Active-specific immunotherapy for melanoma.

Active-Specific Immunotherapy for Melanoma By Malcolm S. Mitchell, William Harel, Raymond A. Kempf, Eddie Hu, June Kan-Mitchell, William D. Boswell, Grace Dean, and Lucy Stevenson Twenty-five patients with metastatic melanoma were treated with a therapeutic vaccine ("theraccine") consisting of allogeneic melanoma lysates and a novel adjuvant, DETOX (Ribi ImmunoChem Research, Inc, Hamilton, MT). Each patient received 200 antigenic units (20 x 106 tumor cell equivalents) subcutaneously on weeks 1, 2, 3, 4, and 6. Clinical responses included one complete remission, three partial remissions, and a long-term (17-month) stability. Two other patients had mixed responses, with partial remissions of numerous subcutaneous nodules. Sites of responsive disease included primarily the skin, but ileal, breast, and a liver metastasis also responded. Removal of residual lesions in patients with partial remissions, whose other lesions had disappeared during treatment, led to long disease-free survivals. The median duration of remission was 17 months, with four of the five responders alive for at least 24 months after treatment. An increase in precursors of cytolytic T cells

ACTIVE-SPECIFIC

immunotherapy may be defined as the use of tumor-associated antigens to immunize a tumor-bearing individual in order to effect rejection of that tumor. We have previously reported the results of a phase I study of a preparation of mechanical homogenates (lysates) derived from two melanoma cell cultures, mixed with the novel immunologic adjuvant DETOX (Ribi ImmunoChem Research, Inc, Hamilton, MT). 1 In that study, we noted not only that the preparation was able to immunize From the Departments of Medicine and Microbiology, University of Southern California School of Medicine and Comprehensive Cancer Center, Los Angeles, CA Submitted June 14, 1989; accepted December 8, 1989. Supported by United States Public Health Service Grant No. RO-1 CA 36233, a grantfrom the Concern Foundation, a contract with Ribi ImmunoChem Research Inc, and gifts from Alan Gleitsman, University of Southern California Cancer Research Associates, the Morey and ClaudiaMirkin Foundation,and Virginia L. Andleman. Address reprint requests to Malcolm S. Mitchell, MD, Department of Medicine, Division of Medical Oncology, University of Southern California Cancer Center, 2025 Zonal Ave, 10-442 GH, Los Angeles, CA 90033. © 1990 by American Society of Clinical Oncology. 0732-183X/90/0805-0006$3.00/0

856

(CTLs) correlated with clinical outcome, when complete, partial, and mixed responses and long-term stability were considered. The CTLs recognized melanoma-associated antigens on many cell lines, but not other types of tumor or normal lymphocytes. Skin-test reactivity to melanoma antigens and serum antibodies against the melanoma cells was unrelated to clinical response. Toxicity was minimal, restricted largely to minor soreness at the site of injection. Only five patients, four of whom were treated with repeated courses, developed severe granulomas. These results confirm that active-specific immunization with allogeneic lysates of melanoma administered with the adjuvant DETOX can induce immunity to melanoma, and can induce regressions of disease in a proportion of patients with metastatic disease with little toxicity. J Clin Oncol 8:856-869. © 1990 by American Society of ClinicalOncology.

50% of 22 patients with melanoma against melanoma-associated antigens, but also that clinical responses ensued as a consequence in five of 17 (29%) patients with advanced, measurable disease. Moreover, we found that cytolytic T lymphocytes (CTLs) were generated during the course of immunization in patients whose lesions regressed, whereas those who had no such increase (as measured in the peripheral blood) uniformly failed to have a clinical response. These T cells were reactive against melanoma-associated antigens but were unrestricted by human leukocyte agglutinin (HLA) class I or II antigens, in contrast to corresponding cytolytic lymphocytes of the mouse. 1' 2 In an attempt to verify the clinical and immunologic activities of immunization with allogeneic melanoma materials and DETOX, we treated 25 melanoma patients with the same dose and schedule of active-specific immunization. The results indicate that it is, in fact, possible to induce clinical remissions in a proportion of melanoma patients with this treatment alone, for which the elicitation of CTLs in the patient appears to be a requisite.

Journalof ClinicalOncology, Vol 8, No 5 (May), 1990: pp 856-869

Downloaded from ascopubs.org by UNIVERSITY of OTAGO on April 22, 2019 from 139.080.135.089 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

857

IMMUNOTHERAPY FOR MELANOMA

Preparationof the Immunizing Materials

METHODS AND STUDY DESIGN Patients Twenty-eight individuals with melanoma were entered onto the study, of whom 25 completed the treatment. The other three patients had rapidly advancing disease and did not complete the series of injections. Eligibility criteria included melanoma histologically confirmed by our Pathology Department's review of slides, a Karnofsky performance status of at least 70%, with adequate liver, renal, and hematologic capacity, and the absence of CNS metastases by computed tomographic (CT) scan or magnetic resonance imaging (MRI). Patients were required to be at least 18 years old, not pregnant, and at least 4 weeks past any previous therapy. They were also required to have delayed type hypersensitivity skin-test reactivity to at least one of the following five microbial skin-test antigens: candida, mumps, trichophyton, intermediate-strength purified protein derivative, and tetanus toxoid. All patients signed an informed consent form and the California Bill of Rights enabling them to resign from the study at any time. The characteristics of the patients, including the treatments they had received, are shown in Table 1.

Since the term "vaccine" usually connotes a substance administered prophylactically, we believe a different word is appropriate for the therapy of a tumor that is already present. Thus, we have referred to our preparation as a "therapeutic vaccine" or "theraccine," a term coined at Ribi ImmunoChem Research, Inc. The detailed preparation of our therapeutic materials was described in our previous publication.' In brief, the theraccine comprised two cultures of melanoma cells, which were begun in our laboratory in the early 1980s from biopsies of subcutaneous nodules, designated as MSM-M-1 (M-l) and MSMM-2 (M-2). We have phenotyped each of the cultures on several occasions, at first with monoclonal antibodies W6/32 and L227 (American Type Culture Collection), and then with sera more specific for individual HLA antigens. Melanoma M-1 was amelanotic, grew rather slowly, expressed glycolipid GD3 but not GD2, and both HLA-A, B, C and HLA-DR antigens. Lymphocytes from the donor of M-1 were serotyped as A2, -, B12, BW62, Cw3, -, DR4, -. The donor of melanoma M-2 died several years ago, precluding the phenotyping of her lymphocytes. Serotyping of the M-1 and M-2 tumors themselves gave somewhat variable results and were considered unreliable. Melanoma M-2 was highly

Table 1. Patients On Study Patient

Age, Sex

R.B. P.H. D.C. C.Q. J V. S.O. H.H N.E. M.M. H.J. E.C. S.H. K.A. H.B. R.B. L.R. S.R. I.V. W.N. S.N. P.D. M.M. W.L. J.L. C.M.

