Introduction
Tumour-associated tissue eosinophilia (TATE) has been increasingly reported, but the exact role of eosinophils in tumours is not yet defined. Many studies have suggested that TATE is associated with favourable prognosis for a variety of carcinomas,Reference Lowe, Jorizzo and Hutt 1 - Reference Nielsen, Hansen, Christensen, Reimert, Brünner and Moesgaard 6 whereas a few other studies have noted that TATE may have a tumour-promoting roleReference Wong, Bowen, Elovic, Gallagher and Weller 7 and association with tumoural invasion of the oral squamous cell carcinomas.Reference Tostes Oliveira, Tjioe and Assao 8 , Reference Said, Wiseman and Yang 9 Eosinophils have been found in several malignant and benign tumours, including oral squamous cell carcinomas, breast carcinomas, gastric cancers, uterine cervix carcinomas, penile cancers, hematologic malignancies, and colonic adenomas.Reference Lowe, Jorizzo and Hutt 1 - Reference Curran and Bertics 12 However, few studies have described TATE in central nervous system (CNS) tumours.Reference Curran and Bertics 12 Two studies from the same group revealed intracavitary eosinophils in malignant astrocytomas of patients who had received an infusion of interleukin 2 (IL-2) combined with ex vivo activated autologous killer cells into the surgical resection cavity. Eosinophils had been absent in the primary operative specimens of those patients, suggesting immunotherapy-induced esoinophila.Reference Hayes, Koslow and Hiesiger 13 , Reference Hayes, Arbit and Odaimi 14 Our recent study showed the infiltration of eosinophils in all four cases of atypical teratoid/rhabdoid tumours (AT/RTs), but not in a small group of glioblastomas.Reference Lu, Wilson, Yong, Pugh and Mehta 15
Pilocytic astrocytoma (PA) is a slow-growing, often cystic astrocytic tumour that occurs predominantly in children.Reference Scheithauer, Hawkins, Tihan, VandenBerg and Burger 16 It is the most common pediatric glial tumour. Pediatric patients with PAs survive longer than adult patients, according to the Central Brain Tumour Registry.Reference Zhang, Kruse and Driggers 17 , Reference Driggers, Zhang, Newcomb, Ge, Hoa and Jadus 18 The discrepancy in outcome between two age groups is attributed to both genetic and nongenetic differences. For example, the glioma-associated antigen precursor protein profile displays the different types between pediatric and adult patients.Reference Zhang, Kruse and Driggers 17 In addition, children’s immune system undergo development that peaks at puberty, which may also help explain why pediatric patients with PAs survive longer.Reference Driggers, Zhang, Newcomb, Ge, Hoa and Jadus 18 Reports have inversely associated glioma risk with atopic diseases.Reference Wiemels, Wiencke and Patoka 19 - Reference Linos, Raine, Alonso and Michaud 21 Eosinophils are established effector cells in atopic diseasesReference Gleich, Adolphson and Leiferman 22 , Reference Hogan, Rosenberg and Moqbel 23 and therefore may be partially responsible for the reported inverse association with atopic diseases and the risk of gliomas.Reference Curran and Bertics 12 In accordance with our observation of eosinophils in the tissues of AR/RTs,Reference Lu, Wilson, Yong, Pugh and Mehta 15 we sought to examine the frequency of eosinophil infiltrates in PAs and determine its clinical-pathological correlation.
Materials and Methods
Patients and Study Design
We received ethics approval from the local institutional Committee on Human Research to complete this study. The study included 44 consecutive cases of patients with PAs diagnosed between 2007 and May 2013. For comparison, we also examined 10 consecutive cases of patients with ependymomas or anaplastic ependymomas diagnosed between 2008 and 2013. Table 1 shows patient characteristics. All patients had undergone surgical procedures at the University of Alberta Hospital. No evidence of concurrent primary infectious/inflammatory disease was present in these cases. Surgical specimens were sent to the pathology department for pathological examination. Tumours were diagnosed and classified according to international guidelines as published in the 2007 World Health Organization’s Classification of Tumours of the Central Nervous System.Reference Scheithauer, Hawkins, Tihan, VandenBerg and Burger 16 We excluded cases with controversial diagnosis.
Patient clinical and pathological features
Abbreviations: mo=months; yr=years; M=male; F=female; FU=follow-up by neuroimaging; RT=radiotherapy.; Bx=biopsy; OR=operation; Re=resection.
