Circulating CD4⁺ T Cells That Produce IL4 or IL17 When Stimulated by Melan-A but Not by NY-ESO-1 Have Negative Impacts on Survival of Patients with Stage IV Melanoma

Melanoma remains a significant clinical problem, still with an annual increase of about 7% and with potentially doubled incidence within the next 20 years (1, 2). Untreated patients with distant metastasis have a poor prognosis with a median overall survival of 9 months (3). However, recently developed systemic treatments such as the B-RAF inhibitors or the CTLA-4 antibody ipilimumab are now licensed as standard therapies and are likely to displace earlier chemotherapies where long-term survival and clinical responses were very rarely observed (4). Treatment with the B-RAF inhibitor vemurafenib results in clinical responses in a high proportion of patients, but recurrence occurs rapidly in essentially all of them (5). In contrast, treatment with ipilimumab can result in prolonged survival and even complete remissions, but it is effective in only a minority of patients (6). Other promising checkpoint inhibitors currently being trialed, such as anti-PD-1 (nivolumab or lambrolizumab) or PD-L1 (MPDL3280A) antibodies, have also yielded remarkable benefits and are expected to become standard therapies quite soon (7, 8). Importantly, combination therapies using modulators targeting different stages of the antitumor immune response are now being tested, with promising first results (9). However, it seems to remain the case that a sizeable proportion of patients fails to respond. Biomarkers that would help to select those patients where therapy is most likely to succeed are not yet available. It would clearly be valuable to be able to predict clinical outcome for patient selection to avoid side effects in patients who will not respond, and to reduce costs. This would also contribute to an understanding of the mechanism of action of such modulators, enabling improvement of therapeutic strategies to advance. Currently, only the amount of plasma lactate dehydrogenase (LDH) is an established biomarker in melanoma (10) and although several other serum markers have been investigated for their prognostic power (11, 12), none is widely accepted so far.

In a previous study, we investigated whether the presence of circulating T cells capable of recognizing common melanoma-associated antigens had a prognostic impact on survival in patients with melanoma. We screened a cohort of 84 stage IV patients and showed that the presence of peripheral NY-ESO-1– and/or Melan-A–reactive memory T cells had a strong independent positive prognostic impact on survival (13). Patients possessing these cells had a longer median survival and a higher 1-year survival rate than those without them. In the first analyses, we stratified patients into responders and nonresponders simply on the basis of whether they possessed cells producing at least one of the measured cytokines after antigen stimulation, regardless of which cytokine was produced (pro- or anti-inflammatory) and which cell type responded (CD4⁺ or CD8⁺). As a group, patients possessing NY-ESO-1– and/or Melan-A–reactive T cells experienced a highly statistically significant survival advantage over those lacking such cells. However, we also identified individual patients who had a short survival time despite the presence of NY-ESO-1– or Melan-A–reactive T cells defined in this manner. Here, we asked whether the responding T-cell phenotype and the nature of the cytokine response further distinguished between shorter- and longer-term survivors within the group of patients, preselected for NY-ESO-1- and/or Melan-A-reactive T cells, who as a whole had a survival advantage compared with patients lacking such T cells. It is important to note that all patients studied had longer than expected survival because of the presence of these reactive T cells, and that here we are examining differences within this long-surviving group. We used intracellular cytokine staining (ICS) to detect 6 cytokines simultaneously, allowing us to perform single-cell analysis of the phenotype (CD4⁺ or CD8⁺) and the function of the response [i.e., production of pro-inflammatory cytokines (IL2, IFNγ, TNF, IL17) and/or anti-inflammatory cytokines (IL4, IL10)]. We show that patient survival can be correlated with these parameters, providing a rationale for modulating precisely and specifically only that type of immune reactivity associated with prolonged survival in individual patients, and potentially revealing some aspects of the mechanistic basis of longer-term survival even in advanced melanoma.

We included 68 patients with melanoma, about half of whom (n = 47) had already been studied earlier (13). Only patients with stage IV melanoma with unresectable distant metastases at the time of blood draw and with available survival follow-up data were included. To study phenotype and function of antigen-specific T cells, reactivity against Melan-A- and/or NY-ESO-1-reactive T cells was a further inclusion criterion. Patients had given their written informed consent for biobanking. The conduct of this study was approved by the Ethics Committee, Tübingen (approvals 147/2011BO2 and 432/2011BO2). All experiments were performed and analyzed centrally in Tübingen.