43M 36F 55M 39M 54M 36M 31F 34F 80F 54M 67M 38M 61M 42M 53M 38M 45F 29F 62M 53F 35M 29M 53M 37M 80F

Prior Rx Surgery Surgery, Surgery, Surgery Surgery Surgery, Surgery Surgery Surgery Surgery Surgery, Surgery, Surgery Surgery, Surgery Su rgery Surgery Surgery, Surgery Surgery, Surgery, Surgery, Surgery Surgery Surgery

RT, Chemo Chemo

RT, Immunotherapy

Immunotherapy* Immunotherapy* Chemo, immunotherapy*

Immunotherapy* Immunotherapy* Immunotherapy* Chemo, immunotherapy*

Clinical Response

Status

None Mixed None None None None PR None None (Minor)t PR None CR None None None (?)t Stable None None None Mixed PR None None Nonet None

Dead Dead Dead Dead Dead Dead Alive, NED Dead Alive Dead Alive Alive Dead Dead Dead Alive Alive ýDead Dead Dead Dead Dead Dead Dead Dead

"*Prior

immunotherapy included: TNF (1), previous melanoma "vaccine" (BCG + irradiated melanoma cells) (2); ricin-conjugated monoclonal antibody (3), interferon-alfa (1), indeterminate (1). tSee text for description of response of patients M.M., R.B. and J.L. CR, complete remission Abbreviations: Rx, therapy; chemo, chemotherapy; RT, radiation therapy; PR, partial remission; CR, complete remission; NED, no evidence of disease; TNF, tumor necrosis factor.


MITCHELL ET AL

858 pigmented, smaller than M-1, grew more rapidly, expressed glycolipid GD2 but not GD3, and had detectable HLA-A, B, C antigens but not HLA-DR. Both cultures were grown to near-confluence in NUNC Cell Factories (NUNC, Roskilde, Denmark), in Roswell Park Memorial Institute (RPMI) 1640 medium with 10% fetal bovine serum without antibiotics. Before harvesting, the cells were grown in RPMI medium without fetal bovine serum for 3 to 4 days to remove most of the fetal proteins from the cell surface. The two cultures of melanoma cells were combined and mechanically disrupted in the cold with a Polytron stainless steel high-speed tissue homogenizer (Tekmar Co, Cincinnati, OH), followed by three cycles of freeze-thawing. No enzymes were used either to remove the cells from the culture flasks or for their lysis. The theraccine included a mixture of M- 1 and M-2 lysates. The batch used for this study was "MAC-4" (melanoma antigen from cultured cells), in which the ratio of M- 1 to M-2 was 1:3. The number of tumor cell-equivalents (tce) was noted, and the number of antigenic units (au) per milliliter was determined by a binding inhibition assay to be described. The lysates were then aliquoted into 2 mL sterile, nonpyrogenic glass vials, at a concentration of 200 au per milliliter, or 20 x 106 tce per milliliter. Vials chosen at random were sent for microbiologic testing at licensed commercial laboratories 0 and the remainder stored at - 70 C. We confirmed that only disrupted melanoma cells were present by microscopic examination of random samples of the lysate mixture. The absence of viable melanoma cells in our preparations was proved by inoculation of 0.2 mL (2-3 x 106 tce) into each of five nude mice and into RPMI medium with 10% fetal bovine serum, which showed no growth of tumor cells over the course of the next 4 weeks. The lysate mixture preparations were proved sterile, mycoplasma-free, nonpyrogenic, endotoxin-free, and devoid of both hepatitis virus and acquired immune deficiency syndrome virus, the latter by direct hybridization and culture with a standard susceptible T-cell line. All investigations for sterility and general safety in test animals were performed in licensed commercial laboratories, in accordance with regulations of the Food and Drug Administration under an investigational new drug (IND) allowance. Assay for Antigenic Content by Binding Inhibition In addition to determining the number of tumor cellequivalents, we measured the content of a melanomaassociated antigen: p250, the target of mouse monoclonal antibody 9.2.27, which was kindly given to us by Dr Ralph Reisfeld of Scripps Clinic and Research Foundation, La Jolla, CA. We used a binding-inhibition assay based on the 3 enzyme immunoassay of Harper et al. Detailed methods of our procedure have been published.' One au was defined as that amount of antigen present in 10 /L of undiluted standard melanoma extract that bound maximally (80% inhibition) to antibody 9.2.27. Immunologic Adjuvant: DETOX DETOX is a novel immunologic adjuvant containing detoxified endotoxin (monophosphoryl lipid A) from Salmo-

4 nella minnesota, cell wall skeletons of Mycobacteriumphlei, squalane oil, and emulsifier. This material is known to stimulate both antibody- and cell-mediated immunity in animals, and has undergone phase I trials in patients. We obtained authorization from Ribi ImmunoChem Research Inc to cross-file on their IND for DETOX, in order to use this adjuvant in our studies.

Immunization In the hospital pharmacy just before administration, 0.25 mL of DETOX (containing 250 ,g of cell wall skeletons and 25 gg of monophosphoryl lipid A) was reconstituted, mixed thoroughly, added to the appropriate amount of melanoma lysate in the vial, and again mixed thoroughly in a syringe. The dose of 200 au (approximately 20 x 106 tce) was administered subcutaneously, divided between two sites, in the triceps region or the buttocks. No injection was given into regions of previous lymphadenectomy or extensive involvement with tumor nodules. Patients were injected on weeks 1, 2, 3, 4, and 6. At least two baseline determinations of immunologic measurements were made, after which immunologic tests were performed weekly, during, and for 2 weeks after the period of immunization.

