EGFR/CEP7 high polysomy is separate and distinct from EGFR amplification in glioblastoma as determined by fluorescence in situ hybridization

Abstract EGFR amplification in gliomas is commonly defined by an EGFR/CEP7 ratio of ≥2. In testing performed at a major reference laboratory, a small subset of patients had ≥5 copies of both EGFR and CEP7 yet were not amplified by the EGFR/CEP7 ratio and were designated high polysomy cases. To determine whether these tumors are more closely related to traditionally defined EGFR-amplified or nonamplified gliomas, a retrospective search identified 22 out of 1143 (1.9%) gliomas with an average of ≥5 copies/cell of EGFR and CEP7 with an EGFR/CEP7 ratio of <2 displaying high polysomy. Of these cases, 4 had insufficient clinicopathologic data to include in additional analysis, 15 were glioblastomas, 2 were IDH-mutant astrocytomas, and 1 was a high-grade glial neoplasm, NOS. Next-generation sequencing available on 3 cases demonstrated one with a TERT promoter mutation, TP53 mutations in all cases, and no EGFR mutations or amplifications, which most closely matched the nonamplified cases. The median overall survival times were 42.86, 66.07, and 41.14 weeks for amplified, highly polysomic, and nonamplified, respectively, and were not significantly different (p =  0.3410). High chromosome 7 polysomic gliomas are rare but our data suggest that they may be biologically similar to nonamplified gliomas.


INTRODUCTION
The 2021 CAP guidelines on molecular biomarker testing for the diagnosis of glioblastoma (GBM) uses EGFR amplification, TERT promoter mutation status, and gain of chromosome 7 with concurrent loss of chromosome10 to help diagnose gliomas that lack necrosis or microvascular proliferation.However, no standardized guidelines exist for determining EGFR amplification or gain of chromosome 7 (1).Since copy number gains of chromosome 7 were first described in astrocytic neoplasms (2), the role of trisomy/polysomy of chromosome 7 as it relates to prognosis has been disputed (2-8).Recent advances have demonstrated that chromosome 7 copy number gain is most useful when considered in the context of the genomic landscape of individual tumors (8).Thus, it becomes essential to have well established guidelines on how to designate polysomy 7. Of equal importance is understanding how EGFR amplification occurs and can be accurately quantified, particularly as it relates to polysomy of chromosome 7.A study performed by Bie� nkowski et al in 2013 utilized qPCR and fluorescence in situ hybridization (FISH) to establish EGFR amplification and polysomy 7 disparately.They determined that increased EGFR amplification was associated with a worse prognosis when in combination with chromosome 7 polysomy as opposed to chromosome 7 polysomy without a relatively increased EGFR amplicon (4).Additional studies have provided support that EGFR amplification and/or polysomy 7 result in differing prognoses and have highlighted the need to determine whether EGFR copy number is increased via gene amplification or polysomy (4-8).The finding that EGFR amplification in and of itself has the highest specificity for glioblastoma lends additional support to the need to determine specific cutoffs when using FISH to establish the presence of an EGFR amplification versus a chromosome 7 polysomy without EGFR amplification (8).Clearly defined criteria for establishing the presence of EGFR amplification can help standardize the diagnosis of GBM and may be critically important as additional therapies targeting the EGFR tyrosine kinase receptor are developed.
Given this lack of guidelines, a wide range of criteria have been used by different institutions and research studies over the years with the most common amplification criteria utilized being an EGFR/CEP7 copy number ratio of �2.0 as assessed by .A recent study compared EGFR FISH with real-time quantitative PCR data in an attempt to determine clinical trial inclusion and suggested using a FISH cut-off of EGFR/CEP7 ratio of �2.0 with at least 50% of nuclei demonstrating EGFR amplification.However, they considered tumors with chromosome 7 polysomy without focal amplification of EGFR as nonamplified and did not further investigate or address this population (12).EGFR has been shown to be amplified in 52% of GBM and in 44% of these cases it is amplified in conjunction with trisomy/polysomy 7 (6).Upon review of the literature, most studies use trisomy and polysomy interchangeably.While CEP7 trisomy is common in gliomas a small population of cases seen at our major reference laboratory display much higher copies numbers (�5) of both EGFR and CEP7 that do not meet the �2.0 EGFR/CEP7 ratio requirement.While the interchangeability of trisomy and polysomy terminology makes evaluation difficult it appears that this population has not been well studied or characterized in gliomas.
Herein, we sought to evaluate the rare cases with an EGFR/ CEP7 ratio of <2 but a copy number of �5 for both EGFR and CEP7 and determine whether a cutoff ratio of �2 would correctly classify these tumors according to tumor biology.