CD4 and CD8 responses against NY-ESO-1 and Melan-A were measured as described previously (13). Briefly, peripheral blood mononuclear cells (PBMC) were stimulated with protein-spanning overlapping peptides (1 μg/mL; PepMix; JPT Peptide Technologies), representing each of the 2 antigens. In addition, cells were also stimulated with overlapping peptides representing the influenza A–derived molecules nucleocapsid protein (NP) and matrix protein 1 (MP1), serving as positive controls. After 12 days culture to amplify the memory cell response, antigen-reactive T cells were restimulated for 12 hours in the presence of Golgi-Plug (1 μL/mL; BD Biosciences) with autologous antigen-pulsed PBMCs (at a ratio of 1:2) that were previously fluorescence-labeled (CFSE: 5 μmol/L, Invitrogen; CPD eFluor 450: 10 μmol/L, eBioscience). Unpulsed PBMCs were used as a negative control. The assay readout was an intracellular staining for 6 different cytokines simultaneously. After coincubation, Fc receptors were blocked with Gamunex (human immunoglobulin; Bayer), and dead cells were labeled with ethidium monoazide (Invitrogen). Cells were fixed and permeabilized with CytoFix/CytoPerm (BD Biosciences) and stained with the following antibodies: CD3–Qdot655 (Invitrogen)/eFluor605 (eBioscience), anti-CD4 + Pacific Orange (Invitrogen), CD4–PerCP (BD Bioscience), CD8–APC-H7 (BD Bioscience), IFNγ–PE-Cy7 (BD Bioscience), TNF-PE (Miltenyi Biotec)/FITC (Biolegend), IL2–Alexa Fluor 700 (BioLegend), IL4–APC (BD Bioscience), IL10–Pacific Blue (eBioscience)/PE (eBioscience), and IL17–PerCP-Cy5.5 (eBioscience). Flow cytometry data were obtained using a 14-color BD LSR II (BD Biosciences) using FACSDiva software (BD Biosciences). Data were analyzed using FlowJo software (Tree Star Inc.).

We used established response criteria to define the presence of antigen-reactive T cells. For details of the gating strategy, see Supplementary Fig. S1. Briefly, after exclusion of duplicates and the autologous stimulator cells, we gated CD4⁺ and CD8⁺ cells within all viable CD3⁺ lymphocytes and analyzed them separately for the production of each cytokine. We evaluated the percentage of cytokine-producing cells among all gated T cells in sample A (restimulated with antigen-pulsed PBMCs) and sample B (restimulated with unpulsed PBMCs). For each cytokine, antigen-responsive T cells were defined as being present if (i) the stimulation index was >2 (sample A divided by sample B) and (ii) a clearly separate cytokine-producing population distinguishable from the nonproducing cells was present in sample A for either CD4⁺ or CD8⁺ T cells (14, 15). To ensure the quality of samples and the functional capacity of the T cells, (iii) a detectable cytokine response after stimulation by influenza-derived peptides was an additional criterion for inclusion.

Phenotype and function of antigen-reactive T cells was analyzed separately for NY-ESO-1, Melan-A, and influenza. Follow-up time was defined from the date when blood was drawn for T-cell analysis to the date of last follow-up or death. Alternatively, we also calculated follow-up time from the date of diagnosed stage IV disease to the date of last follow-up or death. Disease-specific survival probabilities have been calculated and only deaths because of melanoma have been considered, whereas deaths because of other causes were regarded as censored events. Estimates of cumulative survival probabilities according to Kaplan–Meier were described and compared using log-rank tests if sample size was n > 1 in each of the strata. Throughout the analysis, P < 0.05 was considered as statistically significant. Log-rank tests were performed using GraphPad Prism version 5.04 for Windows (GraphPad Software). Multivariate Cox proportional hazards analyses were used to determine the independent effect of prognostic factors. All cytokine responses were considered in multivariate analysis. CD4 and CD8 responses were analyzed separately. Results of the Cox model were described by means of hazard ratios (HR) together with P values based on the Wald test. Throughout the analysis, P values < 0.05 were considered statistically significant. Analyses were carried out using SPSS version 22 (IBM).

After screening for the presence of NY-ESO-1– and/or Melan-A–reactive T cells, 68 patients with unresectable distant metastasis and available clinical follow-up were included in this study. Median age was 55.5 years, with an interquartile range (IQR) of 47 to 67 years. A total of 67% of patients were assigned to the M category M1c. M category data were missing for 1 patient. A total of 67% of the patients were male. The median survival time (MST) of the whole cohort was 14 months (Table 1), as opposed to 6 months for those without such tumor antigen-reactive T cells in a historical cohort of patients (13).

