Immune Gene and Cell Enrichment Is Associated with a Good Prognosis in Ependymoma

Andrew M. Donson, Diane K. Birks, Valerie N. Barton, Qi Wei, Bette K. Kleinschmidt-DeMasters, Michael H. Handler, Allen E. Waziri, Michael Wang and Nicholas K. Foreman
J Immunol December 1, 2009, 183 (11) 7428-7440; DOI: https://doi.org/10.4049/jimmunol.0902811

Abstract

Approximately 50% of children with ependymoma will suffer from tumor recurrences that will ultimately lead to death. Development of more effective therapies and patient stratification in ependymoma mandates better prognostication. In this study, tumor gene expression microarray profiles from pediatric ependymoma clinical samples were subject to ontological analyses to identify outcome-associated biological factors. Histology was subsequently used to evaluate the results of ontological analyses. Ontology analyses revealed that genes associated with nonrecurrent ependymoma were predominantly immune function-related. Additionally, increased expression of immune-related genes was correlated with longer time to progression in recurrent ependymoma. Of those genes associated with both the nonrecurrent phenotype and that positively correlated with time to progression, 95% were associated with immune function. Histological analysis of a subset of these immune function genes revealed that their expression was restricted to a subpopulation of tumor-infiltrating cells. Analysis of tumor-infiltrating immune cells showed increased infiltration of CD4+ T cells in the nonrecurrent ependymomas. No genomic sequences for SV40, BK, JC, or Merkel polyomaviruses were found in nonrecurrent ependymoma. This study reveals that up-regulation of immune function genes is the predominant ontology associated with a good prognosis in ependymoma and it provides preliminary evidence of a beneficial host proinflammatory and/or Ag-specific immune response.

Ependymoma (EPN),3 the third most common brain tumor of children, is treated by surgical resection and radiation therapy (1, 2). Complete resection, often requiring “second-look” surgery, is critical for a favorable outcome (3, 4). Radiation therapy is also standard, and omission of this results in a higher number of tumor recurrences (4, 5). Chemotherapy has so far shown little or no benefit. Unfortunately, >50% of children treated with the standard regimen will suffer from tumor recurrence, which will ultimately result in death (6). This high failure rate represents one of the most significant problems in pediatric neuro-oncology. Despite unfavorable outcome in more than half of pediatric EPN patients, little progress has been made in the past 20 years either in treatment or identification of robust prognostic factors. The ability to identify up-front those EPN patients whose tumor will recur would allow clinicians to try more aggressive treatment regimens, better stratify patients on various treatment protocols, and spare those children whose tumors are unlikely to recur from overly aggressive treatments. Identification of prognostic markers for EPN may have the added benefit of providing insight into the biological mechanisms of tumorigenesis, which could be exploited for the development of more effective therapies.

To date, study of candidate prognostic markers for pediatric EPN have largely been confined to histological grading according to World Health Organization (WHO) tumor classification criteria (7, 8, 9, 10, 11), as well as to molecular markers such as Ki-67 (12, 13), survivin (14, 15), human telomerase reverse transcriptase (16), and nucleolin (4). More recently, global molecular analyses such as array comparative genomic hybridization (17, 18) and gene expression profiling (17, 19, 20, 21) have been employed to discover prognostic chromosomal aberrations or gene expression signatures. These global studies have produced an even wider range of candidate prognostic markers, although none to date have identified a biological mechanism of recurrence. Despite these numerous studies, there remains no predictor of tumor recurrence in EPN that is robustly reproducible from study to study. The driving hypothesis for this study is that gene expression patterns differ between good and bad prognosis EPN, the details of which will allow for better prognostication and provide insights into the biology of recurrence. To achieve this, tumor gene expression profiling combined with gene ontology analysis was used as an unbiased approach to identify sets of functionally related genes that were associated with clinical outcome in EPN clinical samples. Using this approach, it was found that an up-regulation of immune function-related genes was the predominant ontology associated with a complete response to therapy.

Materials and Methods

Patient cohort

Surgical tumor samples were obtained from 19 patients who presented between 1997 and 2007 for treatment at The Children’s Hospital (Denver, CO) who were diagnosed with EPN according to WHO guidelines (22). All patients included in the study were treated uniformly, undergoing complete tumor resection followed by radiation therapy. Samples used in this study were obtained at the time of initial resection and before radiation therapy. Two tumor samples were collected for each patient: one sample was snap-frozen in liquid nitrogen, and one was formalin-fixed paraffin-embedded (FFPE) for routine light microscopy. Outcome data were available for all patients in this study, which was conducted in compliance with Institutional Review Board regulations (COMIRB 95-500 and 05-0149). Patient details are described in Table I⇓.

View this table:
Table I.

Patient cohort demographic and tumor detailsa

Gene expression microarray analysis

Five micrograms of RNA that had been extracted from tumor was amplified, biotin-labeled (Enzo Biochem), and hybridized to Affymetrix HG-U133 Plus 2 microarray chips. Analysis of gene expression microarray data was performed using the Bioconductor R programming language (www.bioconductor.org). Microarray data were background corrected and normalized using the guanine cytosine robust multiarray average (gcRMA) algorithm (23), resulting in log2 expression values. The Affymetrix HG-U133 Plus 2 microarray contains 54,675 probe sets, including multiple probe sets for the same gene. To reduce errors associated with multiple testing, a filtered list containing a single probe set for each gene that possessed the highest gcRMA expression level across all samples used was created (18,624 genes). The microarray data discussed in this publication are Minimum Information About a Microarray Experiment (MIAME) compliant and have been deposited in National Center for Biotechnology Information’s Gene Expression Omnibus (24) and are accessible through Gene Expression Omnibus series accession no. GSE16155 (www.ncbi.nlm.nih.gov/geo/query/).

Gene ontology analyses

Two computer-based ontology analysis tools were used in this study: GSEA (Gene Set Enrichment Analysis: www.broad.mit.edu/gsea) (25) and DAVID (Database for Annotation, Visualization, and Integrated Discovery: http://david.abcc.ncifcrf.gov) (26). Both analyses were used to assess gene lists for enrichment of genes annotated with specific Gene Ontology Project terms (GOTERM; www.geneontology.org) (27). Enrichment is defined as more genes than would be expected by chance that are associated with a specific phenotype or variable.