Immunologic Studies Delayed-type hypersensitivity to the melanoma lysate. A skin test with 16.6 au of lysate mixture (approximately 1.7 x 106 tee) without DETOX was administered before and 1 week after the course of vaccination. A positive skin-test reaction was defined as one in which 5 mm or more diameter of induration was observed at 48 hours. Conversion of a negative skin-test to positive or an increase in more than 25% of the product of the diameters was considered indicative of a change in delayed hypersensitivity to the components of the lysate mixture. Limiting dilution analysis of cytolytic T lymphocyte precursors. As one serial measurement of changes in cellmediated reactivity to melanoma, we chose to measure the frequency of CTLs in the peripheral blood, by our adaptation 5 of the method of Vose. More precisely, because the method with melanoma cells in vitro, it mainly restimulation involves measures precursors of CTLs as well as mature effector cells. We will therefore refer to the lymphocytes whose activity we measured in this assay as "CTL precursors." In brief, peripheral blood lymphocytes prepared from heparinized, nylon wool-column-treated blood by FicollHypaque gradient centrifugation, were washed and resuspended in RPMI-1640 medium with 10% human AB serum ("complete" RPMI). A portion of the peripheral blood mononuclear cells irradiated with 2,000 rads was used as 4 autologous feeder cells (2.0 x 10 /well). Recombinant interleukin-2 (IL-2) (Cetus; Cetus Corp, Emeryville, CA) was added as a growth factor. Several different concentrations of IL-2 were tested at various times during the course of this study. The majority of the assays were performed with 5 or 20 Cetus U/mL. Data from our earlier trial had been obtained with a higher concentration (100 Cetus U/mL) and had several other differences from our present method, making a direct comparison difficult. In this study, data from four


859

IMMUNOTHERAPY FOR MELANOMA patients treated during one period of time were considered invalid because the frequency was calculated from a comparison with control wells containing a high concentration of IL-2. These data, showing no apparent increase in cytolytic activity at any time in any of the four patients, were not included in our final analysis of frequencies of CTL precursors. For the assay, 2 x 104 to 1.25 x 10' responder cells were plated repetitively into 96 well round-bottomed microplates in 0.1 mL of complete RPMI. Thirty-two replicates at each responder cell number were routinely used for statistical accuracy. We added 5 x 103 irradiated tumor cells to wells in which a micro-MLTC reaction was set up. IL-2-induced cytotoxicity was assessed in wells containing lymphocytes, IL-2, and 2.0 x 104 irradiated mononuclear feeder cells alone. The cultures were fed on day 6 with 50 pL/well of fresh medium containing 5 U/mL of IL-2. Plates were cultured for 9 days to determine the frequency of cytotoxic cells. Microcultures of feeder cells alone and feeders with tumor cells and IL-2, served as indicators of background. To test for cytotoxicity, 5 x 10' M-1 melanoma cells labeled with 51Cr-sodium chromate were added to each well in complete RPMI. The reaction mixture was incubated for 6 hours at 370 C, after which 100 gL aliquots of supernate were harvested and their radioactivity measured in a gamma counter. Test wells were considered "positive" when values of cpm released exceed the mean control values, in wells containing autologous feeder cells and stimulator cells, by more than two standard deviations. The number of cells plated that corresponds with 37% negative wells on a plate provides an estimate of frequency, indicating one cytotoxic cell in that population according to the Poisson distribution. The method of calculation we have used is that described by Fazekas de St. Groth,6 an iterative method based on minimization of variance. That method also permits calculation of the 95% confidence interval for the frequency and the X' value of differences between two groups. We defined an increase in frequency of precursors of cytotoxic T lymphocytes as a rise to a range outside the 95% confidence interval of mean frequency of 30 normal control subjects (such as maintenance and clerical personnel, staff scientists, and technicians) assayed concomitantly and representing at least a twofold increase above baseline for that patient. In fact, most increases far exceeded these values. Specificity of the CTL precursors. To analyze the specificity of the CTL precursors generated by immunotherapy, we tested the ability of various unlabeled ("cold") target cells to inhibit the lysis of "5Cr-labeledM-l melanoma cells by the effector lymphocytes. This test was performed on all patients at two times during the immunization: on day 30 at the midpoint of treatment, and on day 60, 2 weeks after the end of the course of treatment. It was conceivable that different types of effector cells were found in the blood at the height of the response, ie, at the midpoint of therapy and during the decline phase of the response in the blood. As with our other assays, this test was described in detail.' Cytotoxic cells were generated in mixed lymphocyte-tumor cell cultures containing 2 x 106 peripheral blood mononuclear cells and 10' irradiated M-1 melanoma cells per milliliter in RPMI medium supplemented with 10% human AB serum and 5 Cetus IL-2 units per milliliter. After 9 days

in culture the effector cells were harvested and cold target competition assays were set up in round-bottomed microculture wells. Each well contained 5 x 10' "Cr-labeled M-1 target cells, 2 x 10' effector cells and various numbers of unlabeled cells as competitors. The ratio of inhibitors to target cells (I:T) ranged from 1.25 to 10. The microplates were spun at 1,000 rpm for 5 minutes and incubated for 6 hours at 370C. To estimate the amount of released "Cr, 100 gL of supernate was harvested and counted in an automated gamma counter. Percent cytotoxicity was calculated as 100 x (cpm in test well-spontaneous release/maximal release-spontaneous release). Percent inhibition was calculated as 100 x [1 (a/b)], where a = cytotoxicity in the presence of cold targets and b = cytotoxicity against labeled target cells alone. The cells used as putative inhibitors in the various experiments included nine melanoma cell lines (M-1 to M-8 and UCLA M21), a squamous cell carcinoma of the lung (Lu-1), two B-cell lymphomas (Daudi and Raji), myeloid leukemia K-562, and concanavalin-A-induced lymphoblasts from normal donors or two patients with melanoma. Serum antibodies. An enzyme immunoassay (EIA) procedure was adapted from Harel and Nelken' and was described in detail.' Fifty microliters per milliliter of serial dilutions of the test sera were added to each well containing washed M-1 melanoma cells. The concentration of immunoglobulin G (IgG) antibodies bound to the melanoma cells was detected by peroxidase-labeled staphylococcal protein A and the substrate o-phenylenediamine. A dose-dependent color intensity developed in the wells containing specific antibodies to the melanoma cells. The specificity of the antibodies was determined in selected patients by absorptions with tumor and normal cells such as melanomas, squamous lung cancer, Raji B lymphoma, and pooled allogeneic normal lymphoblasts.