Sample selection
A retrospective search of brain samples submitted for EGFR FISH performed at a major reference laboratory was performed.A total of 1143 cases were identified.The parametric 95% reference interval upper limit was found to be 5.2 copies of EGFR and CEP7 and we rounded down to the closest whole copy number (�5 copies) and utilized that as our cutoff criteria to capture this highly polysomic outlier population.Statistical analysis was performed on the primary GBM WHO grade 4 highly polysomic cases using an in-house control group of primary GBM WHO grade 4 cases that were submitted for EGFR FISH testing over the same time period (n ¼ 307).Patient information, tumor characteristics, molecular test results, and clinical outcome data were collected and analyzed.All patients had a minimum of 1-year follow-up since the time of diagnosis.Survival times were calculated from the day of original biopsy or resection to the death date.

Fluorescence in situ hybridization
EGFR FISH was performed on 3-to 5-mm-thick formalinfixed, paraffin-embedded (FFPE) tissues.After pretreatment on a Dako PT Link a pepsin digestion step (Agilent Technologies, Santa Clara, CA) was done on a VIP2000.The slides were treated with probes from Abbott Laboratories LSI EGFR SpectrumOrange and LSI CEP7 SpectrumGreen probe.The slides were hybridized on a Thermobrite machine (Abbott Molecular, Des Plaines, IL).Following hybridization, slides were washed with Dako Stringency buffer and dehydrated through a series of ethanol solutions.They were counterstained with DAPI and coverslipped.Each FISH slide was scanned manually in its entirety and a minimum of 50 representative tumor nuclei were enumerated on each case by a technologist and reviewed by a pathologist with expertise in solid tumor FISH.

Immunohistochemistry
IDH1 R132H immunohistochemistry (IHC) staining on tumor tissue sections was performed using the monoclonal mouse H09 antibody clone by Dianova (Geneva, Switzerland) at a 1:200-1:400 concentration on either a Ventana Bench-Mark Ultra or a Leica Bond Autostainer.

Molecular studies
Tumor tissue with adequate tumor cellularity, as evaluated by anatomic and molecular oncologic pathologists, was selected for DNA extraction from FFPE tissue sections.

MGMT methylation
MGMT methylation status was determined by performing a bisulfite conversion on extracted DNA before using PCR amplification and bisulfite-specific primers targeting the MGMT promoter region.Methylation was then detected using either a LightCycler PCR instrument (Roche Diagnostic, Switzerland) or a MassArray MALDI-TOF mass spectrometry (Agena Bioscience, San Diego, CA).

Mutational analysis
Pyrosequencing of extracted DNA to detect IDH1 and IDH2 exon 4 mutations was performed on the Qiagen PyroMark Q24 Pyrosequencer (Qiagen, Germantown, MD).

Next-generation sequencing
All next-generation sequencing (NGS) testing was performed by accredited outside reference laboratories.

Statistics
The statistical analysis was performed using R (R Core Team, 2022) (18).Kruskal-Wallis tests were run to compare overall survival and age, while Wilcoxon tests were run for pairwise comparisons of age and sex (adjusted for multiple comparisons using the Holm method).

RESULTS
With our selection criteria, 1143 cases were identified.

Control cohort
There was a total of 307 IDH-wildtype glioblastoma, WHO grade 4 control cases that were diagnosed during the same time as our high polysomy GBM cases.This control group was composed of 132 EGFR-amplified and 175 EGFR-nonamplified cases.

Demographics
The median age between the 3 groups of amplified, highly polysomic, and nonamplified were 62, 50.5, and 62 years, respectively.Age was statistically significantly different between both amplified and highly polysomic groups (p ¼ 0.031) and between highly polysomic and nonamplified groups (p ¼ 0.036), but not between the amplified and nonamplified groups (p ¼ 0.906).Highly polysomic and nonamplified cases had a broader range of ages (20-85 and 18-87 years old, respectively), compared to the amplified cases (29-84 years old).There were also no significant differences in age when compared to sex for the 3 populations (Fig. 3).

