Preclinical studies have shown that low dose IL-12 can potentiate cytotoxic lymphocyte responses. Since previous trials have demonstrated significant toxicity from high dose recombinant human IL-12 (rhIL-12), we sought to determine an optimal biological dose for rhIL-12 at lower doses when combined with peptide antigens. Two studies were undertaken. The rhIL-12 was administered at doses of 0 (placebo), 10, 30 and 100 ng/kg, subcutaneously in one study and intravenously in the other. Apart from IL-12 dosing, the studies were identical. Subjects had evaluable stage III or IV melanoma which expressed Melan-A by RT-PCR or immunohistochemistry. Melan-A26-35 (EAAGIGILTV) and influenza matrix58-66 (GILGFVFTL) peptides were administered intradermally on weeks 1, 2, 3, 4 and 9. Twenty-eight subjects were enrolled, of whom 24 were evaluable for clinical and immunological responses. Therapy was well tolerated, the main adverse event being influenza-like symptoms. Immunological monitoring included the evaluation of cutaneous reactions and assays for antigen-specific T-cells. Clinical responses included a complete response in a subject with small volume subcutaneous disease, a partial response in a subject with hepatic metastases, and mixed responses in pulmonary, pleural and nodal disease. Biopsies of accessible tumors showed infiltration with CD4+ and CD8+ lymphocytes capable of lysing Melan-A peptide-pulsed targets in vitro. No clear dose-dependent effect of rhIL-12 could be determined. The rhIL-12 given either s.c. or i.v. was well tolerated at doses of 10-100 ng/kg. Clinical and immunological activity has been observed in this study where peptides were administered either with or without low dose rhIL-12.
This article was published in Cancer Immunity, a Cancer Research Institute journal that ceased publication in 2013 and is now provided online in association with Cancer Immunology Research.
Immunotherapy of malignant melanoma entered a new era with the identification of HLA class I-restricted peptide epitopes that can serve as targets for specific cytotoxic T-lymphocytes (CTL). Clinical trials with members of the MAGE family of antigens (1, 2) and melanocyte differentiation antigens (3, 4) have shown clinical and immunological responses, raising the possibility that peptide vaccination strategies might be developed into effective anticancer therapy (5, 6, 7). These early studies have identified some potential limitations to melanoma vaccine efficacy, including heterogeneity of antigen expression and loss of HLA class I expression in tumors, resulting in escape from antigen-specific cytotoxic effectors (8). Furthermore, although peptides alone showed some activity, it was clear that immune responses might be enhanced by the use of adjuvants such as cytokines, helper epitopes and dendritic cells (9).
Cytokines can enhance afferent and efferent aspects of the T-cell immune response. GM-CSF has been shown to enhance immune responses when administered with peptides in murine (10) and human studies (3, 11). This effect is thought to be mediated by dendritic cells, for which GM-CSF is a key growth factor (12). Cytokines that act predominantly on cellular effectors can also be used in an attempt to enhance cellular responses. T-cell growth factors such as IL-2 can enhance anticancer vaccine effects (13), an observation that provides a rationale for using lymphocyte growth factors to improve vaccine efficacy.
IL-12 is an heterodimeric glycoprotein which is a regulator of cell-mediated immunity. It appears to play a critical role in both the initiation and maintenance of cellular immune responses. Receptors for IL-12 are restricted to T- and NK-cells (14). IL-12 has been shown to induce IFN-gamma production by T- and NK-cells (15, 16), enhance the cytolytic activity of NK- and T-cells (15, 17), induce the proliferation of CD4+ and CD8+ cells (17), and direct differentiation of Th1 cells from progenitors (18). It is secreted by dendritic cells (19) and may act to enhance T-cell responses following priming with antigen. Peripheral blood mononuclear cells, particularly monocytes or macrophages, also produce IL-12 after stimulation by bacterial and parasitic products. In murine models, IL-12 can suppress tumor growth in vivo, an effect mediated by CD8+ cells (20).
In phase I clinical trials, rhIL-12 has been administered intravenously and subcutaneously using a variety of schedules (21, 22, 23, 24). In all studies, fever was observed in most subjects at low doses. At the maximum tolerated dose (MTD) of 500 -1,250 ng/kg the main toxicities were transaminitis, leucopenia, fatigue and stomatitis. Tumor responses were observed in subjects with melanoma (21) and renal cell carcinoma (21, 22, 23, 24). In a phase II study, the daily administration of 500 ng/kg rhIL-12 with no preceding test dose caused severe toxicity in several subjects, with asthenia, hypotension, renal impairment, hepatotoxicity, stomatitis and gastrointestinal bleeding being observed (25). Toxicity was associated with very high IFN-gamma levels and alternative strategies for best exploiting the biological effects of rhIL-12 were considered necessary.