Briefly, GSEA takes gene expression profiles that have been assigned a specific phenotype (e.g., nonrecurrent or recurrent) or a continuous variable (e.g., time to progression) and creates a ranked list of genes based on the strength of the association with the phenotype or variable being interrogated. The output is an enrichment score with associated false discovery rate (FDR) adjusted q values and Student’s t test p values for each Gene Ontology term. A Benjamini FDR cutoff of 0.25 was used as recommended by GSEA.

DAVID is a web-based resource that provides Gene Ontology term enrichment scores for lists of genes that, unlike GSEA, have already been identified by the user as significantly associated with a particular phenotype or variable.

Immunohistochemistry (IHC)

IHC was performed on 5-μm FFPE tumor tissue sections. Slides were deparaffinized and then subjected to optimal Ag retrieval protocols. Subsequent steps were performed using the EnVision-HRP kit (Dako) on a Dako autostainer according to standard protocol. Incubation with primary Ab was performed for 2 h. The following dilutions of primary Ab were used, and applied to the sections for 1 h: 1/250 allograft inhibitory factor-1 (AIF-1) (01-1974) from Waco Pure Chemicals; 1/50 HLA-DR (LN3) and 1/40 CD4 (IF6) from Novocastra; 1/100 CD8 (C8/144B), 1/200 CD20 (L26), 1/50 CD45 (2B11 + PD7/26), and 1/100 CD68 (PG-M1) from Dako. Each of these Abs stained a discrete subpopulation of cells that were distributed throughout the parenchyma of the tumor. Slides were analyzed with the Olympus BX40 microscope, ×40 objective lens. Images were captured using an Optronics MicroFire 1600 × 1200 camera and PictureFrame 2.3 imaging software (Optronics). Infiltrating cell abundancies were measured as the mean number of positive staining cells per five fields of view and differential expression between groups was determined using a Student’s t test with a p value cutoff of 0.05.

Quantitative PCR for viral sequences

Quantitative PCR was performed for SV40, BK, JC, and Merkel polyomaviruses (PyV). DNA was extracted from surgical specimens using the Gentrapure DNA extraction kit (Qiagen). All PCR analyses was performed using the ABI 7500 sequence dector (Applied Biosystems). TaqMan primers and probes were synthesized by an Applied Biosystems facility. Probes were dual-labeled at the 5′ end with FAM and the 3′ end with TAMRA. A sequence homology search was performed to ensure the specificity of each primer/probe set. TaqMan PCR amplification data were analyzed with software provided by the manufacturer. All samples were tested in duplicate. Results were expressed as cycle threshold (Ct), which was proportional to the starting copy numbers and was defined as the PCR cycle at which the fluorescence signal of the PCR kinetics exceeds the threshold value of the respective analysis.

Results

Patient demographics, tumor grade, or location do not influence risk of recurrence

In this study the median follow-up for nonrecurrent EPN patients was 5 years 3 mo. The median time to progression (TTP) for recurrent EPN patients was 2 years. No statistically significant difference was seen between recurrent and nonrecurrent EPN patients with respect to tumor WHO grade, location, age at diagnosis, or gender. In those patients with recurrent EPN, anaplastic EPN (WHO grade III) had a significantly shorter TTP than did classic EPN (WHO grade II) (9 mo vs 32 mo, respectively; p = 0.012). A shorter TTP was also seen in supratentorial vs infratentorial tumors (3 mo vs 28 mo, respectively; p = 0.044). No significant correlation was observed between TTP and either age at diagnosis or gender in recurrent patients.

Genes associated with nonrecurrent EPN are predominantly immune-related

Gene expression microarray profiles generated from surgical specimens of EPN at initial presentation were separated into 2 groups: nonrecurrent (n = 9) and recurrent (n = 10). In the first gene ontology analysis, GSEA was used to identify enriched biological function in genes associated with either the nonrecurrent or the recurrent groups, respectively termed “the nonrecurrent phenotype” and “the recurrent phenotype” (Table II⇓). This revealed that “adaptive immune response” was the most highly enriched GOTERM in the nonrecurrent phenotype with a FDR of 0.059. In the recurrent phenotype the most enriched GOTERM was “glutamate signaling pathway”, which did not reach statistical significance by FDR (0.355).

View this table:
Table II.

Ontologic analyses of genes associated with the nonrecurrent and recurrent phenotypes in EPNa

DAVID was used as an additional measure of gene function enrichment. Two lists of genes that were associated either with nonrecurrent or recurrent phenotypes were generated before DAVID analysis as input. One hundred twenty-seven of the 18,624 genes used in this analysis were overexpressed (<2-fold; p < 0.05) in nonrecurrent EPN vs recurrent EPN groups. DAVID demonstrated that the GOTERM “immune response” was the most significantly enriched ontology (FDR = 9.4 × 10−9) in the nonrecurrent phenotype (Table II⇑). In contrast, the most enriched GOTERM in the recurrent EPN phenotype (47 genes) was “multicellular organismal process”, which was not statistically significant by FDR (0.98).

Both GSEA and DAVID identified immune function-related genes as the most enriched ontology in the nonrecurrent EPN phenotype. By contrast, there was no overlap in gene ontology enrichments identified by GSEA and DAVID in the recurrent EPN phenotype, nor did either approach identify any statistically significant enrichment by FDR.

A detailed analysis of the genes associated with nonrecurrent EPN phenotype was performed to elaborate the results of the computer-based ontological analyses described above. All of the 127 genes that were overexpressed in nonrecurrent EPN (>2-fold; p < 0.05) were evaluated for their potential roles in any immune-related process as described in peer-reviewed publications. Fifty-four percent (68 out of127) of these genes had documented immune-related functions. This approach identified a number of immune-related genes beyond those identified by GSEA or DAVID; these genes had erroneously not been assigned an annotation of immune function by GO.

