FoundationOne CDx testing accurately determines whole arm 1p19q codeletion status in gliomas

Abstract Background Molecular profiling of gliomas is vital to ensure diagnostic accuracy, inform prognosis, and identify clinical trial options for primary and recurrent tumors. This study aimed to determine the accuracy of reporting the whole arm 1p19q codeletion status from the FoundationOne platform. Methods Testing was performed on glioma samples as part of clinical care and analyzed up to 395 cancer-associated genes (including IDH1/2). The whole arm 1p19q codeletion status was predicted from the same assay using a custom research-use only algorithm, which was validated using 463 glioma samples with available fluorescence in-situ hybridization (FISH) data. For 519 patients with available outcomes data, progression-free and overall survival were assessed based on whole arm 1p19q codeletion status derived from sequencing data. Results Concordance between 1p19q status based on FISH and our algorithm was 96.7% (449/463) with a positive predictive value (PPV) of 100% and a positive percent agreement (PPA) of 91.0%. All discordant samples were positive for codeletion by FISH and harbored genomic alterations inconsistent with oligodendrogliomas. Median overall survival was 168 months for the IDH1/2 mutant, codeleted group, and 122 months for IDH1/2 mutant-only (hazard ratio (HR): 0.42; P < .05). Conclusions 1p19q codeletion status derived from FoundationOne testing is highly concordant with FISH results. Genomic profiling may be a reliable substitute for traditional FISH testing while also providing IDH1/2 status.

• F1CDx testing may be used in preference to FISH.
• F1CDx detects many of the relevant molecular biomarkers for gliomas in one test.
Gliomas are the most common malignant primary brain tumors, 1,2 comprising 26% of all CNS tumors and 81% of malignant tumors. For these patients, outcomes remain poor, with a 2-year survival rate of only 2% in patients greater than 65 years old and 30% in patients under 45 years old. 3 Historically, histopathological criteria alone separated gliomas into diagnostic categories. However, recent updates to the WHO Classification of CNS tumors have emphasized genomic biomarkers, FoundationOne CDx testing accurately determines whole arm 1p19q codeletion status in gliomas particularly IDH1/2, 1p19q co-deletion, and ATRX. This integrated diagnostic workflow has generated the need for more precise diagnostic entities, which stands to improve the clinical outcomes of patients. 4 IDH1/2 mutated gliomas account for approximately 20% of all glioma samples. 5 Oligodendrogliomas are characterized by an IDH1/2 mutation and whole arm 1p19q co-deletion, while adult lower grade (grade II and III) diffuse astrocytomas frequently harbor an IDH1/2 mutation but lack 1p19q codeletion. 4,[6][7][8] Prognostically, the presence of these two markers, characteristic of oligodendrogliomas, confers a favorable prognosis. [9][10][11][12][13][14][15][16][17][18] A study of 149 WHO grade II glioma samples reported that IDH1/2-mutated, 1p19q codeleted gliomas showed longer overall survival (OS) compared to other molecular subtypes (P < .05). 19 Comparisons of histological subtypes were unable to significantly stratify patients by outcomes (P = . 16), emphasizing the value of molecular stratification of diffuse gliomas over traditional histologic strategies. 7,19 Typically, the IDH1/2 mutation status and 1p19q codeletion status are identified through a combination of FISH to test for 1p19q codeletion and sequencing or immunohistochemistry (IHC) to detect mutations in IDH1/2. Notably, IHC testing is specific to only the IDH1 R132H protein isoform which accounts for approximately 80-85% of all IDH1/2 mutations. 20 With regards to FISH, many commercially available results are reported based on a "minimally deleted region" approach. 21 FISH probes are sensitive but not specific for the detection of 1p19q codeletion because FISH is unable to distinguish loss of the whole chromosome arm from a focal deletion. This distinction is important because overall survival is inferior in those with focal as opposed to whole arm 1p19q codeletion. 18,22 In this study, we sought to assess the accuracy of whole arm 1p19q codeletion calling using F1 or F1CDx in gliomas. We tested our algorithm using samples from 463 glioma samples that were sequenced clinically using F1 or F1CDx and compared the predicted codeletion status to that from FISH. Finally, we analyzed clinical outcomes grouped by their IDH1/2 and F1/F1CDx-derived whole arm 1p19q codeletion status.

