Extended Dose Ipilimumab with a Peptide Vaccine: Immune Correlates Associated with Clinical Benefit in Patients with Resected High-Risk Stage IIIc/IV Melanoma

The clinical development of antibodies that abrogate or augment costimulatory interactions between T cells and antigen-presenting cells has opened a new field with great promise. Cytotoxic T lymphocyte antigen-4 (CTLA-4) is a checkpoint protein expressed on activated T cells and has been shown to bind to B7-1 and B7-2 molecules expressed on mature, activated dendritic cells and other antigen presenting cells, preventing their interaction with CD28 on T cells (1, 2). The expression of membrane-bound CTLA-4 downmodulates T-cell responses. The CTLA-4–B7 interaction serves as a “brake” for T-cell responses and leads to decreases in the PI3 kinase pathway (3, 4). A fully human IgG1 antibody, ipilimumab, binds to and inhibits CTLA-4, leading to increases in T-cell activation and IL-2 secretion. CTLA-4 antibody has been shown to induce rejection of established tumors and decrease relapses in a rodent model of prostate cancer in which spontaneous metastasis occurs. CTLA-4 blockade augmented the clinical activity of murine tumor cell vaccines against established tumors, and depletion of T regulatory cells further increased the activity of the antibody (5–8). These data led to the testing of CTLA-4 antibody alone or with a peptide vaccine in patients after complete surgical resection of high-risk nodal or distant metastatic melanoma. Ipilimumab at doses of 3 mg/kg or greater, alone or with a peptide vaccine, induces regression of metastatic melanoma and has activity in renal cell cancer and prostate cancer (9–14). A recent phase III clinical trial demonstrated that ipilimumab prolonged overall survival in patients with unresectable stages III and IV melanoma after failing at least 1 prior therapy (15). Responses after treatment with ipilimumab are often sustained in the absence of further therapy but have been shown to be associated with immune-related adverse events (irAEs) (11–14). In a prior phase I study, we administered increasing doses of the CTLA-4 antibody ipilimumab and demonstrated that irAEs were inversely correlated with the rate of relapse in patients with resected advanced melanoma (13). No clinical parameter has been consistently found to be a surrogate or predictive marker for response to ipilimumab therapy. To further investigate the activity of ipilimumab and potential factors associated with clinical benefit, we performed a pilot phase II adjuvant study of doses of 3 and 10 mg/kg administered at the extended dosing interval of 6–8 weeks. As in our prior phase I trial, a multipeptide vaccine was also administered to HLA-A*0201+ patients with frequent injections over 1 year. Primary endpoints of the study were toxicity and tolerability, with immune parameters, relapse and time to relapse or death being secondary endpoints. Patients were also followed for overall survival; however, this was outside the scope of the original trial design. Data involving baseline CRP and change in Th17 inducibility of peripheral blood mononuclear cells (PBMC) were obtained and correlated with outcome.

Eligible patients were ≥18 years of age with resected stage IIIc or IV melanoma according to the 2002 modified American Joint Commission on Cancer staging system and were rendered free of disease surgically. Patients were required to have a magnetic resonance imaging or computed tomographic (CT) scan of the brain and CT imaging of the chest, abdomen, and pelvis performed within 4 weeks of initiation of therapy showing no evidence of disease. Eligibility criteria also included adequate renal, hepatic, and hematologic function. To be eligible for vaccination, patients were required to be HLA-A*0201+, and immunohistochemical tumor staining for 1 or more of MART-1/Melan-A, HMB-45, and tyrosinase on at least 10% of tumor cells was required. Fifty patients were HLA-A*0201+ by a DNA polymerase chain reaction (PCR) assay performed at the UCLA Immunogenetics or the American Red Cross Laboratories. Ocular and mucosal melanomas were included. Exclusion criteria included active autoimmune disease, steroid dependence, and prior treatment with ipilimumab. Patients were required to comprehend and sign an informed consent form approved by the Cancer Therapy Research Program of the National Cancer Institute (CTEP/NCI) and the Los Angeles County/University of Southern California or University of South Florida Institutional Review Boards.