In animal studies, Noguchi et al. (26) showed that IL-12 could enhance the generation of tumor-specific CTLs following vaccination with a tumor peptide. This effect was optimal at lower doses of IL-12 and was inhibited at higher doses. The notion that the MTD is the optimal dose for maximum biological effect may thus not apply for cytokines such as rhIL-12.
Since the optimal biologically dose and route of administration of rhIL-12 when combined with peptides are not known, the current studies were designed to evaluate the safety and toxicity of rhIL-12 when given i.v. or s.c. in combination with peptide antigens in patients with advanced melanoma. Secondary endpoints of the study included the evaluation of immunological and tumor responses. Two studies were performed in parallel using different routes of rhIL-12 administration: in Melbourne the i.v. route was used while in Frankfurt the s.c. route was used. These studies are reported together.
Twenty-eight subjects entered two studies: 12 in the i.v. study and 16 in the s.c. study. Four subjects in the s.c. study were replaced because of rapid disease progression requiring their removal from the study within 4 weeks. Subject characteristics are shown in Table 1. There were 17 men and 11 women. The median age was 55.5 years (range 27 - 78). Only 5 had received no prior treatment other than surgery, with 6 receiving radiotherapy, 11 chemotherapy, 5 biochemotherapy, and 16 immunotherapy (11 IFN-alpha, 4 vaccinia oncolysate (27), 2 peptide vaccination, 1 IL-2, and 1 autologous melanoma vaccine). Twenty-three patients had visceral metastases, a poor prognostic factor in patients with metastatic melanoma. rhIL-12 was administered to 21 subjects and seven received placebo. 15 subjects completed all protocol-designated treatments and evaluations. The sole reason for non-completion was disease progression.
All 28 patients were evaluable for the primary endpoint of safety and toxicity. rhIL-12 at doses of 10, 30 and 100 ng/kg combined with intradermal peptide vaccination was well tolerated, with no grade IV toxicities observed. Tables 2 and 3 summarize the toxicities observed. Constitutional side effects were common in both studies, including fever, lethargy, headache and myalgia/arthralgia. Febrile reactions, lethargy and headache were mostly seen with i.v. dosing, fever being more common at increasing rhIL-12 doses. Lethargy, headache, nausea/vomiting and pain (frequently tumor-associated) were frequently observed in subjects receiving placebo. Overall, more toxicities were reported with the i.v. route of administration.
Immune responses - cutaneous responses to peptide antigens
Twenty-five subjects were evaluable for cutaneous immune response to Melan-A and influenza matrix peptides (Tables 4 and 5). Sixteen and 17 subjects had a baseline cutaneous response to Melan-A and influenza matrix peptides respectively. An example of the histopathological appearance of these reactions is shown in Figure 1. Typically, perivascular lymphocyte aggregates were seen in the dermis, which on immunohistochemical analysis contained lymphocytes that expressed CD4 or CD8.
According to the criteria defined in Materials and Methods, 2 patients had a cutaneous reaction response to the Melan-A peptide and 4 to the influenza matrix peptide. Overall there was 1 cutaneous response to either peptide in the i.v. study (a subject who received placebo) and 5 responses in the s.c. study. This difference is not statistically significant. No effect of rhIL-12 dose on the likelihood of a cutaneous response could be discerned.
Responses to the Melan-A peptide before and after vaccination are shown in Table 4. The diameters (in mm) of cutaneous reactions against Melan-A and influenza matrix are shown for two subjects in Figure 2. The first (patient 9) shows a baseline response to both antigens but no enhancement following vaccination. The second (patient 1) shows induction of a strong influenza response and a Melan-A response on a single occasion. Because this Melan-A response was not sustained for more than one reading, this subject was not deemed a Melan-A responder (Table 4). Following vaccination and administration of the rhIL-12 study drug, two subjects showed an increase in the Melan-A DTH score. These were patients 15 and 27, who received rhIL-12 s.c. at the 10 and 100 ng/kg dose levels respectively. In addition, eight subjects with baseline DTH reactions showed evidence of reduced responsiveness at the final timepoint (day 57). Although patient 15 had an initial enhanced response this was followed by a reduction at day 57.
