J Clin Oncol 26: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 26 NUMBER 19 JULY 1 28 JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T Phenotypic and Functional Analysis of Dendritic Cells and Clinical Outcome in Patients With High-Risk Melanoma Treated With Adjuvant Granulocyte Macrophage Colony- Stimulating Factor Adil I. Daud, Noweeda Mirza, Brianna Lenox, Stephanie Andrews, Patricia Urbas, Gui X. Gao, Ji-Hyun Lee, Vernon K. Sondak, Adam I. Riker, Ronald C. DeConti, and Dmitry Gabrilovich From the H. Lee Moffitt Cancer Center, Tampa, FL; and Mitchell Cancer Institute, University of South Alabama, Mobile, AL. Submitted August 8, 27; accepted March 11, 28. Supported by Berlex Oncology (now Bayer Corp). Authors disclosures of potential conflicts of interest and author contributions are found at the end of this article. Corresponding author: Adil I. Daud, MD, Department of Medicine, Division of Hematology-Oncology, University of California San Francisco, 16 Divisadero St, San Francisco, CA ; adaud@medicine.ucsf.edu. 28 by American Society of Clinical Oncology X/8/ /$2. DOI: 1.12/JCO A B S T R A C T Purpose Granulocyte macrophage colony-stimulating factor (GM-CSF) can induce differentiation of dendritic cells (DCs) in preclinical models. We hypothesized that GM-CSF stimulated DC differentiation may result in clinical benefit in patients with high-risk melanoma. Patients and Methods We conducted a prospective trial in patients with high-risk (stage III B/C, IV), resected melanoma, with GM-CSF 1 g/m 2 /d administered for 14 days every 28 days. Patients underwent clinical restaging every four cycles, with DC analysis performed at baseline and at 2, 4, 8, and 12 weeks. Results Of 42 patients enrolled, 39 were assessable for clinical outcome and DC analysis. Median overall survival was 65 months (95% CI, 43 to 67 months) and recurrence-free survival was 5.6 months (95% CI, 3 to 11 months). GM-CSF treatment caused an increase in mature DCs, first identified after 2 weeks of treatment, normalizing by 4 weeks. Patients with decreased DCs at baseline had significant increases in DC number and function compared with those with normal parameters at baseline. No change was observed in the number of myeloid-derived suppressor cells (MDSCs). Early recurrence ( 9 days) correlated with a decreased effect of GM-CSF on host DCs, compared with late or no (evidence of) recurrence. Conclusion GM-CSF treatment was associated with a transient increase in mature DCs, but not MDSCs. Greater increase of DCs was associated with remission or delayed recurrence. The prolonged overall survival observed warrants further exploration. J Clin Oncol 26: by American Society of Clinical Oncology INTRODUCTION Granulocyte macrophage colony-stimulating factor (GM-CSF) is a polypeptide that can stimulate differentiation of hematopoietic progenitor cells. 1 In patients with high-risk, resected melanoma, GM-CSF appears to improve overall survival by almost three-fold compared with matched historical controls. 2 On the basis of these results, the Eastern Cooperative Oncology Group is currently conducting a randomized trial of GM-CSF in patients with high-risk melanoma. One of the major immunostimulatory mechanisms of GM-CSF is through the stimulation of the professional antigen-presenting cells, such as dendritic cells (DCs). GM-CSF increases the mobilization, differentiation, and function of DCs. 3,4 However, the effect of GM-CSF on DC function in cancer patients has not been well studied. Both quantitative and functional defects in DCs have been described in patients with cancer 5 ; these defects may contribute to tumor escape from immune surveillance. 5 In this study, we examined the effects of GM-CSF on DC phenotype and function and its association with clinical outcome in patients with resected, high-risk melanoma. PATIENTS AND METHODS Patient Selection and Treatment Patients had documented stage IIIB/C or IV melanoma. All disease was grossly resected before enrollment. No prior treatment with GM-CSF was allowed. Patients may have had any prior interferon or chemotherapy provided treatment was completed and WBC and platelet 28 by American Society of Clinical Oncology 3235