Thirty-eight patients had Melan-A– and 54 patients NY-ESO-1–reactive T cells; 24 patients possessed T cells reactive against both antigens. MST was 11.5 months for the 54 patients with NY-ESO-1–reactive T cells (1-year survival rate 55%) and 12 months for the 38 patients responding to Melan-A (1-year survival rate 45%). Median follow-up was 26.5 months for 18 patients who were alive at the last follow-up and 11 months for 50 patients who died.

NY-ESO-1 peptides were recognized exclusively by CD4⁺ T cells but not by CD8⁺ T cells in 36 of the 54 responsive patients (67%). In contrast, responses of CD8⁺ T cells alone were seen only in 5 of these patients (9%) and responses of both CD4⁺ and CD8⁺ T cells were recorded in 13 patients (24%; Fig. 1). Thus, a total of 49 CD4⁺ and 18 CD8⁺ T-cell responses were detected upon stimulation with NY-ESO-1 peptides (Fig. 1). Summarized percentages of cytokine-producing cells within the CD4 and CD8 compartment respectively for patients with antigen-specific T cells compared with patients that lacked reactive T cells are shown in Supplementary Table S3. Examples of stainings are shown in Supplementary Fig. S1.

Multicolor flow cytometry allowed us to analyze cytokine production in the CD4 compartment separately from the CD8 compartment. NY-ESO-1–responsive CD4⁺ T cells (in total n = 49 patients) produced mainly the pro-inflammatory cytokines IFNγ (36 patients; 73%), TNF (32 patients; 65%), and IL2 (25 patients; 51%). The anti-inflammatory cytokine IL4 was detected only in a minority (9 patients; 18%). IL10 release by CD4⁺ T cells could not be detected in any patient after stimulation with NY-ESO-1, whereas IL17 was seen in 8 patients (16%; Supplementary Fig. S2A).

CD8⁺ T cells responding to NY-ESO-1 produced TNF in almost all patients (15/18; 83%), with IFNγ produced in 14 (78%), IL2 in 2 (11%), IL4 in 2 (11%), and IL10 in a single patient (6%). IL17 was not produced by NY-ESO-1–reactive CD8⁺ T cells in any of the patients (Supplementary Fig. S2B).

Patients were stratified by different clinical parameters to investigate survival impact by univariate analyses (Table 1). Age, gender, prior treatment, and whether the patient was in the M1a/b versus M1c category of the tumor–node–metastasis staging were found not to have a statistically significant impact on survival in this cohort of longer-surviving patients preselected for T-cell responsiveness to NY-ESO-1 or Melan-A (and thus having extended survival compared with the entire cohort). This is important, as one would expect that there would be a difference between M1a/b and M1c. These data suggest that in this preselected patient cohort, the presence of tumor antigen-reactive T cells overrides any further differences in survival that would otherwise be attributable to the M category.

Although NY-ESO-1 was mainly recognized by CD4⁺ T cells, there was no statistically significant difference in survival depending on whether the patient possessed either CD4⁺ or CD8⁺, or both, types of reactive T cells (Table 2). The production of a particular cytokine (by either CD4⁺ and CD8⁺ T cells) also failed to show any further impact on survival, in both uni- and multivariate analysis (Table 3).

In contrast to findings with NY-ESO-1 peptide stimulation, patients with measurable Melan-A responses most often had reactive T cells limited to the CD8 subset (Fig. 1). Fifteen of 38 patients (40%) showed CD8⁺ T-cell responses, with 13 (34%) having solely CD4 responses and 10 (26%) both CD4⁺ and CD8⁺ T-cell responses. Thus, in total, 23 CD4⁺ and 25 CD8⁺ T-cell responses to Melan-A peptides were observed.

TNF was the cytokine most frequently produced by Melan-A–reactive CD4⁺ cells (17/23 patients, 74%). Other pro-inflammatory cytokines were less frequently produced (IFNγ: 14 patients, 61%; IL2: 9 patients, 39%). The anti-inflammatory cytokine IL4 was produced by 4 patients (17%). IL17 production was seen in 3 patients (13%) and again, IL10 could not be detected in any of them (Supplementary Fig. S2A).

CD8⁺ Melan-A–reactive T cells mainly produced IFNγ (20/25 patients with CD8 responses; 80%) and TNF (14 patients; 56%). IL2 (3 patients; 12%), IL10 (3 patients; 12%), IL17 (1 patient; 4%), and IL4 (1 patient; 4%) detection was rare (Supplementary Fig. S2B).

