CLINICAL
IMMUNOLOGY
Lymphoreticular
AND
IMMUNOPATHOLOGY
18, 415-424 (1981)
Cells within Primary Colorectal A Surface Marker Study’
Carcinoma:
J. A. WERKMEISTER,~E. PIHL, E. S. R. HUGHES, A. M. CUTHBERT~ON, AND A. J. ROLLO Departmenrs Alfred
of Pathology and Immunology, and Surgery, Monash University Hospital, St. Francis Xuvier Cabrini Hospital, und The Avenue Melbourne. Australia
Medical Hospital.
School.
Received July 1, 1980 Stromal lymphoreticular cell infiltration was studied in 60 colorectal carcinomas, using EAC-rosetting and immunofluorescent staining of B lymphocytes. Three distinct but partly overlapping patterns were found: (i) Diffuse leukocytosis concentrated in a bandlike fashion at the tumor edge and comprising plasma cells, eosinophils, macrophages, and lymphocytes, and seen in ah cases. The lymphocytes did not generally react with EAC nor stain with the anti-B-cell serum: (ii) patchy lymphocytosis comprising a mixture of B and non-B lymphocytes and seen in 20 of the 60 tumors; and (iii) perivascular cuffs of small, dark lymphocytes (PLC) which did not generally rosette with EAC, nor stain with anti-B-lymphocyte serum, i.e., putative T lymphocytes, and seen in 16 of the 60 tumors. Stromal leukocytosis of the first type and defined as exceeding 2000 cells/mmZ, related significantly and directly to tumor differentiation and “localized” tumor stage. Both PLC and stromal leukocytosis were significantly (P = 0.028 and 0.009, respectively) associated with a favorable prognosis.
INTRODUCTION
A variety of lymphoreticular cell types and mechanisms have been implicated in anti-tumor in vitro immunoreactivity by blood and regional node lymphocytes (1). Few studies on “intrinsic” host cells in the tumor stroma have been published in spite of their favorable prognostic significance (2-4). In human colorectal carcinoma, Ackerman (5) stated that the “round cell” infiltration at the tumor edge may be an indication of host resistance. Recently, we have isolated intrinsic, cytotoxic T lymphocytes from the stroma of human colorectal carcinomas and also found a direct association between in vitro anti-tumor immunoreactivity and perivascular lymphocyte cuffing at the tumor edge (6). The aim of the present investigation has been to define and classify by immunological markers stromal lymphoreticular cell infiltration in colorectal carcinoma. MATERIALS
AND METHODS
Patients and tumor specimens. Sixty unselected adenocarcinomas removed at operation by the three surgeons in our group were received fresh in the Depart’ Supported by a grant from the Anti-Cancer Council of Victoria. ? Present address: Clinical Immunology Unit, Medical Research Department, Kanematsu Memorial Institute. Sydney Hospital, Sydney, N.S.W. 2000, Australia. 415 0090-1229/81/030415-10$01.00/O Copyright 0 1981 by Academic Press. Inc. All rights of reproduction in any form reserved.
416
WERKMEISTER
El
TABLE
‘Tumor Case code
Dukes’ stage (7)
75lQ63 751075 751082 751130 751149 751182 751184 751186 761007 7610 1 I 761015 761016 761018 761019 761026 761028 761029 761043 761044 761045 761047 761048 761063 761068 761073 761074 761075 76/084 761087 76109 1 761103 761104 761105
B B B A A B B B’ c C B C B C C C B C’ C B c C B B C C’ A B C A A C’. C
Differentiation
AI.
I
Features Leukocytes ~2000imm’
+
PLC 2504
Follow-up in months
t
57 51 42 22 22 26 51 x 31 IO” 30 22 26 47 28 II 0 15 38 27 41 36 31 32 20 22 43 35 23 36 4” 9 0
-+ t + + -
+ + + -
+
+ + -
+ + + t -
+ +
Clinical state N.s.r.” N.s.r. N.s.r. Died” Died” N.s.r. N.s.r. Died” N.s.r. Rec.’ N.s.r. Died” N.s.r. N.s.r. N.s.r. Died” Postop. Died” Died” N.S.I.. N.s.r. N.a.r. N.s.r. N.s.r. Died” N.s.r. N.s.r. N.s.r. N.s.r. N.s.r. Kec .(’ Died” Postop.