2 Daud et al Characteristic Table 1. Baseline Patient Characteristics Study Patients (N 39) No. % Spitler et al 2 (%) Stage IIIB IIIC IVA IVB 7 18 IVC 5 13 Age, years Median 64 Range 36-8 Sex Male Female Location of primary Skin Unknown Breslow thickness, mm years. Range, years. 1.5 mm mm. counts had recovered. A maximum of 6 days from the last surgery that rendered them disease-free was allowed. All patients provided written informed consent, and the institutional review board at the H. Lee Moffitt Cancer Center (Tampa, FL) approved this protocol. GM-CSF was administered at 1 g/m 2 subcutaneously daily for 14 consecutive days followed by 14 days of no treatment for a total of 13 cycles unless there was unacceptable toxicity or recurrence. A physical exam was performed every 4 weeks, and restaging was performed after every 4 cycles. Evaluation of Cell Phenotype Peripheral blood was collected before starting and after 2, 4, 8, and 12 weeks of GM-CSF treatment. Peripheral-blood mononuclear cells (PBMCs) were isolated and cryopreserved in freezing media containing 5% RPMI , 4% FBS, and 1% DMSO and stored in liquid nitrogen. All samples from each patient were analyzed simultaneously. Briefly, PBMCs were thawed, cultured overnight in complete culture medium (RPMI / 1% FCS) and labeled with the appropriate antibodies for 4 minutes on ice. The cocktail of lineage-specific antibodies included phycoerythrin-conjugated antibodies against CD3, CD19, CD56, and CD14 (all from BD Pharmingen, San Jose, CA). In addition, we used peridinin chlorophyll protein conjugated anti human leukocyte antigen (HLA)-DR antibody, allophycocyaninconjugated anti-cd33 and CD11c antibodies, fluorescein isothiocyanate conjugated anti-cd86, CD83, and CD4 antibodies (all from BD Pharmingen). Fluorescein isothiocyanate conjugated antibodies against CCR7 and CD123 were obtained from R&D Systems (Minneapolis, MN) and Miltenyi (Auburn, CA), respectively. The phenotype of the cells was evaluated by multicolor flow cytometry using a FACSCalibur flow cytometer (BD Biosciences, Mountain View, CA). At least, cells were examined from each parameter, with subsequent analysis of the samples described previously. 6 An example of the analysis is shown in Appendix Figure A1 (online only). Evaluation of DC function Allogeneic mixed leukocyte reaction (MLR) was used to evaluate DC function. Mononuclear cells from healthy donors were seeded in U-bottom 96-well plates (1 5 cells per well). Mononuclear cells from patients were isolated as described earlier in this article, irradiated at 2 Gy, washed, and then mixed in triplicates at different ratios (from 1:1 to 1:8) with mononuclear cells isolated from healthy donors. Cell proliferation was assessed after 5-day culture by uptake of 3 [H]-thymidine. One microcurie of 3 [H]-thymidine was added to each well 18 hours before cell collection and thymidine uptake was measured in liquid scintillation counter (Beckman Coulter, Fullerton, CA) and expressed as counts per minute (cpm). Each patient s sample was tested against two different donors, and average for each cell ratio was calculated. All samples from each patient were analyzed simultaneously. Statistical Analysis The distribution of the DC parameters at each time point was explored with the Wilcoxon signed rank test. The changes from the baseline were further analyzed using the same method by the following stratified subgroups: disease status, tumor stage (stage III v stage IV), baseline scores (low v high divided by median value), and time to recurrence (early recurrence [ 9 days] v late recurrence [ 9 days]). The correlation among all immunologic markers was evaluated using Spearman correlation coefficients. We analyzed the association of clinical response to the levels of immunologic markers at each time point utilizing Fisher s exact test. Kaplan-Meier curves were used to plot survival data, and the log-rank test was used to compare survival between subgroups. All statistical tests were two sided, with statistical significance Table 2. Toxicities by Grade Grade 1 Grade 2 Grade 3 Grade 4 Any Grade Adverse Event No. % No. % No. % No. % No. % Anemia Lymphopenia Pain Fever Flu-like symptoms Arthralgia/myalgia Injection-site reaction Fatigue Abdominal cramps GERD/indigestion Insomnia Cough Sweating Abbreviation: GERD, gastroesophageal reflux disease by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

