Abstract
The purpose of this study was to test melanoma vaccines consisting of peptides and immunological adjuvants for optimal immunogenicity and to evaluate laboratory immune monitoring for in vivo relevance. Forty-nine HLA-A2 positive patients with Melan-A positive melanoma were repeatedly vaccinated with Melan-A peptide, with or without immune adjuvant AS02B (QS21 and MPL) or IFA. Peptide-specific CD8 T cells in PBLs were analyzed ex vivo using fluorescent HLA-A2/Melan-A multimers and IFN-gamma ELISPOT assays. The vaccines were well tolerated. In vivo expansion of Melan-A-specific CD8 T cells was observed in 13 patients (1/12 after vaccination with peptide in AS02B and 12/17 after vaccination with peptide in IFA). The T cells produced IFN-gamma and downregulated CD45RA and CD28. T-cell responses correlated with inflammatory skin reactions at vaccine injection sites (P < 0.001) and with DTH reaction to Melan-A peptide (P < 0.01). Twenty-six of 32 evaluable patients showed progressive disease, whereas 4 patients had stable disease. The two patients with the strongest Melan-A-specific T-cell responses experienced regression of metastases in skin, lymph nodes, and lung. We conclude that repeated vaccination with Melan-A peptide in IFA frequently leads to sustained responses of specific CD8 T cells that are detectable ex vivo and correlate with inflammatory skin reactions.
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.
Introduction
Antigen-specific immunotherapy for cancer has potential. Over 50 different tumor-associated peptide antigens binding to HLA-A, -B, or -C (HLA class I molecules) have been identified. So-called ideal peptide antigens are expressed by tumor cells but not by vital normal tissue: They mediate efficient cytolysis and cytokine production by CD8 T cells. Most of the known tumor antigens are derived from nonmutated proteins that immunotherapy can often target in patients (1, 2, 3, 4).
Peptide26-35 of the melanocyte differentiation antigen Melan-A/MART-1 (hereafter referred to as Melan-A) is presented by HLA-A2 and recognized by cytotoxic CD8 T cells (5). Melan-A peptide-based synthetic vaccines were tested in several melanoma trials. One study revealed activation of cytotoxic T cells in PBLs from 3/6 patients treated with 4 weekly intradermal injections of Melan-A peptide in saline (6). Two other studies (7, 8) used Melan-A peptide mixed with IFA, resulting in increased IFN-gamma secretion by peptide-stimulated PBLs. Immune responses were rather weak, as they were only detected after in vitro PBL stimulation for 1 wk or longer (9). In contrast, a recent study found Melan-A-specific T cells that were detectable ex vivo after vaccination with Melan-A peptide in IFA, administered in parallel with progenipoietin, a GM-CSF/FLT-3L agonist (10).
Further phase I/II clinical studies are necessary to identify vaccine formulations capable of inducing strong, sustained T-cell activation. Although it is not yet proven, it may well be that stronger immune responses are associated with improved tumor protection and thus with a better clinical outcome (11, 12).
With respect to immune monitoring, many published results show antigen-specific T cells only after in vitro stimulation (9). It is a major challenge to detect antigen-specific T cells directly ex vivo (10) so that we can characterize the ongoing in vivo immune activity of effector lymphocytes. With the help of fluorescent multimers consisting of HLA/peptide complexes (13), T cells can be quantified ex vivo and analyzed with respect to activation and differentiation state. Using multimers and an IFN-gamma ELISPOT assay, we have developed immune monitoring procedures that satisfy the criteria for high-quality bioanalytical method validation (14).
Here we present the results of a phase I/II study in 49 advanced (stage III and IV) melanoma patients. Our aim was to test stepwise-modified vaccine formulations in order to identify the most immunogenic one(s). Antigenic peptides were given either in saline or mixed with adjuvant AS02B or IFA. Besides Melan-A peptide (natural or analog), each patient received an influenza matrix peptide as a positive control antigen. Two of the vaccine formulations were also tested in conjunction with 7-d treatments with low-dose IL-2 s.c.
Results
Toxicity of vaccine injection
Forty-nine HLA-A2 positive melanoma patients (Table 1) were included in the study. Primary endpoints were toxicity and immune response, and the secondary endpoint was tumor response. Patients who received at least one vaccination were evaluable for treatment toxicity (Table 2). Forty-three patients received at least four vaccines (one cycle, see Figure 1) and were therefore evaluable for immune response.
Patients' characteristics and disease status at study entry. The 49 patients included in the study were sequentially attributed to treatment groups I to IV (see text and Table 2). Twenty-five patients had no measurable disease at study entry due to complete surgical resection. Prestudy treatments were: surgery (all patients, not shown); dacarbazine (DTIC); fotemustine (Fote); platinum (Pt); interferon-alpha (IFN-alpha); and isolated limb perfusion (ILP) without or with TNF-alpha (T), IFN-gamma (I) and/or Melphalan (M). Abbreviation: JWCI, John Wayne Cancer Institute.
