H3K27me3 immunostaining is diagnostic and prognostic in diffuse gliomas with oligodendroglial or mixed oligoastrocytic morphology

Oligodendroglioma is defined by IDH mutation and 1p/19q codeletion. The latter is mutually exclusive to ATRX immunohistochemical loss and has been recently associated with the loss of H3K27me3 immunostaining. We aimed to assess the diagnostic and prognostic value of H3K27me3 immuno-expression in diffuse gliomas with oligodendroglial or mixed oligoastrocytic morphology. H3K27me3 immunostaining was performed in 69 diffuse gliomas with oligodendroglial (n = 62) or oligoastrocytic (n = 7) morphology. The integration with routinely assessed IDH mutations, ATRX immunostaining, and 1p/19q codeletion classified these cases as 60 oligodendroglial and 9 astrocytic. H3K27me3 was lost in 58/60 oligodendrogliomas with retained (n = 47) or non-conclusive (n = 11) ATRX staining, 3/6 IDH-mutant astrocytomas with ATRX loss, and 3/3 IDH-wt astrocytomas. H3K27me3 was retained in 2/60 oligodendrogliomas with retained ATRX, and in 3/6 IDH-mutant astrocytomas, two of which had lost and one retained ATRX. The combination of H3K27me3 and ATRX immunostainings with IDH mutational status correctly classified 55/69 (80%) cases. In IDH-mutant gliomas, ATRX loss indicates astrocytic phenotype, while ATRX retention and H3K27me3 loss identify oligodendroglial phenotype. Only 14 (20%) IDH-mutant cases with retained ATRX and H3K27me3 or inconclusive ATRX immunostaining would have requested 1p/19q codeletion testing to be classified. Furthermore, H3K27me3 retention was associated with significantly shorter relapse-free survival (P < 0.0001), independently from IDH mutation or 1p/19q codeletion (P < 0.005). Our data suggest that adding H3K27me3 immunostaining to the diagnostic workflow of diffuse gliomas with oligodendroglial or mixed morphology is useful for drastically reducing the number of cases requiring 1p/19q codeletion testing and providing relevant prognostic information.


Introduction
The 2016 World Health Organization (WHO) classification of gliomas, by integrating histopathological features with molecular alterations, classifies astrocytic tumours based on the mutational status of IDH1/2 and H3K27M genes, while oligodendrogliomas are defined by the co-occurrence of IDH1/IDH2 mutation and codeletion of whole chromosomal arms 1p and 19q [1]. This classification is prognostically informative, as oligodendrogliomas IDH-mutant and 1p/19q codeleted have the best prognosis among diffuse gliomas, while IDH-wt astrocytic tumours the worst [2]. In addition, diffuse gliomas with 1p/19q codeletion are more likely to respond to chemotherapy [3,4].
IDH mutations in gliomas occur more frequently at residue p.R132 of IDH1 and residue p.R172 of IDH2 [5]. About 90% are IDH1 R132H mutations that can be detected using a commercially available antibody against the mutant epitope [6], while 10% occur in other sites of IDH1 or IDH2 genes and can be currently identified only by DNA mutational analysis [2].
The WHO classification does not indicate a specific method to test 1p/19q codeletion but recommends that the assay should be able to detect whole-arm chromosomal losses [1], as the codeletion is mediated by a balanced whole-arm translocation of chromosomes 1 and 19 followed by the loss of one of the two derivative chromosomes composed of 1p and 19q [7]. Fluorescent in situ hybridization (FISH) is the most common method to assess 1p/19q codeletion [7]. However, this technique has shortcomings. First, it is unable to discriminate between complete and partial deletions of 1p and 19q because commercial probes hybridize to a minimal part of the chromosome arms, at 1p36 and 19q13 loci [8,9]. Second, it cannot establish with certainty 1p/19q codeletion in the context of imbalanced aneuploidy, polyploidy, or polysomy [7,9]. Other molecular methods, such as PCR-based LOH analysis, have higher specificity, but are more labour-intense and require a non-neoplastic control [7]. Therefore, the identification of surrogate markers of 1p/19q codeletion based on immunostaining may facilitate the differential diagnosis between astrocytoma and oligodendroglioma in routine practice.
In this context, it has been reported that the majority of IDH-mutant astrocytomas of all grades harbour truncating mutations in ATRX, which are mutually exclusive to 1p/19q codeletion and can be detected by the immunohistochemical loss of ATRX protein [2,10,11]. Thus, ATRX immunostaining has been suggested as an alternative low-cost marker to d i f f e r e n t i a t e I D H -m u t a n t a s t r o c y t o m a s f r o m oligodendrogliomas [2,10,11]. Nonetheless, a proportion of IDH-mutant astrocytic tumours retain ATRX immunostaining [2,[12][13][14], and ATRX immunostaining has also been reported as non-conclusive in 11.5% of cases [14].
The immunohistochemical loss of histone 3 trimethylated in lysine 27 (H3K27me3) has been proposed to differentiate astrocytic from oligodendroglial neoplasms: diffuse gliomas can be classified astrocytic if retaining H3K27me3 expression, and oligodendroglial when showing H3K27me3 loss in association with mutated IDH and retained or non-conclusive ATRX staining [14]. This proposal stemmed from the finding of the immunohistochemical loss of H3K27me3 in 25/26 IDH-mutant and 1p/19q codeleted oligodendrogliomas and its retention in 120/135 astrocytomas [14]. A following study on 101 gliomas confirmed the significant association between H3K27me3 loss and 1p/19q codeletion but questioned its sensitivity and specificity [15].
We analysed H3K27me3 immunohistochemical expression in 69 diffuse gliomas with oligodendroglial or mixed oligoastrocytic morphology with the aim to clarify its diagnostic and prognostic potential, by assessing its correlation with 1p/19q codeletion and patients' recurrence-free survival (RFS).

Ethics
The study was approved by the local Ethics Committee (Comitato Etico per la Sperimentazione Clinica delle province di Verona e Rovigo; Prot. n. 11335, 02/26/2019) and performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.

Clinical data
Information on tumour localization, extent of surgical resection, and development of recurrence was retrieved using clinical records. Recurrence was defined by means of computerized tomography or magnetic resonance imaging as identification of tumour growth at the site of previous surgery, or as a volumetric increase of tumour residue in case of subtotal or partial surgery. RFS was defined as the length of survival to the detection of a recurrent tumour.
ATRX expression was considered (i) retained, when nuclear staining was observed in both normal (endothelium, neurons) and neoplastic cells; (ii) lost, when staining was absent in neoplastic cells and present in normal cells; and (iii) nonconclusive, when staining was absent in both normal and neoplastic cells.
P53 immunostaining was assessed and scored as previously reported, and only cases with strong staining in at least 10% of neoplastic cells were rated p53 positive [16].
H3K27me3 immunohistochemical expression was classified according to Peckmezci et al. [15]: (i) retained, when nuclear staining was seen in ≥ 5% neoplastic cells; (ii) lost, when staining was absent in > 95% neoplastic cells and present in internal positive controls (endothelium, neurons); and (iii) non-conclusive, when staining was absent in both normal and neoplastic cells.
Chromosome arms 1p/19q codeletion analysis FISH was used to assess 1p/19q codeletion. For each case, two consecutive 3-μm sections were processed with LSI 1p36/19q13 Dual-Colour Probe Sets assay (Vysis/Abbott, Molecular Europe, Wiesbaden, Germany), following manufacturer's protocol. Slides were examined by Olympus BX61 fluorescence microscope equipped with a ×100 oil immersion objective and a triple band pass filter for simultaneous detection of Spectrum Orange, Spectrum Green, and DAPI signals. Two hundred non-overlapping nuclei containing a minimum of 2 reference probe signals were counted.
Cases were classified as (i) 1p/19q codeleted, when showing two reference probe signals (1q and 19p) and one target probe signal (1p and 19q) in at least 50% cells; (ii) noncodeleted, when having two reference probe signals (1q and 19p) and two target probe signals in > 50% cells; and (iii) imbalanced, when > 50% cells had a ratio of reference to probe signal different from 2/2 or 2/1 [18].
Cases with imbalanced 1p/19q status or IDH-wt and1p/19q codeleted at FISH analysis were further analysed using loss of heterozygosity (LOH) analysis. In detail, LOH was performed using tumour and normal DNA pairs extracted from paraffinembedded tissue sections and evaluated by PCR-based LOH assays. Allelic loss analysis was performed using the microsatellite markers D1S508, D1S199, and D1S2734 on chromosome 1p, and D19S412, D19S112, and D19S219 on chromosome 19q. Forward primers were synthesized either with fluorescent tag FAM or HEX. PCR products were subjected to electrophoresis on an Applied Biosystems 310 automated DNA sequencer (Applied Biosystems, Italy) and fluorescent signals analysed using GeneScan software (Applied Biosystems). Allelic imbalance was evaluated by comparing PCR products from tumour and normal DNA. Peak height ratio was calculated, and allelic imbalance resulted from the ratio of normal to tumour signal (N1/N2 over T1/T2). LOH was assigned for values less than or equal to 0.5 and 1.5. Procedures were repeated three times and allelic losses assigned only upon consistency of the findings.

MGMT promoter methylation analysis
The methylation status of O-6-methylguanine-DNA methyltransferase (MGMT) promoter was assessed in all cases. Briefly, 200 ng of DNA was incubated using sodium bisulphate included in EpiTect Plus FFPE Bisulfite Kit (Qiagen) and analysed using pyrosequencing through MGMT Plus kit (Diatech). The assay performs a quantitative analysis of the percentage of methylation of each of the 10 CpG islands located on chr10. Samples were stratified in four groups according to the extension of methylation status that has been related to clinical outcome: unmethylated (< 9%), low (range 9-20%), medium (21-35%), and high level (over 35%) [19]. Considering all the 69 tumours, we also calculated the sensitivity and specificity of ATRX− vs H3K27me3+ for the identification of astrocytomas.

Statistical analysis
RFS was assessed by the Kaplan-Meier method, with the date of primary surgery as the entry data and length of survival to the detection of a recurrent tumour as the end point. The Mantel-Cox log-rank test was applied to assess the strength of association between RFS and each of the parameters as a single variable. A probability (P) value less than 0.05 was considered significant. Statistical analyses were performed using MedCalc 12.1.4.0 statistical software (MedCalc Software, Mariakerke, Belgium).

Results
The results are detailed in Fig. 1 and summarized in Supplementary Figure 1.

Cases
All tumours were newly diagnosed gliomas with cerebral lobe localization, except for case 56, which was localized in the cerebellum. All neoplasms had gross total resection with the exception of case 15 that had subtotal resection. No patient had been treated with neo-adjuvant therapies. Forty-five patients were males and 24 females with mean age of 45 ± 11.6 years.

IDH mutational status
Sixty-six of the 69 gliomas were IDH-mutant: 61 had IDH1 R132H mutations detected by immunohistochemistry (IHC); 5 had IDH2 mutations detected at sequencing analysis, including 2 ODs and 1 AOD with a R172K mutation and 2 AODs with a R172M mutation.

1p/19q codeletion by FISH and LOH analysis
Sixty of the 66 IDH-mutant gliomas were classified 1p/19q codeleted based on FISH and/or LOH analysis.
IDH-mutant tumours with non-conclusive ATRX staining were excluded from the calculation.

Proposed diagnostic algorithm
Based on our results, we propose the diagnostic algorithm illustrated in Fig. 3. IDH1-R132H, ATRX, and H3K27me3 immunostainings should be assessed first. Then, based on the immunohistochemical results, (i) cases with ATRX loss are classified astrocytic; (ii) cases positive for IDH1-R132H with retained ATRX and lost H3K27me3 are classified oligodendroglial; and (iii) cases positive for IDH1-R132H with nonconclusive ATRX or retained ATRX and H3K27me3 should be tested for 1p/19q codeletion. Cases with negative IDH1-R132H staining should be analysed for IDH1/2 mutations to complete the evaluation.
Applying this algorithm to our series, 47 (68%) cases would have been classified astrocytic or oligodendroglial using only immunohistochemistry for IDH1-R132H, ATRX, and H3K2727me3, 8 (12%) cases would have been resolved with IDH1/2 molecular testing, while only 14 (20%) of the 69 cases would have requested the assessment of 1p/19q codeletion by FISH or PCR-microsatellite LOH analysis (Supplementary Figure 2).

Recurrence-free survival analysis
Follow-up data were available for 50 patients. Seven patients had a recurrence, with a RFS ranging between 18 and 47 months (median: 30 months; inter-quartile range: 24-41 months). The remaining 43 patients were free of recurrence in a follow-up period ranging between 6 and 80 months  (Fig. 1). Of the 7 gliomas that relapsed, 5 retained H3K27me3 immunohistochemical expression (P < 0.0001) (Supplementary Table 1).

Discussion
The diagnosis of oligodendroglioma requires the coexistence of IDH1/2 mutation and 1p/19q codeletion [1]. This latter is assessed in most pathology laboratories by FISH, which is time and expertise demanding [7]. Thus, finding surrogate immunohistochemical markers of 1p/19q codeletion could avoid expensive and unnecessary tests [20]. The results of our study suggest that this can be achieved applying an algorithm that starts with the assessment of IDH mutations associated with ATRX and H3K27me3 immunostainings. This approach would reserve 1p/19q codeletion testing only to IDH-mutant gliomas with retention of both ATRX and H3K27me3 or non-conclusive ATRX immunostainings.
The proposed algorithm is based on our observation that ATRX loss was 100% specific to the astrocytic phenotype, which confirms the data repeatedly reported in the literature [2,13,21], and that the co-occurrence of retained ATRX and loss of H3K27me3 in IDH-mutant gliomas was 100% specific to oligodendroglial phenotype.
The utility of H3K27me3 immunostaining in the differential diagnosis between oligodendroglioma and astrocytoma was first reported by Filipski et al. [14], who suggested that H3K27me3 retention indicates astrocytic (non-codeleted) tumours with a predicted probability of 0.9995, while its loss in IDH-mutant gliomas with retained or non-conclusive ATRX stain is 100% specific for oligodendrogliomas [14].   Data availability Data will be available upon request to the corresponding author.

Conflict of interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.