- Email: [email protected]
Autologous Melanoma Vaccine Induces Inflammatory Responses in Melanoma Metastases: Relevance to Immunologic Regression and Immunotherapy
Autologous Melanoma Vaccine Induces Inflammatory Responses in Melanoma Metastases: Relevance to Immunologic Regression and Immunotherapy George F. Murphy, Antoaneta Radu, Michael Kaminer, and David Berd Human primary malignant melanoma is often accompanied by a host response of infiltrating lymphocytes suggestive of tumor antigen – induced immunity and correlated in some tumors with prognosis. Whereas metastatic melanoma deposits typically are not inflamed and contain relatively few lymphocytes and dendritic immune cells, immunization with autologous melanoma-cell vaccine may induce a clinical inflammatory response associated with mononuclearcell infiltration. In this study, we characterize immune responses to dermal and subcutaneous melanoma metastases in dinitrophenyl (DNP)-pre-sensitized patients immunized with DNP-conjugated melanoma cells. Patients so treated develop cutaneous delayed hypersensitivity responses to DNP-conjugated autologous mononuclear cells, and approximately one-half show clinical evidence of inflammation and regression of metastases within 2-4 months. Whereas pre-vaccination biopsies of metastatic melanoma failed to reveal significant infiltration by lymphocytes, biopsies obtained after vaccination and coincident with clinical inflammation were markedly infiltrated preponderantly by T cells with a CD8 þ phenotype. Clustering of these cells about individual degenerating melanoma cells in a manner analogous to ‘‘satellitosis’’ was a consistent feature of this reaction. Enhanced expression of intercellular adhesion molecule-1 (ICAM-1) and human leukocyte antigen (HLA)-DR by melanoma cells were invariably associated with zones of T-cell infiltration, whereas diminished or absent expression was observed in relatively unaffected regions of tumors. Numerous HLA-DR þ , CD4 þ , CD1, Leu-1 dendritic cells were also associated with zones of early T-cell infiltration. These data indicate that clinical inflammation and regression of metastatic melanoma induced by autologous melanoma-cell vaccine involves activated T cells with cytotoxic - suppressor phenotype and dendritic cells putatively capable of local antigen presentation. ICAM-1 upregulation on melanoma cells is a likely mediator of ligand interaction between infiltrating T cells and target cells in this model of antigen-induced host antitumor response. Structural alterations identified in this setting (e.g., tumor cell satellitosis) may provide additional insight into identifying features of naturally occurring host immune responses to primary cutaneous melanomas. J Invest Dermatol 100:335S–341S, 1993
A notion central to the seminal concept of tumor progression of Clark is that developmental evolution of neoplastic cells may be halted by protective host immune mechanisms. Human malignant melanoma has been the subject of intense scrutiny concerning its capability to provoke host immune responses [1]. There is obvious significance to the potential to recapitulate this phenomenon therapeutically. Primary melanoma in early phases of growth frequently is characterized by partially successful host immune reactions that contribute to color variegation and border irregularity [2]. More advanced primary tumors may evoke immune responses at deepest points of dermal invasion. In such Department of Dermatology (GFM, AR, MK), University of Pennsylvania School of Medicine; and Division of Medical Oncology (DB), Thomas Jefferson University, Philadelphia, Pennsylvania, U.S.A. Reprint requests to: Dr. George F. Murphy, 235B Clinical Research Building, 422 Curie Boulevard, Philadelphia, PA 19104. Abbreviations: CY, cyclophosphamide; DFNB, dinitrofluorobenze; DNP, dinitrophenyl; DTH, delayed-type cutaneous hypersensitivity; GSV, grayscale values; HLA, human leukocyte antigen; ICAM-1, intercellular adhesion molecule-1; IL, interleukin; MNC, mononuclear cells; NK, natural killer; PBS, phosphate-buffered saline; TILS, tumor-infiltrating lymphocytes
0022-202X/93/$06.00
instances, quantification of tumor-infiltrating lymphocytes (TILs) may represent a significant independent prognostic variable [3]. The local and systemic mechanisms that account for immune responses directed against melanoma cells are incompletely understood. From a practical viewpoint, little is known concerning structural details of immunologic regression resulting from antigen-specific effector-target cell interactions. Extrapolation from established experimental systems of effector-target-cell interactions, such as acute graft-versus-host disease, suggests that apposition of effector cells of cytotoxic phenotype to degenerating target cells is a characteristic morphologic feature of antigen-directed destructive immune responses [4,5]. This phenomenon, termed ‘‘satellitosis’’ may be observed in certain primary melanomas putatively undergoing immune regression (Fig 1). Until now, experimental model systems to determine whether this histology correlates with antigenic phenotypes characteristic of melanoma-induced cytotoxicity have been lacking. We recently studied clinical and immunologic responses in human melanoma metastases, which ordinarily do not elicit significant host lymphoid infiltration, after administration of
Copyright & 1993 by The Society for Investigative Dermatology, Inc. 335S
336S
MURPHY ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 1. Morphologic characteristics of lymphoid response to primary malignant melanoma. Putative advanced inflammatory regression (A) is often characterized by a zone of lymphocytes (L) and residual melanoma cells (arrows) bordered by lamellar collagen deposition in which pigment-laden macrophages (P) are entrapped. Melanoma cells within the epidermal layer (EPI) are typically spared. Presumed earlier stages of TIL response are characterized by clustering of rare (B) or numerous (C) lymphocytes about individual melanoma cells (arrows) in a manner similar to satellitosis, as classically described in experimental graft-versus-host disease. Pronounced MNC response to a small dermal island of melanoma cells (D) is associated with lymphoid infiltration (larger arrow) and tumor cell necrosis (small arrows). (A, magnification 60; B–D, magnification 200.)
autologous melanoma vaccine. This approach is based on preadministration of cyclophosphamide (CY) to heighten host responses [6–10] to tumor-associated antigens [11–15] to which cancer patients appear to develop immunologic tolerance [16–23]. In clinical trials, it has been shown that the administration of CY prior to injection of autologous tumor vaccine can induce delayed-type cutaneous hyper-sensitivity (DTH) to autologous tumor cells and, sometimes, to regression of metastatic tumors [24]. To further augment DTH to autologous tumor antigens, the vaccine we used was conjugated to dinitrophenyl (DNP), which is known to enhance cell-mediated immunity to tumor-associated antigens [25–28]. In clinical trials using this strategy, it was recently demonstrated that 14 of 24 patients so treated exhibited immune-mediated anti-tumor responses directed against dermal and subcutaneous metastases [29,30]. In this report, we characterize the phenotype of immune responses to melanoma metastases in patients clinically responding to DNP-conjugated autologous melanoma vaccine to better understand the structural and antigenic details of host-mediated immune regression. Our data suggest that satellitosis involving T cells with a suppressor-cytotoxic phenotype is characteristic of immune regression in melanoma induced by this form of therapy. Moreover, these interactions are associated with dendritic immune cell proliferation/infiltration and intercellular adhesion molecule-1 (ICAM-1) induction in responding tumors.
MATERIALS AND METHODS General Experimental Design In the initial treatment protocol, 24 patients with metastatic melanoma were given vaccine consisting of autologous, cryopreserved, irradiated tumor cells conjugated to DNP. All patients had at least one superficial metastasis in skin or subcutaneous tissue that was directly observable and easily biopsied.
Patients were contact sensitized to the hapten by topical application of dinitrofluorobenzene (DNFB) 3 d following low-dose (300 mg/M2) intravenous CY. Two weeks later, and subsequently every 28 d, patients were injected intradermally on the upper arm with DNPhaptenized melanoma cells mixed with bacille Calmette-Gue´rin as adjuvant. CY was administered 3 d before each vaccine injection. Skin testing with DNP-conjugated peripheral blood mononuclear cells (MNC) revealed induction of DTH responses to DNPconjugated MNC in 22 of 24 patients after DNFB sensitization, without DTH to MNC alone or conjugated to a non-cross-reactive hapten, TNP. Between 2 and 4 months after initiation of DNPvaccine therapy (average two to four injections), 14 of 24 patients developed erythema, warmth, and tenderness of metastatic tumors and overlying skin. In 10 patients, one or more biopsies were obtained for immunohistochemical analysis of inflammatory cells, activation antigens, and adhesion molecules. In five patients, biopsies were obtained prior to vaccination and subsequent to vaccination at the time of peak clinical inflammatory response. In addition to qualitative immunophenotypic evaluation, pre-vaccination and post-vaccination melanoma specimens from four patients were also evaluated by computer-assisted image analysis to determine the relative number of labeled T cells infiltrating tumors (6,463–18,534 cells evaluated/lesion). Clinical and Vaccination Protocol We treated 24 patients with metastatic, surgically incurable melanoma, all of whom had at least one metastasis in skin, subcutaneous tissues, or lymph nodes that was directly observable and easily biopsied. Informed consent was obtained. The patients were sensitized to DNP by topical application of 1% DNFB in acetone corn oil on two consecutive days. Intravenous CY (300 mg/M2) was administered 3 d before sensitization.