Treatment and tumor response. The patients are listed in the same sequence in Tables 1, 3, and 4. The vertical colored bars indicate vaccine components. All patients received HLA-A2 binding peptides derived from influenza matrix protein and from Melan-A [either natural (nat) or analog (ana) peptide]. Peptides were given without adjuvant (group I), mixed with AS02B (groups II and IVa), or mixed with IFA (groups III and IVb). For group IV, vaccines were accompanied by 7-d treatments with low dose IL-2 s.c. Abbreviations: PR, partial response; MR, mixed response (regressing metastasis concomitant with progressing metastases); NMD, no measurable disease; PD, progressive disease; SD, stable disease.
Vaccination schedule. Vaccinations were given in cycles; each cycle consisted of four injections (vertical lines). Up to four additional and similar cycles (continued vaccination, not shown) were given to patients without major disease progression, with 2-mo intervals between cycles. Vaccines contained 100 µg each of Melan-A and influenza peptides injected as mixtures in saline or together with 660 µl AS02B or 600 µl IFA. Before starting treatment and 2 wk after the last vaccine of each cycle, peripheral blood cells (PBLs) were withdrawn (vertical arrows) for lymphocyte analyses by flow cytometry and IFN-gamma ELISPOT assays. The patients in treatment group IIIb were on a different schedule (not shown) since vaccines were given biweekly and PBLs were obtained 1 wk after the last vaccine of each cycle.
Administration of peptides in saline caused no significant side effects. In contrast, local inflammatory reactions (grade I or II) lasting over one or more weeks were frequently observed after injections of peptide in IFA: systemic reactions, mild to moderate headache, myalgia, chills, and flu-like symptoms were noticed occasionally after injections with both AS02B and IFA. One patient developed fever for 24 h after an injection of peptides in IFA.
IL-2 (2 x 5 Mio IU/day s.c.) frequently caused systemic grade I or II toxicities that did not require hospitalization. The symptoms were fever, sweating, cough, chills, desquamation, pruritus, nausea, diarrhea, dry mucous, and weakness. Transient eosinophilia (up to 35 times above baseline) was observed in 11/12 patients. Moderate lymphopenia was noticed in 5/12 patients, and mild neutropenia in 6/12 patients. These changes were regularly reproduced upon each IL-2 treatment. Eosinophile counts increased just 1 d after starting therapy and returned to normality 2-3 wk later. Lymphocyte counts dropped within the first day but returned to normality during the 7 d of IL-2 treatment.
Immune responses to Melan-A
Peptide-specific CD8 T cells in PBLs were counted directly ex vivo by multiparameter flow cytometry that allowed us to simultaneously determine CD45RA expression. CD45RA is highly expressed in naive CD8 T cells and is downregulated upon in vivo activation (15). Before vaccination, we detected 0.11% Melan-A-specific (multimer positive) T cells in CD8+ PBLs from patient LAU 750 (Figure 2). Nearly all Melan-A-specific cells (0.10%) were CD45RA+. After vaccination, there were 0.89% multimer positive cells (0.10% CD45RA+ and 0.79% CD45RA-). Thus, Melan-A-specific T cells increased 8-fold and became CD45RA-.
Activation and expansion of tumor antigen-specific T cells in PBLs, detectable ex vivo. PBLs from patient LAU 750 were analyzed by flow cytometry by ex vivo staining with HLA-A2/Melan-A multimers and CD45RA-specific mAbs. The numbers indicate the percentages of multimer positive cells that are CD45RA- (upper left quadrant) or CD45RA+ (upper right quadrant). The dot plots show CD8-sorted T cells (100% corresponds to the total number of CD8+ lymphocytes).
Before vaccination, multimer positive lymphocytes ranged from undetectable/at threshold (less than or equal to 0.01%) to 0.23% Melan-A-specific cells in CD8+ PBLs (Table 3, column 3). Although many patients did not show significant changes in Melan-A-specific cells after vaccination, the values in one patient from group II and eight patients from group III increased more than two-fold after the first vaccination cycle (column 4, colored fields). Patient LAU 337 reached 2.3% (16), and the other patients reached 0.07-0.89% Melan-A-specific cells. After additional vaccine cycles (continued vaccination), there were four more patients with expanded Melan-A-specific cells (column 5). No evidence for T-cell activation was found in patients treated with peptides in saline (group I) or in patients treated with IL-2 (group IV). In total, 13/43 evaluable patients had responding Melan-A-specific CD8 T cells after vaccination. Vaccination with peptides in IFA (group III) was significantly (P < 0.001) associated with the induction of Melan-A-specific T-cell expansion, in contrast to the other vaccination regimens tested.