Ipilimumab was administered intravenously at 3 or 10 mg/kg over 90 minutes every 6 to 8 weeks for 12 months. Reductions in dose were not permitted. Consenting patients who were free from relapse and significant toxicity were continued on maintenance ipilimumab at 10 mg/kg administered every 3 months. For patients who were HLA-A*0201+, antibody infusions were accompanied by 3 separate subcutaneous vaccine injections of 1 mg each of tyrosinase_368–376 (370D), gp100_209–217 (210M), and MART-1_26–35 (27L) peptides emulsified in Montanide ISA 51 VG in alternating lower extremities. The peptide vaccines were administered as outpatient therapy as previously described (16). The intervals between injections were every 2 weeks for the first 6 injections, every 4 weeks for the next 4 injections, and 12 weeks between the final 2 injections. Leukapheresis with exchange of 5 to 7 L to obtain PBMC for immune analyses was performed within 1 week before and at 6 months after the initiation of therapy. Patients were followed until relapse. Overall survival was also determined but was outside the scope of the original trial design.

GMP-grade tyrosinase_368–376 (370D) [NSC# 699048], MART-1_26–35 (27L) [NSC# 709401], and gp100_209–217 (210M) [NSC# 683472] 9 or 10 amino acid epitope peptides were prepared. Peptides were produced by Ben Venue Laboratories, Inc. and were provided by CTEP/NCI. Montanide ISA-51 VG was manufactured by Seppic, Inc.

Ipilimumab (a CTLA4 monoclonal antibody formerly known as MDX-010) is a fully human immunoglobulin (IgG₁) anti-CTLA4 monoclonal antibody (manufactured by Medarex, Inc. and Bristol Myers Squibb).

Pheresis samples were processed to purify PBMCs and frozen at −168°C as previously described (16).

PBMCs were thawed and cultured overnight and then tested in an ELISPOT assay performed as previously described by our group (16). After processing, ELISPOT plates were read on a KS Elispot reader (Carl Zeiss). Values were normalized to spots per 100,000 CD8+ T cells.

PBMC from evaluable patients were stained with fluorophore-conjugated anti-CD3, anti-CD8, anti-CD4, anti-CD25, anti-HLA-DR, anti-peptide/HLA-A2.1 tetramers, anti-CD14, anti-CD19, and anti-CD56 antibodies along with a fluorescent viability marker at 4°C for 30 minutes. Cells were analyzed on a FACS LSR II (BD Biosciences) as previously described (17). Tumor-specific effector T cells were characterized as viable CD4, 14, 19, and 56−, CD3 and CD8+ using FloJo software (Treestar). The frequency of tetramer expression of effector T cells was determined by gating on the tetramer-high population compared to the negative tetramer control-stained cells. Activated helper and effector T cells were characterized as viable CD3 and HLA-DR+, CD14, 19, and 56−, and CD4 or CD8+, respectively.

Immunohistochemical staining of paraffin-embedded sections for gp100, MART-1, and tyrosinase was performed as previously described (13).

Fresh PBMC derived from leukapheresis were stained with anti-CD4 and CD25 antibodies (BD Biosciences), and T regulatory cells were isolated by performing fluorescence-activated cell sorting with a BD FACS-Aria (BD Biosciences). The recovered T regulatory cells were added in increasing concentrations to a mixed lymphocyte reaction (MLR). The MLR comprised co-incubation of T regulatory cells, and a 10:1 ratio of autologous responder cells and allogeneic dendritic cells pretreated with 30 Gy of radiation. Cells were incubated in a 96-well round bottom plate at 37°C/5%CO₂ for 4 days, after which 1 μCi of ³H-Thymidine was added per well. Cells were incubated for an additional 16 hours then washed and transferred onto a filter mat using a Filtermate Harvester (Perkin Elmer). Filter mats were assayed for ³H activity by a MicroBeta Trilux Luminescence Counter (Perkin Elmer). All assays were performed in quintuplicate.

To induce Th-17 cells, available PBMC derived from patients before and 6 months after ipilimumab treatment were cultured with PMA (250 ng/mL) and ionomycin (1 μM) for 2.5 hours, with 4 μL of GolgiStop (BD Biosciences) per 6 mL of culture media added after the first 30 minutes. The cells were recovered, stained with a viability dye, and fluorophore-labeled anti-CD4, and anti-CD8 (BD Biosciences). Cells were then permeabilized (BD Cytofix/Perm kit) and stained with fluorophore-labeled anti-IL17 (eBiosciences) or matched isotype control. Cells were analyzed on a BD FACS Calibur cytometer, and the data were analyzed with FloJo software (Treestar). Th17 inducible frequency was determined by the frequency of total viable CD4+ CD8− cells that stained positively for IL17. The change in Th17 frequency was calculated by subtracting the pretreatment frequency from the posttreatment frequency.