3 DC Phenotype and Clinical Outcome in Melanoma Treated With Adjuvant GM-CSF accepted at the P.5 level. Because of the exploratory nature of this study, multiple comparisons were not adjusted for in the analyses. RESULTS Of 42 patients enrolled, 39 patients were assessable for disease status and for at least one DC analysis. Staging and demographic characteristics are described in Table 1. Comparable data from the Spitler et al trial 2 is shown in Table 1. GM-CSF treatment was well tolerated in this trial, as described in (Table 2). The most common adverse events reported were injection-site reactions (88%). Despite the high-risk characteristics of the patient population under study, the median overall survival was 65 months (95% CI, 43 to 67 months; Fig 1A). This figure is significantly higher than that reported for essentially the same population by Spitler et al 2 or by the American Joint Committee on Cancer staging committee. Median recurrence-free survival was a more modest 5.6 months (95% CI, 3.7 to 11 months; Fig 1B). Patients with stage III B or IIIC disease had a clinical outcome that was similar to those with stage IV disease (median survival, 65 months for stage III, 56.8 months for stage IV; P.7455), as shown in Figure 2. Patients with recurrent disease had significantly shorter survival compared with those who remained in remission (Appendix Fig A2, online A Survival Probability B Survival Probability Time (months) No. of subjects Event Censored Median Survival (95% CL) 39 36% (14) 64% () 65.3 (47.4 to 67.) Time (months) No. of subjects Event Censored Median Survival (95% CL) 39 72% (28) 628% (11) 5.63 (3.47 to 11.7) Fig 1. Kaplan-Meier curves of (A) overall and (B) recurrence-free survival of the assessable cohort. Both parts show the product-limit survival function estimate. Survival Probability.2 Stage III Stage IV Log rank P = Time (months) No. of subjects Event Censored % (7) 33% (7) 61% (11) 67% (14) Stage III Stage IV Median Survival (95% CL) 65.3 (47.4 to 67.) 56.8 (42.8 to NA) Fig 2. Kaplan-Meier Curves of overall survival by initial disease stage of the assessable cohort. Product-limit survival function estimates. NA, not available. only). Control values for the analysis of different DC subsets were established previously using a small cohort of eight healthy volunteers. 6 In the current study, analysis of different subsets of DCs was performed before and during treatment with GM-CSF (a typical DC phenotype analysis is shown in Fig A1). Patients enrolled on this study did not show significant differences in DC parameters compared with healthy volunteers (data not shown). The time course of DC changes is shown in Figure 3. Two weeks of GM-CSF treatment resulted in a significant increase in the proportion of the total population of mature DCs, as defined by the expression of CD86, CD83, CCR7, and CD4 molecules (Fig 3, and Appendix Table A2, online only). GM-CSF did not affect the proportion of immunosuppressive, myeloid-derived suppressor cells (MDSCs) at the early or late time points (Fig 3). Two weeks after GM-CSF was stopped, the proportions of DCs, as well most of their subsets, returned to a pretreatment level. Subsequent cycles of GM-CSF treatment (weeks 8 and 12) did not change the DC presence in peripheral blood (Fig 3). It is important to point out that the number of patients evaluated before the start of treatment and after 2 and 4 weeks of GM-CSF treatment was the same (n 39), whereas analysis could be performed only in 2 patients 8 weeks after start of treatment and in only 16 patients 12 weeks after start of treatment. The total number of DCs was evaluated before and 4 weeks after start of the treatment. No differences from pretreatment levels were found (data not shown). The allogeneic MLR, a hallmark of DC activity, was used to evaluate the function of these cells. Patients enrolled on this study did not show significant differences in DC function compared with healthy volunteers (data not shown). GM-CSF did not cause significant changes in DC function in total cohort of patients (Fig 3). Although the mean values of DC parameters in patients before starting treatment were not different from healthy controls, there was a substantial patient-to-patient variability. We therefore asked whether the effect of GM-CSF on DCs depended on the pretreatment levels of these DC phenotype markers and functional activity. To address this question, patients were split into two groups. included half of the patients with pretreatment levels of DC populations above median values established for the control group. Thus, 28 by American Society of Clinical Oncology 3237