Activation of Melan-A-specific T cells. Forty-three of 49 patients were evaluable for immune responses and are listed according to their treatment groups. Melan-A specific responses were assessed by flow cytometry, IFN-gamma ELISPOT assays, and DTH. The numbers indicate the percentages of positive cells in CD8+ PBLs, except for columns 6-9 which show the percentages of cells negative for CD45RA and CD28 among multimer positive cells. PBLs were withdrawn before vaccination, after the first cycle (1st cyc), and after additional ("continued") vaccination (cont v). Values in colored fields correspond to an increase of more than two-fold or of more than 20% in CD45RA- or CD28- antigen-specific T cells, respectively (comparing values after versus before vaccination). Columns 5, 7, and 9 show the results of the blood sample with the highest value among PBLs withdrawn at various time points after vaccination. Fields are left empty if no additional vaccination cycle was completed (columns 5 and 12), if percentages of multimer positive cells were too low for phenotyping for CD45RA and CD28 (columns 6-9), or if DTH was not performed (columns 13 and 14). Abbreviations: Cyc, cycle; na, not applicable; ni, not interpretable.
Often not just T-cell frequencies, but also the percentages of CD45RA+ multimer positive cells increased after vaccination (columns 6 and 7; colored fields). The exceptions were Melan-A responder patients who had increased percentages of CD45RA- cells before vaccination (e.g., LAU 567 and LAU 444) and did not show a further increase of CD45RA- cells. These patients substantially increased both CD45RA- and CD45RA+ T cells during therapy. Indeed, CD45RA is known to be re-upregulated in strongly activated cytolytic effector T cells (17). These cells were not naive, since they were CCR7- (not shown). The significant CD28 downregulation that is characteristic for CD8 effector T cells (18) was observed in three patients (columns 8 and 9). Downregulation of CD45RA, CCR7, and CD28 by more than 20% of multimer positive cells was found in Melan-A responder, but not in nonresponder, patients.
Increased frequencies of IFN-gamma-secreting Melan-A-specific cells (IFN-gamma ELISPOT assay, columns 11 and 12, colored fields) were found in 8/12 patients with increased percentages of multimer positive cells. Three of these patients showed increased IFN-gamma spots only after additional vaccination. The percentages of IFN-gamma-secreting cells were always lower than those of multimer positive cells, in keeping with the idea that only a fraction of antigen-specific T cells produce IFN-gamma after short-term stimulation (19).
To combine the results of these immune-monitoring assays into a single value describing the magnitude of vaccine-induced immune activation, we applied an arbitrary scoring system comparing pre- with postvaccination results (see Materials and Methods). The responder patient in treatment group II achieved an immune score of 11 points (Figure 3). Responders in group III had scores between 1 and 10. No patient had a negative score. Other than patient LAU 648, responders had scores of 2 or higher and fulfilled at least 2 of our 4 immune response criteria (increased percentage of multimer-positive T cells; increased percentage of IFN-gamma ELISPOT; shift toward CD45RA- T cells; and shift toward CD28- T cells), indicating that the activation of Melan-A-specific T cells was substantial.
Quantitation of Melan-A-specific T-cell activation by immune scoring. No activation was found in 6 patients vaccinated with peptides in saline or in 8 patients vaccinated with peptides in adjuvant (AS02B or IFA) and IL-2. One patient responded to peptides in AS02B, and 12 patients responded to peptides in IFA. Melan-A-specific T cells were assessed by flow cytometry and IFN-gamma ELISPOT assay, and immune scores were calculated as described in Materials and Methods.
DTH reactions to intradermal peptide injection were more often positive after, as compared to before, vaccination (Table 3, columns 13 and 14). Interestingly, Melan-A immune-responder patients were significantly more likely to have increased Melan-A-specific DTH reactions than were nonresponders (P < 0.01).
Immune responses to influenza
Before vaccination, most patients had low but detectable frequencies of influenza-specific T cells (Table 4, column 3) as a result of prior influenza infection (no commercially available influenza vaccines were used during the study). Influenza peptide was given to test the vaccine's ability to boost virus-specific T cells. Nine patients had expanded influenza-specific T cells after vaccination (Table 4, columns 4 and 5, colored fields). Expression of CD45RA and CD28 by influenza-specific cells was systematically tested, but changes in pre- as compared to postvaccination samples were minimal; that is, below 20% (not shown).
Activation of influenza-specific T cells. The results for influenza-specific responses are presented in the same fashion as for Melan-A in Table 3.