Methods
F1/F1CDx testing was performed as part of routine clinical care in a College of American Pathologists (CAP)accredited, Clinical Laboratory Improvement Amendments (CLIA)-certified, New York State-regulated reference laboratory (Foundation Medicine, Inc). All samples underwent central histopathologic review by a board-certified neuropathologist (S.H.R.) using World Health Organization criteria. This study was approved by the Western Institutional Review Board (IRB# 20152817) and includes a waiver of informed consent and a HIPAA waiver of authorization.
At least 50 ng of DNA per specimen was extracted from formalin-fixed paraffin-embedded samples from patients with brain tumors. Next-generation sequencing (NGS) was performed using hybridization-captured, adaptor ligation-based libraries to high, uniform coverage in up to 395 cancer-related genes and the intronic regions of 28 genes commonly involved with rearrangement mutations. Glioma samples were sequenced using one of two assays, F1 or F1CDx, thus results were analyzed separately for each assay. The two assays differ in the baitsets used for hybridization capture to enrich for cancer-related genes. Throughout the manuscript, gene alterations are discussed only if the gene is baited on both baitsets.
Sequence data were analyzed for clinically relevant classes of genomic alterations, defined as alterations that are targetable by anticancer drugs currently available on the market or in registered clinical trials. These alterations include base-pair substitutions, insertions/deletions, copy number alterations, rearrangements/fusions. Tumor mutational burden (TMB) was calculated from ~1 MB of the sequenced genome; patients with TMB > 8.7 mutations per megabase were considered "hypermutated" . 23 We ran a custom research-use only algorithm to assess the whole arm 1p19q codeletion from sequencing data of 463 (162 F1, 148 F1CDx, and 153 UCLA sequenced on F1 or F1CDx) glioma samples. A copy number modeling algorithm utilized the coverage data of baited regions of the genome within each sample, normalized to a processmatched control, to model the copy number of each segment. The minor allele frequencies of up to 59,622 single

Importance of the Study
The integration of genomic biomarkers into brain tumor classification has improved the diagnostic accuracy and led to the development of molecularly stratified clinical trials. Particularly, whole arm 1p19q codeletion status is a valuable diagnostic, prognostic and predictive biomarker in gliomas and has traditionally been performed by FISH. However, FISH is unable to differentiate between whole chromosome arm deletions and smaller focal deletions. This distinction is important given the association of 1p19q whole arm codeletion with improved survival. In this work, we show that comprehensive genomic profiling using F1 or F1CDx testing can accurately detect whole arm loss of 1p and/or 19q, in addition to providing information about genomic alterations (eg IDH1/2, pTERT, TP53) within the sample. Thus, F1 or F1CDx testing may be used in preference to FISH, given their ability to detect many of the relevant molecular biomarkers for gliomas in one test with improved accuracy.

Neuro-Oncology Advances
nucleotide polymorphisms (SNPs) distributed across each segment were used to determine the loss of heterozygosity (LOH) status of each segment. The algorithm then calculated the percentage of the 1p and 19q arms that were monoallelic (under LOH). 24 FISH tests for 1p19q codeletion were performed by the submitting institutions and results were abstracted from clinical pathology reports. Concordance of the 1p19q status from F1/F1CDx vs. FISH was calculated.
Clinical outcomes were assessed for 519 neuro-oncology patients seen at UCLA who received Foundation Medicine's genomic profiling between August 2012 and March 2019. Patients were included whether FISH 1p19q testing was performed or not. There is an overlap of 143 patients between the 519 UCLA samples with available clinical outcomes data and the 153 samples used for the comparison against FISH. All patients provided informed consent under a UCLA Institutional Review Board-approved protocol. OS was defined as the time between the date of initial diagnosis and the date of censor/death. Progression-free survival (PFS) was defined as the time between the date of initial diagnosis and the date of tumor progression following standard of care treatment. The response assessment in neuro-oncology (RANO) criteria was used by treating clinicians to determine tumor progression.

Patient Cohort
Our validation study comprised 463 glioma samples (162 from F1 testing, 148 F1CDx testing, and 153 from UCLA) with available FISH results (Table 1). Unlike the UCLA cohort, samples within the F1 and F1CDx cohort underwent primary selection based on availability of 1p19q FISH status in the submitted pathology report. The median age at testing was 44 years. Tumor types analyzed in this study include oligodendroglioma, oligoastrocytoma, astrocytoma, glioblastoma, glioma (not otherwise specified, NOS) as well as rare ependymoma, medulloblastoma, and low-grade glioma/glioneuronal tumors ( Table 1).
The incidence of pathogenic TP53 alterations was investigated across our cohort and were present in 46% (212/463) of all samples; while alterations in ATRX and pTERT were present in 26% (120/463) and 53% (245/463) of all glioma samples, respectively.