Of the immune-related genes overexpressed in nonrecurrent EPN, a number of genes that are involved in both innate and adaptive immune responses were identified (Table III⇓⇓). Key initiating components of both the classical and lectin complement innate response pathways (C1QC and MASP1, respectively) and downstream complement components C3, C3AR1, and ITGB2 (integrin β2) were identified. Multiple MHC class II alleles were identified (HLA-DMA, HLA-DMB, HLA-DPB1, HLA-DRB5, and CD74). MHC class II is predominantly expressed on APCs, the most predominant of which in the CNS are thought to be the microglia/macrophage population. A number of other genes that are associated with microglia or macrophages were found to be overexpressed in the nonrecurrent EPN phenotype. Among these was AIF1, which is a specific marker of activated microglia/macrophages (28). In the context of adaptive immune function, a number of genes specifically involved in T lymphocyte activity were associated with the nonrecurrent phenotype, including TCR α constant (TRAC), CD37, FYN binding protein (FYB), hepatitis A virus cellular receptor 2 (HAVCR2), hematopoietic cell-specific Lyn substrate-1 (HCLS1), and linker for activation of T cells family member 2 (LAT2). Other notable immune function-related genes associated with the nonrecurrent phenotype are Fc receptors CD64A and B, STAT6, TNF (ligand) superfamily, member 10 (TRAIL), and cytochrome b-245, α and β polypeptides (CYBA and CYBB).

View this table:
Table III.

Immune function related genes overexpressed the nonrecurrent EPN phenotypea

View this table:
Table 3A.

(Continued)

In recurrent EPN, genes that positively correlate with longer time to progression are predominantly immune-related

EPN generally recur within 3 years of initial presentation. Our recurrent EPN cohort had TTP ranging from 1 to 51 mo. To identify genes associated with TTP in EPN that recurred (n = 10), microarray gene expression data were correlated with TTP as a continuous variable using a modified version of the GSEA approach described above. GSEA identified “humoral immune response” as the highest enriched GOTERM in genes that positively correlated with TTP (FDR = 0.0694) (Table IV⇓). In the reverse analysis, “biological process” was the highest enriched GOTERM in genes that negatively correlated with TTP (FDR = 0.223), although these were less statistically significant than was the immune gene correlation.

View this table:
Table IV.

Gene ontology analyses of genes positively and negatively correlated with longer time to progression in EPNa

As an input for DAVID, a list of 395 genes that positively correlated (p < 0.05 estimated by two-sided Pearson correlation test) with TTP in recurrent EPN (n = 10) was created using all 18,624 genes. Using the same approach, a list of 841 genes that were negatively correlated with TTP was also created. Similar to the GSEA results, DAVID confirmed that immune function-related was the most enriched ontology in genes that positively correlated with TTP (Table IV⇑). Cell cycle-related ontologies were found to be enriched in genes that were negatively correlated with TTP (FDR = 2.11 × 10−13), with greater statistical significance than the positive TTP correlate-enriched ontologies (FDR = 7.72 × 10−7).

Detailed analysis of genes that positively correlated with TTP in recurrent EPN was performed to elaborate the results of the above computer-based ontological analyses. Twenty-eight percent (110 out of 395) of the genes positively correlated with TTP in recurrent EPN with statistical significance (p < 0.05) were related to immune function. The results of this analysis, with genes listed and categorized into subgroups according to their documented role in specific immune mechanisms, are provided in Table V⇓⇓. As found in the previous analysis, a number of genes beyond those identified by GSEA or DAVID were found, due to their not having been assigned an annotation of immune function by GO. As seen in the nonrecurrent phenotype, genes whose expression positively correlated with TTP included a number of molecules critically involved in both innate and adaptive immune responses. Some overlap in innate and adaptive immune-related genes was observed between the nonrecurrent phenotype and TTP-positive correlates, analyzed in more detail below. As seen in the nonrecurrent phenotype, multiple components of the complement system (C2, C3, C3AR1, C6, C7, CD53, CD59, CR1, ITGB2) and genes associated with microglia/macrophages (AIF1, CD36, HLA-DMB, LILRA2, LILRB1, LILRB2, LILRB4) and T cells (FYB, HCLS1, LAT2, TAGAP2, TRDV2) were identified in positive correlates of TTP in recurrent EPN. The main difference that distinguished TTP-positive correlates from the nonrecurrent phenotype was the presence of a significant number of genes commonly expressed by B cells. These included multiple Ig genes (IGHA2, IGHG3, IGHM, IGJ, IGKC, IGKV1D-13, IGLC2, IGLJ3, IGLL3, and IGSF6).

View this table:
Table V.

Immune-related genes positively correlated with TTPa

View this table:
Table 5A.

(Continued)

Overlapping genes between the nonrecurrent EPN phenotype and positive correlates of time to progression are almost entirely immune function-related

A number of genes were identified that were associated with both the nonrecurrent EPN phenotype and that were also positively correlated with TTP, emphasizing their involvement in EPN clinical outcome as a whole. Ontological analysis revealed that 95% (19 out of 20) of genes associated with both nonrecurrence and longer TTP have roles in innate and adaptive immune functions, the details of which are outlined in Table VI⇓. These genes are involved in complement activity (C3, C3AR1, and ITGB2), macrophage activity (AIF1), phagocytosis of Ab-coated cells (FCGR1A, FCGR1B, LILRB1, CYBB), Ag presentation (HLA-DMB), lymphocyte tethering and rolling (SELPLG, CORO1A, DOCK2, APBB1IP, ARHGAP4), and lymphocyte activation (HCLS1, LAT2, FYB, GPR65, HCK). Only phosphorylase kinase, γ-1 (PHKG1) had no documented evidence of immune involvement, with its known function being as a key glycogenolytic enzyme. However, the dependence of T-lymphocyte activity on glucose metabolism suggests a potential role in immune function for this gene (29). In the reverse analysis of genes that overlapped between both bad prognosis groups, that is, the recurrent EPN phenotype and TTP negative correlates, only two genes were identified: programmed cell death-6 (PDCD6) and opsin-3 (OPN3), which are known to have roles in TCR-induced apoptosis and photoreception, respectively.

View this table:
Table VI.

Overlapping genes that are associated with both the non-recurrent and long PFS phenotypes in ependymoma showing function, cellular distribution and key reference(s) pertaining to each of the 20 genes

Immune-related genes associated with a good outcome in EPN are expressed by infiltrating cells within the tumor

It was predicted that up-regulated immune-related genes identified by ontological analyses were expressed by tumor-infiltrating immune cells within patient tumor samples. To provide some evidence for this, histology was used to identify individual cells expressing AIF1 and HLA-DR. These immune-related genes associated with the nonrecurrent phenotype are known to be expressed by microglia/macrophages (28). IHC of AIF1 and HLA-DR protein expression was performed in FFPE tissue of nonrecurrent (n = 9) and recurrent (n = 10) EPN. Protein expression of AIF1 (Fig. 1⇓A) and HLA-DR (Fig. 1⇓B) was restricted to a subpopulation of cells in the parenchyma of the tumor with a cellular morphology that resembled microglia/macrophages. These data indicate that at least a subset of immune function gene transcripts identified by microarray analyses are derived from tumor-infiltrating immune cells.