Increased percentages of IFN-gamma ELISPOTs were found in five patients (Table 4, columns 7 and 8, colored fields). Four patients had both increased multimer and ELISPOT values, whereas for five patients the response was only revealed by multimers, and for one patient only by ELISPOT. As with the Melan-A-specific responses, peptide-specific T-cell activation was not observed after IL-2 treatment (group IV). For the remaining patients, activated influenza-specific cells were found in 2/6 patients who received peptides in saline (group I), in 1/12 patients who received peptides in AS02B (group II), and in 7/17 patients who received peptides in IFA (group III). Altogether, activation of influenza-specific cells was observed in 10/43 evaluable patients.
Immune scores were determined in the same manner as were those for Melan-A-specific responses: The two patients who responded to influenza peptide in saline had scores of 4 and 2 (Figure 4). The responder patient in group II had a score of 3, and the 7 responders in group III had scores between 2 and 7. Thus, all responders had scores of 2 or more, demonstrating substantial activation of influenza-specific T cells. Three patients had a score of -1 (not shown), reflecting a slight reduction in the frequency (two patients) or activation stage of influenza-specific T cells (one patient). The infrequency of negative scores suggests that the vaccines did not induce a significant degree of immunological tolerance, an issue that needs to be considered when peptides are used for vaccination (20).
Quantitation of influenza-specific T-cell activation by immune scoring. Two patients responded to peptides in saline, no patient responded to peptides in adjuvant (AS02B or IFA) and IL-2, one patient responded to peptides in AS02B, and seven patients responded to peptides in IFA.
DTH reactions to intradermally injected influenza peptide were frequently positive, and there was a tendency toward increased DTH reactivity after, as compared to before, vaccination (Table 4, columns 9 and 10). Influenza-specific T-cell responder patients were not, however, significantly more likely to have increased influenza-specific DTH reactions after vaccination than were nonresponders.
Tumor response
Thirty-two patients were evaluable for tumor response, and 11 patients had no measurable disease during the study (observation period 5-50 mo, mean 23 mo; Table 2). Among the 32 evaluable patients, 26 had progressive disease, and 4 (LAU 97, 212, 350, and 470) had stable disease over a period of 6, 12, 14, and 22 mo. Two patients showed evidence of objective tumor responses: Patient LAU 337, with multiple metastases, developed a minor response (multiple regressing lesions in skin, lymph nodes, and lung) lasting for 10 mo. However, following that period the disease progressed rapidly, and the patient died 14 mo after entering the study (16). The second patient, LAU 446, developed multiple pulmonary metastases 6 mo after entering the study, all of which disappeared completely (complete remission at 17 mo after study entry) during subsequent vaccination (Figure 5). Brain metastases were found at month 21, and the patient died 23 mo after entering the study.
Regression of pulmonary metastases after immunotherapy. CT scans from patient LAU 446 taken 11 mo (left) and 14 mo (right) after study entry. Bilateral lesions (red circles) were detected at 11 mo. At 14 mo, the lesion in the right lower lobe had regressed partially; a subsequent control CT scan at 17 mo (not shown) revealed complete remission.
Interestingly, patient LAU 337 had the strongest activation of Melan-A-specific cells (immune score of 11), and patient LAU 446 the second highest Melan-A score (score of 10). Thus, the two patients with clinically manifest tumor regressions were the two strongest Melan-A responders. One of the four patients with stable disease had an immune score of 5, whereas the other three were Melan-A nonresponders. Finally, among the 26 patients with progressive disease, we found immune scores of 8, 7, 6, 5, 3, and 2; the remaining 20 patients were Melan-A nonresponders.
Inflammatory reactions at vaccine injection sites
After s.c. injection of peptides in IFA, some patients developed local inflammatory indurations at vaccine sites that persisted for 1-3 wk. The reactions differed in size (affecting an area of skin ranging from 3 to 75 cm2) and were accompanied by swelling and mild to moderate local pain. They were found in 63% of immune-responder patients (that is, those with activated T cells detectable in PBLs), and in only 11% of nonresponders. The difference was highly significant (P < 0.001).
Due to poor resorption of mineral oil, IFA usually persisted at s.c. injection sites, resulting in local, noninflammatory s.c. indurations (0.5-1 cm3 in size) that were observable for several months. Aside from these postvaccine inflammatory reactions, the indurations showed no sign of inflammation. However, in some patients we observed an interesting phenomenon: Upon subsequent vaccination using a different limb, some earlier injection sites developed new inflammatory signs, showing that they were reactivated. Such distant reactivation was found in 50% of immune-responder patients, but only in 7% of nonresponders (P < 0.005). The time interval from vaccination at the distant site to reactivation of the prior vaccination site ranged from 1 to 6 mo. As we will discuss, these observations point to long-lasting vaccine persistence and immunogenicity.