Histopathology
The majority of the highly polysomic GBMs without a prior CNS diagnosis had necrosis (n ¼ 11) and microvascular proliferation (n ¼ 12); however, there were 2 cases (2/14, 14.3%) that lacked both necrosis and microvascular proliferation.One of these 2 cases was reported as a GBM due to the dramatically elevated EGFR copy number of 10.5 with concurrent CEP7 copy number of 12.2, thus with a ratio of 0.9.This result was the only finding used to establish the diagnosis of GBM and highlights the difficulty and necessity of establishing more rigorous criteria for EGFR amplification.The other case was reported as a GBM from an outside institution; the reasoning for the diagnosis was not provided, perhaps necrosis and/or microvascular proliferation was present on an alternative slide from the one we were sent or one of the other 2 GBM-defining molecular alterations (i.e.TERT promoter mutation or a gain in chromosome 7 copy number with concurrent loss of chromosome 10), was known but not provided to our institution.

Molecular characteristics
NGS was available on 3 highly polysomic cases, 22 amplified, and 34 nonamplified cases.All 3 of the highly polysomic cases had a TP53 mutation and only 1 of the 3 cases carried a TERT promoter mutation.).Two of the 3 nonamplified EGFR-mutated cases by FISH correlated with EGFR-amplification by NGS.Upon investigation, their EGFR nonamplification FISH status had been determined on the initial resection and the NGS had been performed on recurrences/resections 5-6 months later.In that time, it is possible that the tumor developed a detectable population with amplification of EGFR.
The mutational burden in the highly polysomic cases was 1.7, 2.5, and 15.8 m/MB.The mutational burden in the amplified and nonamplified groups ranged from 1.0 to 8.0 (mean 4.1) and 0.0 to 34.0 (mean 4.7), respectively.The NGS testing also showed a germline MYH11 p. A1414T mutation in one highly polysomic patient that would correspond to Marfan/Loeys-Dietz syndrome.
Histological features were evaluated in the highly polysomic cases but there were no consistent defining features.However, several cases demonstrated large atypical or giant cell features and/or bizarre multinucleate forms/mitoses (Fig. 4).

Survival
Survival data were available for 102 amplified, 5 highly polysomic, and 131 nonamplified cases (Fig. 5).The median survival was 42.86, 66.07, and 41.14 weeks for the amplified, highly EGFR/CEP7 High Polysomy Glioblastomas � 341 polysomic, and nonamplified groups, respectively.While the median survival was longer for the highly polysomic group, survival times were not significantly different between groups (p ¼ 0.341).

Glioblastoma, IDH-wildtype arising from a low-grade glioneuronal tumor
One of the highly polysomic cases (1/18, 5.6%) was a GBM that appeared to arise from a low-grade glioneuronal tumor diagnosed 16 years prior to the GBM resection in a patient with neurofibromatosis type 1.The GBM demonstrated microvascular proliferation but no necrosis.Unfortunately, no NGS or additional molecular information was available on this tumor.

IDH-mutant astrocytoma results
Two of the highly polysomic cases (2/18, 1.1%) were IDHmutant astrocytomas.IDH1 status was determined using IDH1 R132H IHC.Both cases demonstrated necrosis, although it was focal in one of the cases and that case also showed some features suggestive of a primitive neuronal component.No NGS or additional molecular information was available on these tumors.