4 Daud et al DCs (Lin-HLA-DR+ cells) MDSC (Lin-HLA-DR-CD33+ cells) CD86+ cells among DCs n = 39 n = 39 n = 39 n = 2 n = 16 n = 39 n = 39 n = 39 n = 2 n = 16 n = 39 n = 39 n = 39 1 n = 2 n = *P =.17 P =.36 *P =.1 * * CD4+ cells among DCs n = 39 n = 39 n = 39 n = 2 n = CD83+ cells among DCs n = 39 n = 39 n = 39 n = 2 n = 16 n = 39 n = 39 n = 39 n = 2 n = 16 1 *P =.1 *P =.19 *P =.48 * * * CCR7+ cells among DCs Fig 3. Effect of granulocyte macrophage colony-stimulating factor treatment on (A) dendritic cells (DCs; Lin-HLA-DR cells), (B) myeloid-derived suppressor cells (MD- SCs; Lin-HLA-DR-CD33), (C) CD86 cells among DCs, (D) CD4 cells among DCs, (E) CD83 cells among DCs, (F) CCR7 cells among DCs, and (G) allogeneic mixed leukocyte reaction (MLR) in patients with high-risk melanoma. P values of differences from pretreatment level are shown only if they were less than.5. Lin, lineage; HLA, human leuckocyte antigen; cpm, counts per minute. Allogenic MLR 12, n = 36 n = 34 n = 34 8, 4, 2 4 Time on Treatment (weeks) these patients could be considered as having control level of DC parameters. included half of the patients with pretreatment levels of DC subsets below median values established for control group. GM-CSF treatment did not affect any of the tested cell populations in patients with normal pretreatment levels of DC (Fig 4). In contrast, the proportions of total DCs; CCR7, CD4, and CD83 mature DCs; and allogeneic MLR were significantly increased in patients with low pretreatment levels of these parameters (Fig 4). GM- CSF did not affect the presence of MDSCs regardless of pretreatment levels of these cells (Fig 4). Thus, it appears that GM-CSF treatment increased proportion and function of DCs only in patients who had a low level of these cells before start of the treatment. Next, we asked whether the effect of GM-CSF on DCs could be associated with the clinical outcome of melanoma patients. The first clinical evaluation after 4 weeks of treatment showed no evidence of recurrence for all patients. However, six of 38 patients experienced recurrence within 6 days after the start of GM-CSF therapy, with 11 patients experiencing recurrence within 9 days after the start of the treatment. These were considered early recurrences. A total of 27 patients had no evidence of recurrence during the first 3 months of GM-CSF therapy. These were considered late recurrences. By the time of this analysis, 16 patients eventually experienced recurrence, and 11 patients still have no evidence of recurrence. We evaluated the effect of GM-CSF on DC after 2 weeks of treatment in patients with early and late recurrence. No difference was found in the level of DC subsets between the groups before starting GM-CSF therapy. There was a trend to higher level of DC function in patients with late recurrence (58,847 cpm) compared with patients with early recurrence (49,49 cpm). However, those differences did not reach statistical significance (P.2). Two weeks of GM-CSF treatment resulted in a significant increase in the proportion of DCs and DC:MDSC ratio in patients with late recurrence, but not in those with early recurrence (Fig 5). To investigate the association between DC response to GM- CSF treatment and clinical outcome, we further analyzed patients after 2 weeks of GM-CSF treatment. All patients were split into groups on the basis of the presence of different DC subsets: patients with control levels of DC subsets (greater than median values) and patients with low levels of DC subsets (less than median values). The frequency of patients with favorable course of the disease (late recurrence) was calculated in each group and compared. Patients with control levels of total population of DCs and CD4 mature DCs had significantly higher frequency of late recurrence than patients with low levels of these parameters (P.5 and.16, respectively). It is well established that different populations of DCs are biologically connected with each other. We evaluated this connection using a correlation matrix of 159 pairs of parameters analyzed in this study (parameters are listed in Appendix Table A2). Before starting treatment, we identified statistically significant correlations in 43 pairs of immunologic parameters in patients who had no evidence of recurrence for more than 9 days after starting treatment. The number of pairs with significant correlation remained stable during the first 4 weeks of the GM-CSF treatment by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