Discussion
Ex vivo detectable activation of Melan-A-specific T cells was found in 13/43 evaluable patients, and of influenza-specific T cells in 10/43 patients (Table 5). Six of 13 patients responded to Melan-A but not to influenza, and 3/10 responded to influenza but not to Melan-A. Seven patients responded to both antigens, but only two of them responded within the same months (not shown). Thus, responses to the two peptide antigens occurred largely independently. The data indicate that responsiveness was antigen-specific rather than patient-specific, suggesting that patients did not suffer from general immune suppression.
Immunogenicity of Melan-A and influenza peptides. The numbers correspond to the number of patients with T-cell responses to Melan-A, influenza, or both. In total, 43 patients were evaluable for immune response.
Overall, the majority of vaccinated patients had no ex vivo detectable immune response, which is comparable to many cancer immunotherapy studies performed worldwide (6, 7, 8, 21, 22, 23, 24). Synthetic peptide vaccines do not appear to activate T cells sufficiently to become detectable ex vivo. In contrast, higher immune response rates are revealed by in vitro lymphocyte tests (25, 26, 27), but these depend on T-cell proliferation and have high variability. Ex vivo analysis can give highly reproducible results (14) and can identify the current lymphocyte activation state in the patient, providing crucial information on ongoing effector T-cell activity. We believe that ex vivo analysis should become standard in evaluating therapeutic vaccines.
In most studies, activation of Melan-A-specific cells in PBLs of vaccinated patients was assessed upon in vitro stimulation of T cells (7, 8, 23, 24, 28). Many of our patients had ex vivo detectable activated Melan-A-specific T cells upon vaccination with peptides emulsified in IFA. While nonresponders had relatively stable values, Melan-A responders showed increased postvaccine frequencies of Melan-A-specific T cells with reduced expression of CD28 and/or CD45RA. Thus, changes in the percentage and phenotype of Melan-A-specific T cells occurred in parallel. Furthermore, induction of Melan-A-specific IFN-gamma ELISPOT positive cells was detected in 8/12 multimer responder patients, but in none of the multimer nonresponders.
An important question is whether Melan-A-specific T cells were tumor reactive. We have shown that Melan-A-specific T cells recovered from patient LAU 337 after vaccination were capable of killing melanoma cells by recognizing endogenously expressed Melan-A antigen (29); preliminary results of ongoing studies indicate that T cells from other patients are also tumor reactive. Since we have generated several melanoma cell lines from these patients, we will also be able to assess recognition and lysis of autologous melanoma cells. We also tested T-cell reactivity to the natural Melan-A peptide EAAGIGILTV by ELISPOT assay (not shown), and the results were found to be similar to the ELISPOT data obtained with the analog peptide ELAGIGILTV (Table 3). This finding confirms earlier studies showing that vaccination with the analog peptide activates T cells to recognize natural Melan-A antigen efficiently (30).
In accordance with earlier observations (13, 19, 31), we found that the number of T cells determined by multimers was higher than those determined by IFN-gamma ELISPOT assays. Although the two assays may differ in sensitivity, the results reflect true differences, since many T cells are not secreting cytokines after short-term antigen stimulation, possibly due to insufficient T-cell receptor avidity and/or in vivo activation. Nevertheless, and not surprisingly, the correlation between the percentage of antigen-specific T cells as determined by multimers and ELISPOT assay was statistically significant (14). In addition, patients with high percentages of CD28- cells were not only all ELISPOT positive, but showed the highest ELISPOT values. As expected, the percentages of CD45RA- cells did not correlate with the ELISPOT assay. We also found significant shifts in expression of CCR7, HLA-DR, and 2B4 by Melan-A-specific cells (32), giving us further information on T-cell differentiation. Our results are compatible with the recently established correlation between T-cell surface marker phenotype and function (15, 17, 18, 33, 34, 35). Accordingly, the majority of CD45RA- cells are so-called memory cells that do not secrete IFN-gamma immediately. In contrast, CD28- cells are effector cells, with ongoing cytokine production and cytolytic activity taking place in many of them (18). Although surface markers are useful, it remains a challenge to measure T-cell function directly ex vivo.
Activation of influenza-specific T cells was found in 10 patients, 6 after the first cycle and 4 after additional vaccination. As expected, the percentages of IFN-gamma-secreting cells were lower than the percentages of multimer positive cells. As with healthy adults, most of the patients' influenza-specific cells were CD45RA- and CD28+ at study entry. Throughout the study, expression of both CD45RA and CD28 remained relatively stable (not shown), indicating that the differentiation state of influenza-specific cells had not changed as much as that of Melan-A-specific cells. The number of patients responding to influenza peptide vaccination was small, and responses were no stronger than for Melan-A. This may seem surprising. However, many of the observed strong Melan-A-specific responses were likely due to combined immune stimulation through peptide vaccination plus antigen provided by tumor tissue (36). Furthermore, influenza-specific T cells are usually not, or are only weakly, active in the absence of influenza infection (37 , 38). Thus, our patient population was more likely to generate Melan-A than influenza-specific responses.
