Low Expression of IL-15 and NKT in Tumor Microenvironment Predicts Poor Outcome of MYCN-Non-Amplified Neuroblastoma

Immune tumor microenvironment (TME) in neuroblastoma (NBL) contributes to tumor behavior and treatment response. T cells and natural killer (NK) cells have been shown to play important roles in the neuroblastoma TME. However, few reports address the clinical relevance of natural killer T cells (NKTs) and interleukin-15 (IL-15), one of the crucial cytokines controlling the activation and expansion of NK/NKT cells, in NBL. In this study, we examined NKT immunoscores and IL-15 expression in both MYCN-amplified and MYCN-non-amplified NBL to correlate with clinical outcomes such as event-free survival (EFS) and overall survival (OS). From Gene Expression Omnibus (GEO) datasets GSE45480 (n = 643) and GSE49711 (n = 493), we found that NKT immunoscore and IL-15 expression were both significantly lower in MYCN-amplified NBL, and similar results were observed using our clinical NBL samples (n = 53). Moreover, NBL patients (GEO dataset GSE49711 and our clinical samples) with both lower NKT immunoscore and IL-15 expression exhibited decreased EFS and OS regardless of MYCN gene amplification status. Multivariate analysis further showed that the combination of low NKT immunoscore and low IL-15 expression level was an independent prognostic factor for poor EFS and OS in our NBL patients. These findings provide the rationale for the development of strategy to incorporate IL-15 and NKT cell therapy into the treatment regimen for neuroblastoma.


Introduction
Neuroblastoma (NBL) is a neuroectodermal solid tumor occurring in early childhood with 20~50 cases per million individuals [1]. Disease course and treatment outcome for NBL vary significantly depending on clinical features and molecular risk factors. The International Neuroblastoma Risk Group (INRG) provides a consensus pretreatment classification scheme for risk-stratified therapy according to the stage, age, histology, MYCN (v-myc myelocytomatosis viral related oncogene, neuroblastoma derived (avian)) status, and other

Neuroblastoma (NBL) Samples
Archival samples of cDNA derived from a previous collection of primary tumors from 53 NBL patients were used for the study [24]. These patients were enrolled in studies between 1986 and 1995, and we obtained the samples through the Children's Oncology Group, the Pediatric Oncology Group, and the Cooperative Human Tissue Network. The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Chang Gung Memorial Hospital, Taiwan (protocol code 201701442A3; 11 December 2017). Clinicopathologic information of the participants were summarized in Table 1 and are representative of the NBL population in general. The 5-year event-free survival (EFS) of these patients was 58.8% with a median follow-up time of 3.1 years, while the 5-year overall survival (OS) of these patients was 62.0% with a median follow-up time of 3.47 years.

Real-Time Reverse-Transcription Polymerase Chain Reaction
Total RNA was isolated from NBL specimens using TRIzol (Invitrogen, Carlsbad, CA, USA), and RNA (1 µg) was converted to cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster, CA, USA) in accordance with the manufacturer's protocol. The human IL-15 qPCR primer pair was purchased from OriGene (CAT#: HP233499, Rockville, MD, USA), whereas the Vα24 (5 -CTGGAGGGAAAGAACTGC-3 , 5 -TGTCAGGGAAACAGGACC-3 ) and the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (5 -CCACTCCTCCACCTTT-3 , 5 -ACCACCCTGTTGCTGT-3 ) primer pairs were designed as described by Grose et al. [25]. Expression levels of Vα24 and IL-15 were determined via quantitative RT-PCR with the SYBR Green Real-Time PCR Master Mix using the Applied Biosystems 7500 Fast Real-Time PCR System, with GAPDH serving as the endogenous control. In accordance with the manufacturer's protocol, 10 ng of cDNA template were added for each 10 µL qRT-PCR reaction. The fluorescent signals were analyzed by Applied Biosystems 7500 Software v2.0.6.

Statistical Analysis
Relative levels of Vα24 (a member of the subfamily of the T cell receptor which conserved expressed on NKTs) and IL-15 mRNAs were expressed as −∆Ct after subtracting the Ct of the reference gene, GAPDH, from that of the genes of interest. Their prognostic values were evaluated via receiver operating characteristic (ROC) area under the curve (AUC) analysis, and the Youden index (sensitivity + specificity − 1) was calculated to determine optimal cut-off values for high versus low gene expression levels. Survival curves were plotted using the Kaplan-Meier method, with the log-rank test applied for comparison. The Cox proportional hazards regression model was employed to identify independent prognostic factors. Statistical computations were performed with Prism 7.0 (GraphPad Software, La Jolla, CA, USA) and SPSS V22.0 (IBM, Armonk, NY, USA) software.
Master Mix using the Applied Biosystems 7500 Fast Real-Time PCR System, with GAPDH serving as the endogenous control. In accordance with the manufacturer's protocol, 10 ng of cDNA template were added for each 10 μL qRT-PCR reaction. The fluorescent signals were analyzed by Applied Biosystems 7500 Software v2.0.6.