Eosinophils: − (0 per 10 high-power fields [HPFs]), + (1-~3 per 10 HPFs), ++ (4-15 per 10 HPFs), and +++ (≥16 per 10 HPFs).
*, resection for residual/recurret tumors.
C1–10: control cases of ependyomomas, recurrent ependymomas*, and anaplastic ependymomas#.
Histopathology and Immunohistochemistry
Surgical specimens were formalin-fixed, routinely processed, paraffin-embedded, sectioned at 5 µm, and stained with hematoxylin and eosin and immunohistochemical methods. We examined stained slides for morphological features. To obtain further adjuvant diagnostic information, we performed immunohistochemical analysis on tissue sections by using the EnVision FLEX Mini kit, high pH (Autostainer/Autostainer Plus; Dako, Carpinteria, CA, USA) detection system after the tissue was deparaffinised and rehydrated according to standard protocol. We used the antibodies, including at least glial fibrillary acidic protein 6F2 and MIB-1 (both from Dako), to immunohistochemically confirm glial cell origin and further characterize the tumours.
Two authors (JQL and OR) assessed morphological features of each case or slide with consensus. Eosinophils were identified with their characteristic morphology (Figure 1). Frequency of eosinophils was then assessed semiquantitatively. The number of eosinophils in ten consecutive high-power microscopic fields (HPFs; each original magnification ×400, 0.16 mm2) was scored using the following scheme: − (0 per 10 HPFs), +(1-3 per 10 HPFs), ++ (4-15 per 10 HPFs), and +++ (16 or more per 10 HPFs). This assessment excluded intravascular eosinophils.
Eosinophil infiltrates in a suprasellar pilocytic astrocytoma (case 3 in Table 1). The tumour shows a biphasic pattern with densely fibrillary and microcytic areas, as well as scattered eosinophils (a, arrows) that are morphologically distinguishable from Rosenthal fibres (b, arrowheads; arrows point to eosinophils) and eosinophilic granular body (c, arrowhead; arrows point to eosinophils). Eosinophils are occasionally identified in the perivascular spaces (c, arrows) and associated with associated with extravasated erythrocytes (d, arrows point to eosinophils; rectangle indicates area of the inset with higher magnification). Original magnifications: ×400 (a and d); ×630 (b and c).
Statistical Analysis
We used Fisher’s exact test to evaluate the association between categorical variables. A two-tailed p value of less than 0.05 was considered significant.
Results
Pilocytic Astrocytomas
PAs exhibited characteristic morphologic features, including a biphasic pattern with various proportions of densely fibrillary and microcytic areas (Figure 1a-c), various numbers of Rosenthal fibres (Figure 1b), and eosinophilic granular bodies (Figure 1c). We noted leptomeningeal infiltration in 14 (32%) of 44 PAs. Only three PAs (cases 3, 10, and 41 [Table 1]) exhibited marked perivascular cuffing of lymphoid cells. The frequency and degree of perivascular lymphoid infiltrates in PAs seen here are similar to those of previously published series.Reference Yang, Han, Sughrue, Tihan and Parsa 24 , Reference Hewedi, Radwan, Shash and Elserry 25 While glomeruloid vasculature was often noted in PAs,Reference Scheithauer, Hawkins, Tihan, VandenBerg and Burger 16 a few PAs contained microfoci of extravasated erythrocytes or microhaemorrhages with occasional hemosiderin deposition.Reference Lee, Huh, Sim and Kim 26
Tumour-Infiltrating Eosinophils
We found eosinophils in tumour tissue of 19 (43%) of 44 (cases 1-19 [Table 1]) PAs. Density of intratumoural infiltrating eosinophils was rare to focally scattered (Figure 1a). Eosinophils were morphologically distinguishable from Rosenthal fibres (Figure 1b) and eosinophilic granular bodies (Figure 1c). Sites of PAs containing eosinophils were present throughout the brain (Table 1). Two resections of spinal cord PAs showed no eosinophils. We noted the presence of eosinophils in seven biopsy samples and 12 resections of PA tissues (not statistically significant between two surgical procedures; p=0.16). Eosinophils were present in only 3 of 8 secondary operations for residual/recurrent PAs, compared with those in 16 of 36 original operations of PAs (not statistically different; p=1.00). We found eosinophils in 6 PAs with leptomeningeal infiltration but not in the other 8 PAs with leptomeningeal infiltration (not statistically significant, compared with PAs without leptomeningeal infiltration; p=1.00). No difference (p=0.36) was evident in finding eosinophils between PAs with and without MRI cystic formation. We occasionally identified eosinophils in the perivascular spaces (Figure 1c) and associated with extravasated erythrocytes or microhaemorrhages (Figure 1d).