5 DC Phenotype and Clinical Outcome in Melanoma Treated With Adjuvant GM-CSF DC MDSCC CCR7DC P =.2 P =.17 Before 2 Before 2 Before 2 CD4 DC CD83 DC CD86 DC P =.3 1 P =.5 1 P =.8 Before 2 Before 2 Before 2 Fig 4. Effect of granulocyte macrophage colony-stimulating factor treatment on dendritic cells (DCs) in patients with low and high pretreatment levels of these cells. (A) DCs, (B) myeloid-derived suppressor cells (MDSCs), (C) CCR7 DCs, (D) CD4 DCs, (E) CD83 DCs, (F) CD86 DC, and (G) mixed leukocyte reaction (MLR). Averages for patients with a low pretreatment levels of DC subsets (n 19) and patients with high levels of those parameters (n 19). Mean values are shown; P values are shown only if they are less than.1. cpm, counts per minute. CPM 8, 7, 6, 5, 4, 3, 2, 1, P =.2 MDL Before 2 Time on Treatment (weeks) (P.14). In contrast, patients with early recurrence had only 19 pairs with significant correlation before start of GM-CSF treatment. This was significantly lower compared with patients with more favorable disease outcomes (P.1). After 4 weeks of treatment, significant correlations were detected in only eight of 159 pairs, significantly lower than the number before treatment (P.43) and in those patients with no evidence of recurrence for more than 9 days (P.1). DISCUSSION In preclinical studies, GM-CSF has demonstrated a positive effect on DC differentiation. 7,8 There are, however, limited and conflicting data on GM-CSF effects on DC in patients with cancer. 7,9,1 Therefore, detailed pharmacodynamic analysis of the effect of GM-CSF treatment on DC subsets in cancer patients is needed to help clarify its potential clinical utility. Several groups have examined the role of GM-CSF therapy in high-risk melanoma patients. 11,12 Notably, Spitler et al 2 showed a substantial improvement in median overall survival with GM-CSF therapy in high-risk melanoma patients when compared with matched historical controls. However, this prospective, single-institution trial has not been confirmed by others. To address this issue, we conducted a phase II trial with a similar high-risk melanoma population and a similar GM-CSF administration regimen with pharmacodynamic measurement of DC parameters and function. Our results showed a 65-month median overall survival (95% CI, 43 to 67 months) for patients with high-risk characteristics similar to those treated by Spitler et al 2 (Table 1 and Appendix Table A1, online only). Compared with other melanoma data sets, in this high-risk population, our data show a markedly prolonged survival Although stage migration, patient selection, lead-time bias, and better medical care may potentially account for such differences, our data would argue for a potential positive effect of GM-CSF in melanoma patients. Compared with previously reported data describing decreased DC numbers in other cancers, the aggregate DC numbers were not suppressed in melanoma patients. This could be explained in part by the fact that our patients were tumor free at the time of study enrollment. GM-CSF increased the proportion of total and mature DCs, but not MDSCs. These findings are consistent with our hypothesis that GM-CSF promotes differentiation of DCs. We did not observe an effect of GM-CSF on MDSCs. Studies in mice have shown that GM-CSF induced MDSC expansion only at very high doses Apparently, the lower dose used in our study was not sufficient to 28 by American Society of Clinical Oncology 3239

6 Daud et al Cells (%) Cells (%) Before Before DC Treatment DC/MDSC Treatment P =.31 P <.1 P =.81 P =.35 2 weeks 2 weeks Early Late Early Late Fig 5. Association between the effect of granulocyte macrophage colonystimulating factor on dendritic cell (DC) and myeloid-derived suppressor cell (MDSC) subsets and clinical outcome. Late patients had no evidence of recurrence more than 9 days after start of treatment (n 27); early patients experienced recurrence less than 9 days after start of treatment (n 11). Mean and P values are shown. cause accumulation of these cells. The effect of GM-CSF on DCs was transient in nature, disappearing after 4 weeks on trial. The mechanism of this effect is not clear; it is possible that GM-CSF may downregulate its specific receptors on DCs, rendering them less sensitive to further GM-CSF treatment, or that DCs may migrate out of peripheral blood. We are currently examining this result more carefully. Despite the relatively short-lived effect of GM-CSF on DCs, it may have a direct impact on clinical outcome of melanoma patients. We failed to show a response to GM-CSF in patients who had early recurrence of the disease ( 9 days), whereas patients with late or no recurrence had significant increases in several DC markers. These data were obtained after 2 weeks of treatment. Because no patient demonstrated clinical evidence of recurrence at 4 weeks after the start of GM-CSF treatment, those differences most likely cannot be directly attributed to the disease progression at the time of peripheral-blood collection, suggesting that increased proportions of DCs may have positive impact on disease progression. Although this difference cannot be considered prognostic on the basis of this small data set, changes in DC populations should be further analyzed in large randomized trials. How a transient increase in DCs in peripheral blood could be associated with a positive clinical outcome is not entirely clear. It is possible that DCs migrate from peripheral blood and accumulate in tissues where they can pick up tumor antigen with subsequent stimulation of T cells. 