Statistical Analysis
Relative levels of Vα24 (a member of the subfamily of the T cell receptor which conserved expressed on NKTs) and IL-15 mRNAs were expressed as −ΔCt after subtracting the Ct of the reference gene, GAPDH, from that of the genes of interest. Their prognostic values were evaluated via receiver operating characteristic (ROC) area under the curve (AUC) analysis, and the Youden index (sensitivity + specificity − 1) was calculated to determine optimal cut-off values for high versus low gene expression levels. Survival curves were plotted using the Kaplan-Meier method, with the log-rank test applied for comparison. The Cox proportional hazards regression model was employed to identify independent prognostic factors. Statistical computations were performed with Prism 7.0 (GraphPad Software, La Jolla, CA, USA) and SPSS V22.0 (IBM, Armonk, NY, USA) software.

Lower NKT Immunoscore or IL-15 Expression Level Is Associated with Decreased Event-Free and Overall Survival in NBL Dataset
We next examined the prognostic values of the NKT immunoscore and IL-15 expression level using the GSE49711 dataset. Kaplan-Meier survival analysis showed that NBL patients with low NKT immunoscores exhibited significantly poorer EFS (41.2 versus 71.7%; p < 0.0001) and OS (63.2 versus 81.6%; p < 0.0001) than those with high NKT immunoscores over the course of follow-up (Figure 2A,D). Similarly, a low level of IL-15 expression was significantly associated with decreased EFS (51.9 versus 66.8%, p = 0.0006) and OS (63.2 versus 81.6%, p < 0.0001) compared to high level of IL-15 expression ( Figure 2B,E). To further evaluate the combined effect of the NKT immunoscore and IL-15 expression level, patients were stratified into three groups: both-high, both-low, and others, for Kaplan-Meier survival analysis. As shown in Figure 2C, patients in the both-low cohort had significantly inferior EFS than those in the both-high group (p < 0.0001) at 1 year (62.5 vs. 90.2%), 3 years (37.4 vs. 77.9%), and 5 years (33.5 vs. 75.9%). In addition, patients with both low NKT immunoscore and IL-15 expression had significantly poorer OS than those in the both-high group (p < 0.0001) at 1 year (79.9 vs. 99.5%), 3 years (47.2 vs. 94.1%), and 5 years (47.2 vs. 89.4%) ( Figure 2F).

Lower NKT Immunoscore or IL-15 Expression Level Is Associated with Decreased Event-Free and Overall Survival in NBL Dataset
We next examined the prognostic values of the NKT immunoscore and IL-15 expression level using the GSE49711 dataset. Kaplan-Meier survival analysis showed that NBL patients with low NKT immunoscores exhibited significantly poorer EFS (41.2 versus 71.7%; p < 0.0001) and OS (63.2 versus 81.6%; p < 0.0001) than those with high NKT immunoscores over the course of follow-up (Figure 2A and 2D). Similarly, a low level of IL-15 expression was significantly associated with decreased EFS (51.9 versus 66.8%, p = 0.0006) and OS (63.2 versus 81.6%, p < 0.0001) compared to high level of IL-15 expression ( Figure 2B and 2E). To further evaluate the combined effect of the NKT immunoscore and IL-15 expression level, patients were stratified into three groups: both-high, both-low, and others, for Kaplan-Meier survival analysis. As shown in Figure 2C, patients in the bothlow cohort had significantly inferior EFS than those in the both-high group (p < 0.0001) at 1 year (62.5 vs. 90.2%), 3 years (37.4 vs. 77.9%), and 5 years (33.5 vs. 75.9%). In addition patients with both low NKT immunoscore and IL-15 expression had significantly poorer OS than those in the both-high group (P < 0.0001) at 1 year (79.9 vs. 99.5%), 3 years (47.

Lower Expression of Vα24 or IL-15 Is Associated with Decreased Overall Survival in NBL Patients
To