Further analysis revealed that eosinophils were identified in 18 (62%) of 29 PAs in pediatric patents (age range, 0.5-18 years) but only in 1 (7%) of 15 (significantly less than that of pediatric patients; p=0.0004) PAs in adult patients (age range, 20-72 years). The adult patient with a PA containing eosinophils was 20 years old (case 8 [Table 1]).
In comparison, eosinophils were absent in all ten ependymomas (including recurrences) or anaplastic ependymomas in pediatric patients (age range, 0.5-15 years; case C1-10 [Table 1]).
Discussion
This study has shown TATE in pediatric PAs. We have also found that, in contrast, TATE has been rare in adult PAs and absent in pediatric ependymomas. In combination with previous studies showing TATE in AT/RTsReference Lu, Wilson, Yong, Pugh and Mehta 15 and in malignant astrocytomas,Reference Hayes, Koslow and Hiesiger 13 , Reference Hayes, Arbit and Odaimi 14 these findings suggest that TATE in CNS tumours may be cell origin dependent and age dependent.
The CNS has generally been considered a relatively immunologically privileged organ because of the blood–brain barrier. When CNS injury occurs, antigen-specific cells can traffic to relevant sites in the CNS. With the anatomic complexity of the CNS, researchers have proposed three routes by which immune cells may enter the CNS: from blood to the cerebrospinal fluid via the choroid plexus, from blood to the subarachnoid space, and from blood to the parenchyma.Reference Ransohoff, Kivisäkk and Kidd 27 , Reference Dunn, Dunn and Curry 28 The mechanism of eosinophil entry into the CNS remains unclear. The trafficking route from blood to CSF via the choroid plexus may be disfavoured in PAs, on the basis of our study revealing the absence of eosinophils in ependymomas and many PAs involving the choroid plexus. Eosinophilic infiltrates have been present in various CNS disorders,Reference Curran and Bertics 12 including eosinophilic meningoencephalitis,Reference Graeff-Teixeira, da Silva and Yoshimura 29 Langerhans cell histiocytosis,Reference Davidson, McComb, Bowen and Krieger 30 and chronic subdural hematomas,Reference Sarkar, Lakhtakia, Gill, Sharma, Mahapatra and Mehta 31 which mostly involve the leptomeninges. The location of those disorders containing eosinophil infiltrates appears to favour eosinophils trafficking into the CNS from blood to the subarachnoid space. Although mast cells have been identified in the dura, leptomeninges, choroid plexus, and brain parenchyma,Reference Silver, Silverman, Vitkovic and Lederhendler 32 the presence of eosinophils in the noninfectious process may be attributed to their bidirectional interactions with mast cells.Reference Hogan, Rosenberg and Moqbel 23 , Reference Sarkar, Lakhtakia, Gill, Sharma, Mahapatra and Mehta 31 , Reference Silver, Silverman, Vitkovic and Lederhendler 32 In our study, however, we found no difference in eosinophil infiltrates between PAs with and without leptomeningeal infiltration. Instead, we often observed eosinophils along with extravasated erythrocytes in the perivascular spaces of PAs. This observation suggests that eosinophils are more likely trafficking from blood directly into the CNS tumours, after the vascular structures of “brain–tumour barrier” have been substantially altered in gliomas.Reference Dunn, Dunn and Curry 28
Eosinophils are pleiotropic multifunctional leukocytes involved in the initiation and propagation of diverse inflammatory responses. They are important modulators of innate and adaptive immunity.Reference Hogan, Rosenberg and Moqbel 23 In response to various stimuli, the eosinophils can produce cytotoxic granules, neuromediators, and proinflammatory cytokines, as well as growth factors and profibrotic and angiogenic factors, which are involved in pathogen clearance and tissue remodeling and repair.Reference Curran and Bertics 12 However, once eosinophils have selectively infiltrated inflamed tissues, they release various toxic proteins, including major basic protein, eosinophil cationic protein, eosinophil peroxidase, and eosinophil neurotoxin, which contribute to tissue damage.Reference Navarro, Boix, Cuchillo and Nogués 33 The role of eosinophils in CNS tumours is therefore complex and probably dual, since eosinophils can induce neurotoxicity to adjacent brain tissue and/or apoptosis of tumour cells.Reference Lu, Wilson, Yong, Pugh and Mehta 15 Increasing evidence has suggested that eosinophils may affect each stage of tumour development; for example, cytokines and chemokines produced by tumour cells have been indicated to alter the tumour-suppressive functions of innate immune cells, creating a microenvironment conducive to tumour development. Also, some cytokines have been suggested to induce recruitment and activation of immune cells (including eosinophils) in association with the tumour rejection and enhanced host survival.Reference Curran and Bertics 12 Nevertheless, the exact role of infiltrating eosinophils in glial tumours deserves further investigation.