Primary endpoints of the study were toxicity and tolerability. An irAE was considered to be “significant” if the event required initiation of systemic steroids and/or cessation of therapy. Multiple immune parameters, relapse status, and time to relapse or death were secondary endpoints. Patients were also followed for overall survival; however, this was outside the scope of the original trial design. Differences in the pre- and posttreatment responses to gp100_209–217 (210M), MART-1_26–35 (27L), and tyrosinase_368–376 (370D) peptides (Beckman Coulter) in ELISPOT assays and tetramer expression assayed by flow cytometry were evaluated by the paired Student's t-test or Wilcoxon signed rank test, whichever was appropriate. For some statistical analyses, a cut point was chosen post hoc to divide patients into roughly equal groups. This facilitated statistical analyses of the association of a given covariate with outcome endpoints. The 2-sample t-test (with equal or unequal variances) or Wilcoxon rank-sum test, when appropriate, was used to study the difference of a continuous variable such as change in Th17. Association between a categorical variable and the relapse status was examined using a contingency table and Chi-square or Fisher exact test. The Kaplan–Meier (KM) product limit method was used to estimate the distribution of a time-to-event endpoint such as relapse-free survival (RFS) and overall survival (OS); and the log-rank test was used to informally compare the 2 KM curves between 2 groups of patients. Statistical analyses were performed using SAS version 9.1 or later. A P-value of less than 0.05 was considered to be statistically significant. No multiple testing adjustment was made due to the exploratory nature of the study.

Seventy-five patients were treated in this phase II clinical trial, including 44 (59%) men and 31 (41%) women with a median age of 56 (range: 21–78) years. Forty-six (61%) patients had resected stage IV disease, and the remaining 29 (39%) had resected stage IIIc disease (Table 1). Forty-four (59%) patients had received prior immunotherapy with a cell vaccine, GM-CSF, high-dose IL-2, or interferon-α, 4 had received biochemotherapy, and all patients had undergone surgery (Table 1). Twenty-five (33%) patients received ipilimumab at 3 mg/kg, and 50 (67%) patients received 10 mg/kg.

A total of 28 (37%) of 75 patients on this trial had a “significant” irAE, which was defined as a grade II, III, or IV irAE that required cessation of therapy and/or initiation of systemic steroid therapy. Of these, 11 (15%) patients had hypophysitis, 10 (13%) patients had colitis, and 2 (3%) had dermatitis. In the remaining 5 (7%) patients, 1 case of each of the following irAEs was observed: sarcoidosis, hypercalcemia with renal failure, idiopathic thrombocytopenic purpura, hepatitis, and arthritis. Six patients had a grade II irAE, 21 patients had a grade III irAE, and 1 patient had a grade IV irAE. All 21 patients with grade III irAEs received tapering doses of systemic steroids, and all 12 patients with grade III gastrointestinal and skin toxicities all returned to normal within 3 months of the onset of symptoms. No patients with grade II gastrointestinal irAEs required cessation of therapy or initiation of systemic steroids. By treatment guidelines, these patients were treated symptomatically with loperamide or diphenoxylate and budesonide, a nonabsorbable steroid, by mouth. Two patients developed endoscopically documented colitis that required surgical intervention. One patient required bowel rest with a diverting ileostomy and total parenteral nutrition for 4 months and eventually fully recovered. Another patient developed a bowel perforation requiring partial colectomy and eventually fully recovered to pretreatment status. Of 11 patients with grade II or III hypophysitis, all required replacement corticosteroids, 6 required treatment for at least 2 years, including 1 patient for 5 years who remains well without symptoms. Of note, 3 of the 11 patients with hypophysitis have been successfully weaned from systemic steroids. This indicates that steroid requirement induced by ipilimumab is not necessarily permanent. There were no treatment-related deaths. Toxicities, which were predominantly immune-related, are summarized in Table 2.