It was satisfying that T cells were activated even after numerous booster injections. Protective immunity against cancer likely requires long-lasting immune activity (11). Melan-A-specific T cells remained activated in 12/12 patients throughout the entire observation period (mean 10 mo, range 3-20 mo). Influenza-specific T cells also remained activated throughout the entire observation period in 9/10 patients (mean 12 mo, range 3-32 mo). Only one patient (LAU 444) switched back to influenza nonresponder status at week 40 after having responded during the first vaccination cycle. These findings suggest that the applied vaccines were capable of efficiently boosting specific lymphocytes repeatedly. Recent clinical studies with DCs or recombinant viruses reported that activation of melanoma-specific T cells was only transient (26, 39). In this regard, repeated use of synthetic vaccines may be superior, a possibility that deserves further investigation.
Comparison of the immunogenicity of natural versus analog Melan-A peptides was inconclusive due to the low numbers of responders in the relevant patient groups (groups Ia, IIa versus Ib, IIb). Similarly, no conclusion could be reached when comparing peptide vaccines with or without the QS21-based adjuvant AS02B (group I versus II). A recent melanoma immunotherapy trial demonstrated IFN-gamma ELISPOT T-cell responses after vaccination with tyrosinase368-376 peptide and QS21 injected s.c. (28). It is possible that QS21 and peptide vaccines would lead to immune activation more often when given s.c., as opposed to the i.m. route used in our study.
Vaccine supplementation with low-dose IL-2 s.c. injections did not lead to detectable immune responses comparable to those reported previously (23, 40). It is possible that IL-2 promoted T-cell extravasation and/or T-cell death, or that the latter grew as a result of IL-2 withdrawal (41, 42, 43). IL-2 is thought to be a major effector molecule secreted by helper T cells that contributes to CD8 T-cell immunity (44, 45, 46). Further studies are necessary to define the therapeutic potential of IL-2 for promotion of T-cell responses.
Local inflammatory skin reactions at vaccine injection sites correlated with multimer/ELISPOT results. This was the case for skin reactions occurring the first 2 wk after injection, as well as for the reactivation of past injection sites following booster injections at distant body sites. The latter phenomenon was observed up to 6 mo later, suggesting that the peptide/IFA mix persisted for several months. Indeed, core needle biopsies taken from such inflamed distant sites of our patients led to the identification of selectively accumulated vaccine-specific T cells (not shown), demonstrating a remarkable depot effect associated with IFA. Antigen persistence at vaccine sites may have contributed to IFA's immunogenicity. DTH reactions to peptides were more frequent and more strongly positive after, as compared to before, vaccination. For Melan-A-specific responses, this correlated significantly with multimer/ELISPOT results. Increased influenza-specific DTH reactions were less frequent, probably due to the relatively high rate of responses before study entry, reflecting influenza-specific T-cell immunity in the general population. Collectively, our data are compatible with animal studies demonstrating that peptide-triggered skin reactions are caused by antigen-specific CD8 T cells (47).
Laboratory immune monitoring (multimer and ELISPOT) correlated with clinical observations such as inflammatory skin reactions and DTH. This indicates that the applied laboratory methods for ex vivo T-cell analyses gave relevant results, since they revealed in vivo ongoing immune responses. Furthermore, the two patients with regression of multiple metastases were the ones with the strongest activation of tumor antigen-specific T cells. However, 6 patients had progressive disease despite clearly detectable activation of Melan-A-specific T cells, indicating that activation is often not sufficient to achieve a clinical response. Clinical responses should become more frequent when melanoma-specific T cells are activated more extensively. The latter is a priority and may be achieved through systematic testing of new generation vaccines selected through careful ex vivo analysis of immune activation.
Materials and methods
Patients, study protocol, and eligibility criteria
Forty-nine HLA-A2 positive patients (Table 1) with histologically proven metastatic melanoma of the skin that expressed Melan-A by RT-PCR or immuno-histochemistry were included in the study after giving their informed consent (Sept. 1998-Oct. 2002). The local ethics committee and the Protocol Review Committee of the Ludwig Institute for Cancer Research (LICR, New York, USA.) approved this phase I/II prospective trial of the Ludwig Institute for Cancer Research (LUD 96-010). The inclusion criteria were: a Karnofsky performance status equal to or greater than 70%, normal CBC and kidney-liver function, and no concomitant antitumor therapy or immunosuppressive drugs. The exclusion criteria were: pregnancy, seropositivity for HIV-1 Ab or HBs Ag, brain metastasis, uncontrolled bleeding, clinically significant autoimmune disease, and New York Heart Association (NYHA) class III-IV heart disease. Primary endpoints were toxicity and immune response, and the secondary endpoint was tumor response. Patients who received at least one vaccination were evaluable for treatment toxicity (Table 2). Forty-three patients received at least four vaccines (one cycle) and were therefore evaluable for immune response. Tumor status was assessed by physical examination and CT scans before and after each cycle and upon tumor progression. Eleven patients had no measurable disease at study entry and throughout the entire study period. Among the remaining 38 patients, 32 completed at least the first vaccination cycle and were thus evaluable for tumor response.