Baseline influenza responses were common (Table 5). Seventeen subjects had a baseline response ranging from 4 mm to >8 mm with necrosis. With vaccination there was enhancement of the response in 4 instances and reduced responsiveness in 2. It was not possible to determine an effect of the dose or route of administration of rhIL-12 on the induction or enhancement of cutaneous reactions.
Immune responses - CTLp responses
CTLp responses are shown in Tables 6 and 7 and Figure 3. Scores in the Tables correspond to the number of positive wells out of a maximum of 24. Figure 3 illustrates the data for an unequivocal Melan-A responder (patient 18) who had a single positive well at baseline (Figure 3A), which increased to 24 positive wells on day 57 (Figure 3B). Baseline responses were present in four cases (patients 7, 8, 22 and 25) for Melan-A and in ten cases (patients 1, 5, 8, 13, 16, 18, 20, 21, 26 and 27) for influenza. Responses to Melan-A were enhanced in three subjects who received rhIL-12 i.v. (patients 10, 18 and 24). In two of these the responses were not particularly vigorous but met the definition of increase over baseline. None of the subjects receiving rhIL-12 s.c. showed enhanced responsiveness to Melan-A. In two instances, patients 7 and 22, modest baseline responses were lost over the course of the study. For influenza peptide, enhancement of CTL responses occurred in a larger proportion of subjects. Nine had enhanced responses, of whom most did not have detectable circulating CTLp at baseline. These included subjects who received rhIL-12 by both routes and at each dose level including placebo. Consequently no clear effect of rhIL-12 dose and route of administration could be determined on the enhancement of CTLp responses to Melan-A or influenza.
Tumor regression was seen in some subjects (Table 8), however clinical responses were mostly ‘mixed’. Overall objective responses in the i.v. study were 1 complete response (CR), 1 stable disease (SD) and 9 progressive disease (PD), while in the s.c. study there were 1 partial response (PR), 5 SD and 7 PD. Sites of regressions included lung, lymph nodes, skin, liver and adrenal glands. An example of a mixed response is shown in Figure 4. The graph shows the diameter of six separate pulmonary metastases determined throughout the course of the study. Metastases were documented as having simultaneously regressed, remained stable or progressed.
Figure 5 shows the histopathology and immunopathology of lesions biopsied from patient 1 who had multiple small cutaneous scalp metastases. Complete clinical regression was accompanied by complete pathological regression. Biopsy of a flattened residual pigmented lesion revealed only pigment-laden macrophages, with no evidence of viable melanoma cells being present. Lesions were biopsied in some cases during the course of the study and on occasion demonstrated brisk infiltration by CD8+ T-lymphocytes as shown in Figure 6B (patient 2). Extraction of tumor infiltrating lymphocytes (TILs) from one such lesion demonstrated specificity for Melan-A in a chromium release assay (Figure 6C). It was not practical to perform these studies in all subjects. Nonetheless this example demonstrates that during the course of this study objective regressions were observed and could be shown to be associated with infiltration of metastatic deposits with antigen-specific CTLs.
Several subjects survived well past the median survival for stage IV melanoma which is approximately 6 months. These included patients 1 (1011 days), 23 (860+ days), 24 (638 days), 6, 14, 15, 26 (750+ days each) and 12 (605 days). Of these, patients 1 and 6 received placebo, whereas the other subjects received rhIL-12 at doses of 30 ng/kg i.v., 100 ng/kg i.v., 10 ng/kg s.c., 10 ng/kg s.c., 100 ng/kg s.c. and 10 ng/kg s.c. respectively. Thus there was no discernible effect of rhIL-12 on survival, with two long surviving subjects receiving no rhIL-12 at all. Furthermore, no clear relationship existed between enhanced immune responses (as determined by these assays) and objective clinical response or survival. In one patient (patient 2), Melan-A-specific lymphocytes were extracted from a metastatic subcutaneous deposit yet CTLs were not simultaneously detectable in blood. Antigen-specific effectors may therefore be present in some subjects but may not always be accessible in peripheral blood to evaluate the immune response.
The objectives of these two studies were to determine the safety of rhIL-12 when administered at relatively low doses in conjunction with peptide antigens and to identify an optimal biological dose of rhIL-12. We have established the safety of these schedules for rhIL-12 with peptides. Adverse effects were generally constitutional and low grade and tended to be more frequent and higher grade with increasing rhIL-12 doses. Most reported adverse events were related to the disease rather than the therapy.