FIGURE 1.
FIGURE 1.

Immunohistochemical staining of (A) AIF-1 and (B) HLA-DR in FFPE tumor sections of nonrecurrent EPN with hematoxylin counterstaining (×400). Relative abundancy of (C) AIF-1 and (D) HLA-DR positive infiltrating cells in nonrecurrent (non-rec; n = 9) and recurrent (rec; n = 10) EPN. Cells were scored using the mean of the number of positive staining cells in five fields of view and statistical significance by Student’s t test was defined as p < 0.05. Horizontal bars represent the mean average of scores.

To validate the association of AIF1 and HLA-DR expression with outcome, the frequency of positively immunostaining cells in the parenchyma of nonrecurrent and recurrent EPN was measured. This analysis revealed that AIF1 positive staining cells were significantly more abundant in nonrecurrent EPN (1.91-fold; p = 0.0082) (Fig. 1⇑C). HLA-DR was on average 2-fold more abundant in nonrecurrent EPN but was not significant (2.18-fold; p = 0.082) (Fig. 1⇑D). These data recapitulate the results of gene expression analysis that demonstrated overexpression of AIF1 and HLA-DR5B in nonrecurrent EPN compared with recurrent EPN.

Tumor-infiltrating immune cells are present in EPN and associated with a good outcome

In addition to the microglia/macrophage-associated markers analyzed above, T and B cell-related transcripts were found to be associated with outcome, suggesting a variety of infiltrating immune cell subtypes in EPN. IHC was used to identify CD4+ T cells, CD8+ T cells, CD45+ leukocytes, microphage/microglia (CD68+), and B cells (CD20) in FFPE tissue in nonrecurrent (n = 9) and recurrent (n = 10) EPN. Representative staining of these immune cell subpopulations is depicted in Fig. 2⇓. Microglia/macrophages and CD45+ leukocytes were more abundant than T cells or B cells across all EPN analyzed.

FIGURE 2.
FIGURE 2.

Representative infiltration of (A) CD4+ and (B) CD8+ T cells in nonrecurrent EPN. C, CD45+ leukocytes and (D) CD68+ microglia were observed in greater numbers than T cells across all samples. Immunostaining with hematoxylin counterstain (×400).

Frequency analysis of infiltrating cells showed increased numbers of CD4+ T cells (16-fold; p = 0.045), CD8+ T cells (1.92-fold; p = 0.34), CD45+ leukocytes (1.55-fold; p = 0.16), and microglia/macrophages (3.06-fold; p = 0.18) in nonrecurrent EPN compared with recurrent EPN, although only CD4+ T cells reached statistical significance (Fig. 3⇓). Greater numbers of B cells were observed in recurrent EPN, although this difference was not statistically significant (3.92-fold; p = 0.12).

FIGURE 3.
FIGURE 3.

Tumor-infiltrating immune cells in nonrecurrent (non-rec; n = 9) and recurrent (rec; n = 10) EPN. A, CD45+ leukocytes, (B) CD4+ T cells, (C) CD8+ T cells, (D) CD68+ microglia/macrophages, and (E) CD20+ B cells were identified in paraffin sections of tumor specimens using immunohistochemistry. Cells were scored using the mean of the number of positive staining cells in five fields of view and statistical significance by t test was defined as p < 0.05. Horizontal bars represent the mean average of scores.

Polyomavirus SV40, BK, JC, and Merkel are not present in nonrecurrent EPN

A number of genes associated with the nonrecurrent EPN phenotype are known to be involved in the immune response to viral infection, in particular IFN regulatory factor-7 (IRF7), tripartite motif-containing-22 (TRIM22), and apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like 3G (APOBEC3G) (30, 31, 32). Earlier research found SV40-like PyV sequences in ∼50% of EPN but did not attempt to correlate viral positivity with clinical outcome (33). Based on the fact that the percentage of EPN found to contain viral sequences in this earlier study matched the percentage of patients who did not suffer from recurrence, it was hypothesized that nonrecurrent EPN samples might contain virus, triggering an increase in viral immune response gene expression. Presence of virus has been shown to predict a favorable outcome in other tumor types, such as head and neck cancer, which supported this hypothesis (34). Nonrecurrent (n = 8) and recurrent (n = 9) EPN specimens were therefore screened for the presence of SV40, BK, JC, and Merkel PyV DNA sequences using quantitative PCR. No PyV sequences were found in any of the tumor specimens tested apart from three of nine recurrent EPN that showed weak positivity for SV40. Thus, no association between the nonrecurrent EPN phenotype and the presence of PyV DNA sequences was observed. Despite these data, the possibility cannot be ruled out that some virus other than those tested is present in nonrecurrent EPN.

Discussion

This study provides early circumstantial evidence that in ∼50% of EPN patients there is a host antitumor immune response and/or proinflammatory microenvironment that, when combined with standard therapy, results in complete eradication of remaining residual tumor cells. Additionally, these data provide a novel perspective to the clinical problem of how to identify up-front those children whose EPN will recur by identifying a functional role for genes associated with prognosis, rather than simply listing genes as in previous studies (17, 19, 20, 21). Similar to the results of this study, correlation of lymphoma microarray profiles with outcome demonstrated that immune gene expression was the predominant feature that predicted survival (35). The presence of tumor reactive T and B cells and tumor-infiltrating lymphocytes (TIL) in clinical specimens has been correlated with an improved outcome in a number of tumor types (36, 37). The presence of TIL is a prognostic marker in these tumors and provides a precedent for the correlation of immune cell infiltration with good clinical outcome in EPN seen in the present study. Prospective validation of immune-related factors as an up-front prognostic marker in EPN is clearly warranted based on the results of this study.