VOL.
100, NO. 3, SUPPLEMENT, MARCH 1993
INFLAMMATORY RESPONSES TO MELANOMA
337S
Table I. Monoclonal Antibodies Used in This Study Sourcea
Antibody
Reactivity
Bectin Dickinson
Leu-1
CD5 (mature T cells)
Leu-2
CD8 (cytotoxic-suppressor T cells)
Leu-3
CD4 (helper-inducer T cells; macrophages)
Leu-6
CD1 (Langerhans cells)
Leu-7
Some cytotoxic T; natural killer. (NK) cells
Leu-11
CD16 (Fc immunoglobulin G receptor on NK cells and neutrophils)
Coulter Corp.
Dako Corp.
Leu-M1
CD 15 (monocytes)
Anti-HLA-DR
Class II molecules on B cells, g-interferon-stimulated cells
NKH-1
CD56 (NK cells)
2H4
CD45RA (suppressor -inducer CD4+ T-cell subset)
4B4
CD45RO (helper-inducer CD4+ subset)
TAC
Interleukin-2 receptor
ICAM-1
Intercellular adhesion molecule-1 (lymphocyte function associated antigen-1 ligand)
a Bectin Dickinson Immunocytometry Systems, Mountain View, CA; Coulter Immunology, Hialeah, FL; Dako Corp., Carpinteria, CA.
Two weeks later, patients were again given CY, followed 3 d later by injection of DNP-conjugated melanoma vaccine; CY and DNP vaccine were repeated every 28 d. The vaccine consisted of 10–25 106 cryopreserved, autologous, irradiated (2500 R), DNPconjugated melanoma cells prepared as previously described [31] and mixed with bacille Calmette-Gue´rin. DNP conjugation was performed by the method of Miller and Claman [32]. Patients were tested for cell-mediated immunity to hapten-modified cells by skin testing with DNP-conjugated peripheral blood MNC. Biopsy Selection Initial T-cell antigen screening of 54 excisional biopsies of metastatic melanomas from 24 patients pre-vaccination and post-vaccination revealed inflammatory infiltrates to be restricted to post-vaccination lesions undergoing clinical induration, erythema, and enlargement (n ¼ 14). Of these, four were excluded because of the presence of zonal necrosis (a potential source of secondary, non-specific inflammatory reactions) or the presence of tumor within subcutaneous lymph nodes. As anticipated, cells infiltrating extensively necrotic tumors expressed predominantly histiocytic markers (CD15), and cells infiltrating nodal metastases were predominantly of the CD4 subset and were not associated with individual cell necrosis, human leukocyte antigen (HLA)-DR induction, or satellitosis. Of the 10 remaining cases, five had at least one pre-biopsy and post-biopsy sample, permitting assessment of evolution of inflammation related to vaccination. The postvaccination specimens were obtained 2–4 months after initiation of
Figure 2. Quantification of Leu-1 þ TILs in metastatic melanomas before (solid bars) and after (hatched bars) autologous DNP-conjugated melanoma vaccine (n ¼ 4; p ¼ 0.04, 0.05, 0.005, and 0.03 for patients 1–4, respectively).
vaccine treatment, at points of maximum clinical inflammatory change. The number of inflamed tumors in a single patient ranged from 1 to more than 100 [33]. In patients with multiple superficial metastases, 25–100% of observable lesions became inflamed. Immunocytochemistry Each tissue specimen was subdivided into 1-mm - thick slabs and placed in Michel’s medium for transport and temporary storage. Samples were washed in phosphate-buffered saline (pH 7.4) (PBS), snap-frozen in liquid nitrogen, and cryostat sectioned at 5-mm intervals. For comparative assessment of proportions of inflammatory cells expressing certain epitopes, adjacent tissue sections were studied to minimize artifact resulting from local variation in cell number. Monoclonal antibodies used for initial screening and detailed analyses are enumerated in Table I. Labeling was obtained using an avidin-biotin-peroxidase complex system (Vector Laboratories, Inc., Burlingame, CA) and 3,30 diaminobenzidine as a chromagenic substrate. Briefly, tissue sections were acetone fixed for 10 min at 41C, and primary antibody (or appropriate, isotype-specific irrelevant controls) (dilution range 1: 50–1:200) was placed over each tissue section and incubated in a humidified chamber for 30 min at room temperature. After thorough rinsing with PBS, sections were overlayed with biotinylated antimouse anti-serum (1:200) for 30 min. After washing, avidinbiotinperoxidase complex was added at a concentration of 1:100 for 45 min. Subsequent to thorough PBS rinse, sections were incubated with 3,30 -diaminobenzidine and 3% H2O2 in PBS and counterstained lightly with hematoxylin. Replicate sections were prepared for each experimental condition for evaluation of cellular infiltrates in routine paraffin sections. Image Analysis Quantitative analysis of Leu-1 þ cells was performed on an image-analysis system consisting of a central computer (Sony Communications Products Company, Park Ridge, NJ), color video monitor (Mitsubishi), video camera (Southern Micro Instruments, Inc., Atlanta, GA), and Olympus BH-2 light microscope. Pretreatment and post-treatment biopsy specimens from four individuals
338S
MURPHY ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 3. Immunophenotypic characterization of TIL response in lesions showing clinical inflammation-regression after autologous vaccination. Characteristic fields of metastatic melanoma from a single patient stained with Leu-1 before (A) and after (B) vaccination. Routine microscopy (C) reveals brisk TIL response with foci where lymphocytes surround individual tumor cells (M) in a manner analogous to satellitosis (inset). Subsetting of the TIL response in adjacent tissue sections revealed the majority of cells to be CD8 þ (D) and the minority to be CD4 þ (E). Clustering of TILs about melanoma cell(s) (D and E, arrows) was consistently produced by only the CD8 þ subset (D). (A, B, D, E, magnification 100; C, magnification 200; C inset, magnification 400.)
were analyzed with cell-counting software (Southern Micro Instruments, Inc.). This system used 256 gray-scale values (GSV) from 0 to 255 (0 ¼ black, 255 ¼ white), with the range between black and white subdivided on the basis of relative shades of gray. The working area on the video screen was 466 466 pixels (which represented square meters after calibration), generating 217,516 pixels/field. The system applied a GSV to each pixel. Computer-assisted adjustment to GSV limits enabled segmentation of cells to be counted on the basis of color. Leu-1 þ cells were dark brown and therefore had lower GSV than Leu-1 cells, which were light blue. Calibration was performed to ‘‘teach’’ the computer the number of pixels occupied by a single cell of average area. To determine the number of cells in the field of interest, the total number of segmented pixels on the screen was divided by the calibrated reference pixel count. This measurement was performed to count both Leu-1 þ cells and the total number cells in each field. The accuracy of this method was confirmed by manual counting of positively labeled cells/500 tumor cells in selected specimens with concordance between 5 and 7%. Differences in the pre-vaccination and post-vaccination number of infiltrating Leu-1 þ cells for each patient were evaluated for significance using the two-tailed paired t test.
RESULTS AND DISCUSSION Metastatic Melanoma—Pre-Treatment Biopsies of metastatic melanoma demonstrated malignant cells within the dermis and subcutis, with cytologic features compatible with their known primary malignant melanomas. No specimens showed evidence of significant tumor infiltration by lymphocytes, and immunohisto-chemistry revealed numerous fields devoid of T cells (Leu-1 þ cells) (Figs 2 and 3A). Computer-assisted image analysis revealed between 54 and 421 Leu-1 þ cells/10,000 tumor cells in four pretreatment biopsies (Fig 2) (a fifth case was excluded owing to substantial cellular pigmentation impairing discrimination of immunolabeled inflammatory cells). Only rare HLA-DR þ dendritic cells were present with tumor nodules, and tumor cells expressed HLA-DR and ICAM-1 focally and weakly. These findings are consistent with the paucity of lymphoid inflammatory infiltration typically observed in metastatic melanoma deposits [34,35] and with experimental evidence indicating that cell lines derived from metastatic melanoma, in contrast to early primary lesions, fail to induce autologous T-cell proliferation [36,37]. Metastatic Melanoma—Post-Treatment In contrast to pretreatment biopsies, all specimens of metastatic melanoma (post-treatment, n ¼ 10) were heavily infiltrated by HLA-DR þ and Leu-1 þ T cells (Fig 2 and Fig 3B). Histologic analysis (Fig 3C)
VOL.