Toxicity Careful records were kept of the subjective and objective toxicity of the treatment, aided by weekly laboratory analyses of peripheral blood cells and serum. Standard World Health Organization criteria were used to classify the degree of toxicity.

Evaluation of Clinical Responses By physical examination, plain x-rays, or computed tomographic (CT) scans we measured the longest perpendicular diameters of all tumor masses at weekly intervals before, during, and after treatment. Where numerous lesions (usually skin nodules) were present, we picked at least five of them as sentinel lesions for strict measurement with calipers, but examined all of them. Standard definitions involving both degree and duration of response were used to categorize a clinical response as a complete remission, partial remission, mixed response, or no response. Thus, a "complete remission" was a disappearance of all lesions for at least 4 weeks and a "partial remission" was a decrease of more than 50% in the sum of the products of greatest perpendicular diameters of all lesions, lasting at least 4 weeks. A "mixed response" is not a standard category, and was not formally included in our calculations of overall response rates, but was recorded when


860

MITCHELL ET AL

site-specific responses were obtained. A mixed response by our definition was a decrease of more than 50% in the sum of the products of greatest perpendicular diameters of all lesions in at least one type of site (such as lung, or skin), which lasted at least 4 weeks, while at the same time other sites were unaffected, and the total decrease in volume for all lesions was less than 50%. "No response" was anything less than a partial remission, including minor responses (> 25% to 50% reduction in the sums of diameters) and progressive disease. Stability of disease was not generally considered a response, except in one instance to be described, in which a patient had an exceptionally long progression-free interval.

Table 3. Sites of Response

Skin Breast Nodes Lungs Liver GI Abdominal masses Adrenal, soft tissue, bone

CR

PR

Minor

None

Total

1

4 1 1

2

1

8 1 4 10 5 3 2 5

3 1

1

10 4 2 2 5

Abbreviation: GI, gastrointestinal.

RESULTS

ClinicalResponses Objective major clinical responses (one complete and three partial remissions) were elicited in four patients (16%), and a long-term stability of pulmonary metastases (17 months) was found in a fifth patient (4%) (Table 2). In addition, two patients had a mixed response (8%). In all, evidence for measurable regression of disease was found in seven patients (28%). Sites of disease at which responses were noted are shown in Table 3. Complete disappearance of a single right posterior scalp subcutaneous nodule 0.5 cm in diameter was elicited on two occasions in one individual, S.H., a 38-year-old man. His previous therapy was surgical removal of subcutaneous nodules, which was required frequently. This patient had had three recurrences of disease within a year before undergoing treatment. While the usual course of his nodules was to grow during a 2-month period, during the first course of treatment the nodule became impalpable after 4 Table 2. Responses Responses

No. (%)

Complete remissions Partial remissions Long-term stability Mixed responses No response

1 3 1 2 18

(4) (12)* (4) (8)t (72)$

Total

25(100)

*Two patients were surgically rendered NED, one from August 19, 1987 to the present (and the other from September 16, 1988 to January 24, 1989 after their final skin nodule of four and five original nodules, respectively) was excised. had a PR in approximately 10 subcutaneous sites, while abdominal masses grew. *One patient had 33% shrinkage of liver nodule, while axillary lymph node mass increased; in another, a 1.2 cm liver lesion (? cyst) disappeared.

"tBoth

weeks, and during a subsequent course of five injections the same thing occurred. While on a "maintenance" program of monthly injections, the nodule had remained barely palpable (approximately 1 to 2 mm in diameter). No new lesions appeared during a period of 17 months during monthly injections and a complete five-dose course of immunization. In September 1988 an 8 to 9 mm lung nodule was seen on a CT scan and an 8 mm subcutaneous nodule of the right lateral chest was palpated. The patient was therefore formally removed from the study. Yet, the progress of his tumor may have been slowed, since both the lung nodule and the subcutaneous metastasis remained the same size into late January 1989. Other than the patient's following a macrobiotic diet, no treatment was administered during that time. For that reason, a further course of theraccine was given "off-protocol" in February and March 1989 without DETOX because of previous granulomas at the sites of injection. There was a consequent shrinkage of the chest wall nodule to 2 mm, and of the posterior scalp lesion to 4 mm, while the lung nodule remained 8 mm. In the absence of more promising therapy with similarly low toxicity, we plan to continue treatment with theraccine until unequivocal evidence of loss of immunologic control is noted. Partial remissions were further augmented by removal of residual nodules in two of three patients who sustained such a remission. Patient H.H., a 31-year-old woman, had four nodules in her breasts and subcutaneous tissue of the chest, abdomen, and left calf. She had had recurrences of the nodules at 4 to 6 monthly intervals during the year before treatment was begun, most removed surgically. Three of the nodules disappeared during treatment, with no clinical or histologic evidence of inflammation (Fig 1A and


861


Fig 1. Histology of subcutaneous lesions in patient H.H. A subcutaneous nodule excised before therapy is shown in (A). The lesion in (B) was regressing during treatment, without obvious signs of inflammation, while the one in (C) was a residual nodule that became enlarged and inflamed, accompanied by a hyperplastic regional lymph node (proved by biopsy). Note the degeneracy and necrosis of the melanoma cells in (B) with a scanty degree of cellular infiltration. This appearance was for more typical of the histopathology of nodules of responding patients than the intense reaction in (C). Most nodules showed lymphocytic infiltrates at the periphery of the tumor, and in the perivascular region.