DISCUSSION
Highly polysomic EGFR and CEP7 cases as defined above with an EGFR/CEP7 ratio of <2 and a copy number of �5 for both probes are rare and only comprised �1.9% of cases  submitted to a major reference lab for EGFR FISH testing.Although they demonstrated an elevated copy number of EGFR, in our small sample group they appear to be dissimilar to amplified tumors defined by an EGFR/CEP7 ratio of �2.This is consistent with previously reported studies assessing EGFR amplification as it relates to chromosome 7 copy number (4-7).Highly polysomic cases in this study had a better overall median survival than the traditionally amplified group, although this finding was not statistically significant.High polysomic cases were also seen in younger patients (<29 years old), than cases in the amplified group.However, the sample size of the highly polysomic group (n ¼ 18) was small, which may have limited the statistical power to detect significant differences and similarities between the amplified and nonamplified groups.
Molecularly, the highly polysomic cases were more similar to the EGFR nonamplified group with all 3 of the NGS-tested cases harboring a TP53 mutation and only 1 containing a TERT promoter mutation.Moreover, TP53 mutations and EGFR overexpression have been previously reported to be mutually exclusive (19).This pattern of mutual exclusivity was also seen in our control samples with only 3.6% (2/56) of our NGS-tested control cases having both a reported TP53 mutation and EGFR amplification.
The fact that high polysomy of EGFR and CEP7 was also seen in IDH-mutant astrocytomas, which rarely have an EGFR amplification, and a glioblastoma appearing to evolve from a low-grade glioneuronal tumor would seem to further indicate that high polysomy is not synonymous with EGFR amplification and that an EGFR/CEP7 ratio is more useful for establishing the presence of EGFR amplification by FISH testing than solely relying on the EGFR copy number.
Although the present data are not conclusive, these findings support distinguishing high polysomy cases from outright EGFR amplified cases.We classify high polysomy cases without definitive EGFR amplification as those that have copy numbers of EGFR and CEP7 � 5 with a ratio of <2.In these cases, we recommend not outright designating a case as positive for EGFR amplification but rather as nonamplified or indeterminate with a comment describing that if the diagnosis or treatment depend on the EGFR status additional testing such as DNA sequencing, RNA sequencing, and/or expression profiling should be undertaken to establish the amplified or unamplified status of EGFR.
Limitations of this study included the small overall sample size of 22 high polysomy cases total with only 3 of these having NGS results available; survival data were also available on only a limited subset of cases.However, EGFR-CEP7 high polysomy is extremely rare and thus necessitated including reference laboratory cases, which generally have limited clinical information.Sequencing was not undertaken for our outside cases due to the lack of tissue to perform additional testing, and the tissue being returned to the outside institution; also, there was no availability for patient consent.Additionally, while NGS is worthwhile for looking at mutations across other genes, it can be challenging to ascertain polysomy or EGFR/CEP7 ratios for direct comparisons with FISH.Hence, evaluation of publicly available databases, (although these may be fraught with similar issues to using NGS for a direct head-to-head comparison), or additional multicenter studies with a larger subset of highly polysomic EGFR and CEP7 patients are needed to establish the exact useful EGFR/ CEP7 ratio.However, these methods are outside the scope of the present report.The data presented herein support using an EGFR/CEP7 ratio to establish the presence of EGFR amplification over using the EGFR copy number in isolation and distinguishing in clinical reports between EGFR amplification and EGFR-CEP7 high polysomy.

Figure 2 .
Figure 2. Flowchart of cases.(A) All cases represented from initial identification (n ¼ 1143) to those included in the study.(B) Pie chart of composition of 18 high polysomy cases included in the study.

Figure 3 .
Figure3.Age and sex by EGFR status.Box and whisker plot with female cases in red and male cases in blue.

Figure 4 .
Figure 4. FISH images and corresponding histology in representative tumors.(A, B) Case with average of �7 EGFR CN/cell.(C, D) Case with average of �8 EGFR CN/cell and numerous giant cells.(E, F) Case in which the majority of cells showed �5 EGFR CN/Cell but with �5%-10% of cells that had >10 copies of both EGFR and CEP7 and occasional giant cells.(G, H) Case with average of >10 EGFR and CEP7 CN/cell.Mag: FISH images, 100�; H&E images, 20�.

Figure 5 .
Figure 5. Overall survival by EGFR status.Survival was calculated, in weeks, from date of biopsy/excision to the date of death.
While the sample sizes were limited, these molecular results more closely aligned with the EGFR nonamplified group than the EGFR amplified group because TP53 mutations were seen in only 9% (2/22) of the amplified group but comprised 32.4% (11/34) of the nonamplified group.Conversely, TERT promoter mutations were seen in 95.5% (21/ 22) of the amplified group and only 61.8% (21/34) of the nonamplified group.No EGFR amplifications or mutations were detected in the 3 cases with NGS from the high polysomic group, whereas 21/22 (95.5%) and 2/34(5.9%) of the EGFR amplified and nonamplified FISH group, respectively, were classified as EGFR amplified by NGS.Furthermore, nearly all amplified cases (20/22, 90.9%) also contained an EGFR mutation while the nonamplified group only had 3 cases of an EGFR mutation (3/34, 8.8%