Of the 75 patients, 37 (49%) finished all 7 doses of ipilimumab, 17 (23%) stopped treatment early because of dose-limiting toxicity, 20 (27%) patients stopped treatment early due to relapse during the first year of treatment, and 1 (1%) patient withdrew consent before completing 7 doses. Twenty (27%) patients continued on maintenance ipilimumab after the initial 7 doses, 1 of whom has continued for 3.5 years without significant toxicity or relapse. There have been 33 (44%) relapses to date and a 2-year RFS rate of 56% (95% CI: 44–67%) with a median follow-up of 29.5 (range: 13.8–68.4) months as of June 1, 2010 (Fig. 1A). Of 33 patients who relapsed, 9 (12%) are again free of disease after subsequent treatment and 8 (11%) are alive with disease. There were 16 (21%) deaths, all attributable to disease, and the 2-year OS was 86% (95% CI: 75–92%, Fig. 1B). There was a trend in favor of the stage IIIc patients for longer duration of RFS compared to stage IV patients (log-rank, P = 0.20, Fig. 1C). Of 33 patients with a relapse, only 8 (24%) had significant irAEs; 20 (48%) of 42 without relapse had a significant irAE (OR = 0.35, 95% CI: 0.13–0.96, Chi-square, P = 0.038). The median RFS and OS without a significant irAE were 20 and 58 months, respectively, whereas the median RFS and OS with a significant irAE have not been reached after a median follow-up of 29.5 months. Development of significant irAEs appeared to be positively associated with a longer duration of RFS (HR = 0.43, 95% CI: 0.19–0.91; log-rank, P = 0.031, Fig. 1D). Further follow-up may help determine whether overall survival is significantly associated with irAE.

Sixty-one of 75 patients completed leukapheresis both prior to, and 6 months after initiating the trial. For HLA-A*0201+ patients, peptide-specific immune responses to MART-1, gp100, and tyrosinase were assayed using functional, γ-interferon ELISPOT assays as well as flow cytometry for tetramer expression on matched PBMC before and 6 months after starting ipilimumab treatment. For ELISPOT assays, 40 patients of 50 with available PBMC were evaluated. Only 10 (25%) of the 40 evaluable patients had evidence of an immune response to MART-1 or gp100 (data not shown). There was no clear correlation of ELISPOT reactivity with relapse status or with irAEs, although the limited number of patients precludes making definitive correlations. No significant ELISPOT reactivity to tyrosinase was observed (data not shown). There were 23 patients who received vaccine and who had available PBMC evaluable for tetramer expression by flow cytometry. In contrast to the ELISPOT results, there was a significant increase in MART 27L tetramer expression 6 months after treatment compared to baseline (median expression posttreatment was 0.30 vs. median pretreatment expression of 0.17, Wilcoxon, P < 0.0001, Fig. 2A).

Because ipilimumab did not consistently increase antigen-specific circulating T cells in vitro, we examined the effect of ipilimumab on suppressive T regulatory cells. T regulatory cells that were CD4+, CD25+ high were measured by flow cytometry, and no difference before treatment compared to 6-month posttreatment initiation (after 4 doses of CTLA-4 antibody) was observed (data not shown). This is consistent with prior reports in metastatic melanoma (17, and data not shown). Fresh PBMC pre- and 6 months posttreatment from 8 patients were also sorted and recovered by flow cytometry on the basis of CD4 positivity and high levels of CD25 to perform assays measuring T regulatory activity. The sorted cells were titrated into an MLR with irradiated allogeneic dendritic cells and autologous effector cells. Their inhibitory activity was evaluated by suppression of an allogeneic MLR as their ratios to effectors were increased. No differences in MLR reactivity were observed in the pre- versus posttreatment samples as T regulatory cell numbers were increased. These data shown in Figure 2B suggest that administration of ipilimumab did not affect the number or function of natural CD4+ CD25+ high, T regulatory cells.

We assayed the number and activation status of overall circulating immune cells before and after treatment. Flow cytometry analyses were performed on a subset of PBMC specimens at the time of the first, second, and third ipilimumab infusions as above. CD3+, CD4+, CD8+ T cells, CD56+ NK cells, and CD19+ B cells were enumerated. There were no consistent changes over time in those subsets after ipilimumab treatment (data not shown). The activation marker HLA-DR was also measured on T cells. A significant increase over time in both CD4+/HLA-DR+ and CD8+/HLA-DR+ T cells (Wilcoxon, P < 0.001) was noted, as shown in Figure 2C, reflecting an increased population of activated T helper and cytolytic cells.