100, NO. 3, SUPPLEMENT, MARCH 1993
INFLAMMATORY RESPONSES TO MELANOMA
339S
Figure 4. Immunophenotypic characterization of ICAM-1 expression and of dendritic cells in lesions showing clinical response after vaccination. (A,B) Adjacent sections of subcutaneous melanoma, one stained with Leu-1 (A) and the other stained for ICAM-1 (dotted line demarcates junction between stroma above and tumor cells below). Zone of most pronounced infiltration by small T lymphocytes (A, upper third of tumor nodule) corresponds to most intense ICAM-1 expression by larger tumor cells (B). Less pronounced ICAM-1 expression (B, lower two-thirds of tumor nodule) corresponds to diminished TIL response (A, lower two-thirds of tumor nodule). Staining for HLA-DR in adjacent sections of the upper third (B, upper arrow), middle third (B, middle arrow), and lower third (B, lower arrow) is shown in C–E. Note the zones of diffuse class II expression by larger tumor cells and dendritic cells (C) and exclusive expression by numerous (D) and less numerous (E) dendritic cells. These zones correlate directly with areas of most pronounced (top) to least pronounced (bottom) Leu-1 þ T-cell infiltration (A) and ICAM-1 expression (B). Dendritic cells from zones represented in C and D expressed CD4 (F), although the majority were negative for Leu-Ml (G). (A and B, magnification 60; C–E, magnification 200; F and G, magnification 100.)
demonstrated small lymphoid cells diffusely infiltrating among large, anaplastic melanoma cells in a pattern identical to that described for TILs in primary melanoma [3]. Degenerative melanoma cells were frequently surrounded by peripheral clusters of lymphocytes in a pattern similar to satellitosis, as described in acute graft-versus-host disease [4,5] (Fig 3C, inset).
Computer-assisted image analysis (Fig 2) confirmed the difference in the number of TILs between pre-treatment specimens (54–421 Leu-1 þ cells/10,000 cells) and post-treatment specimens (1160–2559 Leu-1 þ cells/ 10,000 cells) (p ¼ 0.03–0.005). Immunophenotypic analysis for T-cell subclasses in adjacent tissue sections revealed that the majority of infiltrating T cells
340S
MURPHY ET AL
were CD8 þ (Fig 3D), and a minority were CD4 þ (Fig 3E ). Moreover, peripheral aggregation of CD8 þ T cells about melanoma cells in a pattern corresponding to the satellitosis typical of routine histologic preparations (Fig 3C, inset) was frequently observed (Fig 3D). Only rare cells were Leu 7 þ or Leu 11 þ , indicating that natural killer (NK) cells do not account for a significant proportion of the TIL response induced by autologous vaccine. The majority of TILs did not express immunoreactive interleukin-2 (IL-2) receptors using the Tac antibody, an observation in keeping with previously reported fluorescence-activated cell sorting analyses of inflamed tumors post-vaccination [33] and with our observations using this labeling system in experimental cutaneous DTH [38]. Whether this represents an artifact of antibody used, sensitivity of immunohistochemical localization, timing of biopsy, or the result of receptor saturation by IL-2, remains to be determined. Because TAC antibody labels only the a-chain of the IL-2 receptor, studies using recently described monoclonal reagents to the b-chain of the IL-2 receptor [39] are now clearly indicated. In vivo, TILs were recently shown to be of the CD3 þ CD8 þ phenotype, with an oligoclonal (rather than polyclonal) pattern of T-cell receptor gene rearrangement [40]. Cytotoxicity against melanoma cells was recently demonstrated to reside in cloned populations of CD8 þ T cells that recognize autologous tumor cells through the T-cell receptor and in an HLA class I-restricted manner [41]. It was not possible to reliably segregate the minority population of CD4 þ TILs into 4B4 þ (helper-inducer) and 2H4 þ (suppressor-inducer) subgroups because tumor cells demonstrated apparent expression of the CD45RO epitope (defined by 4B4 antiserum), precluding comparative enumeration of associated labeled lymphoid elements. Occasional lymphoid cells were 2H4 þ , however, representing infiltration of tumors by a suppressor- inducer CD4 þ subset that may be impaired by CY, accounting partially for its immunoenhancing effect [42]. Presumed early T-cell infiltration of metastatic melanoma at the periphery of nodules was spatially associated with tumor cell ICAM-1 expression (Fig 4A,B). As in experimental cutaneous DTH responses in human interfollicular epidermis [38,43], T-cell migration into melanoma nodules was invariably associated with enhanced expression of melanoma cell ICAM-1, indicative of possible g-interferon effect. ICAM-1 regions of tumor were never infiltrated by T cells in excess of numbers seen in pre-vaccination specimens. These observations raise the possibility that interaction between TILs and melanoma cells may be partially mediated by ICAM-1/lymphocyte function-associated antigen-1 ligand interactions. Although such interaction would potentially facilitate efficacy of immune responses, it has also been suggested that ICAM-1 expression by melanoma cells could enhance ‘‘passenger’’ dissemination via heterotopic attachment between malignant cells and leukocytes actively trafficking across vessel-tissue interfaces [44]. In zones of maximal ICAM-1 immunoreactivity (Fig 4C ), melanoma cells strongly expressed HLA-DR, whereas areas of minimal ICAM-1 staining (Fig 4D) revealed HLA-DR expression only by infiltrating CD1 dendritic cells. ICAM-1 zones of tumor (Fig 4E ), which typically were devoid of infiltrating T cells, displayed fewer, less intensely stained dendritic HLA-DR þ cells. The majority of dendritic cells infiltrating vaccine-treated melanomas were CD4 þ (Fig 4F ), although these cells were clearly separable morphologically and antigenically from more rounded CD5 þ T cells, which also may express class II molecules
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
on activation (Fig 4G). The HLA-DR þ , CD1, and CD4 þ phenotypes of these dendritic cells suggests that they belong to the family of dermal dendritic immune cells, as originally described in 1985 by virtue of differential immunoreactivity for antibodies to certain b2-microglobulin glycoproteins shared by Langerhans cells and endothelial cells [45]. Recently, dermal ‘‘dendrocytes’’ exhibiting positivity for the fibrin-stabilizing factor, factor XIIIa, have been characterized [46]. These cells probably represent a congruent population of dermal dendritic immune cells. It is intriguing to speculate that some of these dendritic cells may be responsible for fibrin–fibronectin polymerization, antigen presentation, or cytokine production that augments the T-cell migration or activation directed against melanoma cells after immunization. The findings in this study support the concept that immunologic responses to metastatic malignant melanoma after administration of autologous, hapten-conjugated vaccine involve infiltrative lymphoid populations predominantly of the CD8 þ phenotype. Tumors so affected may demonstrate ICAM-1 and HLA-DR expression, presumably induced by g-interferon, a lymphokine that has been shown to regulate T-cell responses to melanoma in vitro [34]. Moreover, HLA-DR þ , CD4 þ , and CD1 dendritic immune cells appear integral to T-cell infiltration of melanomas after autologous vaccination. These observations in a relevant clinical model of antigen-induced melanoma regression may aid in understanding naturally occurring structural alterations typical of tumor-directed host immunity in primary melanomas, as well as the cellular interactions that underpin efficacy of melanoma immunotherapy in responsive recipients. This work was supported by grants CA-40358 and CA-39248 from the National Institutes of Health.