death
death
ment of Pathology for histological examination. Part of the tissue was used for the immunocytological studies described below. The patients’ age range was 35-86 years (mean 61) and 28 were male. Histological assessment. Tumor and lymph node tissue were examined macroscopically, fixed in formalin, embedded in paraffin, and classified according to Dukes (7), (Table 1); six tumors confined to the bowel wall were staged as A, 22 had invaded through the muscle coat and were staged as B, and 32 which had spread to regional lymph nodes were staged as C. Tumor differentiation was classified into two groups: Group 1, good; Group 2, poor, the latter including also tumors of intermediate differentiation according to conventional criteria (8). The number of stromal lymphoreticular cells comprising plasma cells,
COLORECTAL
CARCINOMA TABLE Tumor
LYMPHORETICULAR I-Continued
Features
Case code
Dukes’ stage (7)
76/l 10 76/l 12 76/l 13 76/l 14 761118 761121 761126 761131 761132 761134 761143 761144 761159 761165 761174 76/1?7 761184
C” B C B B B C B C’ C C B B C C C C
-
+ -
+ -
761185 761189 761192
B C C
-
+ +
+ -
761194 761195 761198 761206 761209 7612 I 1 761226
B C C C’ C B C
-
+ + + -
+ + -
” ” ’ ” ’
Leukocytes >2OOO/mm’
Differentiation
No sign of recurrence. Died of tumor. Spread to adjacent or distant Lost to follow-up. Alive with recurrence.
-
+ + + -
+ + + +
+ +
organ
417
CELLS
(palliative
PLC a5%
t t i + -
-
Follow-up in months 25 46 35 26 33 44 13 19 15‘ 26 35 27 30 27 22 25 30 3” 38 23 20 26 40 15 40 26 20
Clinical state Died” N.s.r. Rec.” N.s.r. N.s.r. N.s.r. N.s.r. Died* Died” Died” N.s.r. N.s.r. Diedh Diedh N.s.r. N.s.r. N.s.r. N.s.r. N.s.r. N.s.r. N.s.r. N.s.r. Died” Died” Died” N.s.r. Died”
op.).
eosinophils, macrophages, and lymphocytes was counted in 10 areas at the tumor edges; these fields, using the 100 oil-immersion objective, were selected so that each field was approximately equidistant from the next, along the whole lateral and deep tumor edges. Cell infiltrates within the actual tumor, neutrophilic leukocytes and lymphocytes surrounding blood vessels, were ignored. The number of lymphoreticular cells, using a Reichert stage micrometer for calibration of the microscope, was expressed as a mean per mm*. In the final analysis it was found that 2000 lymphoreticular cells/mm2 was a convenient cutoff point, this figure being exceeded in 20 of the tumors (Table 1) and hereafter referred to as stromal leukocytosis. The degree of lymphocytic cuffing at the deep edge of the tumors was assessed
418
WERKMEISTER
E I 41..
quantitatively standardized to assess the around blood intersections)
as described previously (3, 4, 6). Briefly, a microscope with a square graticule with 25 intersections (Zeiss, Oberkochen) was used relative size of areas occupied by “cuffs“ of small dark lymphocytes vessels. These were expressed as a percentage of the total (2200 counted at x 4 objective magnification. EAC-rosrtting on .fro:en .sec.tions. The degree and patterns of stromal lymphoreticular cell distribution in the stroma of the 60 colorectal carcinomas was related to the presence or absence of complement (C3) receptors on the surfaces of these cells by EAC-rosetting on frozen sections. Fresh, unfixed tissue selected from the deep lateral edge of each of the 60 colorectal, was snap-frozen in a liquid nitrogen-isopentane slurry as described by Naim (9). EAC indicator cellular intermediates (sheep erythrocytes coated with rabbit anti-erythrocyte antibody and mouse complement) were prepared as described previously (6). Rosetting on frozen sections was performed according to the method of Jaffe et uf. (10). Sections were cut on a cryostat microtome at 6 pm, air dried, and overlaid with a 5% suspension of EAC. After incubation for 30 min at room temperature, the sections were suspended face down in 0.15 M saline for 10 min and carefully dipped twice in saline to complete the washing. The sections were fixed in 2.5% glutaraldehyde for 30 min and stained with hematoxylin and eosin. Experiments were designed so that lymph node sections, in which the topographic distribution of T and B lymphocyte areas is known ( 11, 12), were treated in parallel with and in the same way as the tumor sections. Absence of EAC-rosetting lymphocytes in the tumor sections was given credence only when germinal centers in the colorectal lymph node sections stained with EAC (Fig. 1). ImmunoJluor~s~.en~~~ OH fio:rrr tissue sections. Six-micrometer-thick frozen sections were cut from 20 consecutive tumors, air dried, and stained with a
FIG. 1. Frozen ment cell markers
sections (EAC).
of human lymph node treated with sheep erythrocyte-antibody-compleThere is strong adherence of EAC to germinal center (H & E, X 17%.
