG, D.A. & SLOR, H. (1988) The use of ELISA for the measurement of autoantibodies to nuclear antigens: comparison with other methodologies. Isr. J. med. Sci. 24, 747. GOLDENBERG, D.M. (1988) Targeting of cancer with radiolabeled antibodies prospects for imaging and therapy. Arch. Pathol. Lab. Med. 112, 580. GOLDENBERG, D.M. (1989) Imaging and therapy of cancer with radiolabeled monoclonal antibodies. In Immunity to Cancer II(ed. by M. S. Mitchell) p. 413. Alan R. Liss, New York. GONEN, B., RAN, M., WRESCHNER, D., CAHALON, L., ANAVI, R., SHLOMO, Y. & WITZ, I.P. (1988) Some immunological properties of high and low tumorigenic cellular variants of c-H-ras transformed 3T3 cels. Nat. Immun. Cell Growth Regul. 7, 144. HENDERSON, B.E., Ross, R.K., PIKE, M.C. & CASAGRANDE, J.T. (1982) Endogenous hormones as a major factor in human cancer. Cancer Res. 42, 3232. HERZENBERG, L.A., STALL, A.M., LALOR, P.A., SIDMAN, C., MOORE,
423
W.A. & PARKS, D.R. (1986) The Ly-l B cell lineage. Immunol. Rev. 93,81. Kipps, T.J. (1989) the CD5 B cell. Adv. Immunol. 47, 117. MACKAY, I.R. & GERSHWIN, M.E. (1989) Molecular basis of mitochondrial autoreactivity in primary biliary cirrhosis. Immunol. Today, 10, 315. MACKENZIE, M.R., PAGLIERONI, T.G. & WARNER, N.L. (1987) Multiple Myeloma: an immunologic profile. IV. The EA rosette-forming cell is a Leu-l positive immunoregulatory B cell. J. Immunol. 139, 24. TAN, E.M. (1989) Antinuclear antibodies: diagnostic markers for autoimmune diseases and probes for cell biology. Adv. Immunol. 44,93. TOUGH, D.F. & CHOW, D.A. (1988) Tumorigenicity of murine lymphomas selected through fluorescence-detected natural antibody binding. Cancer Res. 48, 270. WITZ, I.P. & AGASSY-CAHALON, L. (1987) Do naturally occurring antibodies play a role in the progression and proliferation of tumor cells? Int. Rev. Immunol. 3, 133. WITZ, I.P., CAHALON, L., SMORODINSKY, N.I. & RAN, M. (1988) The bearing of certain tumors may result in autoimmunity. Isr. J. med. Sci. 24, 744. YRON, I., SCHICKLER, M., FISCH, B., PINKAS, H., OVADIA, J. & WITZ, I.P. (1986) The immune system during the precancer and the early cancer period. IL-2 production by PBL from postmenopausal women with and without endometrial carcinoma. Int. J. Cancer, 38, 331. ZURGIL, N., SHAFRIR, S., SLOR, H. & SHOENFELD, Y. (1988) Identification of anti-Sm and anti-RNP autoantibodies by immunoblotting and immunoassay. Isr. J. med. Sci. 24, 751.