Lower Expression of Vα24 or IL-15 Is Associated with Decreased Overall Survival in NBL Patients
To validate the results of our in silico analysis of transcriptomic data, we performed correlative analysis between expression levels of  Furthermore, univariate Cox regression analysis of various parameters associated with EFS and OS in NBL revealed that in addition to well-known prognostic factors such as INSS stage, risk category, and age at diagnosis, low expression of Vα24 (p = 0.012), low expression of IL-15 (p = 0.011), and concurrently low expressions of Vα24 and IL-15 (p = 0.003) were all found to be significant predictors of poor EFS (Table 2). In addition, low expression of Vα24 (p = 0.03) and concurrently low expressions of Vα24 and IL-15 (p = 0.03) were also significant predictive factors of poor OS (Table 3). However, upon multivariate Cox regression analysis, the patient's risk category (p = 0.003 and 0.015, respectively) and low expression of IL-15 (p = 0.012 and 0.027, respectively) in combination with low expression of Vα24 remained the only independent predictors of EFS and OS (p = 0.009 and 0.045, respectively) (Tables 2 and 3).  Furthermore, univariate Cox regression analysis of various parameters associated with EFS and OS in NBL revealed that in addition to well-known prognostic factors such as INSS stage, risk category, and age at diagnosis, low expression of Vα24 (p = 0.012), low expression of IL-15 (p = 0.011), and concurrently low expressions of Vα24 and IL-15 (p = 0.003) were all found to be significant predictors of poor EFS (Table 2). In addition, low expression of Vα24 (p = 0.03) and concurrently low expressions of Vα24 and IL-15 (p = 0.03) were also significant predictive factors of poor OS (Table 3). However, upon multivariate Cox regression analysis, the patient's risk category (p = 0.003 and 0.015, respectively) and low expression of IL-15 (p = 0.012 and 0.027, respectively) in combination with low expression of Vα24 remained the only independent predictors of EFS and OS (p = 0.009 and 0.045, respectively) (Tables 2 and 3).

Lower Expression of Vα24 or IL-15 Is Associated with Inferior Overall Survival in MYCN-Non-Amplified NBL Patients
We next evaluated the prognostic effects of Vα24 and IL-15 expression levels in MYCNnon-amplified NBL patients. Kaplan-Meier survival analysis indicated that MYCN-nonamplified NBL patients with low Vα24 expression had significantly worse EFS (56.4 versus 100%; p = 0.007) and OS (66.9 versus 100%; p = 0.03) than those with high Vα24 expression ( Figure 5A,D). Low IL-15 expression was associated with poorer EFS with borderline significance but was not significantly associated with OS (p = 0.05 and 0.41, respectively) compared to high IL-15 expression ( Figure 5B,E). We also evaluated the combined prognostic value of Vα24 and IL-15 expression levels by stratifying MYCN-non-amplified NBL patients into three groups as described above for Kaplan-Meier survival analysis. As shown in Figure 5C,F, patients with concurrently low Vα24 and IL-15 expressions tend to have inferior EFS (50.9 versus 100%; p = 0.003) and OS (66.6 versus 100%; p = 0.003) compared to those in the both-high group.

Lower NKT Immunoscore Is Associated with Inferior Overall Survival in MYCN-Amplified NBL Dataset
Kaplan-Meier survival analysis of MYCN-amplified NBL patients in GSE49711 showed that those with low NKT immunoscores had significantly worse OS than their high NKT immunoscore counterparts (30.0 versus 57.0%; p = 0.03), although there was only a trend for shorter EFS (p = 0.23) (Supplementary Figure S3D). The low IL-15 expression group showed a similar trend for adverse EFS and OS (p = 0.15 and 0.20, respectively) (Supplementary Figure S3B,E). Patients who were low for both NKT immunoscores and IL-15 expression level had significantly worse OS than those in the both-high group (p = 0.03) at 1 year (70.9 vs. 84.5%), 3 years (26.9 vs. 73.5%), and 5 years (26.9 vs. 71.3%) (Supplementary Figure S3F). As shown in Supplementary Figure S3C, there was a similar trend for the worse EFS in the both-low group (p = 0.13).