We investigated the effect of ipilimumab on a novel class of helper T cells termed Th17 cells. Th17 cells have been associated with autoimmune and antitumor immunity in murine models and in human autoimmune diseases (18–20). A report on a different CTLA-4 blocking antibody, tremelimumab, previously demonstrated a treatment-induced increase in the inducibility of Th17 cells (21). In our patients, ipilimumab treatment significantly increased the inducibility of Th17 cells after stimulation with PMA and ionomycin (Wilcoxon, P < 0.0001, Fig. 2D). We subsequently explored the association of Th17 inducibility with relapse, discussed later.

No factors have been consistently associated with clinical benefit from ipilimumab. A number of serologic and hematologic parameters were assessed at baseline and at varying time points after initiating ipilimumab. Patients whose baseline C-reactive protein (CRP) was 2 mg/L or greater had a significantly improved freedom from relapse (OR of relapse comparing <2 vs. ≥2: 0.34, 95% CI: 0.13–0.90; Fisher exact, P = 0.035, Figure 3A) as well as a marginally significantly improved RFS (HR: 0.50; 95% CI: 0.24–1.11, P = 0.081). As described earlier, treatment with ipilimumab was associated with a significant increase in Th-17 inducibility. Th-17 inducibility was also associated with outcome, as a higher change in Th17 inducibility from baseline to 6 months was positively associated with freedom from relapse in evaluable patients (n = 47; median Th17 change: 0.04 and 0.09 for those who never relapsed vs. those who did during the study, respectively; Satterwaithe t-test, P = 0.047, Fig. 3B). The small number of deaths precluded an assessment regarding overall survival. There were no significant correlations between any chemistry or hematologic parameter and outcome analyzed, including baseline and change in absolute lymphocyte count (Fig. 3C, and data not shown).

In this clinical trial, we treated 75 patients with high-risk resected melanoma with extended dose ipilimumab every 6–8 weeks. HLA-A*0201 patients also received multiple peptide vaccinations. Overall clinical results were favorable, with only 16 (21%) deaths and 33 (44%) relapses in 75 patients. Nine (12%) patients that relapsed were rendered free of disease with subsequent therapy. Median relapse-free and overall survivals for all patients have not yet been reached after a median follow-up of 29.5 months in a population of patients with a poor expected outcome. These results are promising compared to historical reports. Two recent reports, 1 of resected stage IV melanoma patients and 1 involving resected stage IIIb, IIIc disease, reported median relapse-free survivals of 7.2 and 8.8 months, respectively (22, 23). In this study, the median relapse-free survival of stage IV patients was 40.5 months and was not reached for the stage IIIc patients after a median of 29.5 months of follow-up. However, larger randomized trials will be required to definitively establish the utility of ipilimumab in the adjuvant setting.

Data derived from patients receiving ipilimumab doses of 3 and 10 mg/kg were pooled. Pooling enhanced the robustness of the statistical analysis due to the small number of patients in the 2 dosing cohorts. We did not observe a significant difference in toxicity or outcome when comparing the 2 dosing regimens. Wolchok and colleagues recently reported a randomized phase II clinical trial involving patients with unresectable advanced melanoma who were randomized to 0.3, 3, and 10 mg/kg (24). Although there was a favorable difference in response rate in the 10 mg/kg group, consistent with our findings, the authors reported that overall survival and the disease control rate were not significantly different between the 3 and 10 mg/kg groups.

There was an inverse association between the development of significant irAEs and relapse, supporting results from a prior trial (13) and consistent with published reports in patients with metastatic melanoma (14). There were 2 episodes of grade III colitis requiring surgery. One patient required a diverting loop ileostomy and bowel rest with TPN, the other had a perforation that required colectomy. Both patients eventually recovered fully to pretreatment status, and both patients are currently free of disease. There were no treatment-related deaths. A 37% rate of significant irAEs was observed, the majority of which were managed with outpatient steroids and cessation of ipilimumab. A potential concern for the use of ipilimumab in the adjuvant setting involves the development of permanent steroid dependence due to hypophysitis that could compromise future immunotherapy in the event of a subsequent relapse. In this report, 6 (8%) of 75 patients required replacement steroids due to persistent hypophysitis for more than 2 years. Only 1 of these patients had a relapse but was rendered free of disease with subsequent resection. Thus, steroid requirement associated with irAE on this trial did not significantly impair subsequent treatment after relapse. Three of 11 patients requiring hormone replacement for hypophysitis were successfully weaned, indicating that hypophysitis induced by ipilimumab is reversible in some patients.