IB) while the last remained. A second course of treatment was given, during which the remaining nodule, in the right breast became larger and inflamed, and a right axillary lymph node became palpable (approximately 2 cm in diameter). When the nodule was removed, it contained a large number of inflammatory cells, lymphocytes, macrophages, and plasma cells (Fig 1C) and the lymph node showed hyperplasia without evidence of tumor cells. This patient has not had a recurrence of disease since August 19, 1987 (as of January 1, 1990) when the last residual disease was removed, after which she has been maintained on maintenance therapy of monthly injections of theraccine, supplemented by approximately semiannual complete courses of immunization. The second patient, P.D., a 35-year-old man, had two 1.3 cm lesions of his ileum (Fig 2A

and B), as well as five subcutaneous nodules on his right lower extremity. Within 10 weeks after beginning therapy, after two courses of treatment, is ileal lesions had disappeared (Fig 2C and D) and remained absent for nearly 2 years, the ensuing duration of his life. All but two of his subcutaneous nodules also disappeared, and one of the two, his largest at 2 cm, had only necrotic cells on fine needle aspiration. The histology of his subcutaneous nodules before and cytology by fine needle aspiration after treatment is shown in Figs 3A and 3B. When the residual lesions were finally removed on September 16, 1988, intact melanoma cells were present, but no subcutaneous nodule recurred until January 24, 1989. This patient had received maintenance injections of theraccine monthly and a total of three complete courses of treatment. He died in late


862

MITCHELL ET AL

Fig 2. Ileal lesions in patient P.D. on a CT scan with ingested contrast material. Note the two 1.3 cm diameter lesions indicated by the white arrows (A and B), which disappeared within 10 weeks during therapy (C and D)and did not recur by the time of death in late September 1989.

September 1989, 24 months after beginning treatment. Patient H.J., a patient with large lymph node masses on both sides of his neck, had a partial remission measured (by two independent observers) by volume changes rather than diameters. By the diameters approximation, there was a 41% decrease in the masses, whereas there was more than 50% shrinkage by volume. In fact, the masses became flat after 2 weeks of active immunotherapy, and remained flat for a total of 6 weeks. A single, small liver metastasis and a small lung metastasis remained stable during that time. The total volume of all tumor masses decreased by more than 50%, therefore, this response was classified as partial remission rather than a mixed response. Patient L.R., a 38-year-old man, had five lung nodules, which were measured on repeated CT

scans. This patient had essentially no change in the size of the nodules for at least 17 months, and they were little increased as of October 6, 1989. However, on January 17, 1989, four new lesions were found in the brain, which forced us to abandon immunotherapy. Remarkably, the patient is still alive (as of January 1, 1990); having had no growth of two of the brain lesions, and surgical resection of the other two after a gradual increase in size. An 81-year-old woman, M.M., had shrinkage of her subcutaneous nodules by more than 50% within 4 weeks. These lesions grew slightly and remained at 68% of their original size (ie, a 32% shrinkage) by the end of the treatment. Although this constituted a "minor response" in some classifications, (25% to 50% shrinkage for at least 4 weeks), this was classified by our criteria as "no response." Two other patients had a mixed re-



863

Fig 3. Histology (A) before treatment and fine needle cytology (B) after treatment of subcutaneous lesions from the lower calf of patient P.D. There was some lymphocytic infiltrate even before therapy, but the melanoma cells were viable, with gradual enlargement of the nodule. After therapy, no viable cells were seen on fine needle aspiration cytology. However, the patient declined to have the lesion resected, and there was subsequent regrowth at the site. The nodule was resected in September 1988, and the patient had no evidence of disease until January 24, 1989.

sponse, with more than 50% shrinkage of approximately 10 skin nodules but growth in massive abdominal masses. Of the remainder of the patients with no response, one patient (J.L.) showed a 33% decrease in a large liver metastasis after one course of treatment. Unfortunately, he had more than 25% increase in the size of an axillary mass, and was taken off the study. Although we consider this response in the liver encouraging, because the liver mass did not decrease by at least 50%, we did not designate his

response to treatment as a mixed response. Patient R.B. showed disappearance of one of two liver lesions, a 1.2 cm diameter hypodense region, during treatment, with stability in lung and skin lesions, but because the radiologist could not be certain that the lesion was not a cyst we did not classify that as a response. Durationsof Response Table 4 illustrates the durations of response in the five patients who achieved what we consid-


864

MITCHELL ET AL Table 4. Durations of Response

Patient

Sites

Rx Begun

PR

H.H. H.J. S.H.

Breast subcutaneous Nodes, lung, livert Subcutaneous

4/8/87 5/13/87 6/30/87

4/22/87* 5/20/87 7/7/87

L.R. P.D.

Lung, previous subcutaneous Ileum subcutaneous

9/8/87 9/30/87

(Stable) 12/2/87*

CR

PD or Recurrence

Duration (months)

3/8/88

NED 6/30/87 9/13/88

30 + 1.5 15 (PR) 6 (CR) 17 16

1/7/89 1/24/89

NOTE: All patients with a long duration of response have been treated with maintenance vaccination; monthly injections supplemented by semiannual full courses of immunization. Abbreviations: Rx, therapy; PR, partial remission; CR, complete remission; PD progressive disease; NED, no evidence of disease. *Patients H.H. and P.D. were surgically rendered free of disease on 8/19/87 and 9/16/88, respectively. H.J.'s small lung and liver nodules were unchanged during treatment.

"tPatient

ered a useful clinical response to the theraccine. All of the patients have been treated with maintenance courses of monthly injections, with a full course of immunization semiannually. The median duration of remission was 17 months. Rendering patients "disease-free" by resecting residual lesions after complete regression of all others has probably been useful in this setting of continual immunization against the tumor, in that two patients (and one from our phase I study) have continued without evidence of disease for prolonged periods of time. In fact, four of five responders lived _ 24 months from the start of therapy, three of whom are still alive as of January 1, 1990. Toxicity Most patients had local soreness at the site of injection for 24 hours, beginning 1 day after the injection. Fever and malaise consistently occurred in one patient, P.D., who achieved a good partial remission. Otherwise the therapy was largely without major side effects or toxicity. Not only were people able to continue their daily activities without interruption, several of the patients who had demanding occupations, such as a professional jockey and a preparatory school headmaster, pursued their vigorous activities with little or no diminution of ability. In fact, the only toxicity of any severity that we encountered was inflamed granulomas at the sites of injection in five patients. Sterile abscess formation occurred in all five patients, which usually required surgical lancing after initial spontaneous drainage occurred. In all but one of the patients, the severe granulomas were manifest only after repeated courses of immunization, and usually after at least 10 injections. In other