26 In this case a transient increase of DCs in peripheral blood may cause a long-lasting effect. More detailed studies, including biopsies of tissues in addition to circulating DCs are needed to directly test this possibility. It appears that the effect of GM-CSF depends on basal levels of DC subsets. GM-CSF increased the proportion of DC only in patients who initially had a low level of these cells, having no effect for those patients with a normal level of DCs. These results suggest that GM-CSF treatment may be beneficial only to those patients with reduced DC numbers at baseline. The total tumor burden seems to directly affect the proportion of DCs in cancer patients. 5 A recent study evaluated the effect of GM-CSF alone or in combination with autologous tumor-derived heat shock protein peptide vaccine in 16 patients with metastatic melanoma. The authors observed an accumulation of immunosuppressive CD14 /HLA- DR negative/low cells. This accumulation was not observed with vaccine alone. 27 We did not find such an effect of GM-CSF on MDSCs in our study. There are two important differences between this study and our trial, namely that our patients were tumor free at the start of GM-CSF treatment and that the dose and schedule of GM-CSF used in our trial was different. Although in this study we have independently replicated the clinical results observed by Spitler et al, 2 ultimately, the Eastern Cooperative Oncology Group E4697 trial, which recently randomly assigned more than 8 patients with resected, high-risk melanoma in a 2 2 design to adjuvant GM-CSF versus placebo and (for HLA-A2 positive patients only) to a peptide vaccine versus placebo, will establish whether or not adjuvant GM-CSF therapy influences the natural history of these patients. Given both our findings and those of Spitler et al, it will be of particular interest to see whether the overall survival impact of GM-CSF therapy may be greater than the recurrence-free survival benefit a situation at odds with the findings in adjuvant interferon trials. 28 Our observations regarding the effect of this agent on DC phenotype and function could provide potential surrogate markers for future trials and offer avenues to further extend this benefit. Our results also suggest that perhaps only a subset of patients treated with GM-CSF obtain a benefit. AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a U are those for which no compensation was received; those relationships marked with a C were compensated. For a detailed description of the disclosure categories, or for more information about by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

7 DC Phenotype and Clinical Outcome in Melanoma Treated With Adjuvant GM-CSF ASCO s conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors. Employment or Leadership Position: None Consultant or Advisory Role: None Stock Ownership: None Honoraria: Adil I. Daud, Berlex Oncology Research Funding: Adil I. Daud, Berlex Oncology Expert Testimony: None Other Remuneration: None AUTHOR CONTRIBUTIONS Conception and design: Adil Daud, Ronald C. De Conti, Dmitry Gabrilovich Administrative support: Adil Daud, Vernon K. Sondak Provision of study materials or patients: Adil Daud, Brianna Lennox, Vernon K. Sondak, Adam I. Riker, Ronald C. De Conti Collection and assembly of data: Adil Daud, Noweeda Mirza, Brianna Lennox, Stephanie Andrews, Patricia Urbas, Ronald C. De Conti Data analysis and interpretation: Adil Daud, Noweeda Mirza, Brianna Lennox, Stephanie Andrews, Gui X. Gao, Ji-Hyun Lee, Vernon K. Sondak, Adam I. Riker, Dmitry Gabrilovich Manuscript writing: Adil Daud, Vernon K. Sondak, Adam I. Riker, Ronald C. De Conti, Dmitry Gabrilovich Final approval of manuscript: Adil Daud, Stephanie Andrews, Vernon K. Sondak, Adam I. Riker, Dmitry Gabrilovich REFERENCES 1. Cairo MS: Review of G-CSF and GM-CSF: Effects on neonatal neutrophil kinetics. Am J Pediatr Hematol Oncol 11: , Spitler LE, Grossbard ML, Ernstoff MS, et al: Adjuvant therapy of stage III and IV malignant melanoma using granulocyte-macrophage colonystimulating factor. J Clin Oncol 18: , 2 3. Everly JJ, Lonial S: Immunomodulatory effects of human recombinant granulocyte-macrophage