The results of this study provide preliminary evidence for involvement of both the innate and adaptive arms of the immune response in host control of EPN. The innate immune system uses a diversity of pathways to recognize and respond to Ags, including potential cancer-specific Ags. The complement system is the major humoral component of the innate immune system, and multiple complement system genes were found to be associated with a good outcome in EPN (C1QC, C2, C3, C6, C7, C3AR1, CR1, CD53, CD59, ITGB2, MASP1, SERPING1). Complement-dependent cytotoxicity is thought to be one of the most important mechanisms of action of therapeutic mAbs against cancer (38). In animal studies of rituximab-mediated tumor control, the presence of C1Q was found to be critical for effective complement-dependent cytotoxicity. C1QC, a key initiating molecule of the classical, Ab-dependent complement pathway, was associated with the nonrecurrent EPN phenotype, but not with a long TTP in recurrent EPN.

A number of genes specifically associated with activity of microglia/macrophages, the key cellular component of the innate immune system, were found to correlate with good outcome in EPN. These included AIF1 (28), multiple MHC class II alleles (HLA-DMA, HLA-DMB, HLA-DPB1, HLA-DRB5 and CD74), and leukocyte Ig-like receptor, subfamily b1 (LILRB1). IHC analysis of AIF1 and HLA-DR demonstrated that these molecules are restricted to tumor-infiltrating cells. Based on the morphology of AIF1 and HLA-DR staining, as compared with macrophage/microglia staining in matched samples, it appears that AIF1 and HLA-DR are being expressed by tumor-infiltrating microglia/macrophages as expected. The increased expression of AIF1 in nonrecurrent EPN vs recurrent EPN was demonstrated by both microarray analysis (2.62-fold; p = 0.0073) and IHC (1.8-fold; p = 0.0082). Consistent with our data, MHC class II expression positively correlates with a favorable outcome in a variety of non-CNS tumors such as diffuse large B cell lymphoma and hepatocellular carcinoma (39, 40).

The association of microglia/macrophage-specific transcripts with improved outcome in EPN is contrary to a number of reports of compromised microglia/macrophage activity, including reduced MHC class II expression, in other CNS tumors (41, 42). Furthermore, there is growing evidence that tumor-infiltrating macrophages promote tumor activity in the brain and elsewhere (43, 44). Note that most studies of tumor-infiltrating microglia/macrophages in the CNS have been performed in glioblastoma, which has a highly immunosuppressive tumor microenviroment and a uniformly dismal outcome. Direct comparison of infiltrating microglia/macrophages in good outcome EPN and glioblastoma may shed light on this disparity.

The up-regulation of numerous adaptive immune response related genes was observed in good prognosis EPN. In previous studies of CNS microglia, innate immune system activation was characterized by up-regulation of type-1 IFN and MHC class II expression, resulting in cross-presentation of viral epitopes to CD4+ T cells (45). Consistent with this, in nonrecurrent EPN we observed overexpression of IFN-induced genes (e.g., IFIT1, IFIT3), multiple MHC class II genes, genes specifically associated with T cell activation (e.g., TRAC, CD37, FYB, HAVCR2, HCLS1), and increased frequency of tumor-infiltrating CD4+ T cells. A number of other examples of specific adaptive immune response activities are implied by EPN outcome-associated transcripts. These include the observation that B cell-associated transcripts are correlated with delayed recurrence, but are not found in the nonrecurrent phenotype. Although preliminary, this result suggest that an Ab response affords some resistance to tumor recurrence, but a T cell-specific response is required for complete tumor elimination. The presence of a number of T cell function-related transcripts elaborate specific T cell functions in good outcome EPN. For example, polarization of nonrecurrent EPN infiltrating T cells to the Th1 phenotype is implied by the presence of HAVCR2 (TIM3) (46). Taken together, these data provide preliminary evidence that, beyond the simple presence of an immune infiltrate, the phenotype and function of that infiltrate may influence clinical outcome in EPN. This conclusion is consistent with the report by Galon et al. demonstrating that the type (specifically Th1), density, and location of immune cells within human colorectal tumors predict clinical outcome better than current staging criteria (47).

Finally, the interactive role between standard surgery, radiotherapy, and chemotherapy and the host immune system cannot be overstated. There is increasing evidence that antitumor immune responses may contribute to the control of cancer after conventional chemotherapy, by modulating the equilibrium between the tumor and the immune system (48, 49). This theory may apply to our findings, whereby in those EPN that harbor a host immune response, surgery and radiation therapy may shift the balance of the equilibrium in favor of the host by critically increasing the immune/tumor cell ratio. This would then result in elimination of remaining residual tumor by the immune system, resulting in a favorable outcome in the patient. In those patients who do not receive a complete tumor resection or radiation therapy, the equilibrium remains in favor of the tumor, resulting in the poor outcome that is observed in such scenarios. In those patients that receive standard therapy but lack an antitumor immune response, residual tumor continues to grow unhindered despite receiving standard therapy, resulting in tumor recurrence.

Despite the promising results in animal studies of CNS cancer immunotherapy, clinical trials using immunotherapy in humans have had limited success (50, 51, 52, 53). This failure suggests that knowledge of the antitumor immune response in the human CNS cannot be extrapolated from animal models as previously assumed. A more rational approach to successfully implementing immunotherapy would be to design strategies based on data taken from direct clinical studies of human host anti-CNS tumor immune responses. This report potentially illustrates just such a response, underscoring the value and potential impact of these findings.

Acknowledgments

We thank Patsy Ruegg at IHCtech for assistance with immunohistochemistry and Liza Litzenberger for photographic expertise.

Disclosures

The authors have no financial conflicts of interest.

Footnotes

  • 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.

  • ↵1 This work was supported by the Tanner Seebaum Foundation.

  • ↵2 Address correspondence and reprint requests to Andrew M. Donson, Department of Pediatrics, University of Colorado Denver, Mail Stop 8302, P.O. Box 6511, Aurora, CO 80045. E-mail address: andrew.donson{at}ucdenver.edu

  • ↵3 Abbreviations used in this paper: EPN, ependymoma; AIF-1, allograft inhibitory factor-1; DAVID; Database for Annotation, Visualization, and Integrated Discovery; FDR, false discovery rate; FFPE, formalin-fixed paraffin-embedded; gcRMA, GeneChip robust multiarray average; GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; IHC, immunohistochemistry; TIL, tumor-infiltrating lymphocyte; TTP, time to progression; GOTERM, Gene Ontology Project term.