We observed evidence that immune responses could be enhanced against two HLA A2-restricted peptide antigens: Melan-A26-35 (EAAGIGILTV) and influenza matrix58-66 (GILGFVFTL). Cutaneous DTH reactions were associated with CD4+ and CD8+ T lymphocyte infiltration and tumor regressions were associated with the presence of antigen-specific CTLs, which could be isolated from regressing lesions. Nonetheless, no clear relationship could be identified between these responses and the dose of rhIL-12 administered or the route of administration. Furthermore, objective clinical regressions were seen, and in some cases, patients survived for prolonged periods. However these clinical outcomes were also observed in subjects who received peptides with placebo.
These results contrast with those of Lee at al. (28) who demonstrated that immune responses could be enhanced by IL-12. In that study intradermal IL-12 (30 ng/kg) was co-injected at peptide injection sites in 48 patients with resected stage III or IV melanoma. The peptides, derived from tyrosinase and gp100, were administered in incomplete Freud's adjuvant (IFA) over a 20-week period. That study differed from our own in several important regards, including the method of IL-12 administration, the duration of the treatment protocol, the disease status of the patients, the use of IFA and the number of patients in each of the treatment groups.
Although we were unable to demonstrate a dose-response relationship between rhIL-12 and immune enhancement, our findings do not preclude an effect of IL-12. Such relationships were observed in animal studies (26) where the variability between experimental subjects was small. Similarly, they do not preclude an effect of the route of administration. Rather, they highlight the challenge of identifying such relationships in patients with advanced melanoma. In several cases, clinical outcomes interfered with proper immune evaluations. In the s.c. study, three subjects were not evaluable for immune evaluation because of rapid disease progression. Although these were replaced, a further four progressed within the month during which the first four peptide doses were injected. Similarly in the i.v. study, five subjects failed to complete all protocol requirements for vaccination and evaluation because of disease progression. Disease progression in these cases is consistent with the natural history of advanced metastatic disease. It follows that these are not ideal subjects for studies where endpoints are primarily immunological and where such responses may evolve slowly. Further, to observe any impact on the clinical course of the disease, the effects of treatment would need to be dramatic, or subjects would need to be randomized in larger studies.
Additional variability may occur at the level of the baseline response to antigens. Melan-A responses have previously been described in melanoma patients and control subjects (29, 30). In this study, these and responses to influenza at baseline varied considerably from subject to subject. In addition, differences in immune repertoire, tumor burden, or even exposure to cross-reacting viral epitopes (31) may confound interpretation of any response. To assess the impact of rhIL-12 on responses to these antigens would therefore require larger numbers of subjects in order to overcome this variability.
As a consequence, we propose a re-evaluation of clinical trial design for cancer vaccines. Such studies could be viewed in two broad classes: (i) those with clinical endpoints and (ii) those with immunological endpoints. For studies where the endpoints are clinical, evaluable disease remains a prerequisite for inclusion. Such studies might include immune evaluations, so as to draw correlations between surrogate immune endpoints and clinical outcomes for example. These studies can benefit from careful analysis of antigen expression in individual lesions, tumor infiltrating lymphocytes and the interplay between immunity and cancer. For the other class of study, the primary focus is the quantitation of immune endpoints. Such studies are required to determine optimal vaccine doses, route of administration, schedules, choice of adjuvant, or to define the role of concurrent cytokine administration by a variety of doses, routes, combinations and schedules. These studies are most likely to succeed if subject variability can be reduced and if subjects can reliably complete the treatment course. This is best achieved by selecting subjects without substantial tumor burden, such as fully resected disease. As a result, clinical response information will be lost. If it can be established that these treatments are safe, inclusion of patients with minimal residual disease can be justified ethically, and later studies with clinical endpoints would then be required to test those approaches which appear to be most successful at inducing immune responses.
Materials and methods
Eligible subjects had unresectable, Melan-A-expressing stage III or stage IV melanoma. Each subject had to be HLA-A2 positive and have a Karnofsky performance status of 70% or more. Patients with active central nervous system metastases, major organ dysfunction, autoimmune disease (except vitiligo) or active peptic ulcer disease were excluded from the study. Concurrent corticosteroids or non-steroidal anti-inflammatory drugs were not permitted during the study treatment and evaluation periods. Subjects who did not complete at least four weeks of the study protocol due to a reason unrelated to the study treatment were not evaluable for immune responses and were replaced. All subjects gave informed consent.