Study groups and treatment
Patients were divided among four distinct vaccine regimen groups, I to IV, which were each further divided into two subgroups, a and b. In each treatment group, approximately two-thirds of the patients had American Joint Committee on Cancer (AJCC) stage III melanoma and the others had stage IV melanoma, except group IIb, which had 4/7 stage IV patients (Table 1). Patients received the vaccines shown in Table 2; that is, group I received Melan-A and influenza peptides (group Ia, natural Melan-A peptide; group Ib, Melan-A analog peptide). Group II received Melan-A and influenza peptides in adjuvant AS02B (IIa, natural Melan-A; IIb, Melan-A analog). Group III received Melan-A analog and influenza peptides in adjuvant IFA (IIIa, at monthly intervals, like all the other groups; IIIb, at biweekly intervals). Finally, group IV received Melan-A analog and influenza peptides in adjuvant, accompanied by 7-d treatments with low-dose recombinant human Interleukin-2 (rhIL-2) s.c. (IVa, AS02B; IVb, IFA). Four vaccines were given per vaccination cycle (Figure 1). Patients received a maximum of 5 cycles; that is, 20 vaccinations at most. Peptide amino acid sequences were GILGFVFTL (influenza matrix protein58-66 peptide), EAAGIGILTV (Melan-A26-35 natural peptide), and ELAGIGILTV (Melan-A26-35 analog peptide). The analog peptide has increased A2 binding affinity and immunogenicity, and preserved T-cell receptor (TCR) fine-specificity recognition by natural Melan-A peptide-specific T cells (30). Peptides were synthesized by Multiple Peptide Systems San Diego (USA), and formulated (330 µg/ml in PBS/30 DMSO) by the Biological Production Facility, LICR Melbourne (Australia). AS02B (GlaxoSmithKline Biologicals, Rixensart, Belgium) contained QS21 (a natural saponin) and monophosphoryl lipid A (MPL) formulated in an oil-in-water emulsion. Adjuvant IFA (Montanide ISA-51, Seppic, France) contained mineral oil (Drakeol) and anhydro mannitol octadecanoate. Melan-A and influenza peptides (100 µg each) were always mixed together. Peptide solutions (300 µl + 300 µl) were mixed with 600 µl adjuvant. Mixtures with IFA were aspirated 10-20 times in a syringe to obtain a stable emulsion. Vaccines with AS02B were given i.m.; vaccines with IFA s.c., according to the manufacturer's instructions. Recombinant human Interleukin-2 (rhIL-2, Roche Pharma, Basel, Switzerland) was given s.c. at low dose (2 x 5 MIU rhIL-2/day) for 7 days, starting the day before vaccination, and for the first but not subsequent cycles.
Multimer and IFN-gamma ELISPOT assays
Ficoll-Paque centrifuged PBLs (1-2 x 107) were cryopreserved in RPMI 1640, 40% FCS, and 10% DMSO. Phycoerythrin-labeled HLA-A*0201/peptide multimers (13, 48) (originally called tetramers) were made with Melan-A26-35 analog peptide ‘ELA’ (ELAGIGILTV) or influenza matrix protein58-66 peptide (GILGFVFTL). Five color stains were carried out with anti-CD28FITC, multimers, anti-CD45RAECD, anti-CD8APC, and DAPI reagents. Pre- and postimmune samples were thawed and tested in the same experiment. A MiniMACS device (Miltenyi Biotech Inc.) was used to purify CD8+ T cells by positive sorting, resulting in >97% CD3+ CD8+ cells. Cells (106) were incubated with multimers (1 µg/ml, 60 min, RT) and then with fluorescent antibodies (anti-CD8 and -CD28, Becton-Dickinson, Mountain View, CA; anti-CD45RA, Immunotech, Marseille, France) for 30 min at 4°C. Dead cells were stained with DAPI (1 µg/ml, 30 min; Molecular Probes, Eugene, OR). Then data for 5 x 105 CD8+ T cells/sample were acquired in a FACSVantage™ machine (Becton Dickinson, Mountain View, CA), and analyzed using CellQuest software. IFN-gamma ELISPOT assays were performed using IFN-gamma-specific antibodies (Diaclone, Biotest, Switzerland) (49). Briefly, plates were coated overnight with antibody to human IFN-gamma and washed 6 times. We added 1.66 x 105 PBLs/well in 200 µl Iscove medium supplemented with 8% human serum and 10 µg/ml peptide and incubated them for 16 h at 37°C. Assays were performed in 6 replicates, including cultures without peptide and with HIV-1 polymerase peptide ILKEPVHGV. Cells were removed and plates were developed with a second (biotinylated) antibody to human IFN-gamma and streptavidin-alkaline phosphatase (Diaclone, Biotest, Switzerland). The spots were revealed with BCIP/NBT substrate (Sigma Tablets) and counted with an automatic reader (Bioreader® 2000, BioSys GmbH, Karben-Frankfurt Germany). The percentage of CD3+ CD8+ cells in PBLs was determined by flow cytometry on the same batch of cryopreserved cells, which allowed us to calculate percentage spots per CD8+ T cells.
