Sensitive and rapid detection of TERT promoter and IDH mutations in diffuse gliomas

Abstract

Background: Mutations in telomerase reverse transcriptase promoter (TERTp) and isocitrate dehydrogenase 1 and 2 (IDH) offer objective markers to assist in classifying diffuse gliomas into genetic subgroups. However, traditional mutation detection techniques lack sensitivity or have long turnaround times or high costs. We developed GliomaDx, an allele-specific, locked nucleic acid-based quantitative PCR assay to overcome these limitations and sensitively detect TERTp and IDH mutations.

Methods: We evaluated the performance of GliomaDx on cell line DNA and frozen tissue diffuse glioma samples with variable tumor percentage to mimic use in clinical settings and validated low percentage variants using sensitive techniques including droplet digital PCR (ddPCR) and next-generation sequencing. We also developed GliomaDx Nest, which incorporates a high-fidelity multiplex pre-amplification step prior to allele-specific PCR for low-input formalin-fixed paraffin embedded (FFPE) samples.

Results: GliomaDx detects the TERTp and IDH1 alterations at an analytical sensitivity of 0.1% mutant allele fraction, corresponding to 0.2% tumor cellularity. GliomaDx identified TERTp/IDH1 alterations in a cohort of frozen tissue samples with variable tumor percentage of all major diffuse glioma histologic types. GliomaDx Nest is able to detect these hotspot mutations with similar sensitivity from pre-amplified samples and was successfully tested on a cohort of clinical FFPE samples. Testing of a cohort of previously identified TERTpWT-IDHWT gliomas (by Sanger sequencing) revealed that 26.3% harbored low-percentage mutations. Analysis by ddPCR and whole exome sequencing of these tumors confirmed the low mutant fraction of these alterations and overall mutation-based tumor purity.

Conclusions: Our results show that GliomaDx can rapidly detect TERTp/IDH mutations with high sensitivity, identifying cases that might be missed due to the lack of sensitivity of other techniques. This approach may facilitate more objective classification of diffuse glioma samples in clinical settings such as intraoperative diagnosis or in testing cases with low tumor purity.

Keywords: IDH mutation; TERT promoter mutation; glioma classification; qPCR; sequencing.

Fig. 1

Overview of GliomaDx for detecting TERT promoter and IDH1 mutations. (A) GliomaDx is based on amplification in a mutant allele-specific manner. Placement of the locked nucleic acid (LNA) base over the 3ʹ end of the primer greatly improves the discrimination of the primer toward mutant alleles, enabling mutation-specific amplification even in the presence of elevated wildtype back background. (B) The LNA is a nucleic acid analog containing a 2ʹ-O, 4ʹ-C methylene bridge that restricts the flexibility of the ribofuranose ring, enabling enhanced stability and specificity for allelic discrimination.

Fig. 2

GliomaDx detects TERT/IDH hotspot mutations at 0.1% mutant allele fraction. (A) Cell line gDNA with target mutation was diluted with wildtype gDNA. All tests done in triplicate with controls: wildtype DNA (black, 0%), NTC (green). RFU is log scale. (B) The Ct difference between each mutation dilution and a wildtype sample is shown, revealing discrimination even down to 0.1% mutant allele fraction. (C) Example Sanger sequencing chromatograms of mutant dilutions showing that at 10%, the mutation is not reliably detectable over background. NTC = non-template control (water); RFU = relative fluorescence units; horizontal line indicates threshold RFU value.

Fig. 3

GliomaDx identifies recurrent TERT/IDH alterations in a cohort of frozen glioma tumor tissues with variable tumor cellularity. (A) A cohort of diffuse glioma frozen tumor samples were assayed for TERT and IDH mutations by GliomaDx (N = 39). The H&E from frozen sections for each case was examined by a neuropathologist and tumor percentage estimated, with significant portions exhibiting suboptimal characteristics, including possible tumor, infiltrating tumor, or necrosis.

Fig. 4

GliomaDx Nest for detection of TERT/IDH mutations in limited-input samples. (A) GliomaDx Nest works by high-fidelity multiplex pre-amplification of the TERT promoter, IDH1/2 mutation loci followed by amplicon dilution and nested allele-specific LNA-based qPCR. The pre-amplification allows for sensitive testing of many mutations off a single sample. (B) GliomaDx Nest retains similar sensitivity to that of GliomaDx, using the same allele-specific LNA primers. (C) A cohort of FFPE cases (N = 29) with known mutation status was assayed with GliomaDx Nest. Mutation results were consistent between both the clinical result (immunohistochemistry and Sanger sequencing based) and GliomaDx Nest. * indicates IDH2 mutation (R172K); IDH1/2 boxes without letters indicate IDH1 R132H mutations; “S” indicates IDH1 R132S; “C” indicates IDH1 R132C.

Fig. 5

Low percentage TERT/IDH variants identified by GliomaDx represent likely low tumor purity gliomas. (A) A cohort of TERTp^WT-IDH^WT diffuse gliomas was classified based on Sanger sequencing of TERTp and IDH in over 400 tumors. (B) GliomaDx identified many cases harboring low frequency alterations in the TERT promoter and IDH1 R132H. These alterations largely matched the expected associated histology. (C) Survival analysis of the new TERTp^WT-IDH^WT group (N = 25, median survival 21 mo) compared with the GliomaDx detected TERT promoter mutant cases (N = 8, median survival 8.52 mo) reveals a significantly worse survival (***P < 0.0005, log-rank test). The detected IDH mutant cases (N = 2) had a median survival of 103.43 months (not shown due to low sample number). (D) The Sanger sequencing chromatogram is shown for Case J, which appears wildtype due to the low sensitivity of traditional sequencing. (E) IDH1 R132H droplet digital PCR validated that Case J was IDH1 R132H–mutant. (F) Whole exome sequencing revealed that most exomic alterations were low mutation percentage (<30%). Dots represent the mutant read percentage for each variant identified. Middle lines represent the median values. Outer lines represent the range of values.