REFERENCES 1. Rowden G: Immunology of melanoma. In: Mihm MC Jr, Murphy GF, Kaufman N (eds.). Pathobiology and Recognition of Malignant Melanoma. Williams & Wilkins, Baltimore, 1988, pp 50–78 2. Clark WH Jr, Elder DE, Van Horn M: The biologic forms of malignant melanoma. Hum Pathol 17:443–450, 1986 3. Clark WH Jr, Elder DE, Guerry D, Braitman LE, Track BJ, Schultz D, Synnestvedt M, Halpern AC: Model predicting survival in stage I melanoma based on tumor progression. J Natl Cancer Inst 81: 1893–1904, 1989 4. Woodruff JM, Hansen JA, Good RA, Santos GW, Slavin RE: The pathology of the graft-versus-host reaction (GVHR) in adults receiving bone marrow transplants. Transplant Proc 8:675–684, 1976 5. Guillen FJ, Ferrara J, Hancock WW, Messadi DV, Fonferko E, Burakoff SJ, Murphy GF: Acute cutaneous graft-versus-host disease to minor histocompatibility antigens in a murine model: evidence that large granular lymphocytes are effector cells in the immune response. Lab Invest 137:1874–1877, 1986 6. Maguire HC Jr, Ettore VL: Enhancement of dinitrochlorobenzene (DNCB) contact sensitization by cyclophosphamide in the guinea pig. J Invest Dermatol 48:39–43, 1967 7. Turk JL, Parker D: Effect of cyclophosphamide on immunological control mechanisms. Immunol Rev 65:99–113, 1982 8. Schwartz A, Askenase PW, Gershon RK: Regulation of delayed-type hypersensitivity reactions by cyclophosphamide-sensitive T cells. J Immunol 121:1573–1577, 1978 9. Yoshida S, Nomoto K, Himeno K, Takeya K: Cooperation between cyclosphosphamide tumoricidal activity and host anti-tumor immunity in the cure of mice bearing large MOPC-315 tumors. Cancer Res 41:2163–2168, 1981
VOL.
100, NO. 3, SUPPLEMENT, MARCH 1993
INFLAMMATORY RESPONSES TO MELANOMA
341S
10. Berd D, Mastrangelo MJ, Engstrom PF, Paul A, Maguire H: Augmentation of the human immune response by cyclophosphamide. Cancer Res42:4862–4866, 1982
30. Murphy GF, Radu A, Mastrangelo M, Berd D: Tumor infiltrating T cells in metastatic melanoma: induction by immunization with autologous, DNP-conjugated tumor cells (abstr). Lab Invest 62:70A, 1990
11. Reisfeld RA, Cheresh DA: Human tumor antigens. Adv Immunol 40:323–377, 1987 12. Old LJ: Cancer immunology: the search for specificity. Cancer Res 41:361–375, 1981
31. Berd D, Maguire HC, McCue P, Mastrangelo MJ: Treatment of metastatic melanoma with an autologous tumor cell vaccine: clinical and immunological results in 64 patients. J Clin Oncol 8:1858–1867, 1990
13. Mukherji B, Guha A, Chakraborty M, Sivanandham M, Nashed AL: Clonal analysis of cytotoxic and regulatory T cell responses against human melanoma. J Exp Med 169:1961–1976, 1989
32. Miller SD, Claman HN: The induction of hapten-specific T cell tolerance by using hapten-modified lymphoid cells. I. Characteristics of tolerance induction. J Immunol 117:1519–1526, 1976
14. Topalian SL, Solomon D, Rosenberg SA: Tumor-specific cytolysis by lymphocytes infiltrating human melanomas. J Immunol 142:3714–3725, 1989
33. Berd D, Murphy GF, Maguire HC, Mastrangelo MJ: Immunization with haptenized, autologous tumor cells induces inflammation of human melanoma metastases. Cancer Res 51:2731–2734, 1991
15. Houghton AN, Mintzer D, Cordon-Cardo C, Welt S, Fliegel B: Mouse monoclonal IgG3 antibody detecting GD3 ganglioside: a phase I trial in patients with malignant melanoma. Proc Natl Acad Sci USA 82:1242–1246, 1985
34. Guerry D, Alexander MA, Elder DE, Herlyn MF: Interferon-g regulates the T cell response to precursor and nevi and biologically early melanoma. J Immunol 139:305–310, 1987
16. Berd D, Maguire HC Jr, Mastrangelo MJ: Induction of cell-mediated immunity to autologous melanoma cells and regression of metastases after treatment with a melanoma cell vaccine preceded by cyclophosphamide. Cancer Res 46:2572–2577, 1986 17. Mitchell MS, Kan-Mitchell J, Kempf RA, Harel W, Shau H: Active specific immunotherapy for melanoma: phase I trial of allogeneic lysates and a novel adjuvant. Cancer Res 48:5883–5893, 1988 18. Berd D, Mastrangelo MJ: Active immunotherapy of human melanoma exploiting the immunopotentiating effects of cyclophosphamide. Cancer Invest 6:337–349, 1988 19. Takei F, Levy J, Kilburn DG: Characterization of suppressor cells in mice bearing syngeneic mastocytoma. J Immunol 118:412–417, 1977 20. Berendt MJ, North RJ: T-cell-mediated suppression on anti-tumor immunity. Human tumor antigens. J Exp Med 151:69–80, 1980 21. Mullen CA, Urban JL, Van Waes C, Rowley DA, Schrieber H: Multiple cancers. Tumor burden permits the outgrowth of other cancers. J Exp Med 162:1665–1682, 1985 22. Fujimoto S, Green MI, Sehon AH: Regulation of the immune response to tumor antigens. I. Immunosuppressor cells in tumor-bearing hosts. J Immunol 116:791–796, 1976 23. McBride WH, Howie SEM: Induction of tolerance to a murine fibrosarcoma in two zones of dosage—the involvement of suppressor cells. Br J Cancer 53:707–712, 1986 24. Berd D, Maguire HC, McCue P, Mastrangelo MJ: Treatment of metastatic melanoma with an autologous tumor-cell vaccine: clinical and immunologic results in 64 patients. J Clin Oncol 8:1858–1867, 1990 25. Mitchison NA: Immunologic approach to cancer. Transplant Proc 2:92–103, 1970 26. Fujiwara H, Aoki H, Yoshioka T, Tomita S, Ikegami R: Establishment of a tumor-specific immunotherapy model utilizing TNP-reactive helper T cell activity and its application to the autochthonous tumor system. J Immunol 133:509–514, 1984
35. Kornstein MJ, Brooks JSJ, Elder DE: Immunoperoxidase localization of lymphocyte subsets in the host response to melanoma and nevi. Cancer Res 43:2749–2753, 1983 36. Alexander MA, Bennicelli J, Guerry D: Defective antigen presentation by human melanoma cell lines cultured from advanced, but not biologically early disease. J Immunol 142:4070–4078, 1989 37. Guerry D, Alexander MA, Herlyn MF, Zehngebot LM, Mitchell KF, Zmijewski CM, Lusk J: HLA-DR histocompatibility leukocyte antigens permit cultured human melanoma cells from early but not advanced lesions to stimulate autologous lymphocytes. J Clin Invest 73:267–271, 1984 38. Waldorf HA, Walsh LJ, Schechter NM, Murphy GF: Early cellular events in evolving cutaneous delayed hypersensitivity in humans. Am J Pathol 138:477–486, 1991 39. Ohashi Y, Takeshita T, Nagata S, Mori K, Sugamura K: Differentiation expression of the II-2 receptor subunits p55 and p75 on various populations of primary peripheral blood mononuclear cells. J Immunol 143:3584–3595, 1989 40. Belldegrun A, Kasid A, Uppenkamp M, Topalian SL, Rosenberg SA: Human tumor infiltrating lymphocytes. Analysis of lymphokine mRNA expression and relevance to cancer immunotherapy. J Immunol 142:4520–4526, 1989 41. Gervois N, Heuze F, Diez E, Jotereau F: Selective expansion of a specific anti-tumor CD8 þ cytotoxic T lymphocyte clone in the bulk culture of tumor-infiltrating lymphocytes from a melanoma patient: cytotoxic activity and T cell receptor gene rearrangements. Eur J Immunol 20:825–831, 1990 42. Berd D, Mastrangelo MJ: Effect of low dose cyclophosphamide on the immune system of cancer patients: Depletion of CD4 þ , 2H4 þ suppressor-inducer T-cells. Cancer Res 48:1671–1675, 1988 43. Lewis RE, Buchsbaum M, Whitaker D, Murphy GF: Intercellular adhesion molecule expression in the evolving human cutaneous delayed hypersensitivity reaction. J Invest Dermatol 93:672–677, 1989
27. Fujiwara H, Hamaoka T, Shearer GM, Yamamoto H, Terry WD: The augmentation of in vivo and in vitro tumor-specific T cell-mediated immunity by amplifier T cells. J Immunol 124:863–869, 1980
44. Johnson JP, Stade BG, Holzmann B, Schwable W, Riethmuller G: De novo expression of intercellular-adhesion molecule 1 in melanoma correlates with increased risk of metastasis. Proc Natl Acad Sci USA 86:641–644, 1989
28. Flood PM, Schreiber H, Ron Y: Protective immunity to progressive tumors can be induced by antigen presented on regressor tumors. J Immunol 138:3573–3578, 1987
45. Murphy GF, Bronstein BR, Knowles RW, Bhan AK: Ultrastructural localization of M241 glycoprotein on dendritic and endothelial cells in normal human skin. Lab Invest 52:264–269, 1985
