High Frequency of H3 K27M Mutations in Adult Midline Gliomas
Total Page:16
File Type:pdf, Size:1020Kb
Journal of Cancer Research and Clinical Oncology (2019) 145:839–850 https://doi.org/10.1007/s00432-018-02836-5 ORIGINAL ARTICLE – CANCER RESEARCH High frequency of H3 K27M mutations in adult midline gliomas Azadeh Ebrahimi1,2,8,10 · Marco Skardelly3,4,5,8 · Martin U. Schuhmann3 · Martin Ebinger9 · David Reuss2,10 · Manuela Neumann1,8 · Ghazaleh Tabatabai4,5,6,7,8 · Patricia Kohlhof‑Meinecke11 · Jens Schittenhelm1,8 Received: 11 October 2018 / Accepted: 27 December 2018 / Published online: 4 January 2019 © Springer-Verlag GmbH Germany, part of Springer Nature 2019 Abstract Purpose Diffuse midline gliomas, H3 K27M-mutant were introduced as a new grade IV entity in WHO classification of tumors 2016. These tumors occur often in pediatric patients and show an adverse prognosis with a median survival less than a year. Most of the studies on these tumors, previously known as pediatric diffuse intrinsic pontine glioma, are on pediatric patients and its significance in adult patients is likely underestimated. Methods We studied 165 cases of brain tumors of midline localization initially diagnosed as diffuse astrocytomas, oligo- dendrogliomas, pilocytic astrocytomas, supependymomas, ependymomas and medulloblastomas in patients with an age range of 2–85. Results We identified 41 diffuse midline gliomas according WHO 2016, including 12 pediatric and 29 adult cases, among them two cases with histological features of low grade tumors: pilocytic astrocytoma and subependymoma. 49% (20/41) of the patients were above 30 years old by the first tumor manifestation including 29% (11/41) above 54 that signifies a broader age spectrum as previously reported. Our study confirms that H3 K27M mutations are associated with a poorer prognosis in pediatric patients compared to wild-type tumors, while in adult patients these mutations do not influence the survival significantly. The pattern of tumor growth was different in pediatric compared to adult patients; a diffuse growth along the brain axis was more evident in adult compared to pediatric patients (24% vs. 15%). Conclusion H3 K27M mutations are frequent in adult midline gliomas and have a prognostic role similar to H3 K27M wild- type high-grade tumors. Keywords Diffuse midline glioma · H3F3A · H3 K27M mutations Introduction of the H3F3A (encoding histone H3.3), HIST2H3C (encod- ing histone H3.2) and HIST1H3B/C (encoding histone H3.1) Diffuse midline gliomas H3 K27M mutant are aggressive genes (Castel et al. 2015; Hoffman et al. 2016). They also brain tumors of short median survival, usually less than a show a distinct pattern of global DNA methylation placing year, despite current multimodal therapies (Buczkowicz them into a separate entity (Sturm et al. 2012). Replacement et al. 2014a; Robison and Kieran 2014). These gliomas are of lysine by methionine at codon 27 of the gene encoding mainly located in thalamus, brain stem and spinal cord and histone variant H3 is the most frequent and typical muta- are attributed mostly to the pediatric age group. However, tion in these tumors that can be detected by sequencing they have been reported in young adults and even rarely in as well as immunohistochemistry (IHC) either directly or patients older than 50 years (Buczkowicz et al. 2014a; Daoud through its consequences, i.e. the global loss of trimeth- et al. 2018). These gliomas are characterized by specific ylation (H3K27me3) on this residue (Venneti et al. 2014). genetic and epigenetic alterations such as somatic mutations Initial studies have associated diffuse midline gliomas with astrocytic high-grade morphology (i.e. high mitotic activity, endothelial proliferation and necrosis) and midline localiza- * Azadeh Ebrahimi tion. However, these tumors can show a broader spectrum [email protected] of histological entities (Joyon et al. 2017; Pages et al. 2016; * Jens Schittenhelm Solomon et al. 2016). The midline localization has been sug- [email protected] gested to be associated with specific molecular pathways Extended author information available on the last page of the article Vol.:(0123456789)1 3 840 Journal of Cancer Research and Clinical Oncology (2019) 145:839–850 in these tumors other than those known for classical dif- brain stem and also interhemispheric tumors with expan- fuse gliomas of other localizations; these molecular changes sion to the midline structures) in the Department of Neuro- make the histology of midline gliomas in treatment and sur- surgery, University Hospital of Tuebingen, between 1996 vival prediction rather insignificant (Paugh et al. 2011; Puget and 2018, including three outpatient consultation cases from et al. 2012). So far, the majority of studies on diffuse midline Department of