colony-stimulating factor (rhugm-csf): Evidence of antitumour activity. Expert Opin Biol Ther 5: , 4. Shi Y, Liu CH, Roberts AI, et al: Granulocytemacrophage colony-stimulating factor (GM-CSF) and T-cell responses: What we do and don t know. Cell Res 16: , Gabrilovich D: Mechanisms and functional significance of tumour-induced dendritic-cell defects. Nat Rev Immunol 4: , Mirza N, Fishman M, Fricke I, et al: All-transretinoic acid improves differentiation of myeloid cells and immune response in cancer patients. Cancer Res 66: , Dranoff G: GM-CSF-secreting melanoma vaccines. Oncogene 22: , Leong SP, Enders-Zohr P, Zhou YM, et al: Recombinant human granulocyte macrophagecolony stimulating factor (rhgm-csf) and autologous melanoma vaccine mediate tumor regression in patients with metastatic melanoma. J Immunother 22: , Hasskamp JH, Elias EG, Zapas JL: In vivo effects of sequential granulocyte-macrophage colony stimulating factor (GM-CSF) and interleukin-2 (IL-2) on circulating dendritic cells (DC) in patients with surgically resected high risk cutaneous melanoma. J Clin Immunol 26: , Parmiani G, Castelli C, Pilla L, et al: Opposite immune functions of GM-CSF administered as vaccine adjuvant in cancer patients. Ann Oncol 18: , Morton D: Results of multicenter randomized phase III trial of Canvaxin versus BCG in patients with resected stage III or stage IV melanoma. Presented at the 59th Annual Cancer Symposium of the Society for Surgical Oncology, San Diego, CA, March 23-26, Lawson D, Kirkwood JM: Granulocytemacrophage colony-stimulating factor: Another cytokine with adjuvant therapeutic benefit in melanoma? J Clin Oncol 18: , Homsi J, Kashani-Sabet M, Messina JL, et al: Cutaneous melanoma: Prognostic factors. Cancer Control 12: , 14. Hsueh EC, Essner R, Foshag LJ, et al: Prolonged survival after complete resection of disseminated melanoma and active immunotherapy with a therapeutic cancer vaccine. J Clin Oncol 2: , Morton DL: Immune response to postsurgical adjuvant active immunotherapy with Canvaxin polyvalent cancer vaccine: Correlations with clinical course of patients with metastatic melanoma. Dev Biol (Basel) 116:29-217; discussion , Balch CM, Buzaid AC, Soong SJ, et al: Final version of the American Joint Committee on Cancer staging system for cutaneous melanoma. J Clin Oncol 19: , Andrews S, Robinson L, Cantor A, et al: Survival after surgical resection of isolated pulmonary metastases from malignant melanoma. Cancer Control 13: , Cormier JN, Xing Y, Feng L, et al: Metastatic melanoma to lymph nodes in patients with unknown primary sites. Cancer 16:212-22, Eton O, Legha SS, Moon TE, et al: Prognostic factors for survival of patients treated systemically for disseminated melanoma. J Clin Oncol 16: , Khan MA, Andrews S, Ismail-Khan R, et al: Overall and progression-free survival in metastatic melanoma: Analysis of a single-institution database. Cancer Control 13: , Fricke I, Gabrilovich DI: Dendritic cells and tumor microenvironment: A dangerous liaison. Immunol Invest 35: , Fricke I, Mirza N, Dupont J, et al: Vascular endothelial growth factor-trap overcomes defects in dendritic cell differentiation but does not improve antigen-specific immune responses. Clin Cancer Res 13: , Almand B, Clark JI, Nikitina E, et al: Increased production of immature myeloid cells in cancer patients: A mechanism of immunosuppression in cancer. J Immunol 166: , Almand B, Resser JR, Lindman B, et al: Clinical significance of defective dendritic cell differentiation in cancer. Clin Cancer Res 6: , 2. Serafini P, Carbley R, Noonan KA, et al: Highdose granulocyte-macrophage colony-stimulating factor-producing vaccines impair the immune response through the recruitment of myeloid suppressor cells. Cancer Res 64: , Pinedo HM, Buter J, Luykx-de Bakker SA, et al: Extended neoadjuvant chemotherapy in locally advanced breast cancer combined with GM-CSF: Effect on tumour-draining lymph node dendritic cells. Eur J Cancer 39: , Filipazzi P, Valenti R, Huber V, et al: Identification of a new subset of myeloid suppressor cells in peripheral blood of melanoma patients with modulation by a granulocyte macrophage colonystimulation factor based antitumor vaccine. J Clin Oncol :46-53, Wheatley K, Ives N, Hancock B, et al: Does adjuvant interferon-alpha for high-risk melanoma provide a worthwhile benefit? A meta-analysis of the randomised trials. Cancer Treat Rev 29:241-2, 23 Appendix The Appendix is included in the full-text version of this article, available online at It is not included in the PDF version (via Adobe Reader ) by American Society of Clinical Oncology 3241