29. Berd D, Mastrangelo MJ, Green C, Clark C, Hart E: Tumor inflammatory response induced with autologous melanoma cells conjugated to dinitrophenol (DNP) (abstr). Proc Am Assoc Cancer Res 30:382A, 1989
46. Cerio R, Griffiths CE, Cooper KD, Nickoloff BJ, Headington JT: Characterization of factor XIIIa positive dermal dendritic cells in normal and inflamed skin. Br J Dermatol 121:421–431, 1989
A notion central to the seminal concept of tumor progression of Clark is that developmental evolution of neoplastic cells may be halted by protective host immune mechanisms. Human malignant melanoma has been the subject of intense scrutiny concerning its capability to provoke host immune responses [1]. There is obvious significance to the potential to recapitulate this phenomenon therapeutically. Primary melanoma in early phases of growth frequently is characterized by partially successful host immune reactions that contribute to color variegation and border irregularity [2]. More advanced primary tumors may evoke immune responses at deepest points of dermal invasion. In such Department of Dermatology (GFM, AR, MK), University of Pennsylvania School of Medicine; and Division of Medical Oncology (DB), Thomas Jefferson University, Philadelphia, Pennsylvania, U.S.A. Reprint requests to: Dr. George F. Murphy, 235B Clinical Research Building, 422 Curie Boulevard, Philadelphia, PA 19104. Abbreviations: CY, cyclophosphamide; DFNB, dinitrofluorobenze; DNP, dinitrophenyl; DTH, delayed-type cutaneous hypersensitivity; GSV, grayscale values; HLA, human leukocyte antigen; ICAM-1, intercellular adhesion molecule-1; IL, interleukin; MNC, mononuclear cells; NK, natural killer; PBS, phosphate-buffered saline; TILS, tumor-infiltrating lymphocytes
0022-202X/93/$06.00
instances, quantification of tumor-infiltrating lymphocytes (TILs) may represent a significant independent prognostic variable [3]. The local and systemic mechanisms that account for immune responses directed against melanoma cells are incompletely understood. From a practical viewpoint, little is known concerning structural details of immunologic regression resulting from antigen-specific effector-target cell interactions. Extrapolation from established experimental systems of effector-target-cell interactions, such as acute graft-versus-host disease, suggests that apposition of effector cells of cytotoxic phenotype to degenerating target cells is a characteristic morphologic feature of antigen-directed destructive immune responses [4,5]. This phenomenon, termed ‘‘satellitosis’’ may be observed in certain primary melanomas putatively undergoing immune regression (Fig 1). Until now, experimental model systems to determine whether this histology correlates with antigenic phenotypes characteristic of melanoma-induced cytotoxicity have been lacking. We recently studied clinical and immunologic responses in human melanoma metastases, which ordinarily do not elicit significant host lymphoid infiltration, after administration of
Copyright & 1993 by The Society for Investigative Dermatology, Inc. 335S
336S
MURPHY ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 1. Morphologic characteristics of lymphoid response to primary malignant melanoma. Putative advanced inflammatory regression (A) is often characterized by a zone of lymphocytes (L) and residual melanoma cells (arrows) bordered by lamellar collagen deposition in which pigment-laden macrophages (P) are entrapped. Melanoma cells within the epidermal layer (EPI) are typically spared. Presumed earlier stages of TIL response are characterized by clustering of rare (B) or numerous (C) lymphocytes about individual melanoma cells (arrows) in a manner similar to satellitosis, as classically described in experimental graft-versus-host disease. Pronounced MNC response to a small dermal island of melanoma cells (D) is associated with lymphoid infiltration (larger arrow) and tumor cell necrosis (small arrows). (A, magnification 60; B–D, magnification 200.)
autologous melanoma vaccine. This approach is based on preadministration of cyclophosphamide (CY) to heighten host responses [6–10] to tumor-associated antigens [11–15] to which cancer patients appear to develop immunologic tolerance [16–23]. In clinical trials, it has been shown that the administration of CY prior to injection of autologous tumor vaccine can induce delayed-type cutaneous hyper-sensitivity (DTH) to autologous tumor cells and, sometimes, to regression of metastatic tumors [24]. To further augment DTH to autologous tumor antigens, the vaccine we used was conjugated to dinitrophenyl (DNP), which is known to enhance cell-mediated immunity to tumor-associated antigens [25–28]. In clinical trials using this strategy, it was recently demonstrated that 14 of 24 patients so treated exhibited immune-mediated anti-tumor responses directed against dermal and subcutaneous metastases [29,30]. In this report, we characterize the phenotype of immune responses to melanoma metastases in patients clinically responding to DNP-conjugated autologous melanoma vaccine to better understand the structural and antigenic details of host-mediated immune regression. Our data suggest that satellitosis involving T cells with a suppressor-cytotoxic phenotype is characteristic of immune regression in melanoma induced by this form of therapy. Moreover, these interactions are associated with dendritic immune cell proliferation/infiltration and intercellular adhesion molecule-1 (ICAM-1) induction in responding tumors.
MATERIALS AND METHODS General Experimental Design In the initial treatment protocol, 24 patients with metastatic melanoma were given vaccine consisting of autologous, cryopreserved, irradiated tumor cells conjugated to DNP. All patients had at least one superficial metastasis in skin or subcutaneous tissue that was directly observable and easily biopsied.
Patients were contact sensitized to the hapten by topical application of dinitrofluorobenzene (DNFB) 3 d following low-dose (300 mg/M2) intravenous CY. Two weeks later, and subsequently every 28 d, patients were injected intradermally on the upper arm with DNPhaptenized melanoma cells mixed with bacille Calmette-Gue´rin as adjuvant. CY was administered 3 d before each vaccine injection. Skin testing with DNP-conjugated peripheral blood mononuclear cells (MNC) revealed induction of DTH responses to DNPconjugated MNC in 22 of 24 patients after DNFB sensitization, without DTH to MNC alone or conjugated to a non-cross-reactive hapten, TNP. Between 2 and 4 months after initiation of DNPvaccine therapy (average two to four injections), 14 of 24 patients developed erythema, warmth, and tenderness of metastatic tumors and overlying skin. In 10 patients, one or more biopsies were obtained for immunohistochemical analysis of inflammatory cells, activation antigens, and adhesion molecules. In five patients, biopsies were obtained prior to vaccination and subsequent to vaccination at the time of peak clinical inflammatory response. In addition to qualitative immunophenotypic evaluation, pre-vaccination and post-vaccination melanoma specimens from four patients were also evaluated by computer-assisted image analysis to determine the relative number of labeled T cells infiltrating tumors (6,463–18,534 cells evaluated/lesion). Clinical and Vaccination Protocol We treated 24 patients with metastatic, surgically incurable melanoma, all of whom had at least one metastasis in skin, subcutaneous tissues, or lymph nodes that was directly observable and easily biopsied. Informed consent was obtained. The patients were sensitized to DNP by topical application of 1% DNFB in acetone corn oil on two consecutive days. Intravenous CY (300 mg/M2) was administered 3 d before sensitization.
VOL.
100, NO. 3, SUPPLEMENT, MARCH 1993
INFLAMMATORY RESPONSES TO MELANOMA
337S
Table I. Monoclonal Antibodies Used in This Study Sourcea
Antibody
Reactivity
Bectin Dickinson
Leu-1
CD5 (mature T cells)
Leu-2
CD8 (cytotoxic-suppressor T cells)
Leu-3
CD4 (helper-inducer T cells; macrophages)
Leu-6
CD1 (Langerhans cells)
Leu-7
Some cytotoxic T; natural killer. (NK) cells
Leu-11
CD16 (Fc immunoglobulin G receptor on NK cells and neutrophils)
Coulter Corp.
