Induction of potent antitumor T cell immunity is considered to be crucial for the therapeutic efficacy of immunotherapy. Here, we report on cytokine expression profiles from the PBMCs of 13 patients treated with neoadjuvant ipilimumab at baseline and at week six. We examined responses to two melanoma lineage (gp100, MART-1) and a cancer testis (NY-ESO-1) antigen, all commonly expressed by the majority of melanoma tumors [11–13] and all known to be spontaneously immunogenic. We chose to assess expression levels of cytokines and chemokines with widely varied functions, and we show that at baseline (prior to treatment with ipilimumab), cytokine profiles are differentially expressed in response to antigen stimulation. We find that PBMC supernatants express 17 cytokines at higher levels when stimulated with NY-ESO-1 versus gp100 or MART-1, in line with the known spontaneous immunogenicity of NY-ESO-1 in eliciting CD4 and CD8 T cell responses  in addition to antibody responses. On stimulation with NY-ESO-1, expression of IL-1β, MIP1-β, IL1-RA, VEGF, IL-13, IL-17, MIP1-α, GM-CSF, MCP-1, IL-5, IL2-R, IL-4, IL-10, IFN-γ, TNF-α, IL-8 and IL-2 was significantly increased. Interestingly, these cytokines fall into varying functional groups, including pro-inflammatory (IL-1β, MIP-1β, IL-17, IFN- γ, TNF-α and IL-8) and anti-inflammatory (IL-1RA, IL-10). We also find TH1 (IFN-γ, IL-2,) and TH2 (IL-4, IL-13) cytokines. These findings suggest that PBMCs of patients with advanced melanoma are polyfunctional, with expression of cytokines across varied pathways.
NY-ESO-1 is a member of the cancer testis antigen (CTA) family, a group of tumor-associated antigens that are expressed on multiple different tumor types, with expression in normal tissues restricted to placenta and testis . NY-ESO-1 mRNA is expressed in 17–42% of melanomas [15–17], and it is thought to be the most immunogenic CTA . Up to 50% of patients with advanced melanoma whose tumors express NY-ESO-1 are seropositive for anti-NY-ESO-1 antibodies , and even in the absence of detectable antigen expression, NY-ESO-1 may induce NY-ESO-1 antibodies in 10% of patients . NY-ESO-1 has also been shown to stimulate spontaneous and synchronized humoral and cell-mediated immune responses. In one study, more than 90% of patients with circulating anti-NY-ESO-1 antibodies also had an NY-ESO-1-specific CD8+ T-cell response, whereas this effector T cell response was absent in patients without detectable anti-NY-ESO-1 antibody . Therefore, many clinical trials of vaccines employing NY-ESO-1 are ongoing, with several already reported [21–24]. Thus, detection of multiple cytokines at elevated expression levels in response to NY-ESO-1 antigen stimulation suggests that patients with advanced disease (in our cohort, 92% had stage IIIC melanoma) have an NY-ESO-1 specific immune response, though this is likely functionally dampened in the tumor microenvironment by immune checkpoints such as PD-1/PD-L1  and CTLA-4 .
When PBMC cytokine expression levels were tested at week 6, after 2 cycles of ipilimumab, we find that 15 cytokines have increased expression when stimulated with NY-ESO-1 versus gp-100 or MART-1. Specifically, expression of IL-1β, VEGF, G-CSF, HGF, IL-13, IL-17, GM-CSF, MCP-1, IL-5, IL-7, IL-4, IL-10, IFN-γ, IL-8 and IL-2 was significantly increased. Increased expression in response to NY-ESO-1 versus gp-100 or MART-1 at week 6, as we saw at baseline, again suggests increased activated cell specificity for the NY-ESO-1 antigen. In addition, we again note cytokine and chemokine expression across multiple effector functions. Five proteins were elevated at baseline in response to NY-ESO-1 antigen, but not at week 6: IL-2R, IL-1RA, MIP-1α, MIP-1β, and TNF-α. At week 6, we noted an increase in G-CSF, IL-7 and HGF, which had not been noted at baseline. Though difficult to draw conclusions from this small cohort, an increase in growth factors may indicate increased CD4 and CD8 T cell activation, though G-CSF is a key modulator of MDSC [27, 28], and G-CSF expression has correlated with increased MDSC and worse overall survival in murine models .
We had previously reported that in 24 patients treated with neoadjuvant ipilimumab, for whom tumor tissue was available, CD8 + tumor infiltrating lymphocytes were increased after treatment . Furthermore, a significant increase in circulating CD4+, CD8+ and T regulatory cells (Treg; CD4+ CD25+ Foxp3+) was noted at week 6, compared to baseline, and this correlated with improved PFS . Numerous subsets of Treg have diverse functions in relation to the anti-tumor responses , and our findings of higher pro- and anti-inflammatory cytokines in response to antigen stimulation with NY-ESO-1 after 6 weeks of treatment suggests that ipilimumab does not only function by a pro-inflammatory mechanism. Moreover, the pro-inflammatory response appears to be the most dominant, based on increases in several pro-inflammatory cytokines, including IL-1β, IL-6, MIP-1α, IL-17, IL-8, IL-2, TNF-α and IFN-γ, and this may be the mechanism leading to the clinical benefit seen in a subset of patients. In an exploratory analysis, we assessed the association between Treg and MDSC with our cytokine data, and no significant associations were found after adjusting for multiple testing, likely due to the small sample size that we are reporting on. We clustered patients at baseline and at week 6 into two groups based on their cytokine expression profile. At baseline, seven patients were in cluster 1 and had higher cytokine expression when stimulated with NY-ESO-1, while six patients were in cluster 2. Poor RFS was defined as patients who had died or relapsed within 1 year, whereas patients with good RFS had not died or relapsed within 1 year. Notably, the cluster into which patients were placed based on their cytokine expression profile was found to correlate with RFS at 12 months. While this did not reach statistical significance, there was a sharp trend towards improved RFS in cluster 1, where 5 of 7 patients had good RFS, whereas in cluster 2, 5 of 6 patients had poor RFS. These differences in RFS based on cytokine level clustering were noted both at baseline and at week 6. This suggests that having higher cytokine expression in respond to NY-ESO-1 compared to gp-100 and MART-1 may have some association with better RFS.
Our study has several limitations. The small cohort size limits the conclusions that can be drawn, and may be why we do not see a statistically significant difference when we compare RFS in cluster 1 versus cluster 2. Furthermore, the cytokine expression levels were determined in supernatants of PBMC, clearly limiting our ability to differentiate between expression profiles of specific cell subsets. Our earlier study testing intracellular cytokine expression confirmed that both CD4+ and CD8+ T cells produced IFN-γ, but many of these circulating proteins have other sources. To our knowledge, this is the first study to report on cytokine expression profiles from PBMCs of patients treated with ipilimumab. Furthermore, we compare levels at baseline and after 6 weeks of treatment. We show that at baseline, 17 cytokines are expressed at higher levels in response to NY-ESO-1 than in response to gp-100 or MART-1. We show that this difference is maintained after 6 weeks of treatment of ipilimumab. Importantly, we find that when patients are clustered into groups based on their cytokine expression profile, patients in cluster 1 had superior RFS compared to patients in cluster 2. These findings warrant validation across a larger sample size, and it would also be important to determine if these differences are also seen in patients with metastatic melanoma who are treated with ipilimumab. If our findings persist in larger cohorts, this may help determine which patients are likely to derive RFS benefit from ipilimumab.