  • Received August 25, 2009.
  • Accepted October 5, 2009.

References

  1. ↵
    Rickert, C. H., W. Paulus. 2001. Epidemiology of central nervous system tumors in childhood and adolescence based on the new WHO classification. Childs Nerv. Syst. 17: 503-511.
    OpenUrlCrossRefPubMed
  2. ↵
    Central Brain Tumor Registry of the United States (CBTRUS). 2005. Statistical report: prmary brain tumors in the United States, 1998–2002. Central Brain Tumor Registry of the United States. Available at www.cbtrus.org/reports//2005–2006/2006report.pdf
  3. ↵
    Foreman, N. K., S. Love, S. S. Gill, H. B. Coakham. 1997. Second-look surgery for incompletely resected fourth ventricle ependymomas: technical case report. Neurosurgery 40: 856-860. discussion 860.
    OpenUrlCrossRefPubMed
  4. ↵
    Ridley, L., R. Rahman, M. A. Brundler, D. Ellison, J. Lowe, K. Robson, E. Prebble, I. Luckett, R. J. Gilbertson, S. Parkes, et al 2008. Multifactorial analysis of predictors of outcome in pediatric intracranial ependymoma. Neurol. Oncol. 10: 675-689.
    OpenUrlAbstract/FREE Full Text
  5. ↵
    Rousseau, P., J. L. Habrand, D. Sarrazin, C. Kalifa, M. J. Terrier-Lacombe, C. Rekacewicz, A. Rey. 1994. Treatment of intracranial ependymomas of children: review of a 15-year experience. Int. J. Radiat. Oncol. Biol. Phys. 28: 381-386.
    OpenUrlCrossRefPubMed
  6. ↵
    Agaoglu, F. Y., I. Ayan, Y. Dizdar, R. Kebudi, O. Gorgun, E. Darendeliler. 2005. Ependymal tumors in childhood. Pediatr. Blood Cancer 45: 298-303.
    OpenUrlCrossRefPubMed
  7. ↵
    Pollack, I. F., P. C. Gerszten, A. J. Martinez, K. H. Lo, B. Shultz, A. L. Albright, J. Janosky, M. Deutsch. 1995. Intracranial ependymomas of childhood: long-term outcome and prognostic factors. Neurosurgery 37: 655-666. discussion 666–657.
    OpenUrlCrossRefPubMed
  8. ↵
    Robertson, P. L., P. M. Zeltzer, J. M. Boyett, L. B. Rorke, J. C. Allen, J. R. Geyer, P. Stanley, H. Li, A. L. Albright, P. McGuire-Cullen, et al 1998. Survival and prognostic factors following radiation therapy and chemotherapy for ependymomas in children: a report of the Children’s Cancer Group. J. Neurosurg. 88: 695-703.
    OpenUrlCrossRefPubMed
  9. ↵
    Figarella-Branger, D., M. Civatte, C. Bouvier-Labit, J. Gouvernet, D. Gambarelli, J. C. Gentet, G. Lena, M. Choux, J. F. Pellissier. 2000. Prognostic factors in intracranial ependymomas in children. J. Neurosurg. 93: 605-613.
    OpenUrlPubMed
  10. ↵
    Korshunov, A., A. Golanov, V. Timirgaz. 2002. Immunohistochemical markers for prognosis of ependymal neoplasms. J. Neurooncol. 58: 255-270.
    OpenUrlCrossRefPubMed
  11. ↵
    Tihan, T., T. Zhou, E. Holmes, P. C. Burger, S. Ozuysal, E. J. Rushing. 2008. The prognostic value of histological grading of posterior fossa ependymomas in children: a Children’s Oncology Group study and a review of prognostic factors. Mod. Pathol. 21: 165-177.
    OpenUrlPubMed
  12. ↵
    Bennetto, L., N. Foreman, B. Harding, R. Hayward, J. Ironside, S. Love, D. Ellison. 1998. Ki-67 immunolabelling index is a prognostic indicator in childhood posterior fossa ependymomas. Neuropathol. Appl. Neurobiol. 24: 434-440.
    OpenUrlCrossRefPubMed
  13. ↵
    Preusser, M., H. Heinzl, E. Gelpi, R. Hoftberger, I. Fischer, I. Pipp, I. Milenkovic, A. Wohrer, F. Popovici, S. Wolfsberger, J. A. Hainfellner. 2008. Ki67 index in intracranial ependymoma: a promising histopathological candidate biomarker. Histopathology 53: 39-47.
    OpenUrlCrossRefPubMed
  14. ↵
    Altura, R. A., R. S. Olshefski, Y. Jiang, D. R. Boue. 2003. Nuclear expression of survivin in paediatric ependymomas and choroid plexus tumours correlates with morphologic tumour grade. Br. J. Cancer 89: 1743-1749.
    OpenUrlCrossRefPubMed
  15. ↵
    Preusser, M., S. Wolfsberger, T. Czech, I. Slavc, H. Budka, J. A. Hainfellner. 2005. Survivin expression in intracranial ependymomas and its correlation with tumor cell proliferation and patient outcome. Am. J. Clin. Pathol. 124: 543-549.
    OpenUrlAbstract/FREE Full Text
  16. ↵
    Tabori, U., J. Ma, M. Carter, M. Zielenska, J. Rutka, E. Bouffet, U. Bartels, D. Malkin, C. Hawkins. 2006. Human telomere reverse transcriptase expression predicts progression and survival in pediatric intracranial ependymoma. J. Clin. Oncol. 24: 1522-1528.
    OpenUrlAbstract/FREE Full Text
  17. ↵
    Modena, P., E. Lualdi, F. Facchinetti, J. Veltman, J. F. Reid, S. Minardi, I. Janssen, F. Giangaspero, M. Forni, G. Finocchiaro, et al 2006. Identification of tumor-specific molecular signatures in intracranial ependymoma and association with clinical characteristics. J. Clin. Oncol. 24: 5223-5233.
    OpenUrlAbstract/FREE Full Text
  18. ↵
    Pezzolo, A., V. Capra, A. Raso, F. Morandi, F. Parodi, C. Gambini, P. Nozza, F. Giangaspero, A. Cama, V. Pistoia, M. L. Garre. 2008. Identification of novel chromosomal abnormalities and prognostic cytogenetics markers in intracranial pediatric ependymoma. Cancer Lett. 261: 235-243.