Dako Corp.
Leu-M1
CD 15 (monocytes)
Anti-HLA-DR
Class II molecules on B cells, g-interferon-stimulated cells
NKH-1
CD56 (NK cells)
2H4
CD45RA (suppressor -inducer CD4+ T-cell subset)
4B4
CD45RO (helper-inducer CD4+ subset)
TAC
Interleukin-2 receptor
ICAM-1
Intercellular adhesion molecule-1 (lymphocyte function associated antigen-1 ligand)
a Bectin Dickinson Immunocytometry Systems, Mountain View, CA; Coulter Immunology, Hialeah, FL; Dako Corp., Carpinteria, CA.
Two weeks later, patients were again given CY, followed 3 d later by injection of DNP-conjugated melanoma vaccine; CY and DNP vaccine were repeated every 28 d. The vaccine consisted of 10–25 106 cryopreserved, autologous, irradiated (2500 R), DNPconjugated melanoma cells prepared as previously described [31] and mixed with bacille Calmette-Gue´rin. DNP conjugation was performed by the method of Miller and Claman [32]. Patients were tested for cell-mediated immunity to hapten-modified cells by skin testing with DNP-conjugated peripheral blood MNC. Biopsy Selection Initial T-cell antigen screening of 54 excisional biopsies of metastatic melanomas from 24 patients pre-vaccination and post-vaccination revealed inflammatory infiltrates to be restricted to post-vaccination lesions undergoing clinical induration, erythema, and enlargement (n ¼ 14). Of these, four were excluded because of the presence of zonal necrosis (a potential source of secondary, non-specific inflammatory reactions) or the presence of tumor within subcutaneous lymph nodes. As anticipated, cells infiltrating extensively necrotic tumors expressed predominantly histiocytic markers (CD15), and cells infiltrating nodal metastases were predominantly of the CD4 subset and were not associated with individual cell necrosis, human leukocyte antigen (HLA)-DR induction, or satellitosis. Of the 10 remaining cases, five had at least one pre-biopsy and post-biopsy sample, permitting assessment of evolution of inflammation related to vaccination. The postvaccination specimens were obtained 2–4 months after initiation of
Figure 2. Quantification of Leu-1 þ TILs in metastatic melanomas before (solid bars) and after (hatched bars) autologous DNP-conjugated melanoma vaccine (n ¼ 4; p ¼ 0.04, 0.05, 0.005, and 0.03 for patients 1–4, respectively).
vaccine treatment, at points of maximum clinical inflammatory change. The number of inflamed tumors in a single patient ranged from 1 to more than 100 [33]. In patients with multiple superficial metastases, 25–100% of observable lesions became inflamed. Immunocytochemistry Each tissue specimen was subdivided into 1-mm - thick slabs and placed in Michel’s medium for transport and temporary storage. Samples were washed in phosphate-buffered saline (pH 7.4) (PBS), snap-frozen in liquid nitrogen, and cryostat sectioned at 5-mm intervals. For comparative assessment of proportions of inflammatory cells expressing certain epitopes, adjacent tissue sections were studied to minimize artifact resulting from local variation in cell number. Monoclonal antibodies used for initial screening and detailed analyses are enumerated in Table I. Labeling was obtained using an avidin-biotin-peroxidase complex system (Vector Laboratories, Inc., Burlingame, CA) and 3,30 diaminobenzidine as a chromagenic substrate. Briefly, tissue sections were acetone fixed for 10 min at 41C, and primary antibody (or appropriate, isotype-specific irrelevant controls) (dilution range 1: 50–1:200) was placed over each tissue section and incubated in a humidified chamber for 30 min at room temperature. After thorough rinsing with PBS, sections were overlayed with biotinylated antimouse anti-serum (1:200) for 30 min. After washing, avidinbiotinperoxidase complex was added at a concentration of 1:100 for 45 min. Subsequent to thorough PBS rinse, sections were incubated with 3,30 -diaminobenzidine and 3% H2O2 in PBS and counterstained lightly with hematoxylin. Replicate sections were prepared for each experimental condition for evaluation of cellular infiltrates in routine paraffin sections. Image Analysis Quantitative analysis of Leu-1 þ cells was performed on an image-analysis system consisting of a central computer (Sony Communications Products Company, Park Ridge, NJ), color video monitor (Mitsubishi), video camera (Southern Micro Instruments, Inc., Atlanta, GA), and Olympus BH-2 light microscope. Pretreatment and post-treatment biopsy specimens from four individuals
338S
MURPHY ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 3. Immunophenotypic characterization of TIL response in lesions showing clinical inflammation-regression after autologous vaccination. Characteristic fields of metastatic melanoma from a single patient stained with Leu-1 before (A) and after (B) vaccination. Routine microscopy (C) reveals brisk TIL response with foci where lymphocytes surround individual tumor cells (M) in a manner analogous to satellitosis (inset). Subsetting of the TIL response in adjacent tissue sections revealed the majority of cells to be CD8 þ (D) and the minority to be CD4 þ (E). Clustering of TILs about melanoma cell(s) (D and E, arrows) was consistently produced by only the CD8 þ subset (D). (A, B, D, E, magnification 100; C, magnification 200; C inset, magnification 400.)
were analyzed with cell-counting software (Southern Micro Instruments, Inc.). This system used 256 gray-scale values (GSV) from 0 to 255 (0 ¼ black, 255 ¼ white), with the range between black and white subdivided on the basis of relative shades of gray. The working area on the video screen was 466 466 pixels (which represented square meters after calibration), generating 217,516 pixels/field. The system applied a GSV to each pixel. Computer-assisted adjustment to GSV limits enabled segmentation of cells to be counted on the basis of color. Leu-1 þ cells were dark brown and therefore had lower GSV than Leu-1 cells, which were light blue. Calibration was performed to ‘‘teach’’ the computer the number of pixels occupied by a single cell of average area. To determine the number of cells in the field of interest, the total number of segmented pixels on the screen was divided by the calibrated reference pixel count. This measurement was performed to count both Leu-1 þ cells and the total number cells in each field. The accuracy of this method was confirmed by manual counting of positively labeled cells/500 tumor cells in selected specimens with concordance between 5 and 7%. Differences in the pre-vaccination and post-vaccination number of infiltrating Leu-1 þ cells for each patient were evaluated for significance using the two-tailed paired t test.
RESULTS AND DISCUSSION Metastatic Melanoma—Pre-Treatment Biopsies of metastatic melanoma demonstrated malignant cells within the dermis and subcutis, with cytologic features compatible with their known primary malignant melanomas. No specimens showed evidence of significant tumor infiltration by lymphocytes, and immunohisto-chemistry revealed numerous fields devoid of T cells (Leu-1 þ cells) (Figs 2 and 3A). Computer-assisted image analysis revealed between 54 and 421 Leu-1 þ cells/10,000 tumor cells in four pretreatment biopsies (Fig 2) (a fifth case was excluded owing to substantial cellular pigmentation impairing discrimination of immunolabeled inflammatory cells). Only rare HLA-DR þ dendritic cells were present with tumor nodules, and tumor cells expressed HLA-DR and ICAM-1 focally and weakly. These findings are consistent with the paucity of lymphoid inflammatory infiltration typically observed in metastatic melanoma deposits [34,35] and with experimental evidence indicating that cell lines derived from metastatic melanoma, in contrast to early primary lesions, fail to induce autologous T-cell proliferation [36,37]. Metastatic Melanoma—Post-Treatment In contrast to pretreatment biopsies, all specimens of metastatic melanoma (post-treatment, n ¼ 10) were heavily infiltrated by HLA-DR þ and Leu-1 þ T cells (Fig 2 and Fig 3B). Histologic analysis (Fig 3C)
VOL.