Univariate analyses revealed that neither the presence nor absence of a Melan-A–specific CD8⁺ T-cell response (P = 0.1217) nor the presence or absence of a combined CD4⁺ and CD8⁺ T-cell response (P = 0.2752) was associated with a further survival benefit in this preselected group of patients with longer-than expected survival. However, patients who possessed Melan-A–specific CD4⁺ T cells had a relatively reduced survival time compared with patients without these cells in this group (P = 0.0190; Table 2). Patients with CD4⁺ T-cell responses against Melan-A had a MST of 11 months, compared with 27 months for the remaining patients without CD4⁺ Melan-A–reactive T cells (Fig. 2A). Next, we clustered patients into 3 groups (Fig. 2B): solely CD4⁺ T-cell responses versus solely CD8⁺ T-cell responses versus combined CD4⁺ and CD8⁺ responses. Although not being statistically significant (P = 0.1180), the unfavorable survival of those patients with a CD4⁺ response, either limited to CD4 alone or combined with CD8 (both MST: 11 months), suggests the negative impact of CD4⁺ responses to be dominant over potentially beneficial CD8⁺-mediated effects (MST: 27 months). Strikingly, however, it was not the presence of CD4⁺ T-cell reactivity to Melan-A per se that correlated with a relative survival disadvantage in this group, but the nature of this response. Univariate analysis revealed that patients whose CD4⁺ T cells responded to Melan-A stimulation by producing IL17 or IL4 both had a reduced survival time compared with patients whose Melan-A–reactive CD4⁺ T cells did not produce these cytokines (P = 0.0016 and 0.0245, respectively; Fig. 3 and Table 3). This was also true if follow-up time was calculated from the date of diagnosed stage IV disease to the date of last follow-up or death (see Supplementary Table S2). Patients whose CD4⁺ T cells produced IL17 had a MST of only 5 months, and those that produced IL4 a MST of only 6 months compared with 14 and 12 months, respectively. We subsequently performed multivariate analyses for cytokine responses within the CD4 and CD8 compartment separately (Table 3). Cox proportional hazards analyses revealed that within the CD4 compartment, IL17 production remains as an independent risk factor. Patients that did not possess IL17-producing Melan-A–reactive T cells within the CD4 compartment showed an HR of only 0.13 (P = 0.028). Interestingly, Cox proportional hazards analyses also revealed a survival benefit for patients that possess CD8⁺ IFN-γ–producing T cells specific for Melan-A (HR, 3.41 for patients lacking IFNγ; P = 0.033).

The general immune competence of all patients was controlled for by stimulation with influenza-derived peptides, against which all adults respond. In 67 of the 68 patients (99%), influenza antigens were recognized by both CD4⁺ and CD8⁺ T cells at the same time. The remaining patient showed a solely CD4⁺ response (Fig. 1).

CD4⁺ T cells responded to influenza stimulation mainly by producing IFNγ (67/68 patients (99%). TNF was detected in 60 patients (88%) and IL2 in 59 (87%). The cytokines IL4 and IL10 could be detected in 26 patients (38%) and 3 patients (4%), respectively. IL17 was detected in 17 patients (25%; Supplementary Fig. S2A).

IFNγ was detected in 66 of 67 patients (99%) with a CD8⁺ T-cell response. TNF was detected in 63 (94%), and IL2 in 22 patients (33%). IL4 was detected in 13 patients (19%), IL10 in 7 (10%) and IL17 in 12 patients (18%; Supplementary Fig. S2B).

None of the tested parameters of influenza reactivity had any impact on melanoma patients' survival (see Supplementary Table S1).

We previously reported that the presence of circulating NY-ESO-1– and/or Melan-A–reactive T cells was strongly associated with prolonged survival in patients with stage IV melanoma in general (13). Here, we characterized in detail the phenotype and the function of these antigen-specific T cells to identify determinants associated with the observed positive impact on survival.

Analyzing the phenotype of T cells targeting NY-ESO-1, we found that it is mainly a CD4 response that is present in patients with melanoma with favorable survival. Similar results were reported before. Valmori and colleagues were able to detect CD8⁺ NY-ESO-1–reactive T cells (using intracellular cytokine staining after in vitro stimulation) in only 6 of 11 vaccinated melanoma patients, whereas CD4⁺ responses could be detected in all but one patient (16). Interestingly, adoptively transferred NY-ESO-1–specific CD4⁺ T cells resulted in impressive clinical responses (17). However, even if naturally occurring CD4 responses targeting NY-ESO-1 were more frequently observed, meaningful CD8 responses were likewise detected after vaccination with full length protein (18). Therefore, class I as well as class II epitopes should be considered for NY-ESO-1–specific cancer immunotherapy.