Immune cell infiltrates in tumours often vary with tumour type and size.Reference Verbik and Joshi 34 , Reference Mantovani, Allavena, Sica and Balkwill 35 TATE is common and occurs in several non-CNS tumour types, particularly tumours of epithelial origin in the colon and breast.Reference Lowe, Jorizzo and Hutt 1 , Reference Samoszuk 10 , Reference Cormier, Taranova and Bedient 36 Autologous neuroblastoma cells modified to secrete IL-2 and given to pediatric patients with advanced neuroblastoma generated local and systemic antitumour immune responses, including infiltration of eosinophils, as well as dendritic cells, CD4+, and CD8+ lymphocytes.Reference Russell, Strother and Mei 37 In two studies, after the original operation of malignant astrocytomas, intracavitary injection of IL-2 plus ex vivo activated autologous killer cells induced eosinophilia in the intracavitary fluid, tissue, and CSF.Reference Hayes, Koslow and Hiesiger 13 , Reference Hayes, Arbit and Odaimi 14 This eosinophilia appeared to correlate with enhanced patient survival. However, both studies found no eosinophils in the original operative specimens of patients before the IL-2–killer cell immunotherapy, suggesting that eosinophilia is immunotherapy induced in malignant astrocytomas. We recently showed infiltration of eosinophils in all four resections of AT/RTs that are malignant embryonic tumours with divergent differentiation, but absence in four original resections of glioblastomas.Reference Lu, Wilson, Yong, Pugh and Mehta 15 The present study revealed the presence of eosinophils in PAs but not in ependymomas. On the basis of these observations, eosinophil infiltrates are probably limited to some CNS tumours with certain cell origins, particularly in astrocytomas or tumours containing an astrocytic component/differentiation. Although the pathogenesis of cell origin-dependent TATE is unclear, it may be at least partially attributed to different types of tumour-specific antigens present in those CNS tumours.Reference Zhang, Kruse and Driggers 17 , Reference Driggers, Zhang, Newcomb, Ge, Hoa and Jadus 18 , Reference Dunn, Dunn and Curry 28 , Reference Dunn, Fecci and Curry 38 With the presence of this cell origin-dependent TATE and its dissociation from other peripheral blood elements, the infiltration of eosinophils in astrocytomas seems to be actively involved in their pathogenesis other than a passive reactive process.Reference Curran and Bertics 12 , Reference Hewedi, Radwan, Shash and Elserry 25
The major finding of our study is that eosinophils are commonly present in pediatric PAs but rarely (only one 20-year-old patient) in adult PAs. This age-dependent finding is consistent with that of our other study showing eosinophils in AT/RTs of all four patients younger than 2 years.Reference Lu, Wilson, Yong, Pugh and Mehta 15 Two possibilities exist to interpret this age-dependent CNS TATE in pediatric tumours: One is the developing immune system in children, since their immunity peaks around puberty. The other is the difference in tumour-specific antigens between pediatric and adult gliomas.Reference Zhang, Kruse and Driggers 17 , Reference Driggers, Zhang, Newcomb, Ge, Hoa and Jadus 18 The exact mechanisms of the age-dependent TATE require further study.
Conclusion
Our results suggest that the presence of eosinophils is a common feature of pediatric PAs but not of adult PAs. This finding may be at least partially attributed to developing immune system and different tumour-specific antigens in children. Since increasing evidence has suggested that TATE is associated with favourable prognosis in a few tumours, including malignant astrocytomas,Reference Lowe, Jorizzo and Hutt 1 - Reference Nielsen, Hansen, Christensen, Reimert, Brünner and Moesgaard 6 , Reference Hayes, Koslow and Hiesiger 13 , Reference Hayes, Arbit and Odaimi 14 the presence of infiltrating eosinophils in PAs might be related to the much longer survival of pediatric PA patients than that of adults. The infiltration of eosinophils may play a functional role in the development of pediatric PAs, as well as some other pediatric CNS tumours such as AT/RTs.
Disclosures
The authors report no conflicts of interest.