100, NO. 3, SUPPLEMENT, MARCH 1993
INFLAMMATORY RESPONSES TO MELANOMA
339S
Figure 4. Immunophenotypic characterization of ICAM-1 expression and of dendritic cells in lesions showing clinical response after vaccination. (A,B) Adjacent sections of subcutaneous melanoma, one stained with Leu-1 (A) and the other stained for ICAM-1 (dotted line demarcates junction between stroma above and tumor cells below). Zone of most pronounced infiltration by small T lymphocytes (A, upper third of tumor nodule) corresponds to most intense ICAM-1 expression by larger tumor cells (B). Less pronounced ICAM-1 expression (B, lower two-thirds of tumor nodule) corresponds to diminished TIL response (A, lower two-thirds of tumor nodule). Staining for HLA-DR in adjacent sections of the upper third (B, upper arrow), middle third (B, middle arrow), and lower third (B, lower arrow) is shown in C–E. Note the zones of diffuse class II expression by larger tumor cells and dendritic cells (C) and exclusive expression by numerous (D) and less numerous (E) dendritic cells. These zones correlate directly with areas of most pronounced (top) to least pronounced (bottom) Leu-1 þ T-cell infiltration (A) and ICAM-1 expression (B). Dendritic cells from zones represented in C and D expressed CD4 (F), although the majority were negative for Leu-Ml (G). (A and B, magnification 60; C–E, magnification 200; F and G, magnification 100.)
demonstrated small lymphoid cells diffusely infiltrating among large, anaplastic melanoma cells in a pattern identical to that described for TILs in primary melanoma [3]. Degenerative melanoma cells were frequently surrounded by peripheral clusters of lymphocytes in a pattern similar to satellitosis, as described in acute graft-versus-host disease [4,5] (Fig 3C, inset).
Computer-assisted image analysis (Fig 2) confirmed the difference in the number of TILs between pre-treatment specimens (54–421 Leu-1 þ cells/10,000 cells) and post-treatment specimens (1160–2559 Leu-1 þ cells/ 10,000 cells) (p ¼ 0.03–0.005). Immunophenotypic analysis for T-cell subclasses in adjacent tissue sections revealed that the majority of infiltrating T cells
340S
MURPHY ET AL
were CD8 þ (Fig 3D), and a minority were CD4 þ (Fig 3E ). Moreover, peripheral aggregation of CD8 þ T cells about melanoma cells in a pattern corresponding to the satellitosis typical of routine histologic preparations (Fig 3C, inset) was frequently observed (Fig 3D). Only rare cells were Leu 7 þ or Leu 11 þ , indicating that natural killer (NK) cells do not account for a significant proportion of the TIL response induced by autologous vaccine. The majority of TILs did not express immunoreactive interleukin-2 (IL-2) receptors using the Tac antibody, an observation in keeping with previously reported fluorescence-activated cell sorting analyses of inflamed tumors post-vaccination [33] and with our observations using this labeling system in experimental cutaneous DTH [38]. Whether this represents an artifact of antibody used, sensitivity of immunohistochemical localization, timing of biopsy, or the result of receptor saturation by IL-2, remains to be determined. Because TAC antibody labels only the a-chain of the IL-2 receptor, studies using recently described monoclonal reagents to the b-chain of the IL-2 receptor [39] are now clearly indicated. In vivo, TILs were recently shown to be of the CD3 þ CD8 þ phenotype, with an oligoclonal (rather than polyclonal) pattern of T-cell receptor gene rearrangement [40]. Cytotoxicity against melanoma cells was recently demonstrated to reside in cloned populations of CD8 þ T cells that recognize autologous tumor cells through the T-cell receptor and in an HLA class I-restricted manner [41]. It was not possible to reliably segregate the minority population of CD4 þ TILs into 4B4 þ (helper-inducer) and 2H4 þ (suppressor-inducer) subgroups because tumor cells demonstrated apparent expression of the CD45RO epitope (defined by 4B4 antiserum), precluding comparative enumeration of associated labeled lymphoid elements. Occasional lymphoid cells were 2H4 þ , however, representing infiltration of tumors by a suppressor- inducer CD4 þ subset that may be impaired by CY, accounting partially for its immunoenhancing effect [42]. Presumed early T-cell infiltration of metastatic melanoma at the periphery of nodules was spatially associated with tumor cell ICAM-1 expression (Fig 4A,B). As in experimental cutaneous DTH responses in human interfollicular epidermis [38,43], T-cell migration into melanoma nodules was invariably associated with enhanced expression of melanoma cell ICAM-1, indicative of possible g-interferon effect. ICAM-1 regions of tumor were never infiltrated by T cells in excess of numbers seen in pre-vaccination specimens. These observations raise the possibility that interaction between TILs and melanoma cells may be partially mediated by ICAM-1/lymphocyte function-associated antigen-1 ligand interactions. Although such interaction would potentially facilitate efficacy of immune responses, it has also been suggested that ICAM-1 expression by melanoma cells could enhance ‘‘passenger’’ dissemination via heterotopic attachment between malignant cells and leukocytes actively trafficking across vessel-tissue interfaces [44]. In zones of maximal ICAM-1 immunoreactivity (Fig 4C ), melanoma cells strongly expressed HLA-DR, whereas areas of minimal ICAM-1 staining (Fig 4D) revealed HLA-DR expression only by infiltrating CD1 dendritic cells. ICAM-1 zones of tumor (Fig 4E ), which typically were devoid of infiltrating T cells, displayed fewer, less intensely stained dendritic HLA-DR þ cells. The majority of dendritic cells infiltrating vaccine-treated melanomas were CD4 þ (Fig 4F ), although these cells were clearly separable morphologically and antigenically from more rounded CD5 þ T cells, which also may express class II molecules
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
on activation (Fig 4G). The HLA-DR þ , CD1, and CD4 þ phenotypes of these dendritic cells suggests that they belong to the family of dermal dendritic immune cells, as originally described in 1985 by virtue of differential immunoreactivity for antibodies to certain b2-microglobulin glycoproteins shared by Langerhans cells and endothelial cells [45]. Recently, dermal ‘‘dendrocytes’’ exhibiting positivity for the fibrin-stabilizing factor, factor XIIIa, have been characterized [46]. These cells probably represent a congruent population of dermal dendritic immune cells. It is intriguing to speculate that some of these dendritic cells may be responsible for fibrin–fibronectin polymerization, antigen presentation, or cytokine production that augments the T-cell migration or activation directed against melanoma cells after immunization. The findings in this study support the concept that immunologic responses to metastatic malignant melanoma after administration of autologous, hapten-conjugated vaccine involve infiltrative lymphoid populations predominantly of the CD8 þ phenotype. Tumors so affected may demonstrate ICAM-1 and HLA-DR expression, presumably induced by g-interferon, a lymphokine that has been shown to regulate T-cell responses to melanoma in vitro [34]. Moreover, HLA-DR þ , CD4 þ , and CD1 dendritic immune cells appear integral to T-cell infiltration of melanomas after autologous vaccination. These observations in a relevant clinical model of antigen-induced melanoma regression may aid in understanding naturally occurring structural alterations typical of tumor-directed host immunity in primary melanomas, as well as the cellular interactions that underpin efficacy of melanoma immunotherapy in responsive recipients. This work was supported by grants CA-40358 and CA-39248 from the National Institutes of Health.
REFERENCES 1. Rowden G: Immunology of melanoma. In: Mihm MC Jr, Murphy GF, Kaufman N (eds.). Pathobiology and Recognition of Malignant Melanoma. Williams & Wilkins, Baltimore, 1988, pp 50–78 2. Clark WH Jr, Elder DE, Van Horn M: The biologic forms of malignant melanoma. Hum Pathol 17:443–450, 1986 3. Clark WH Jr, Elder DE, Guerry D, Braitman LE, Track BJ, Schultz D, Synnestvedt M, Halpern AC: Model predicting survival in stage I melanoma based on tumor progression. J Natl Cancer Inst 81: 1893–1904, 1989 4. Woodruff JM, Hansen JA, Good RA, Santos GW, Slavin RE: The pathology of the graft-versus-host reaction (GVHR) in adults receiving bone marrow transplants. Transplant Proc 8:675–684, 1976 5. Guillen FJ, Ferrara J, Hancock WW, Messadi DV, Fonferko E, Burakoff SJ, Murphy GF: Acute cutaneous graft-versus-host disease to minor histocompatibility antigens in a murine model: evidence that large granular lymphocytes are effector cells in the immune response. Lab Invest 137:1874–1877, 1986 6. Maguire HC Jr, Ettore VL: Enhancement of dinitrochlorobenzene (DNCB) contact sensitization by cyclophosphamide in the guinea pig. J Invest Dermatol 48:39–43, 1967 7. Turk JL, Parker D: Effect of cyclophosphamide on immunological control mechanisms. Immunol Rev 65:99–113, 1982 8. Schwartz A, Askenase PW, Gershon RK: Regulation of delayed-type hypersensitivity reactions by cyclophosphamide-sensitive T cells. J Immunol 121:1573–1577, 1978 9. Yoshida S, Nomoto K, Himeno K, Takeya K: Cooperation between cyclosphosphamide tumoricidal activity and host anti-tumor immunity in the cure of mice bearing large MOPC-315 tumors. Cancer Res 41:2163–2168, 1981
VOL.