COLORECTAL
CARCINOMA
LYMPHORETICULAR
CELLS
419
fluorescent conjugated anti-B-lymphocytic leukemia (CLL) cells, absorbed once with human blood group AB erythrocyte “envelopes” and twice with human fetal thymus cells as described by Nairn (9); specific reactivity against CLL and approximately 20% of normal peripheral blood lymphocytes (B lymphocytes). Lymph node sections were again used as positive controls in parallel with the tumor sections: the conjugate stained germinal centers and primary follicles (Fig. 2). Clinicuf data und survival analysis. Clinical findings at resection, recurrence and survival data, the latter recorded at regular outpatient visits to the three surgeons in our team, were made available for analysis after completion of the immunopathological assessment (Table 1). Seven of the patients had evidence of tumor spread to adjacent or distant organs, at the operation and were treated with a palliative resection. Fifty-three patients underwent a potentially curative resection. Two patients died from postoperative complications; a total of 19 have subsequently died of clinically confirmed recurrent tumor, two had confirmed recurrences but were lost to follow-up, three are alive with recurrent tumor. Thirty-five patients remained free of disease at the time of last follow-up. Survival and recurrence data were plotted graphically according to the method of Kaplan and Meier (13). Differences between survival curves were evaluated statistically by the generalized Wilcoxon test according to Gehan (14). RESULTS
Strom& Leukocytosis and Tumor Morphologic& Features There was a significant association between stromal leukocytosis and absence of tumor spread (Table 2); stromal leukocytosis (>20OO/mm*) being present in 14
FIG. 2. Frozen lymphocyte serum.
sections of human lymph There is strong fluorescent
node stained with staining of germinal
fluorescein-conjugated center area (x340).
anti-B-
420
WERKMEISTER
E’T Al.
of 28 (50%) tumors stages A and B, in comparison with 6 of 32 (19%) stage C tumors (P < 0.025). Similarly. 7 of 15 (47%) well-differentiated tumors showed stromal leukocytosis in comparison with I3 of 45 (29%~) of the poorly differentiated tumors (P < 0.05). EAC-Resetting on Frozen Tissue Sections Three different patterns of lymphoreticular cell infiltration could be seen in the stroma and at the lateral and deep edges of the tumor: (i) Diffuse leukocytosis within the tumor stroma extending to and forming a bandlike infiltrate immediately outside tumor glands at the edge. This was dominated by plasma cells, eosinophils, varying numbers of macrophages, and only few lymphocytes. Neither cell type formed rosettes with, or bound EAC (Fig. 3). This cell pattern was seen to a varying extent in all tumors. (ii) Patchy lymphocytosis comprising medium-sized lymphocytes, mostly found peripherally within the actual stroma and at the tumor edge. Such infiltrates formed EAC-rosettes to a varying extent, generally between 40 and 50% (Fig. 4). This pattern was linked with the diffuse and bandlike type described above and seen in 20 of the 60 cases. (iii) Perivascular infiltration by small lymphocytes, mostly around postcapillary venules in the vascular plexus of the pericolic/perirectal fat, i.e., outside the actual tumor stroma. Such perivascular lymphocyte cuffs of 5% or more of the pericolic/perirectal tissue were found in 16 of 60 (27%) cases. The lymphocytes did not generally bind EAC (Fig. 5). Immunojluorescent Stuining of Frozen Tissue Sections Only the patchy lymphocyte infiltrates previously shown to bind EAC, stained with fluorescein-labeled antibody. There was no staining of perivascular lymphocyte cuffs (PLC) nor of the diffuse cell infiltrates, or of the bandlike cell infiltrate at the tumor edges. Lymphoreticular Cell Infiltration in Relation Survival curve analysis showed a significant vival, stromal leukocytosis. and PLC (Table limited and our series contains censored data,
TABLE RELATIONSHIP
BETWEEN
TUMOR
features
Stages A and B Stage C Good differentiation Poor differentiation
2
STAGF, DIFFERENTIATION. IN COLOREX CAL CARCINOMA
Number Tumor
to Survivul favorable association between sur3). Although the follow-up time is i.e., patients who are still alive or
of cases with
STROMAI.
