Expanding the spectrum of TERT promoter mutations in CNS tumors: A case series of non-canonical mutations

To the Editor:

TERT promoter (TERTp) mutations are a molecular biomarker in multiple central nervous system (CNS) tumor types and are primarily represented by the canonical C228T and C250T activating hotspot missense mutations.¹ These canonical mutations generate a new consensus sequence for binding of E26 transformation-specific (ETS) and recruit transcription factor complex GA-binding protein, increasing TERT expression and contributing to telomere maintenance.² Other clinically relevant TERTp sequence variants with evidence of cancer association and functional experimental data supporting oncogenicity have been reported and are hereafter referred to as “non-canonical TERTp mutations”.³ Examples of non-canonical TERTp mutations include the c.-57A>C mutation exclusively described in non-CNS tumors and the TERTp activating duplications recurrently reported in glioblastoma, IDH-wildtype (GBM).^3–5 Herein, we describe the clinicopathological, molecular, and epigenetic characterization of a series of 12 CNS tumors (11 gliomas and 1 meningioma) harboring previously reported and potentially novel non-canonical TERTp mutations, expanding the spectrum of TERTp mutations associated with primary CNS tumors.

Among CNS tumors routinely clinically tested at the Mayo Clinic Laboratory of Genetics and Genomics (2020-2024) using 2 custom DNA targeted neuro-oncology next-generation sequencing (NGS) panels, we identified 12 cases with non-canonical TERTp mutations. The NGS panels targeted TERT RefSeq NM_198253 positions c.-95 through c.-225 (n = 5), and c.-25 through c.-246 (n = 7). A subset of cases had data for OncoScan CNV Plus chromosomal microarray (n = 10) and/or Illumina Infinium MethylationEPIC v1.0/2.0 methylation array (n = 9) using the NCI/Bethesda v2 classifier. Methylation classifier results were interpreted as “match” (when both superfamily and class calibrated scores ≥0.9), “suggestive” (superfamily score ≥0.9 and class score 0.5-0.9) and “no match” (superfamily score <0.9, regardless of class score). The final integrated diagnosis for each case was based on histopathological, mutation, copy number, and methylation profiling.

Clinicopathological, molecular, and epigenetic data associated with our detected non-canonical TERTp mutations are detailed in Table 1. These non-canonical mutations occurred at variant allele frequencies in keeping with somatic clonal origin (mean, 34%; range, 18-46) and were not observed in conjunction with canonical TERTp or ATRX mutations. CNS tumors harboring these non-canonical mutations occurred in adults (mean age, 60 years; range, 36-88) and were mostly (11 of 12) morphologically high-grade (Figure 1A–D); GBM was the most frequent (8 of 12) integrated diagnosis. The c.-57A>C was the most common non-canonical TERTp mutation and was observed in 5 cases, including 2 oligodendrogliomas, IDH-mutant and 1p/19q-codeleted; 1 GBM; 1 anaplastic meningioma; and 1 diffuse glioma, IDH-mutant, not elsewhere classified (NEC). The c.-57A>C mutation was originally identified in familial melanoma, prompting investigation and discovery of the canonical TERTp hotspot mutations in sporadic melanoma.^3,6 This mutation generates a consensus sequence for ETS binding like the canonical TERTp mutations and has been shown to increase TERT expression.^3,7,8 Although not previously reported in CNS tumors, the c.-57A>C mutation has been observed in other cancer types, supporting an oncogenic role.⁹ Four cases, all GBMs, had either the c.-100_-79dup (n = 2) or c.-110_-89dup (n = 2) duplications. These TERTp duplications have recently been reported in GBM and shown to be functionally and mechanistically equivalent to canonical TERTp hotspot mutations.^4,5 The remaining variants are potentially novel non-canonical TERTp mutations (c.-92_-91ins GGCGGCCCCGCCCCTTCCTTTC, c.-118_-117insTCCCCGGCCCAGCCCCTTCCGGG, and c.-123_-76dup), and were detected in 1 tumor each, which were also all GBMs. These potential non-canonical mutations have not been previously reported but their sizes and sequence structures suggest that they may be functionally equivalent to the C228T mutation or generate favorable TERTp activation motif conformations like the TERTp activating duplications⁵ (Figure 1E).

Table 1.

Clinicopathological, molecular, and epigenetic findings.