words, this was a complication generally restricted to successfully treated patients. However, one patient developed abscesses after only five injections, although she had never received related substances previously. The granulomatous response was to the DETOX portion of the theraccine, because we were able to continue therapy in the four patients who had sustained a clinical response with theraccine alone without further complications. Immunologic Responses Precursorsof CTLs. We made several modifications of our techniques for measuring CTL precursors, particularly a change in the concentration of IL-2 in the medium during the 9-day incubation period. Our best results here were obtained with 5 Cetus U of IL-2 per milliliter in test wells, and 0 and 5 U in controls, but our recent evidence suggests 20 to 25 Cetus U/mL as optimal for generating specific CTLs without eliciting nonspecific lymphokine-activated killer cells. Precursors of CTLs were increased in frequency in 14 (66%) of 21 patients in whom a complete series of studies was performed (Tables 5 and 6). Four patients, including one responder, had a series of assays performed that we considTable 5. Increase in CTL Precursors Versus Complete, Partial, or Mixed Responses, and One Long-Term Stability

CTL increased CTL not increased Total

Responders*

Nonresponders

Total

6 0 6

8 7 15

14 7 21

NOTE: P = .05 by Fisher's exact test. *One partial responder excluded from analysis on whom data were considered invalid.


865

IMMUNOTHERAPY FOR MELANOMA Table 6. Increase in CTL Precursors Versus Response: Complete or Partial Remissions and One Long-Term Stability

CTL increased CTL not increased Total

Responders*

Nonresponders

Total

4 0 4

10 7 17

14 7 21

NOTE. P = .16 by Fisher's exact test. *One partial responder excluded from analysis on whom data were considered invalid.

ered unreliable and were excluded from the analysis. Baseline values were commonly 5 1:50,000 by our current assays, which we feel are more specific but less sensitive than earlier versions. The mean baseline reactivity of all patients was 1.05/104 lymphocytes (SD, 1.4). In contrast, 30 normal individuals had a mean value of 0.23 (SD, 0.15). The mean peak level achieved during immunization was 6.88/104 cells (SD, 10.8). Peak values of one precursor per 5,000 or 10,000 circulating lymphocytes were frequently found in responding patients. The relative increase of at least fivefold to 25-fold was on the same order as that found with our previous methods, and two of our responders (H.H. and P.D.) had increases of 30- and 38-fold. The peak of the CTL increase in peripheral blood occurred as early as day 8, but was found just as frequently on day 22. In either case, the rise and fall was abrupt, with a diminution towards baseline values during the next 2 to 3 weeks. Patients followed during second and third courses of treatment showed the same rise and fall as during their first series of injections. As shown in Table 5, there appeared to be a correlation (Fisher's exact test, two-tailed, P = .05) between an increase in CTL evoked by active immunotherapy and a clinical response if one included all responses, eg, major and mixed responses and long-term stability. In contrast with our phase I study,' in which there was a statistically significant correlation between major responses (complete and partial remissions) alone and an increase in CTL precursors (P = .04), we could not demonstrate such a correlation here, even when the long-term responder was considered a "responder" (Table 6). Nevertheless, confirming our earlier observations, none of the patients who lacked an increase in CTL precursors had a clinical response. Specificity. Cold target competition assays

on the 10 patients in whom successful assays were performed on two occasions, and who also had more than 30% cytotoxicity at a ratio of 40:1, are summarized in Fig 4. With M-l as the target, a variety of allogeneic cell lines of melanoma inhibited the interaction of effectors and labeled melanoma cells. We tested seven different cell lines in our library, and all exerted more than 30% inhibition. As illustrated in the figure, unlabeled M-1, M-3, and M-21 melanoma cells inhibited the interaction of effectors and labeled target cells by a mean of 40% to 50%. Squamous lung carcinoma and B lymphomas were not inhibitory, nor were normal allogeneic lymphoblasts from patients with melanoma. We analyzed, by Student's paired t-test, the degree of inhibition by various types of inhibitor cells. Melanomas as a group differed from the nonmelanomas (P < .001). Each individual melanoma also differed from nonmelanoma cells, when analyzed individually or as a group (P < .01), except melanoma M-2. It differed from the other melanomas in its degree of inhibition (P < .05). M-2, in fact, was not significantly more inhibitory than K-562 (P = .31), but was more competitive than either the lung carcinoma or B lymphoma (P = .00013 and .024, respectively). K-562, which we thought to be the best cold target competitor among the nonmelanoma cells, nevertheless was significantly less inhibitory than all of the melanomas except M-2. In addition, specific statistical comparisons showed it to be no different from Daudi B lymphoma (P = .056), although it was a better inhibitor than squamous lung carcinoma Lu-1 (P = .015). Inhibition by anti-CD3 (Leu 4) proved that the cytolytic lymphocytes were T cells. However, monoclonal antibodies to class I (W6/32) and class II (L227) HLA antigens failed to inhibit the killing, even though our bulk cultures contained both CD4+ and CD8+ CTLs as subsequently proved by cloning (W. Harel and M.S. Mitchell, data to be published). This lack of involvement of the major histocompatibility complex in the interaction of CTLs and melanoma cells reinforces the results of cold target competition, and confirms our previous results.' Antibody synthesis. None of the patients had a titer of antibodies against melanoma cells before treatment. Eleven of the 25 patients (44%) produced serum antibodies reacting with mela-


MITCHELL ET AL

866 "^

50 -

40 C 0

r_

30 -

20 -

4-JR

10 -

a

II

I K562

I

M-1

0

M-3

M21

M-2

Lu- 1

BL

I NL

Cell Unes

M

1st CTI (day 30)

EM

) I

I

AntiCD3

AntiAntiHLA-II HLA-1

2nd CTI (day 60)

Fig 4. Cold target competition assays of bulk cultures of CTLs. Means + SE are depicted. Reactivity against target melanoma was inhibited by several allogeneic melanomas and anti-CD3, but generally not by other tumors or anti-HLA class I or II.Specificity was the same on days 30 and 60.