Pathology, Stuttgart. The study was author- gliomas have been performed on pediatric age group while ized by the ethics board of University Hospital of Tuebin- a group of adult patients also present with H3K27M mutant gen (permission number 494/2016BO2). The patients with midline gliomas (Daoud et al. 2018; Laigle-Donadey et al. a minimum age of 18 years at diagnosis were considered for 2008; Meyronet et al. 2017). The clinically significant and adult cohort. Histological diagnosis and tumor grading in all histopathological characteristics of H3 K27M mutations in samples were reviewed primarily on full slides and revised adult vs. pediatric patients have not yet been studied com- according to the 2016 WHO classification system for CNS prehensively. Current diagnostic guidelines according to tumors by at least two experienced neuropathologists (Louis recommendations of WHO classification of CNS tumors et al. 2016). 2016 on diffuse midline gliomas have extended the relevant molecular examinations for histone mutations from pediatric Immunohistochemistry patients to adult patients younger than 54 years old based on a few findings in adults according to previous studies (Foster H3K27M, ATRX, IDH1R132H and BRAF V600E immu- et al. 2016; Louis et al. 2018). In order to address the impact nostains were performed on an automated immunohisto- of H3F3A mutation in adult patients compared to pediat- chemistry system (BenchMark, Ventana Medical Systems, ric patients, we retrospectively investigated a cohort of 165 Strasbourg, France), as previously described (Schittenhelm patients with neuroepithelial brain tumors of midline locali- et al. 2011). Settings for the immunostains were as follows: zation, aging from 2 to 85 years at first diagnosis. The status OptiView method: CC1 pretreatment for 40 min, primary of H3K27M mutation along with other relevant molecular antibody incubation for 20 min at 42 °C, counterstaining markers such as ATRX, BRAF V600E and IDH1 and IDH2 with hematoxylin. Antibodies were used as listed in Table 1 mutations, as well as methylation status of MGMT promoter, and all other staining reagents were provided from Ventana were investigated in this study. In a subset of these tumors, Medical Systems, Strasbourg, France. The immunohisto- NGS panel sequencing and DNA methylation array analy- chemical staining was performed when sufficient material sis were carried out. We characterized the histopathological was available or when after histological review a relevant aspects of these tumors along with clinical significance of differential diagnosis had to be excluded. 17 out of 165 cases H3 K27M mutations in pediatric vs. adult patients. were analyzed as tissue microarrays (TMA). TMAs were provided as previously described (Ebrahimi et al. 2016) and used for BRAF V600E immunostaining. These cases Materials and methods included ependymomas of midline localization. All 17 tissue cores contained more than 95% representative tumor with at Tissue samples least 80% tumor cell content. 165 neuroepithelial brain tumor samples were enrolled in Molecular studies this study. The samples were obtained from the patients that underwent surgery for neuroepithelial tumors in mid- Further molecular analyses were performed when ade- line structures. In order to avoid a histology bias, the cases quate tissue for DNA extraction was available. Using a were selected by their location based on preoperative imag- BlackPREP FFPE kit (Analytik Jena, Germany), DNA ing studies (thalamus, brainstem, spinal cord, cerebellar of adequate quality and quantity was extracted from the vermis, third and fourth ventricles with expansion to the tumor tissue according to the manufacturer’s instructions. Table 1 List of antibodies used for immunohistochemistry Antibody Antibody dilution Incubation time Provider Rabbit anti human Histone H3 (K27M mutant) 1:500 32 min Merck Millipore, Billeria, MA, USA Mouse IgG2a IDH1 R132H (H09) LOT 13211/19 1:400 32 min Dianova, Hamburg, Germany Rabbit anti-ATRX 1:400 40 min Sigma St. Louis, MO, USA Supernatant (Klon VE1), Heidelberg-BRAFV600E 1:5 30 min DKFZ, Dr. Capper Rabbit anti human Tri-Methyl-Histone H3 (K27) (C36B11) 1:200 32 min Cell signaling, Cambridge, UK 1 3 Journal of Cancer Research and Clinical Oncology (2019) 145:839–850 841 Tissue was selected from the regions of paraffin blocks MGMT promoter methylation calling was 50%. We defined that presented sufficient tumor content in microscopy a cut-off value of 9% to classify MGMT methylated versus (minimum of 50% tumor cell content). All samples lack- nonmethylated cases (Quillien et al. 2016). ing IDH1R132H mutation based on immunohistochemical staining were further sequenced for IDH1 and IDH2 muta- Generation of DNA methylation array data and gene tions in patients younger than 55 years