Case	Age/sex	Final integrated diagnosis	CNS WHO grade	TERTp mutation	VAF	Methylation superfamilya	Methylation class (calibrated score)
1	61/F	GBM, IDH-wildtype	4	c.-100_-79dup	0.18	GBM	GBM_RTK_I (0.863)
2	65/F	GBM, IDH-wildtype	4	c.-100_-79dup	0.2	N/A	N/A
3	75/M	GBM, IDH-wildtype	4	c.-92_-91insGGCGGCCCCGCCCCTTCCTTTC	0.25	N/A	N/A
4	68/F	GBM, IDH-wildtype	4	c.-118_-117insTCCCCGGCCCAGCCCCTTCCGGG	0.3	GBM	GBM_RTK_II (0.969)
5	88/F	GBM, IDH-wildtype	4	c.-123_-76dup	0.25	GBM	GBM_RTK_I (0.998)
6	54/M	GBM, IDH-wildtype	4	c.-110_-89dup	0.41	GBM	GBM_RTK_II (0.998)
7	73/F	GBM, IDH-wildtype	4	c.-110_-89dup	0.45	GBM	GBM_RTK_I (0.995)
8	36/M	Oligodendroglioma, IDH-mutant and 1p/19q-codeleted	3	c.-57A>C	0.38	IDH_glioma	O_IDH (0.992)
9	42/M	Diffuse glioma, IDH-mutant, NEC	2	c.-57A>C	0.39	IDH_glioma	O_IDH (0.992)
10	48/F	GBM, IDH-wildtype	4	c.-57A>C	0.37	GBM	GBM_MES_TYP (0.993)
11	70/M	Anaplastic meningioma	3	c.-57A>C	0.39	Meningioma	MNG_MAL (0.792)
12	40/M	Oligodendroglioma, IDH-mutant and 1p/19q-codeleted	3	c.-57A>C	0.46	N/A	N/A

^aMethylation superfamily calibrated score ≥0.9 in all cases.

Abbreviations: F, female; GBM, glioblastoma; M, male; N/A, not available; NEC, not elsewhere classified; VAF, variant allele frequency.

Figure 1.

Histopathological findings. (A) Glioblastoma, IDH-wildtype (Case 7). (B) Oligodendroglioma, IDH-mutant and 1p/19q-codeleted, CNS WHO grade 3 (Case 8). (C) Anaplastic meningioma (Case 11). (D) Diffuse glioma, IDH-mutant, NEC (Case 9). (H&E; A, 100×; B, C, 400×; D, 200×). (E) Genomic architecture of potentially novel non-canonical TERTp mutations.

By genome-wide methylation profiling, all 9 tested cases matched (ie, high confidence result with calibrated score ≥0.9) to a methylation superfamily, with 7 also matching to a methylation class. Methylation class was suggestive in the remaining 2 cases, a GBM (Case 1; GBM_RTK_I) with grade 4 histological features (necrosis and microvascular proliferation) and mutations in EGFR and PTEN, and an anaplastic meningioma (Case 11; MNG_MAL) with histological features of anaplasia (cytological atypia and >20 mitoses/10 high power fields), NF2 mutation, 22q loss, and CDKN2A/B homozygous deletion. Overall, methylation superfamily and class taxonomy data supported the diagnosis in most (8 of 9) cases based on concordance with overall histological and available mutation and copy number data. The only exception was a histologically low-grade infiltrating glioma with oligodendroglial-like features (Case 9) that lacked the cytologic uniformity of a classic oligodendroglioma. This tumor was found to harbor whole arm 1p loss but only partial loss of 19q, as well as mutations of IDH1 (R132H) and NF1, and matched to methylation class oligodendroglioma (O_IDH) with high confidence. This unusual case was designated as a diffuse glioma, IDH-mutant, NEC, CNS WHO grade 2 (Figure 1D).

In this series, non-canonical TERTp mutations occurred primarily in CNS tumors with the typical histological, molecular, cytogenetic, and epigenetic features of tumor types that often harbor canonical TERTp mutations as a molecular biomarker. The sequence structure of such mutations suggests that they functionally converge with the canonical mutations in creating new ETS binding sites. The overlapping clinicopathological features and sequence structures in this context support the impression that these non-canonical TERTp mutations may be biologically equivalent to the canonical TERTp mutations. It is noteworthy that the c.-57A>C variant was not previously detectable using our original targeted neuro-oncology NGS panel, which only targeted positions c.-95 through c.-225. Our updated targeted neuro-oncology NGS panel is extracted from a pan-cancer targeted master NGS panel, which was designed to target positions c.-25 through c.-246, allowing for the detection of the c.-57A>C mutation. We hypothesize that the non-canonical mutations described herein may not have been widely recognized since a considerable portion of the literature on TERTp mutations in CNS tumors is based on studies that only targeted the canonical mutations. Additionally, laboratories may be utilizing NGS panels or PCR-based assays designed to specifically target only the canonical TERTp mutations. From a classification standpoint, recognition and inclusion of non-canonical TERTp mutations as one of the TERTp activating mechanisms alternative to canonical mutations would expand the spectrum of diagnostically relevant TERTp mutations that can be considered a molecular biomarker for CNS tumors and potentially rescue a subset of cases that have been so far considered “TERTp-wildtype.”

Funding

None declared.

Conflicts of interest

None declared.