Validation of multimer and IFN-gamma ELISPOT assays
Special emphasis was given to standardizing and validating multimer and ELISPOT assays. This is described elsewhere (14) and is not a clinical study report, but rather a quality assessment of tetramer and ELISPOT analyses through compiled data analysis from nonselected blood samples (n = 180) obtained from nonselected populations of healthy donors and melanoma patients. Briefly, for both multimer and IFN-gamma ELISPOT assays, the detection limit was 100 cells in 106 CD8 T cells (0.01%), as determined with PBLs from healthy HLA-A*0201 positive individuals (19). Repeated experiments with A2/Melan-A and A2/influenza multimers (compiled data) showed variations of 15 ± 16% (mean ± SD). Good reproducibility was also found for the ELISPOT assay (30 ± 21% for Melan-A and 44 ± 31% for influenza). In addition, multiple blood samples per individual were tested in order to assess longitudinal intraindividual result variability. Variation coefficients (SD expressed as a percentage of the mean value) were 20 ± 27% for Melan-A and 28 ± 31% for influenza multimers. Additional controls were performed with double multimer staining using PE-labeled A2/Melan-A multimers and APC-labeled A2/influenza multimers. More than 99% of multimer positive cells did not stain with the nonspecific multimer, confirming a high level of specificity. For IFN-gamma ELISPOT assays, negative controls were carried out for each PBL sample, where we found 0.0019 ± 0.0024% spots in CD8 T cells (mean ± SD of all tests of this study), confirming that the background was low; that is, very much below the detection limit of 0.01%. Taken together, these results support the notion that a greater than two-fold increase in T-cell frequency reflects a true T cell response in vivo.
Combined immune response score
To quantitate immune activation, we applied an arbitrary scoring system. Values after vaccination were compared to those before vaccination. For increases in the percentage of T cells detected with multimers or ELISPOT, the scoring was: 1 point for an increase of >2- to 3-fold; 2 points for >3- to 10-fold; and 3 points for >10-fold. For shifts in phenotype; that is, for increases in CD45RA- or CD28- T cells, the scoring was: 1 point for >20 to 40%; 2 points for >40 to 60%; and 3 points for >60% negative cells. Negative score points were assigned in the same manner for decreased values after, as compared to before, vaccination.
DTH reactions
Ten micrograms of peptide in 30 µl PBS was injected intradermally, using separate skin sites for influenza and for Melan-A. Classification at 48 h was "++" for over 8-mm diameter induration; "+" for 4 to 8 mm; and "-" for less than 4 mm. DTH was performed at study entry and at the end of vaccination cycle 1.
Acknowledgments
We are grateful to all our patients for their generous collaboration and confidence. We are indebted to the Protocol Review Committee of the Ludwig Institute for Cancer Research and to the local medical and ethical committees for their suggestions for improving the protocol, and to E. Hoffman and L. Pugliese for supervising the trial. We gratefully acknowledge GlaxoSmithKline Biologicals for AS02B, Seppic for Montanide ISA-51 (IFA), Roche Pharma for IL-2, R. Murphy for peptides, F.-A. Le Gal, O. Michielin, and S. Leyvraz for collaboration and advice, and V. Aubert, J.-M. Tiercy, and B. Mazzi for HLA typing. We are also grateful for the excellent technical and secretarial help of D. Minaïdis, C. Baroffio, R. Milesi, N. Montandon, K. Muehlethaler, M. van Overloop, S. Salvi, and A. Porret. This work was sponsored by the Ludwig Institute for Cancer Research.
- Received March 1, 2004.
- Accepted April 27, 2004.
- Copyright © 2004 by Daniel E. Speiser
References
Volume 4, Issue 1