    OpenUrlPubMed
  19. ↵
    Sowar, K., J. Straessle, A. M. Donson, M. Handler, N. K. Foreman. 2006. Predicting which children are at risk for ependymoma relapse. J. Neurooncol. 78: 41-46.
    OpenUrlCrossRefPubMed
  20. ↵
    Lukashova-v Zangen, I., S. Kneitz, C. M. Monoranu, S. Rutkowski, B. Hinkes, G. H. Vince, B. Huang, W. Roggendorf. 2007. Ependymoma gene expression profiles associated with histological subtype, proliferation, and patient survival. Acta Neuropathol. 113: 325-337.
    OpenUrlCrossRefPubMed
  21. ↵
    de Bont, J. M., J. M. Kros, M. M. Passier, R. E. Reddingius, P. A. Sillevis Smitt, T. M. Luider, M. L. den Boer, R. Pieters. 2008. Differential expression and prognostic significance of SOX genes in pediatric medulloblastoma and ependymoma identified by microarray analysis. Neurol. Oncol. 10: 648-660.
    OpenUrlAbstract/FREE Full Text
  22. ↵
    McLendon, R. E., O. D. Wiestler, J. M. Kros, A. Korshunov, H.-K. Ng. 2007. Ependymoma. P. Kleihues, and W. K. Cavenee, eds. Pathology and Genetics of Tumours of the Nervous System 74-80. IARC Press, Lyon, France.
  23. ↵
    Wu, Z., R. A. Irizarry, R. Gentleman, F. Martinez-Murillo, F. Spencer. 2004. A model-based background adjustment for oligonucleotide expression arrays. J. Am. Stat. Assoc. 99: 909-917.
    OpenUrlCrossRef
  24. ↵
    Edgar, R., M. Domrachev, A. E. Lash. 2002. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 30: 207-210.
    OpenUrlAbstract/FREE Full Text
  25. ↵
    Mootha, V. K., C. M. Lindgren, K. F. Eriksson, A. Subramanian, S. Sihag, J. Lehar, P. Puigserver, E. Carlsson, M. Ridderstrale, E. Laurila, et al 2003. PGC-1α-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat. Genet. 34: 267-273.
    OpenUrlCrossRefPubMed
  26. ↵
    Dennis, G., Jr, B. T. Sherman, D. A. Hosack, J. Yang, W. Gao, H. C. Lane, R. A. Lempicki. 2003. DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol. 4: P3
    OpenUrlCrossRefPubMed
  27. ↵
    Ashburner, M., C. A. Ball, J. A. Blake, D. Botstein, H. Butler, J. M. Cherry, A. P. Davis, K. Dolinski, S. S. Dwight, J. T. Eppig, et al 2000. Gene ontology: tool for the unification of biology: The Gene Ontology Consortium. Nat. Genet. 25: 25-29.
    OpenUrlCrossRefPubMed
  28. ↵
    Yamada, M., K. Ohsawa, Y. Imai, S. Kohsaka, S. Kamitori. 2006. X-ray structures of the microglia/macrophage-specific protein Iba1 from human and mouse demonstrate novel molecular conformation change induced by calcium binding. J. Mol. Biol. 364: 449-457.
    OpenUrlCrossRefPubMed
  29. ↵
    Cham, C. M., G. Driessens, J. P. O'Keefe, T. F. Gajewski. 2008. Glucose deprivation inhibits multiple key gene expression events and effector functions in CD8+ T cells. Eur. J. Immunol. 38: 2438-2450.
    OpenUrlCrossRefPubMed
  30. ↵
    Song, Y. J., K. M. Izumi, N. P. Shinners, B. E. Gewurz, E. Kieff. 2008. IRF7 activation by Epstein-Barr virus latent membrane protein 1 requires localization at activation sites and TRAF6, but not TRAF2 or TRAF3. Proc. Natl. Acad. Sci. USA 105: 18448-18453.
    OpenUrlAbstract/FREE Full Text
  31. ↵
    Tissot, C., N. Mechti. 1995. Molecular cloning of a new interferon-induced factor that represses human immunodeficiency virus type 1 long terminal repeat expression. J. Biol. Chem. 270: 14891-14898.
    OpenUrlAbstract/FREE Full Text
  32. ↵
    Sheehy, A. M., N. C. Gaddis, J. D. Choi, M. H. Malim. 2002. Isolation of a human gene that inhibits HIV-1 infection and is suppressed by the viral Vif protein. Nature 418: 646-650.
    OpenUrlCrossRefPubMed
  33. ↵
    Bergsagel, D. J., M. J. Finegold, J. S. Butel, W. J. Kupsky, R. L. Garcea. 1992. DNA sequences similar to those of simian virus 40 in ependymomas and choroid plexus tumors of childhood. N. Engl. J. Med. 326: 988-993.
    OpenUrlCrossRefPubMed
  34. ↵
    Lindel, K., K. T. Beer, J. Laissue, R. H. Greiner, D. M. Aebersold. 2001. Human papillomavirus positive squamous cell carcinoma of the oropharynx: a radiosensitive subgroup of head and neck carcinoma. Cancer 92: 805-813.
    OpenUrlCrossRefPubMed
  35. ↵
    Dave, S. S., G. Wright, B. Tan, A. Rosenwald, R. D. Gascoyne, W. C. Chan, R. I. Fisher, R. M. Braziel, L. M. Rimsza, T. M. Grogan, et al 2004. Prediction of survival in follicular lymphoma based on molecular features of tumor-infiltrating immune cells. N. Engl. J. Med. 351: 2159-2169.
    OpenUrlCrossRefPubMed
  36. ↵
    Clemente, C. G., M. C. Mihm, Jr, R. Bufalino, S. Zurrida, P. Collini, N. Cascinelli. 1996. Prognostic value of tumor infiltrating lymphocytes in the vertical growth phase of primary cutaneous melanoma. Cancer 77: 1303-1310.
    OpenUrlCrossRefPubMed
  37. ↵
    Cho, Y., M. Miyamoto, K. Kato, A. Fukunaga, T. Shichinohe, Y. Kawarada, Y. Hida, T. Oshikiri, T. Kurokawa, M. Suzuoki, et al 2003. CD4+ and CD8+ T cells cooperate to improve prognosis of patients with esophageal squamous cell carcinoma. Cancer Res. 63: 1555-1559.