stromal
LEUKOcYlOSlS
leukocytes -z 2,OOOimm’ _--_~-
>2,000/mms 14 6 7 13
AND
14 26 8 32
x2 = 5.23; P < 0.025 x1’ = 3.91: P < 0.05
-
COLORECTAL
CARCINOMA
LYMPHORETICULAR
CELLS
421
FIG. 3. Frozen sections of the lymphoreticular cell infiltrate peripheral to the edge of a human colonic carcinoma (not illustrated). The section was “stained” with EAC. There is no binding of the marker to the lymphoid cells, mainly macrophages, eosinophils. and plasma cells (x 175).
have been lost to follow-up, cancer specific survival at the time of the last recurrence death was 89 -+ 7 and 82 + 12% for patients whose primary tumors showed stromal leukocytosis or PLC, respectively. The corresponding survival figures in cases without these immunomorphological features were 34 t 12 and 36 5 13%.
FIG. 4. Frozen sections of a patchy lymphoreticular cell infiltrate at the edge of human colonic adenocarcinoma. EAC marker binds to approximately half the lymphoid cells, mainly small- to medium-sized lymphocytes (x340).
WERKMEISTER
Flc;. 5. Frozen sections of perivascular human 1colonic carcinoma. The section wax mainly small dark lymphocytes ( i 340)
Ef
Al.
lymphoreticular treated
cell
with
EAC
infiltrate
which
did
at the not
hind
mesenteric
edge
to the lymphoid
of a t:ells.
DISCUSSION
The patterns and immunological marker characteristics of stromal lymphoreticular cell infiltrates were studied in unfixed frozen sections of 60 colorectal adenocarcinomas. using a complement receptor rosetting technique (EAC) and immunofluorescence by means of an anti-B-lymphocyte serum. Three distinct patterns were found: first, diffuse leukocytosis in the tumor stroma concentrated in a bandlike fashion at the tumor edge and comprising plasma cells, eosinophils, macrophages, and a small number of mostly nonrosetting, non-B type lymphocytes. This pattern was seen in all tumors: in 20 of these the number of TABLE REL.ATIONSHIP
BETWEE.N
STKOMAI. AND
Lymphoreticular
Stromal
CANC.EK-SPECIFIC
cell
Died
infiltration
3 PEKIVASCLILAR
Ltukocu-row.
LYMPHOC)
it
CUPFING
(PLCt.
SURVIVAL
of
Median
recurrence/total
in months
‘120
= 0.10
N.r.”
17137
= 0.46
38
2116
= 0.13
I7141
= 0.41
p
leukocytes: 12OOO/mm’
0.009
PLC
3x7
N.r. 0.028
4% I’ Generalized ’ Not reached.
Wilcoxon
analysis
of survival
curves
39 (14)
COLORECTAL
CARCINOMA
LYMPHORETICULAR
CELLS
423
lymphoreticular cells at the edge exceeded 20OO/mm* and such marked infiltration was significantly more often apparent in well-differentiated stage A and B tumors and was associated with a favorable prognosis. Second, in one-third of the tumors, patchy lymphocytosis was seen, mainly at the tumor periphery, in which 40-X% of the lymphocytes bound to EAC and to a similar extent stained with the antiB-lymphocyte serum. Third, perivascular subserosaYperirecta1 lymphocyte cuffs (PLC) in the vascular plexus at the deep tumor edge were found, seen in 27% of the tumors; these lymphocytes did not generally rosette with EAC nor stain with the anti-B-cell serum. PLC was also associated with significantly longer survival. Our finding that stromal lymphoreticular cells in colorectal carcinoma represent a mixture of B and non-B, i.e., putative T lymphocytes (15) appears to be a valid observation because both techniques used for the demonstration of B lymphocytes gave similar results. The biological significance of our results is that immunocytes of both humoral and cell-mediated immune reactions are present in the stroma of colorectal carcinomas. The exact functional nature of the lymphocytes which did not react with EAC nor stain with the anti-B-cell serum is not clear from our results, and several possibilities are open, i.e., these lymphocytes may be either T cells, natural killer (NK) cells, or mediate antibody-dependent cell-mediated cytotoxicity (ADCC). We have previously demonstrated both cytotoxic T lymphocytes and cells capable of mediating ADCC among the lymphoreticular cell populations intrinsic to colorectal carcinoma (6). We were also able to demonstrate a direct statistical association between intrinsic E-rosetting lymphocyte-mediated cytotoxicity against primary cultures of colorectal carcinomas, and PLC (6). These findings, together with the marker studies presented here, confirm the predominantly non-B, i.e.. putative T-lymphocytic nature of PLC. It is still unclear whether any of the in vitro functions ascribed to the various lymphocyte subpopulations intrinsic to the stroma of colorectal carcinomas relate to in vivo behavior of these tumors. Our group has demonstrated that serum antibody to the cytoplasm of autochthonous colorectal cancer cells is associated with a major survival advantage (16) but we have no proof whether or not this antibody is of primary importance and is produced locally. The association between the degree of stromal leukocytosis and “localized“ tumor stages, described in this paper, is consistent with previous studies (4, 17). We have no data to indicate whether or not such cell infiltration is an “early” event in the natural history of colorectal carcinomas, or whether its presence is a morphological expression of immune homeostasis mechanisms causally related to prevention of spread of a certain population of these tumors. Furthermore, a third possibility must be considered, i.e., whether stromal leukocytosis may conceivably have occurred as a response to tumor necrosis or infection. The latter possibility is very unlikely, because of the easily recognizable neutrophilic and macrophage response to the latter, while the stromal leukocytosis described by us is immunocytic in nature, comprising in order of predominance and with little variation, PLC at the deep edge and the characteristic bandlike peripheral accumulation of plasma cells, eosinophils, macrophages, and occasional lymphocytes. These features relate directly to the “localized” tumor stages A and B, good
424
WERKMEISTER
ET Al.