noma cells as a result of active immunotherapy. Included in this group were those whose antibody titer exceeded 1:20; two other individuals had a titer of 1:20. None of the patients had a titer before treatment. To test for specificity, absorptions were performed on eight patients. Six of the eight had antibodies that were absorbed not only by melanoma cells but by other tumors, particularly lung carcinoma, B lymphomas, and K-562. In four of the six patients, normal fibroblasts also absorbed the serum reactivity. In the two patients whose serum antibodies reacted only against melanoma, absorption of both sets of sera with M-l removed reactivity. In one, M-21 also removed reactivity. No other melanoma or tumor cell line absorbed reactivity from these two sets of sera. What target epitopes were being identified by the nonspecific antibodies in the majority of the patients is uncertain. It was not due to fetal bovine serum, because that substance did not absorb the reactivity. Not surprisingly, considering the general lack of specificity of the antibody production, we

failed to find a correlation between antibody production and a clinical response (Table 7). While three of the seven responders (including long-term stability and mixed responses), developed antibodies, eight of the 18 nonresponders also did so. Four of the seven responders failed to make antibodies, as did 10 of the 18 nonresponders. Delayed-type hypersensitivity. Delayed-type hypersensitivity skin tests to the melanoma lysates were negative in all individuals before treatment. After treatment, three patients (S.O., J.L., R.B.) showed a conversion of reactivity to positive. None of these patients was a responder. Furthermore, after maintenance courses of treatment, the responding patients were repeatedly Table 7. Serum Antibodies and Clinical Response

Antibodies increased Antibodies unchanged Total

Responder

Nonresponders

Total

3 4 7

8 10 18

11 14 25

NOTE. P = .34 by Fisher's exact test.


867


retested and invariably failed to exhibit a positive skin test. DISCUSSION The results of this phase II trial of activespecific immunotherapy in 25 patients treated with a standard dose of melanoma immunogens derived from allogeneic lysates, together with the novel adjuvant DETOX, 4 generally confirm the immunologic and clinical activity of this type of preparation. In our first report, a phase I study in which doses ranging from 5 x 106 tee to 4 x 10' tee, and in which low-dose cyclophosphamide was administered to one-half the patients in each group, we could not be certain whether the responses were ascribable to a particular dose, and in fact whether the theraccine (lysates plus DETOX) alone could consistently elicit those responses. Here we have used a constant dose of 2 x 107 tce (200 antigenic) without cyclophosphamide and found 20% objective responses, with a median duration of 17 months. This includes a patient with no regression but more than a year's stability of pulmonary disease. Mixed responses, with partial remission of large numbers of skin nodules, were noted in two other patients, and at least one "nonresponder" showed a 33% shrinkage in his large liver lesion after only one course of treatment. Whether low-dose cyclophosphamide can improve the regimen will be studied specifically in the future. Yet, our results have shown that cyclophosphamide is not an essential adjunct to active-specific immunotherapy with allogeneic materials in order to achieve immunization and clinical regressions of disease. With weaker immunogens, such as autologous tumor cells, cyclophosphamide may be more important.8 Toxicity was minimal, without anaphylactic reactions, and was principally the consequence of granuloma formation in response to the DETOX component. The patients who achieve a longterm response and receive more than 10 injections to maintain it are those in whom we have usually encountered such granulomas. In all, a totall of 13 of 85 patients (15%) treated with various batches of our allogeneic melanoma lysates and DETOX have had clinically significant granulomas, including five with draining sterile abscesses, but also eight others with red, tender, firm masses. This can be ameliorated to some

degree with antiinflammatory drugs, but is best handled by reducing the DETOX to 0.125 mL or less, or omitting that portion entirely from subsequent treatments. An increase in CTL precursors in the blood during immunization was correlated with clinical response, if one includes all responders such as a patient with long-term stability and two with mixed responses. Unfortunately complete and partial responses were too few in this trial to permit a statistically significant correlation to be drawn between effector cell generation and a major objective response. However, as we noted in our first trial,' the failure to generate CTLs was invariably associated with a failure to cause shrinkage of tumor masses. In other words, only the subgroup of patients in whom an appropriate T-cell-mediated immunity was elicited had the potential to reject their tumor. Whether the CTL was the actual and only effector cell is not certain, and, in fact, other correlations might exist between helper or delayed-type hypersensitivity T cells and a clinical response. Nevertheless, it is clear that the determination of the frequency of precursors of CTLs is a valuable measurement in helping to predict a clinical response. While a fuller discussion of the unusual major histocompatibility complex (MHC)-unrestricted CTLs will be presented in a subsequent communication, we should note that they are not lymphokine-activated killer cells, which are not T cells but are instead derived from natural killer cells. Our CTLs are true (CD4+) T cells, but seem to recognize a melanoma-associated antigen independent of HLA antigens, in contrast to mouse CTL in which specific antigens and MHC antigens are corecognized by T-cell receptors. Antibody synthesis has not correlated with clinical response in either of our two clinical trials, nor could we demonstrate any significant degree of tumor-specificity here. Neither was a change in skin-test reactivity to melanomaassociated antigens associated with a clinical response, and in fact only three patients showed reactivity after treatment. Our inability to induce delayed-type hypersensitivity to the immunizing materials contrasts somewhat with results of others, 9 and we have no immediate explanation for the discrepancy. We have observed that patients who sustain a remission of disease as a result of immunologic