Discussion
In this study, we demonstrated that lower expression of both NKTs (Vα24) and IL-15 is significantly associated with poor EFS and OS in NBL patients as well as those in the MYCNnon-amplified NBL subgroup. Moreover, in MYCN-amplified NBL, lower expression of both NKTs and IL-15 significantly correlates with worse OS. These findings provide novel insights for the risk stratification of NBL and suggest that strategies to enhance either NKTs or IL-15 may improve clinical outcomes for high-risk NBL. In addition, association of MYCN amplification with advanced tumor stage and disease progression in NBL has long been known [7,29]. Studies of the MYCN transgenic mouse model have shown that MYCN is sufficient to drive murine NBL tumorigenesis [30]. MYCN is a multifaceted transcription regulator in NBL that can activate genes sustaining growth and repress genes driving differentiation [31]. Furthermore, MYCN in NBL could influence immune surveillance by modulating the infiltration of T cells [9,32], NK cells [10], and NKTs [15], thus supporting the role of MYCN as an immunosuppressive oncogene in high-risk NBL patients. Although MYCN is distinct from MYC (c-myc) [33,34], they are known to share prominent but incomplete redundancy [31]. Recent studies indicate that MYC not only acts as an oncogene but also directly regulates immune responses by various mechanisms that facilitate immune evasion and immunosuppression [35][36][37]. However, whether MYCN in NBL exerts a similar effect on immune regulation as MYC does in other cancers remains to be investigated.
There is currently no report on the clinical relevance of IL-15 in NBL. We are the first to show that lower IL-15 expression is significantly associated with worse EFS and OS in NBL patients. This is in line with previous reports that IL-15 is an independent prognostic marker for patients with prostate cancer [38], breast cancer [39], and lung adenocarcinoma [40]. We are also the first to report the association between IL-15 and MYCN, although the molecular mechanism remains obscure. IL-15 and IL-2 are known to have similar biologic properties due to shared receptor signaling components (IL-2 receptor β and γc), but they differ in their distinctive high-binding-affinity α-chain receptors [41]. While the combination of IL-2 and anti-GD2 has been shown to enhance NK cell-mediated ADCC against NBL [42], IL-2-related toxicities remain a clinical challenge [3]. Compared to IL-2, IL-15 does not provoke activation-induced cell death (AICD), induces less regulatory T cell expansion, and is not associated with capillary leak syndrome [43]. Given these advantages, IL-15 has been used as a monotherapy or add-on strategy for cancer treatment [44,45]. The substitution of IL-15 for IL-2 has led to significant preclinical antitumor activity that exceeded the effect of IL-2 in an orthotopic patient-derived xenograft (PDX) model of NBL [20], and co-expression of IL-15 with GD2-specific chimeric antigen receptor (CAR) has enhanced the therapeutic efficacy of CAR-NKTs or CAR-Ts in xenograft mouse models of NBL [18,46]. Taken together, the demonstrated benefit of combining IL-15 with immunotherapy is in line with our findings that high IL-15 expression significantly correlates with better clinical outcomes in NBL patients.
In a study of 98 NBL patients, NKT level as assessed by Vα24 expression is higher in MYCN-non-amplified than MYCN-amplified cases, with higher expression associated with better clinical outcomes [16]. However, the prognostic significance of IL-15 expression in NBL has not been explored. Low levels of NKTs in the TME are associated with worse outcome in this study, which is in agreement with previous reports [16,47] and in line with the cold tumor hypothesis [5] that lack of infiltrating NKTs in TME predicts poor outcome in NBL. However, this is the first study to further analyze NKT expression in the NBL TME according to the tumor's MYCN status, demonstrating that fewer NKTs were associated with worse EFS and OS in MYCN-non-amplified NBL and with worse OS in MYCN-amplified NBL. Although little is known regarding the interaction between NKTs and MYCN in NBL, Metelitsa et al. has shown that MYCN regulates the localization of NKTs to the site of disease in NBL and that NKT infiltration is preferentially found in MYCN-non-amplified NBL in a CCL2-dependent manner [16]. NKTs have the capacity to mount strong anti-tumor responses and have thus become a major focus in the development of effective cancer immunotherapy. It has been reported that IL-15 protects antigen-activated NKTs from suppression by tumor-conditioned hypoxic tumor-associated macrophages [48]. Moreover, IL-15 can expand and activate NK, NKTs, and memory CD8 + T cells in TME, leading to tumor destruction [49]. The positive impacts of IL-15 on NK and T cells, in addition to NKT cells, may account for the greater power of combined NKT and IL-15 in predicting outcome of MYCN-non-amplified NBL. Novel strategies to enhance NKT activity by the incorporation of IL-15 in GD2-based immunotherapy, such as anti-GD2 antibody or GD2-specific CAR-Ts or CAR-NKs, are being actively pursued in a phase 1 trial of autologous NKTs engineered to co-express a GD2-specific CAR with IL-15 in three children with relapsed or resistant NBL (NCT03294954). Currently, no dose-limiting toxicities have been observed, and objective response with regression of metastatic bone lesions has been noted in one patient [50]. However, in view of the extremely high cost of CAR-Ts, CAR-NKs, and CAR-NKTs cell therapies, the combination of NKT activators with IL-15 might offer a less expensive alternate approach. We previously reported αgalactosyl ceramide (α-GC) and its analogs as strong activators of the anti-cancer effects of NKTs [51,52], and combining α-GC or its analog with IL-15 may be a promising add-on approach for anti-GD2 immunotherapy in high-risk NBL.

Conclusions
This study shows that in NBL patients with MYCN-non-amplified tumors, lower expression of NKTs (Vα24) or IL-15 alone or in combination are independent prognostic factors for poor EFS, and lower expression of IL-15 alone or in combination with low NKTs (Vα24) are independent predictors for worse OS. These findings support the incorporation of IL-15 into various immunotherapeutic strategies for the treatment of neuroblastoma. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The GEO data presented in this study are openly available in GSE49711 and GSE45480. The archival sample data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethical.