CDKN2A/B deletion in IDH-mutant astrocytomas: An evaluation by Fluorescence In-Situ Hybridization

Introduction: CDKN2A/B homozygous-deletion is one of the de�ning features of grade 4 in IDH-mutant astrocytic tumours. Aim: To evaluate CDKN2A/B-deletion in IDH-mutant astrocytic tumours and its clinico-pathological impact. Materials and methods: CDKN2A/B-deletion was evaluated by Fluorescence in-situ hybridisation (FISH) and interpreted by Ambros et al. (method 1) and Marker et al. (method 2) methods. Results: Eighty-three out of 94 cases (histologically - grade 2: 3, grade 3: 46, grade 4: 34) were interpretable on FISH. Concordant CDKN2A/B-deletion was observed in 71% (27/38) of lower-grade tumours (n = 49) and 90% (27/30) of histological grade 4 tumours. Both the interpretation methods showed good agreement (Kappa = 0.75). CDKN2A/B-deletion showed inverse correlation for < 10% MIB-1 labeling index (p = 0.01) while by method-2 showed signi�cant correlation for grade 4 (p = 0.02). No signi�cant correlation was observed for any other clinico-pathological parameters. Twenty-four patients showed progression/recurrence (including deaths), no signi�cant difference in frequency of CDKN2A/B deletion was observed among cases with disease progression across different histological grades. Conclusions: CDKN2A/B-deletion was observed across all the histological grades of IDH-mutant astrocytic tumours, expectedly more in higher grade. FISH, as a method can be used for detection of CDKN2A/B homozygous-deletion, when there is concordant interpretation.


Introduction
Cyclin Dependent Kinase inhibitor (CDKN2A/B) genes, play a pivotal role in the surveillance of cell cycle progression via p16/INK4a, p14/ARF and p15 proteins [1].Inactivating alterations like homo/hemizygous deletion, point mutation, epigenetic changes etc. of these genes can induce unabated proliferative signalling, thus facilitating tumourigenesis [2].In IDH-mutant diffuse gliomas, CDKN2A/B homozygous deletion confers aggressive biological behaviour, is one of the de ning features for grade 4 in the current (5th edition) WHO classi cation of CNS tumours [3,4].
The latter methods are superior; while FISH is considered less accurate but is easier to implement.
This study has been undertaken to evaluate the e cacy of FISH for detection (comparing the two more recently accepted methods of interpretation) of CDKN2A/B deletion and its clinico-pathological impact across the all grades of IDH-mutant astrocytomas.

Cases
IDH-mutant astrocytomas of all the histological grades with immunohistochemical loss of alpha thalassemia/mental retardation syndrome X-linked (ATRX) protein expression diagnosed during June 2019-December 2020, formed the study cohort.
Immunohistochemistry for IDH1R132H (1:100 dilution; H09; Dianova; Germany) and ATRX (1:750 dilution; Polyclonal; Sigma; USA) was performed using Ventana Benchmark XT autoimmunostainer.IDH1R132H immunohistochemistry negative cases were subsequently evaluated by targeted Sanger sequencing for other IDH1/2 mutations.In these cases, DNA was extracted from the representative formalin xed para n embedded (FFPE) tissue sample using the QIAamp DNA FFPE Tissue kit (Qiagen, Germany), quanti ed using the Nano Drop spectrophotometer (Thermo Scienti c, USA).PCR was performed using appropriate primers for exon 4 of IDH1R132 and IDH2R172.Sanger sequencing was performed using a BigDye Terminator Cycle Sequencing Kit v3.1 (Applied Biosystems, USA) on ABI3500 Genetic Analyzer (Applied Biosystems, USA).The electropherograms were analysed by Sequence analysis software Chromas 4.3.4.

Evaluation of CDKN2A/B deletion by uorescence in-situ hybridisation (FISH)
All the cases were evaluated using 'ZytoLight SPEC CDKN2A/CEN9 Dual Color Probe', on the positively charged slides with 3-4µ thick sections from the representative formalin-xed para n-embedded (FFPE) tumour tissue blocks.The probe was a mixture of green uorochrome labelled polynucleotides targeting 9p21.3 (CDKN2A and CDKN2B genes) and orange uorochrome labelled polynucleotides targeting 9q12 (centromere of chromosome 9).Orange and green signals were evaluated in the vascular endothelial cells (an internal control) to rule out hybridization failure.When both green and orange signals were absent in the tumour as well as endothelial cell nuclei, the case was considered as 'uninterpretable'.The cases with interpretable signals in ≥ 80% of the tumour cells were considered interpretable and a minimum 100 non-overlapping tumour cell nuclei with recordable signals were assessed.The number of green and orange signals were recorded in each individual tumour cell nucleus.
Two different interpretation methods were employed and criteria for those were as follows: Method-1: Criteria employed by Ambros et al. [5,6] Cases were classi ed into the following groups: True deletion: Green to orange ratio < 0.8 and > 50% of nuclei harbouring ≤ 1 green signal in presence of orange signal/s.Imbalance: Green to orange ratio < 0.8 and ≤ 50% of nuclei harbouring ≤ 1 green signal in presence of orange signal/s.
No-deletion: Green to orange ratio ≥ 0.8.Method-2: Criteria employed by Marker et al. [7] The tumour cell nuclei signal patterns were initially categorised as follows (Fig. 1 Based on the percentage of tumour nuclei in 'homozygous deletion' signal pattern group, the cases were classi ed into two following categories: Homozygous deletion: ≥30% tumour nuclei in the homozygous deletion group; irrespective of green to orange ratio.
No-deletion: <30% tumour nuclei in the homozygous deletion group; irrespective of green to orange ratio.

Pathological and clinical parameters
Histological and immunohistochemical features (grade & MIB-1 labeling index), type of IDH-mutation, MGMTpromoter methylation status (by gel-based methylation speci c PCR method) [8], and clinical features (age, gender, follow-up details) were recorded.The overall survival (OS) was calculated from the date of primary surgery to the last date of follow-up while progression-free survival (PFS) was calculated from the date of primary surgery to the date of recurrence/progression or death; whichever is earlier.

Statistical Analysis
The statistical analysis was performed using the IBM SPSS software version 26.Comparative evaluation between CDKN2A/B deletion and clinicopathological parameters by applying chi-square test and Fisher's exact test for univariate analysis.OS and PFS were calculated using reverse Kaplan-Meier method and correlated with CDKN2A/B deletion using log rank test.
Correlation with clinical and pathological parameters (n = 83): The homozygous deletion group of method-2 showed signi cant association for grade 4 astrocytomas than those of lower histological grade (p = 0.02).However, the same was not observed for deletion group of method- MGMT-promoter methylation status was available for 17 cases (all grade 4), 10 were methylated and 7 unmethylated.MGMT-promoter methylation did not show signi cant statistical correlation with CDKN2A/B deletion (p = 1.00 and p = 0.54 when interpreted by method 1 and 2 respectively).
Eighty-two patients had undergone debulking surgery and only single patient had undergone stereotactic biopsy.Adjuvant treatment details were available in 61 patients, all had received concurrent radiotherapy and temozolomide; while 60 had received adjuvant chemotherapy (in the form of single agent temozolomide in 57 patients; PCV regimen in 3 patients).
Data of follow-up period of > 6 months was available in 75 cases.Median follow-up period was 21 months, with IQR of 12-29 months.Twenty-one patients were deceased while 54 patients were alive (3 with disease progression) on the last date of follow up.Of the 24 cases with recurrence/progression (including deaths), 1 was of histological grade 2 (showed concordant 'no-deletion'), 6 were of histological grade 3 (concordant deletion and no-deletion in 3 and 1 case/s respectively, and 17 were of grade 4 (concordant deletion and nodeletion in 15 and 1 case/s respectively (Table 3).No statistically signi cant difference was observed in the frequency of CDKN2A/B deletion for the groups with and without documented disease progression (including deaths) across different histological grades when results were interpreted by either of the methods (p = 0.74, p = 0.27) and also among concordant cases (p = 0.52).
Median OS (n = 58) and PFS (n = 14) was 75.14 months and 45.17 months respectively (mean OS: 60.36 months, Mean PFS: 42.29 months).There was no signi cant correlation for CDKN2A/B deletion by either of the methods without consideration of histological grade (p = 0.30 and p = 0.13 by method 1 and method 2 respectively).Correlation of CDKN2A/B deletion among lower grade IDH-mutant astrocytomas could not be assessed due to lack of uniform and detailed follow-up data.

Discussion
The CDKN2A/B deletion is known to reduce both PFS and OS, especially in patients with adult-type lower-grade IDH-mutant diffuse gliomas [9][10][11][12][13].It has been assessed using various methods, however, practical challenges in the implementation of appropriate method remain largely unaddressed in the consensus literature.
FISH is a commonly used method to detect chromosomal copy number variation and structural alterations in routine diagnostics of glial tumours and also has been used for detection of CDKN2A/B deletion in various studies [2, 7, 9, 11, 14-18].However, non-uniformity exists among various studies in terms of type of tissue (FFPE/fresh/frozen) used, number of nuclei in which signals were counted and the cut-off values for determining CDKN2A/B deletion (Table 4).The cut-off values for abnormal CDKN2A/CEN9 ratio as that beyond +/-2 standard deviation have been derived, based on the signal pattern observed in the autopsy brain tissue, or comparing the results with another testing method such as quantitative PCR or DNA microarray [14].In few studies, cut-off values are derived using multivariate analysis models taking natural history of disease and prognosis into consideration [7].This study was undertaken to understand the impact of two different interpretation cut-off values used.methods, but there are enough cases with discordance, especially in lower-grade, where it is all the more important.The cohort of imbalance cases of method 1 (n = 13) showed homozygous deletion (n = 8) and nodeletion (n = 5) with method 2. Two cases of deletion by method 1 showed contradictory observation of nodeletion by method-2 [Table 2 (a)].This variance/discordance can be due to hemizygous or complex deletion pattern of signals in these cases, as the interpretation criteria used in method-2 essentially incorporates homozygous deletion pattern of signals; while method-1 encompass all patterns.Thus, the cohort of discordant and imbalance cases probably represent the cases with complex deletion or those which are borderline/uncertain for CDKN2A/B deletion and may require other advance molecular methods for con rmation.Whereas, when the interpretation is concordant between the two methods, then probably can be considered as true for CDKN2A/B deletion.
On excluding the cases with imbalance and discordant results, 1/3 histologically grade 2 and, 26/46 histologically grade 3 IDH-mutant astrocytomas showed concordant deletion.Thus, on applying the current WHO criteria, grade upgradation of lower-grade IDH-mutant astrocytomas would be 55.1%.As compared to prior studies (Table 4), the higher upgradation rate in this study may be explained by very low representation of lower-grade tumours, possibly represent more of peripheral in ltrating part confounding the interpretation of grade, and probably different racial/geographical distribution.
The incidence of CDKN2A/B homozygous deletion in IDH-mutant gliomas range from 9 to 43% (median 22%) [12].Korshunov et al. have shown homozygous deletion in 43% of grade 4 IDH-mutant astrocytomas, whereas in the current study, 27/30 (90%) grade 4 IDH-mutant astrocytomas showed concordant deletion [19].This frequency of CDKN2A/B deletion in grade 4 tumours in the current study, is exceedingly high as compared to the contemporary literature and is not explainable, would need to be established in much larger cohort.In the current study, a negative correlation was observed for MIB-1 LI of < 10%, which probably can be used to triage for selection of the cases for testing CDKN2A/B deletion in resource constraint setting.
CDKN2A/B deletion is now well-known poor prognostic factor in astrocytomas.However, this study has failed to establish signi cant correlation between OS/PFS and CDKN2A/B deletion, even among cases with concordant results; likely due to non-uniform follow-up details, assumptive OS/PFS calculations and heterogeneity in the cases (i.e. the study cohort is more of higher grade tumours), thus cannot further comment on the signi cance of FISH testing for CDKN2A/B deletion, which can be answered in well controlled prospective studies.
Similarly, loss of expression for p16 protein on immunohistochemistry has been shown to have high sensitivity and negative predictive value with a limited speci city for detecting CDKN2A/B deletion [2,20,21].Speci city for predicting CDKN2A/B homozygous deletion is better when < 5% p16 positive cells nuclei and excluding homozygous deletion with > 20% positive cells [22].However, loss of p16 expression can be seen because of different molecular alterations (including point mutation, promoter hypermethylation) apart from homozygous deletion.Thus, inconsistency in association of loss of p16 expression with worse patient outcome exists in the literature [9,12,16].Also, it may become di cult to interpret due to marked admixture of reactive glial cells especially in lower-grade or peripheral in ltrating part of the gliomas limiting practical utility of p16 immunohistochemistry. Recently, the loss of MTAP immunostaining has been shown to have good positive predictive value and better speci city than p16 immunohistochemistry for detecting homozygous CDKN2A/B deletion.However, the literature addressing utility in gliomas is sparse [15,23].
Quantitative PCR and FISH are considered reliable techniques, with certain limitations.PCR is likely to miss small deletion when there is signi cant admixture of normal cells or clonal heterogeneity and FISH may miss small deletion unless cosmid probes are used.Perry et al. showed concordance rate between CGH array and FISH or PCR as 94% and 100% respectively [14].CDKN2A/B deletion also showed correlation with reduced OS when assessed using FISH [7,9,12].
To conclude, CDKN2A/B deletion noted across all grades of IDH-mutant astrocytomas, more frequent in higher grade astrocytomas (i.e.histological grade 3 and 4).The trend suggests a possible negative association for MIB-1 LI of < 10%.It has no correlation with the type of IDH1R132 mutations and there is no difference in the disease recurrence patterns among cases having CDKN2A/B deletion and those without deletion.
FISH as a methodology for detecting CDKN2A/B deletion need to be clinically validated in a larger cohort of samples before considering this as a clinically acceptable modality in patient management.However, based on this study FISH can be used as a method to assess the deletion, but the interpretation of deletion should be made when both the criteria for deletion by methods 1 & 2 are ful lled.Discordant results, however should be tested using more sensitive and speci c methods for clinical applications.
): Homozygous deletion: No green signals seen in presence of ≥ 1 orange signal Hemizygous deletion: Single green signal with two orange signals Complex deletion: ≤2 green signals and > 2 orange signals Hyperploid: ≥3 green and orange signals and/or more green signals as compared to orange signals Normal: Equal proportion of orange and green signals seen.

Declarations
This work was nancially supported by neuro-oncology disease management group and institutional ethics committee of Tata Memorial Hospital, Mumbai.This study was for presented (as an oral presentation) at International Congress of Neuropathology (ICN 2023) held at Berlin, Germany in September 2023.Author (Dr.Manali) has received travel grants from organizing committee of ICN 2023 to encourage presentation of this study.

Table 1
Distribution of clinicopathological parameters across CDKN2A/B deleted and non-deleted cases.

Table 2 (
a & b): Comparison of CDKN2A/B FISH results when interpreted by two methods and concordance for deletion and no-deletion by method-1 and method-2 viewpoint.
*Excluding cases with imbalance by method-1 (p = 0.07).No signi cant difference in frequency of CDKN2A/B deletion was observed between grade 2 and grade 3 tumours when deletion was interpreted by either of the methods (p = 0.06; p = 0.20 respectively).
1, irrespective of combining imbalance with deletion (p = 0.14) or no-deletion (p = 0.69).The concordant CDKN2A/B deletion group also did not show signi cant association with grade 4 astrocytomas than those of lower histological grade

Table 3
Grade-wise distribution of survival and disease progression (including deaths).

Table 4
Overview of studies utilising FISH for assessing CDKN2A/B deletion in IDH-mutant Astrocytomas.Even though the current study showed good agreement (kappa = 0.75) between Ambros et al. andMarker et al.