differentiation, and favorable survival data as shown both previously (3. 4) and in this paper-the results suggest that this is either an “early” phenomenon in the natural history of a carcinoma or occurs as a response to well-expressed tumor antigens. Whatever the mechanisms involved, PLC and other types of stromal lymphoreticular cell patterns described here reflect a favorable immune reactivity in the patient with colorectal cancer. Its various in vitro expressions, i.e., antiautochthonous tumor cell cytotoxicity of humoral, T-lymphocyte, macrophage, NK cell, or ADCC type mechanisms will now have to be related to clinical, cancer-specific survival data. So far we have demonstrated intrinsic Tlymphocyte anti-tumor cytotoxicity (6) and the prognostic significance of its morphological expressions (3, 4). A further understanding of the functions and interrelationship of these and other intratumoral host cell types described here may conceivably form a basis for the development of more effective adjuvant treatment regimens aimed at stimulating beneficial lymphoreticular cell response to tumor cells. ACKNOWLEDGMENTS We thank Mrs. A. B. Price for help with computation.
REFERENCES 1. Hellstrom, I., Advan. Immunol. 18, 209, 1974. 2. Underwood, J. C. E., Brit. J. Cancer 30, 538, 1974. 3. Pihl, E., Malahy, M. A., Khankhanian, N., Hersh, E. M., and Mavligit, G. M., Cuncer Res. 37, 4145, 1977. 4. Pihl, E., Nairn, R. C., Hughes, E. S. R., Cuthbertson. A. M., and Rollo, A. J.. Pathology 12, 15, 1980. 5. Ackermann. L. W., “Ackerman and Del Regato’s Cancer: Diagnosis, Treatment and Prognosis.” Mosby, St. Louis, MO.. 1970. 6. Werkmeister, J., Pihl, E., Nind, A., Flannery, G., and Nairn, R. C.,Brir. J. Cancer 40,839, 1979. 7. Dukes. C. E., In “Cancer of the Rectum” (D. W. Smithers and C. E. Dukes, Eds.), pp. 59-68. Livingstone, Edinburgh, 1960. 8. Pihl, E., Hughes, E. S. R., Nind, A. P. P., and Nairn, R. C., Brit. Med. .I. 3, 742. 1975. 9. Naim, R. C., “Fluorescent Protein Tracing.” Churchill, Livingstone, Edinburgh, 1976. 10. Jaffe, E. S., Shevach. E. M., Frank, M. M., Bernard. C. W.. andGreen, I., N. EngI. J. Med. 290, 813, 1974. 11. Cottier, H., Turk, J.. and Sobin, L., Bul/. WHO 47, 375, 1972. 12. Pihl, E., Naim, R. C., Muller, H. K.. Hughes, E. S. R.. Cuthbertson. A. M., and Rollo, A. J., Cancer Res. 36, 3665, 1976. 13. Kaplan, E. L., and Meier, P., J. Amer. Statist. Assoc. 53, 457. 1958. 14. Gehan, E. A., Biometrika 52, 203, 1965. 15. Jondal, M., Holm, E., and Wigzell. H., J. Exp. Med. 136, 207. 1972. 16. Nind, A. P. P., Naim, R. C., and Pihl, E., Cancer Immunol. Immunother. 7, 257, 1980. 17. Watt, A. G., and House, A. K., Cancer 41, 279. 1978.