MITCHELL ET AL

868 therapy frequently have a considerable duration of response even when the remission was only partial (Table 4). In fact, long-term stability (> 1 year) of previously progressive melanoma is a feature not only of patients whose disease measurably decreases but also of some who have no shrinkage or responses less than partial remissions. Patients rendered free of tumor by resection of a residual tumor mass that has failed to disappear during treatment while other nodules have are more akin to patients with complete remissions than those artificially rendered diseasefree by surgery alone. In contrast to chemotherapy-induced responses, it is possible that a new equilibrium between the patient and the tumor can be induced by immunotherapy-perhaps through the induction of a larger number of memory T cells-which is manifested more by longevity than by the objective degree of regression. Statistical data supporting this contention, linking levels of CTL precursors with survival regardless of whether a clinical response was noted, have been derived from our two clinical trials with theraccine (Von Eschen K et al, data to be published). We are now combining active-specific immunotherapy with IL-2, to try to enhance the level of CTLs created by the theraccine, as well as with a pretreatment with cyclophosphamide to diminish the level of tumor-specific T-suppressor cells that might exist. The latter are very difficult to measure precisely in humans, but the assumption in our study is that they may potentially interfere with our ability to immunize our patients. Other cytokines might reasonably be combined in the future, particularly IL-1, which McCune et al (personal communication) have found active in increasing immunity to renal cell carcinoma in mice. Further refinement of active-specific immunotherapy itself will involve a purification of the

relevant antigens, through two complementary approaches. We hope to deduce which antigens are immunogenic in humans by studying the ability of peptides from melanoma lysates, separated by Western immunoblotting, to stimulate cloned T lymphocytes derived from successfully immunized patients.I'0l, The ability of those peptides to block CTLs from reacting against target melanoma cells will also be used as a test for their immunogenicity. By an alternative approach, we are also attempting to produce fusion proteins from complementary DNA (cDNA) clones derived by subtractive hybridization techniques from melanoma cells. Messenger RNA (mRNA) from squamous lung carcinoma Lu-1, a tumor shown by our assays not to have antigens in common with melanomas, will be "subtracted" from melanoma mRNA to narrow the range of derived proteins to be screened for immunogenicity. Our panel of human monoclonal antibodies against melanoma 2,'13 and, perhaps more relevantly, our T-cell clones, will be used for the screening of those proteins. It may eventually be possible to produce a predetermined mixture of appropriate immunogens to include in the theraccine instead of the crude mixture of immunogenic and, undoubtedly, superfluous nonimmunogenic materials now used. Since the effectiveness of this therapeutic vaccine has been demonstrated in two clinical trials in advanced disease, we feel that it may be even more useful in treating microscopic residual disease as an adjunct to resection of a primary deeply invasive melanoma and any lymph nodes involved with tumor. ACKNOWLEDGMENT It is a pleasure to acknowledge the expert technical assistance of Carl Hadley and Janet Stahlberg in the performance of immunologic assays.

REFERENCES 1. Mitchell MS, Kan-Mitchell J, Kempf RA, et al: Active specific immunotherapy for melanoma: Phase I trial of allogeneic lysates and a novel adjuvant. Cancer Res 48:58835893, 1988 2. Zinkernagel RM, Doherty PC: Restriction of in vitro T cell-mediated cytotoxicity in lymphocytic choriomeningitis within a syngeneic or semiallogeneic system. Nature 248:701702, 1974 3. Harper J, Bumol T, Reisfeld R: Serological and biochemical analyses of monoclonal antibodies to human melanoma-associated antigens. Hybridoma 1:423-432, 1982

4. Ribi E, Cantrell JL, Takayama K, et al: Lipid A and immunotherapy. Rev Infect Dis 6:567-572, 1984 5. Vose BM: Quantitation of proliferative and cytotoxic precursor cells directed against human tumors: Limiting dilution analysis in peripheral blood and at the tumor site. Int J Cancer 30:135-142, 1982 6. Fazekas de St. Groth S: The evaluation of limiting dilutions assays. J Immunol Methods 49:R11-R23, 1982 7. Harel W, Nelken D: An enzyme immunoassay for detection of anti-sperm antibodies. Amer J Reprod Immunol and Microbiol 8:137-140, 1985


869

IMMUNOTHERAPY FOR MELANOMA 8. Berd D, Maguire HC Jr, Mastrangelo MJ: Potentiation of human cell-mediated and humoral immunity by low-dose cyclophosphamide. Cancer Res 44:5439-5444, 1984 9. Bystryn JC, Jacobsen S, Harris M, et al: Preparation and characterization of a polyvalent human melanoma antigen vaccine. J Biol Response Mod 5:211-224, 1986 10. Young DB, Lamb JR: T lymphocytes respond to solid-phase antigen: A novel approach to the molecular analysis of cellular immunity. Immunology 59:167-171, 1986 11. Hersey P, MacDonald M, Werkman H: Western blot

analysis of antigens on melanoma cells recognized by cytotoxic T cells. J Natl Cancer Inst 80:826-835, 1988 12. Kan-Mitchell J, Imam A, Kempf RA, et al: Human monoclonal antibodies directed against melanoma tumorassociated antigens. Cancer Res 46:2490-2496, 1986 13. Imam A, Mitchell MS, Modlin RL, et al: Human monoclonal antibodies that distinguish cutaneous malignant melanomas from benign nevi in fixed tissue sections. J Invest Derm 86:145-148, 1986


Intralesional immunotherapy for melanoma.

Intratumoral immunotherapy for melanoma.

Specific active immunotherapy for melanoma.

Combination immunotherapy breakthrough for melanoma.

Advances in immunotherapy for melanoma.

Immunotherapy for human malignant melanoma.

Melanoma immunotherapy.

Predictive factors for immunotherapy in melanoma.

Immunotherapy for advanced melanoma: future directions.

Advances in immunotherapy for melanoma management.

Metastatic melanoma and immunotherapy.

Malignant melanoma and immunotherapy.

Active immunotherapy in malignant melanoma.

Melanoma immunotherapy dominates the field.

Melanoma-associated antigen synthesized in vitro for active specific immunotherapy.

Concurrent radiotherapy and ipilimumab immunotherapy for patients with melanoma.

Localized CD47 blockade enhances immunotherapy for murine melanoma.

Sequential chemotherapy and immunotherapy for the treatment of metastatic melanoma.

Beyond CTLA-4: novel immunotherapy strategies for metastatic melanoma.

Toxicity management of immunotherapy for patients with metastatic melanoma.

Immunomodulatory Monoclonal Antibodies in Combined Immunotherapy Trials for Cutaneous Melanoma.

Stereotactic radiosurgery and immunotherapy for metastatic spinal melanoma.

Advances in Immunotherapy for Melanoma: A Comprehensive Review.

Non-BRAF-targeted therapy, immunotherapy, and combination therapy for melanoma.