Furthermore, immune monitoring using MHC-I multimers, especially after systemic treatments like chemotherapy or application of cytokines, would yield a very incomplete picture of the patient's response repertoire. Unmeasured CD4⁺ responses (in our study 67% of all NY-ESO-1–specific T cells) might be one of the reasons for the deficiency of data showing correlations between clinical outcome and immunologic findings. For example, Jäger and colleagues found only 3 of 15 patients with stage IV melanoma with detectable CD8⁺ NY-ESO-1–specific T cells, using tetramer technology (NY-ESO-1 p_157-165; ref. 19).

Stratification of patients according to detection of either CD4⁺ or CD8⁺ NY-ESO-1–specific T cells, or both, did not reveal any differences in the survival advantage. Patients with T cells reactive to NY-ESO-1 show prolonged survival compared with patients lacking these responses no matter what type of T-cell reactivity was present. Within this selected cohort, no other immunologic parameter had any impact on survival, confirming that NY-ESO-1 would be a good target for immunotherapy, in that any resulting response would be expected to correlate with a positive benefit. Some limited clinical trial data are consistent with this notion. Yuan and colleagues observed that 6 of 8 ipilimumab-treated patients with evidence of clinical benefit (partial response, complete response, stable disease) showed CD4⁺ and CD8⁺ T-cell responses to NY-ESO-1; only 1 of 5 patients that progressed showed a response on NY-ESO-1 (20). Hunder and colleagues reported durable clinical remissions, which were mediated by CD4⁺ T-cell clones with specificity for NY-ESO-1, in advanced melanoma (17).

In contrast to the findings with NY-ESO-1–reactive T cells in our patients, the ratio of CD4⁺ versus CD8⁺ antigen-reactive T cells was more equally distributed for Melan-A reactivity. Strikingly, the phenotype of the Melan-A–specific T cells did have a strong impact on survival time. Although the presence of CD8⁺ and/or CD4⁺ T cells targeting NY-ESO-1 was associated with favorable survival, the presence of CD4⁺ T cells targeting Melan-A was not associated with a survival benefit and even abolished the favorable effect of CD8⁺ T cells targeting Melan-A if both were detectable. This suggests a suppressive impact of these CD4⁺ T cells.

One possibility to explain these differences is that NY-ESO-1 belongs to the group of cancer/testis antigens. Expression is restricted to germline tissue and different types of cancers (21). Because of the lack of MHC-I and MHC-II expression in germline tissues, there is no presentation of these antigens to T cells in the physiologic situation. Immune tolerance mechanisms limiting these responses are not assumed. In contrast, Melan-A is constitutively expressed by melanoma cells as well as by normal melanocytes. To prevent autoimmunity, immune tolerance mechanisms are assumed. Among those mechanisms regulatory T cells play an important role.

We recently showed that the frequency of CD4⁺CD25⁺FoxP3⁺ regulatory T cells did not influence survival in patients with late-stage melanoma (22). However, recent data imply that Tregs exert their suppressive function in an antigen-specific manner (23). The CD4⁺ T cells targeting Melan-A, which abrogated the favorable effect of CD8⁺ T cells targeting Melan-A could represent such regulatory T cells. IL10, a cytokine that was reported to be secreted by antigen-specific Tregs (23) was not detected in our cohort. However, it was reported that their suppressive activity was rather related to cell–cell contact than mediated by secreted IL10 and that Tregs produce IL4 as well as pro-inflammatory cytokines, for example IFNγ (24). The same cytokines were likewise secreted by CD4⁺ Melan-A–specific T cells in our study. Therefore, our findings do not rule out the possibility that the antigen-specific CD4⁺ cells observed here include immunosuppressive regulatory T-cell populations. One way to address this issue would be the additional analysis of FoxP3, which was not investigated initially and could finally not be performed because of lack of PBMCs in our patients. This needs to be clarified in future studies.