Predicting Whole Arm 1p19q Codeletion Status from Foundation Testing
Copy number and LOH status were determined for 1p and 19q in all 463 glioma samples, as illustrated in an example in Figure 2A. A sample was considered computationally 1p19q codeleted if >50% of both the 1p and 19q arms were monoallelic, i.e. lost heterozygosity in 1p and 19q ( Figure  2B). We compared results from our next-generation sequencing algorithm for whole arm 1p19q codeletion to those from FISH, and concordance was assessed ( Figure 3A)

Analysis of Discordant Samples
The genomic profile of all samples used in this study is outlined in Figure 3C. We specifically noted 14 discordant samples, that were called positive for codeletion by FISH and negative by our NGS-based algorithm ( Figure  3D). We saw no evidence to indicate that tumor purity impacted concordance. Six discordant samples were all negative for mutations involving IDH1/2, CIC, and FUBP1 which would be uncharacteristic of oligodendroglial lineage tumors harboring a true whole arm 1p19q codeletion (Table S1). Eight discordant samples were IDH1 mutant. Manual review of the copy number data for these samples revealed that six cases harbored partial or complete loss of one arm. Furthermore, these samples harbored co-occurring alterations involving TP53 and ATRX, a genomic profile characteristic of astrocytic lineage adult diffuse gliomas. These findings suggest that the 14 discordant samples reported as 1p19q codeleted by FISH are not true oligodendrogliomas.

Discussion
Gliomas represent a spectrum of tumors with varying lineages, histologic grades, clinical courses, and prognosis. Based on current WHO guidelines, the distinction between oligodendrogliomas and diffuse astrocytomas necessitates the detection of several molecular markers, primarily the IDH1/2 mutation status and whole arm 1p19q codeletion status. 4 In this study, we investigated the feasibility and accuracy of detecting the whole arm 1p19q codeletion status through comprehensive genomic profiling, instead of the traditional FISH testing. Our results showed that the computationally derived whole arm 1p19q codeletion status was highly concordant with FISH results. This was true when we assessed all glioma samples and when we restricted the analysis to only include IDH1/2 mutated samples. In general, assessment and reporting of the codeletion status should be reserved for cases with IDH1/2 mutations, since it has been shown that 1p19q codeletion status has no impact on the survival of IDH1/2 WT tumors, such as glioblastomas. 26

Neuro-Oncology Advances
alterations (1p and/or 19q loss, 52% and 70% of OAs), or astrocytoma-like alterations (TP53 mutations, 32%). 27 Other studies have indicated gliomas previously characterized as OAs contain subsets driven by mutations to the TERT promoter region. 28 Given that OAs are no longer a recognized diagnostic entity, identifying the molecular subpopulations that exist can inform the practical reclassification of this group.
Importantly, patients with F1/F1CDx-derived whole arm 1p19q codeletion showed increased overall survival compared to non-codeleted counterparts. Specifically, patients with IDH1/2 mutations and whole arm 1p19q codeletion had better overall survival than those with IDH1/2 mutations but no codeletion. This is in line with findings from multiple groups, showing that patients with FISHderived 1p19q codeletion have better survival outcomes compared to patients with non-codeleted tumors. [9][10][11][12][13][14][15][16][17][18] Furthermore, 1p19q codeletion is associated with improved temozolomide (TMZ) sensitivity, and the use of IDH1/2 inhibitors is being investigated in clinical trials. 29,30 Because FISH targets a single locus, positive results may indicate whole or partial arm deletion. The benefit of a comprehensive genomic profiling approach is that it can distinguish between whole and partial arm deletion. This distinction is important given the association of whole arm 1p19q codeletion with improved survival. 22 The frequency of partial 1p19q codeletion is estimated to be at 3.6% 31 and in our study, 6 out of the 14 total discordances were likely due to partial arm deletions, leading to a potential misdiagnosis of these cases. Our results suggest that F1 or F1CDx testing is a reliable substitute for FISH to detect the 1p19q codeletion status given its ability to distinguish partial vs. whole arm loss, in addition to