    OpenUrlAbstract/FREE Full Text
  38. ↵
    Glennie, M. J., R. R. French, M. S. Cragg, R. P. Taylor. 2007. Mechanisms of killing by anti-CD20 monoclonal antibodies. Mol. Immunol. 44: 3823-3837.
    OpenUrlCrossRefPubMed
  39. ↵
    Rimsza, L. M., R. A. Roberts, T. P. Miller, J. M. Unger, M. LeBlanc, R. M. Braziel, D. D. Weisenberger, W. C. Chan, H. K. Muller-Hermelink, E. S. Jaffe, et al 2004. Loss of MHC class II gene and protein expression in diffuse large B-cell lymphoma is related to decreased tumor immunosurveillance and poor patient survival regardless of other prognostic factors: a follow-up study from the Leukemia and Lymphoma Molecular Profiling Project. Blood 103: 4251-4258.
    OpenUrlAbstract/FREE Full Text
  40. ↵
    Matoba, K., N. Iizuka, T. Gondo, T. Ishihara, H. Yamada-Okabe, T. Tamesa, N. Takemoto, K. Hashimoto, K. Sakamoto, T. Miyamoto, et al 2005. Tumor HLA-DR expression linked to early intrahepatic recurrence of hepatocellular carcinoma. Int. J. Cancer. 115: 231-240.
    OpenUrlCrossRefPubMed
  41. ↵
    Flugel, A., M. S. Labeur, E. M. Grasbon-Frodl, G. W. Kreutzberg, M. B. Graeber. 1999. Microglia only weakly present glioma antigen to cytotoxic T cells. Int. J. Dev. Neurosci. 17: 547-556.
    OpenUrlCrossRefPubMed
  42. ↵
    Hussain, S. F., D. Yang, D. Suki, E. Grimm, A. B. Heimberger. 2006. Innate immune functions of microglia isolated from human glioma patients. J. Transl. Med. 4: 15
    OpenUrlCrossRefPubMed
  43. ↵
    Markovic, D. S., R. Glass, M. Synowitz, N. Rooijen, H. Kettenmann. 2005. Microglia stimulate the invasiveness of glioma cells by increasing the activity of metalloprotease-2. J. Neuropathol. Exp. Neurol. 64: 754-762.
    OpenUrlCrossRefPubMed
  44. ↵
    Biswas, S. K., L. Gangi, S. Paul, T. Schioppa, A. Saccani, M. Sironi, B. Bottazzi, A. Doni, B. Vincenzo, F. Pasqualini, et al 2006. A distinct and unique transcriptional program expressed by tumor-associated macrophages (defective NF-κB and enhanced IRF-3/STAT1 activation). Blood 107: 2112-2122.
    OpenUrlAbstract/FREE Full Text
  45. ↵
    Olson, J. K., A. M. Girvin, S. D. Miller. 2001. Direct activation of innate and antigen-presenting functions of microglia following infection with Theiler’s virus. J. Virol. 75: 9780-9789.
    OpenUrlAbstract/FREE Full Text
  46. ↵
    Monney, L., C. A. Sabatos, J. L. Gaglia, A. Ryu, H. Waldner, T. Chernova, S. Manning, E. A. Greenfield, A. J. Coyle, R. A. Sobel, et al 2002. Th1-specific cell surface protein Tim-3 regulates macrophage activation and severity of an autoimmune disease. Nature 415: 536-541.
    OpenUrlCrossRefPubMed
  47. ↵
    Galon, J., A. Costes, F. Sanchez-Cabo, A. Kirilovsky, B. Mlecnik, C. Lagorce-Pages, M. Tosolini, M. Camus, A. Berger, P. Wind, et al 2006. Type, density, and location of immune cells within human colorectal tumors predict clinical outcome. Science 313: 1960-1964.
    OpenUrlAbstract/FREE Full Text
  48. ↵
    Danna, E. A., P. Sinha, M. Gilbert, V. K. Clements, B. A. Pulaski, S. Ostrand-Rosenberg. 2004. Surgical removal of primary tumor reverses tumor-induced immunosuppression despite the presence of metastatic disease. Cancer Res. 64: 2205-2211.
    OpenUrlAbstract/FREE Full Text
  49. ↵
    Zitvogel, L., L. Apetoh, F. Ghiringhelli, F. Andre, A. Tesniere, G. Kroemer. 2008. The anticancer immune response: indispensable for therapeutic success?. J. Clin. Invest. 118: 1991-2001.
    OpenUrlCrossRefPubMed
  50. ↵
    Farkkila, M., J. Jaaskelainen, M. Kallio, G. Blomstedt, R. Raininko, P. Virkkunen, A. Paetau, H. Sarelin, M. Mantyla. 1994. Randomised, controlled study of intratumoral recombinant γ-interferon treatment in newly diagnosed glioblastoma. Br. J. Cancer 70: 138-141.
    OpenUrlPubMed
  51. ↵
    Liau, L. M., R. M. Prins, S. M. Kiertscher, S. K. Odesa, T. J. Kremen, A. J. Giovannone, J. W. Lin, D. J. Chute, P. S. Mischel, T. F. Cloughesy, M. D. Roth. 2005. Dendritic cell vaccination in glioblastoma patients induces systemic and intracranial T-cell responses modulated by the local central nervous system tumor microenvironment. Clin. Cancer Res. 11: 5515-5525.
    OpenUrlAbstract/FREE Full Text
  52. ↵
    Plautz, G. E., G. H. Barnett, D. W. Miller, B. H. Cohen, R. A. Prayson, J. C. Krauss, M. Luciano, D. B. Kangisser, S. Shu. 1998. Systemic T cell adoptive immunotherapy of malignant gliomas. J. Neurosurg. 89: 42-51.
    OpenUrlCrossRefPubMed
  53. ↵
    Riva, P., G. Franceschi, A. Arista, M. Frattarelli, N. Riva, A. M. Cremonini, G. Giuliani, M. Casi. 1997. Local application of radiolabeled monoclonal antibodies in the treatment of high grade malignant gliomas: a six-year clinical experience. Cancer 80: 2733-2742.
    OpenUrlCrossRefPubMed
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools

Jump to section