In this study we identified 2 markers associated with clinical benefit from ipilimumab. A higher frequency of Th17 inducible cells after 6 months of therapy was a surrogate marker for freedom from relapse. A previous study involving a different CTLA-4 blocking antibody, tremelimumab, in unresectable stage IV melanoma patients, reported that low baseline CRP levels were associated with overall survival (25). However, in our report, a higher baseline CRP level appeared to be positively associated with freedom from relapse (odds of having no relapse for those whose baseline CRP < 2.0 was reduced by 66% compared to those with a baseline CRP ≥ 2.0, P = 0.035). In contrast to a prior report involving ipilimumab treatment of unresectable stage IV melanoma, we did not observe a correlation of increase in absolute lymphocyte count to clinical benefit (26). The difference in the CRP data could be explained by the use of a different class of CTLA-4 blocking antibody in the first study. More likely, however, the observed differences in both results involving baseline CRP and absolute lymphocyte count could be due to the inherent difference in the biology of advanced measurable disease compared to that of resected disease.

Antigen-specific immune assays did not reveal any association with clinical benefit, despite a significant increase in tetramer MART-positive circulating T cells after vaccination. CD4+/HLA-DR and CD8+/HLA-DR+ T cells were increased in number, reflecting a general state of immune activation and a pharmacodynamic marker for the effect of ipilimumab, as shown in past trials (13, 14). Assays for the function of T regulatory cells using sorted CD4+CD25+ high T cells did not reveal any differences after treatment with CTLA-4 antibody. These data confirm prior in vitro findings (27, 28, and Weber and colleagues, unpublished observations) and published experiments from patients with metastatic melanoma (17). However, ipilimumab may influence T regulatory cells in the tumor microenvironment rather than in the circulation. In patients with localized bladder cancer, Liakou and colleagues observed a consistent decrease in FoxP3+ T regulatory cells in tumor-infiltrating lymphocytes after CTLA-4 blockade (P < 0.05), whereas the effect on circulating T regulatory cells was inconsistent (29). Because our study was conducted in the adjuvant setting after surgical resection, analysis of the tumor microenvironment was precluded.

Important questions remain about CTLA-4 abrogation as a cancer therapy. By what immune mechanism do CTLA-4 antibodies induce clinical benefit in melanoma? It is possible that CD4 cells may be the effectors that mediate clinical benefit with ipilimumab, and that it is the provision of augmented nonspecific T-cell help that is responsible for the antimelanoma immune response. Trials combining CTLA-4 antibody with class I peptide vaccines have not documented increased peptide-specific CD8 immune responses in peripheral blood samples (13, 14). However, if CTLA-4 abrogation does indeed act at the level of the CD8+ cytolytic T cell, the effects may best be evaluated in the draining lymph nodes and in tumor-infiltrating T cells rather than in peripheral blood.

Alternatively, epitope-spreading to antigens not included in the vaccine may occur. The cancer testis antigen, NY-ESO-1, represents a potential candidate. A recent study reported 15 patients with metastatic melanoma treated with ipilimumab, and treatment response was correlated with de novo NY-ESO-1 reactivity without prior NY-ESO-1 immunization. Five of the 8 patients with a clinical response to CTLA-4 blockade were sero-positive for NY-ESO-1, whereas all 7 nonresponders were sero-negative (30). Finally, the peptide vaccination itself may not be important for the induction of antitumor immunity, and CTLA-4 blockade alone may be sufficient to derive clinical benefit.

Important therapeutic issues for CTLA-4 abrogating antibodies are the value of long-term maintenance therapy and the toxicities associated with it. This has been partly answered by the current trial, because 32% of patients exhibited grade III toxicity with a dosing interval of 8 weeks. These toxicity data are similar to a previous study with approximately the same proportion of grade III toxicity when the drug was given with a similar vaccine at the same dose but at a frequency of every 3 weeks to patients with unresectable stage IV melanoma (31). The onset of toxicity was delayed compared to this study, but the eventual proportion of patients with grade III irAEs was similar. The value of long-term maintenance therapy is still unknown, but maintenance therapy is currently being evaluated in trials of patients with stage IV unresectable and stage III resected disease.

No potential conflicts of interest were disclosed.

We thank Rasa Hamilton at the Moffitt Cancer Center for editorial assistance, and Jodi Kroeger and Kate Shapland from the Moffitt Flow Cytometry core for their technical advice and assistance. Support from grant NCI CA 109307 and from the Donald A. Adam Comprehensive Melanoma Research Center is acknowledged.

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.