Tumor antigen detection
Informed consent was obtained for enrollment in these Institutional Review Board-approved immunization protocols sponsored by the Ludwig Institute for Cancer Research. In both studies, subjects were treated in three rhIL-12 dose cohorts, at doses of 10, 30 and 100 ng/kg. In each cohort, subjects were randomized to receive rhIL-12 or placebo at a ratio of 3:1. The rhIL-12 or placebo was administered during weeks 1 and 4. In Melbourne, subjects received rhIL-12 or placebo as an i.v. dose on days 1 and 22. In Frankfurt, the study drug was administered s.c.. For each cohort both subjects and treating physicians were blinded as to whether rhIL-12 or placebo was given. In both studies peptides (100 µg) were injected intradermally on days 1, 8, 15, 22 and 57.
The Melan-A26-35 (EAAGIGILTV) and influenza matrix58-66 (GILGFVFTL) peptides were synthesized by Multiple Peptide Systems, San Diego, CA. They were >98% pure as confirmed by HPLC. 100 µg of each peptide was formulated in 0.3 ml of phosphate-buffered saline, pH 7.4, with 5% dimethylsulfoxide and dispensed as single dose aliquots of 100 µg at the Ludwig Institute Biological Production Facility, Melbourne, Australia. The rhIL-12 was manufactured by the Genetics Institute and reconstituted with bacteriostatic sterile water.
Evaluation of safety and toxicity
Subjects who received at least one dose of rhIL-12, Melan-A and influenza matrix peptides were evaluable for safety and toxicity. The Common Toxicity Criteria of the National Cancer Institute (October 1993) were used to grade toxicities.
Evaluation of immune responses
Subjects who remained in the study for at least 29 days were evaluable for immune responses. These were determined by cutaneous reactivity and measurement of peptide-specific cytotoxic lymphocyte precursors (CTLps).
Erythema and induration at vaccine sites were measured 48 hours after intradermal peptide injection. The maximum diameter of the erythema or induration was used in a cutaneous reaction scoring system (see Tables 4 and 5). A cutaneous reaction response was defined as an increase, sustained for at least two vaccinations, in one score level over baseline.
In vitro assay for cytotoxic lymphocyte precursors
A split-well assay was employed to evaluate CTLps (34). Briefly, peripheral blood mononuclear cells (PBMCs) from 100 ml blood samples taken on days 0 (pre-study), 15, 29 and 57 were used to lyse peptide-pulsed T2 cells following 2 rounds of stimulation with specific peptide. Effectors were split into 24 wells with peptide targets and 24 paired control wells without peptide, providing a semi-quantitative evaluation of CTLp frequency. A positive well was defined as one in which the percentage of T2 cell lysis was more than twice that of its paired control well and more than the mean plus one standard deviation of the values of all 24 control wells. Induction of a CTL response was defined as a doubling of the number of positive wells over baseline or an increase of at least four wells if the baseline number of positive wells was 0-4.
Alternative assays were used in some subjects, depending on the availability of cells. These included an ELIspot assay (35), modified to detect IFN-gamma-secreting CD8+ cells, and a ‘bulk culture’ CTL assay (29). The peptides used in these studies were the Melan-A26-35 (EAAGIGILTV), influenza matrix58-66 (GILGFVFTL), tyrosinase368-376 (YMDGTMSQV), tyrosinase1-9 (MLLAVLYCL) and MAGE-3 A2271-279 (FLWGPRALV) peptides (Multiple Peptide Systems, San Diego, CA).
Evaluation of tumor responses
Subjects who remained in the study for at least 29 days were evaluable for tumor responses. Standard objective response criteria were used. For the purpose of correlating immunological and tumor responses, a non-conventional approach to analysis was taken since mixed responses were observed and regression of individual lesions might indicate an immune effect. Melan-A antigen expression could not be assessed in all metastases. Furthermore, since Melan-A antigen expression may be heterogeneous (32), any objective regression in individual lesions was considered worth documenting. For this purpose metastases were documented separately and scored as responses even if progression occurred in other lesions. Such responses are classified as mixed responses.
Patient survival was measured in days from the date of study entry to death.
The authors would like to acknowledge the support and inspiration provided to this project by Dr. Lloyd Old, Ludwig Institute for Cancer Research, New York.
- Received May 11, 2003.
- Accepted June 16, 2003.
- Copyright © 2003 by Jonathan Cebon