The functional analysis measuring 6 different cytokines after stimulation with NY-ESO peptides showed a pattern of released cytokines similar to what was reported before (25). Nevertheless, no significant associations between released cytokine and survival were found. In contrast, a negative impact was observed for CD4⁺ T cells producing IL4 or IL17 after stimulation with Melan-A peptides. Patients that possessed Melan-A-=reactive CD4⁺ T cells producing the anti-inflammatory cytokine IL4 on stimulation these patients had an MST of only 6 months compared with 12 months for patients without IL4 responses (P = 0.0245). Interestingly, clinical trials administering IL4 for metastatic melanoma have failed (26). In addition, patients with T cells producing the (pro-inflammatory) cytokine IL17 upon stimulation with Melan-A also had a markedly lower survival time (MST 5 months vs. 14 months; P = 0.0016). The role of IL17-producing cells in cancer is still ambiguous (27). Derhovanessian and colleagues reported that high frequencies of a subset of IL17⁺ CD4⁺ T cells were associated with a reduced time to progression in vaccinated prostate cancer patients (28). Tosolini and colleagues found that colorectal cancer patients with low expression of Th17-associated gene expression had a survival benefit compared with patients with high gene expression levels (29). However, accumulating data suggest that IL17-producing cells might also mediate antitumor effects under certain conditions (30). For melanoma, only mouse models have been studied so far (31, 32).

Although multivariate analyses confirmed that a lack of IL17-producing CD4⁺ Melan-A–reactive T cells was independently associated with prolonged survival, low patient numbers clearly emphasize the need for confirmatory studies in a larger cohort. Nevertheless, depletion of IL17- and IL4-producing CD4⁺ T cells targeting Melan-A might be considered in adoptive T-cell transfer protocols, for example by using immunomagnetic separation techniques (33).

Thus, we believe that our data indicate that the nature of the antitumor T-cell response is different for different tumor antigens and has potentially impact on patient's survival. The results on Melan-A reactivity presented here may have important implications for the development of future immunotherapeutic protocols. Many clinical trials, especially using DNA, RNA, or whole protein vaccination, aim at inducing both CD4⁺ and CD8⁺ T-cell responses, believed always to be desirable. However, according to the results presented here for Melan-A, this may be counterproductive depending on the nature of the T-cell induced. Hence, vaccination with class I epitopes or adoptive T-cell transfer of previously generated, antigen-specific CD8⁺ T cells might be the most promising way of exploiting immune responses against this particular target antigen.

Concerning NY-ESO-1 reactivity, we confirmed the equally important role of either CD4⁺ or CD8⁺-mediated responses for patient survival, further supporting the importance of NY-ESO-1 as an immunotherapeutic target. CD4 responses targeting this antigen were more frequently observed than CD8 responses. This provides a rationale to further apply strategies promoting anti NY-ESO-1 CD4 responses in patients with melanoma.

T.K. Eigentler is a consultant/advisory board member for Bristol-Myers Squibb. M. Maio is a consultant/advisory board member for Bristol-Myers Squibb, GlaxoSmithKline, and Roche. No potential conflicts of interest were disclosed by the other authors.

Conception and design: H. Zelba, B. Weide, M. Maio, D. Schadendorf, C. Garbe, G. Pawelec

Development of methodology: H. Zelba, B. Weide, E. Derhovanessian, M. Maio, G. Pawelec

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Zelba, B. Weide, A. Martens, J.K. Bailur, C. Kyzirakos, A. Pflugfelder, T.K. Eigentler, A.M. Di Giacomo, M. Maio, E.H.J.G. Aarntzen, J. de Vries, A. Sucker, D. Schadendorf, G. Pawelec

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Zelba, B. Weide, A. Martens, E. Derhovanessian, J.K. Bailur, C. Kyzirakos, A.M. Di Giacomo, M. Maio, E.H.J.G. Aarntzen, P. Büttner, C. Garbe, G. Pawelec

Writing, review, and/or revision of the manuscript: H. Zelba, B. Weide, A. Martens, E. Derhovanessian, J.K. Bailur, C. Kyzirakos, A. Pflugfelder, T.K. Eigentler, A.M. Di Giacomo, M. Maio, E.H.J.G. Aarntzen, D. Schadendorf, P. Büttner, C. Garbe, G. Pawelec

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Weide, A. Sucker

Study supervision: H. Zelba, B. Weide, C. Garbe, G. Pawelec

This work was supported by the following grants: Graduiertenkolleg 794 (to H. Zelba), Sonderforschungsbereich 685 from the Deutsche Forschungsgemeinschaft (to B. Weide and C. Garbe), and project 01EI1401 ISPE-BREAST from the Federal Ministry of Education and Research (BMBF) (to G. Pawelec).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.