100, NO. 3, SUPPLEMENT, MARCH 1993
INFLAMMATORY RESPONSES TO MELANOMA
341S
10. Berd D, Mastrangelo MJ, Engstrom PF, Paul A, Maguire H: Augmentation of the human immune response by cyclophosphamide. Cancer Res42:4862–4866, 1982
30. Murphy GF, Radu A, Mastrangelo M, Berd D: Tumor infiltrating T cells in metastatic melanoma: induction by immunization with autologous, DNP-conjugated tumor cells (abstr). Lab Invest 62:70A, 1990
11. Reisfeld RA, Cheresh DA: Human tumor antigens. Adv Immunol 40:323–377, 1987 12. Old LJ: Cancer immunology: the search for specificity. Cancer Res 41:361–375, 1981
31. Berd D, Maguire HC, McCue P, Mastrangelo MJ: Treatment of metastatic melanoma with an autologous tumor cell vaccine: clinical and immunological results in 64 patients. J Clin Oncol 8:1858–1867, 1990
13. Mukherji B, Guha A, Chakraborty M, Sivanandham M, Nashed AL: Clonal analysis of cytotoxic and regulatory T cell responses against human melanoma. J Exp Med 169:1961–1976, 1989
32. Miller SD, Claman HN: The induction of hapten-specific T cell tolerance by using hapten-modified lymphoid cells. I. Characteristics of tolerance induction. J Immunol 117:1519–1526, 1976
14. Topalian SL, Solomon D, Rosenberg SA: Tumor-specific cytolysis by lymphocytes infiltrating human melanomas. J Immunol 142:3714–3725, 1989
33. Berd D, Murphy GF, Maguire HC, Mastrangelo MJ: Immunization with haptenized, autologous tumor cells induces inflammation of human melanoma metastases. Cancer Res 51:2731–2734, 1991
15. Houghton AN, Mintzer D, Cordon-Cardo C, Welt S, Fliegel B: Mouse monoclonal IgG3 antibody detecting GD3 ganglioside: a phase I trial in patients with malignant melanoma. Proc Natl Acad Sci USA 82:1242–1246, 1985
34. Guerry D, Alexander MA, Elder DE, Herlyn MF: Interferon-g regulates the T cell response to precursor and nevi and biologically early melanoma. J Immunol 139:305–310, 1987
16. Berd D, Maguire HC Jr, Mastrangelo MJ: Induction of cell-mediated immunity to autologous melanoma cells and regression of metastases after treatment with a melanoma cell vaccine preceded by cyclophosphamide. Cancer Res 46:2572–2577, 1986 17. Mitchell MS, Kan-Mitchell J, Kempf RA, Harel W, Shau H: Active specific immunotherapy for melanoma: phase I trial of allogeneic lysates and a novel adjuvant. Cancer Res 48:5883–5893, 1988 18. Berd D, Mastrangelo MJ: Active immunotherapy of human melanoma exploiting the immunopotentiating effects of cyclophosphamide. Cancer Invest 6:337–349, 1988 19. Takei F, Levy J, Kilburn DG: Characterization of suppressor cells in mice bearing syngeneic mastocytoma. J Immunol 118:412–417, 1977 20. Berendt MJ, North RJ: T-cell-mediated suppression on anti-tumor immunity. Human tumor antigens. J Exp Med 151:69–80, 1980 21. Mullen CA, Urban JL, Van Waes C, Rowley DA, Schrieber H: Multiple cancers. Tumor burden permits the outgrowth of other cancers. J Exp Med 162:1665–1682, 1985 22. Fujimoto S, Green MI, Sehon AH: Regulation of the immune response to tumor antigens. I. Immunosuppressor cells in tumor-bearing hosts. J Immunol 116:791–796, 1976 23. McBride WH, Howie SEM: Induction of tolerance to a murine fibrosarcoma in two zones of dosage—the involvement of suppressor cells. Br J Cancer 53:707–712, 1986 24. Berd D, Maguire HC, McCue P, Mastrangelo MJ: Treatment of metastatic melanoma with an autologous tumor-cell vaccine: clinical and immunologic results in 64 patients. J Clin Oncol 8:1858–1867, 1990 25. Mitchison NA: Immunologic approach to cancer. Transplant Proc 2:92–103, 1970 26. Fujiwara H, Aoki H, Yoshioka T, Tomita S, Ikegami R: Establishment of a tumor-specific immunotherapy model utilizing TNP-reactive helper T cell activity and its application to the autochthonous tumor system. J Immunol 133:509–514, 1984
35. Kornstein MJ, Brooks JSJ, Elder DE: Immunoperoxidase localization of lymphocyte subsets in the host response to melanoma and nevi. Cancer Res 43:2749–2753, 1983 36. Alexander MA, Bennicelli J, Guerry D: Defective antigen presentation by human melanoma cell lines cultured from advanced, but not biologically early disease. J Immunol 142:4070–4078, 1989 37. Guerry D, Alexander MA, Herlyn MF, Zehngebot LM, Mitchell KF, Zmijewski CM, Lusk J: HLA-DR histocompatibility leukocyte antigens permit cultured human melanoma cells from early but not advanced lesions to stimulate autologous lymphocytes. J Clin Invest 73:267–271, 1984 38. Waldorf HA, Walsh LJ, Schechter NM, Murphy GF: Early cellular events in evolving cutaneous delayed hypersensitivity in humans. Am J Pathol 138:477–486, 1991 39. Ohashi Y, Takeshita T, Nagata S, Mori K, Sugamura K: Differentiation expression of the II-2 receptor subunits p55 and p75 on various populations of primary peripheral blood mononuclear cells. J Immunol 143:3584–3595, 1989 40. Belldegrun A, Kasid A, Uppenkamp M, Topalian SL, Rosenberg SA: Human tumor infiltrating lymphocytes. Analysis of lymphokine mRNA expression and relevance to cancer immunotherapy. J Immunol 142:4520–4526, 1989 41. Gervois N, Heuze F, Diez E, Jotereau F: Selective expansion of a specific anti-tumor CD8 þ cytotoxic T lymphocyte clone in the bulk culture of tumor-infiltrating lymphocytes from a melanoma patient: cytotoxic activity and T cell receptor gene rearrangements. Eur J Immunol 20:825–831, 1990 42. Berd D, Mastrangelo MJ: Effect of low dose cyclophosphamide on the immune system of cancer patients: Depletion of CD4 þ , 2H4 þ suppressor-inducer T-cells. Cancer Res 48:1671–1675, 1988 43. Lewis RE, Buchsbaum M, Whitaker D, Murphy GF: Intercellular adhesion molecule expression in the evolving human cutaneous delayed hypersensitivity reaction. J Invest Dermatol 93:672–677, 1989
27. Fujiwara H, Hamaoka T, Shearer GM, Yamamoto H, Terry WD: The augmentation of in vivo and in vitro tumor-specific T cell-mediated immunity by amplifier T cells. J Immunol 124:863–869, 1980
44. Johnson JP, Stade BG, Holzmann B, Schwable W, Riethmuller G: De novo expression of intercellular-adhesion molecule 1 in melanoma correlates with increased risk of metastasis. Proc Natl Acad Sci USA 86:641–644, 1989
28. Flood PM, Schreiber H, Ron Y: Protective immunity to progressive tumors can be induced by antigen presented on regressor tumors. J Immunol 138:3573–3578, 1987
45. Murphy GF, Bronstein BR, Knowles RW, Bhan AK: Ultrastructural localization of M241 glycoprotein on dendritic and endothelial cells in normal human skin. Lab Invest 52:264–269, 1985
29. Berd D, Mastrangelo MJ, Green C, Clark C, Hart E: Tumor inflammatory response induced with autologous melanoma cells conjugated to dinitrophenol (DNP) (abstr). Proc Am Assoc Cancer Res 30:382A, 1989
46. Cerio R, Griffiths CE, Cooper KD, Nickoloff BJ, Headington JT: Characterization of factor XIIIa positive dermal dendritic cells in normal and inflamed skin. Br J Dermatol 121:421–431, 1989