TERT Promoter Mutation Analysis for Blood-Based Diagnosis and Monitoring of Gliomas

Liquid biopsy for the detection and monitoring of brain tumors is of significant clinical interest. Upon clinical presentation, patients typically undergo imaging followed by biopsy alone or biopsy with resection for diagnosis and determination of histopathologic classification (1). While tissue biopsy is invasive and can, in some cases, be high risk, liquid biopsy can offer a less invasive sampling approach that still affords significant clinical information for diagnosis. In addition, liquid biopsy can be performed more frequently to allow for longitudinal treatment monitoring. The detection of glioma mutations via liquid biopsy of cerebrospinal fluid has been demonstrated (2–6); however, the ability to detect mutations in cell free DNA (cfDNA) in plasma with similar sensitivities has been limited to date (5, 7, 8).

TERT promoter mutations are among the most common genetic alterations in glioma (6, 9, 10).The two most common mutations are C228T (45% of all TERT promoter mutations) and the C250T (15% of all TERT promoter mutations), occurring 124 and 146 bp, respectively, upstream of the TERT transcriptional start site (11). These mutations are typically heterozygous, mutually exclusive, and due to the GC-rich nature of the promoter, yield an identical 11bp “CCCGGAAGGGG” ETS/TCF binding motif that is associated with increased transcriptional activity of TERT, and therefore increased telomerase activity (12, 13). Somatic mutations in TERT promoter are common across all grades of gliomas with a frequency of 8% in pilocytic astrocytomas (grade I), 15% in diffuse astrocytomas (grade II), 45% in oligodendrogliomas (grade II), 54% in anaplastic oligodendrogliomas (grade III), and 62% in glioblastoma (grade IV; refs. 9, 10, 14, 15). Notably, the presence of TERT promoter mutation is a poor prognostic marker in high-grade IDH wild-type glioblastoma (9, 10, 15), but TERT promoter mutation confers a positive prognosis in lower grade tumors, distinguishing between TERT/IDH comutated oligodendroglioma and IDH-mutant astrocytoma (9). To date, PCR amplification of TERT has proven technically difficult due to the high GC content in the promoter region (>80%; refs. 16, 17). To address these issues, several additive and nonadditive-based approaches including Q-sol, 7-deaza-dGTP, and locked nucleic acid (LNA)-enhanced probes have been reported previously (11, 18–20).

In this report, we describe the development of a novel ddPCR probe-based assay (Fig. 1A) that can detect two TERT promoter mutations (C228T and C250T) in the plasma of glioma patients simultaneously. In an analysis of totaling 157 patients, we describe the sensitivity of this assay for detecting TERT promoter mutations in plasma, and highlight potential clinical applications, for diagnosis and therapeutic monitoring of glial tumors using this blood test.

The study population (n = 157) included patients 18 years or older with pathology confirmed TERT mutant or wild-type gliomas who underwent surgery at either the Massachusetts General Hospital (MGH, Boston, MA), Henry Ford Health System (HFHS), or at Washington University St. Louis (WUStL, ST. Louis, MO) and age-matched healthy controls (Fig. 2A; Table 1).The study population was divided into a discovery cohort (n = 83; 46 TERT mutant, 17 TERT wild-type, 20 matched healthy controls) and a blinded multi-institution validation cohort (n = 74; 42 TERT mutant, 9 TERT WT, 14 matched healthy controls, 9 nontumor MRI positive). For glioma cohorts, exclusion criteria consisted of history of other primary or metastatic cancers, active infectious disease, current or previous enrollment in clinical trials, and hemolyzed plasma samples. All healthy control subjects were screened for pertinent oncologic and neurologic medical histories. Individuals with a history of cancer, neurological disorders, and infectious diseases were excluded from the study. All samples were collected with written informed consent after the patient was advised of the potential risks and benefits, as well as the investigational nature of the study. Our studies were conducted in accordance with principles for human experimentation as defined in the U.S. Common Rule and were approved by the Human Investigational Review Board of each study center under Partners institutional review board (IRB)-approved protocol number 2017P001581. BRISQ guideline reports are included in Supplementary Table S1. Samples were taken from patient population undergoing treatment at MGH, HFHS, or WUStL. Some samples may still be available and could be shared with appropriate documentation (IRB, MTA etc.). ExoLution PLUS is a proprietary kit available from Exosome Diagnostics (a BioTechne brand).

Tumor tissue was microdissected and suspended in RNAlater (Ambion) or flash-frozen and stored at −80°C.

Whole blood was collected using K2 EDTA tubes with an inert gel barrier (BD Vacutainer Blood Collection Tubes), from preoperatively placed arterial lines or venipuncture. Within 2 hours of collection, samples were centrifuged at 1,100 × g for 10 minutes at 20°C to separate the plasma from the hematocrit and filtered using 0.8-μm filters. One-milliliter aliquots were stored at -80°C for later downstream analysis. With the exception of the longitudinal samples, all baseline samples were collected prior to surgical resection.

Human carcinoma cell line A431 (ATCC CRL-1555; RRID:CVCL_0037) was cultured in DMEM with high glucose (DMEM; Gibco, Invitrogen Cell Culture), containing 10% FBS (Life Technologies Corporation) and 1% penicillin/streptomycin (Life Technologies). Human glioma cell line U87 (ATCC HTB-14; RRID:CVCL_0022) was cultured in DMEM with high glucose, containing 10% FBS and 1% penicillin/streptomycin. Human brain microvascular endothelial cells (HBMVEC) were kindly provided by Xandra O. Breakefield (Massachusetts General Hospital, Boston, Massachusetts) and cultured using endothelial basal medium (EGM-2 MV Microvascular Endothelial Cell Growth Medium-2 BulletKit, Lonza). All cell lines were grown to 50%–70% confluency prior to gDNA extraction to minimize cell death and optimize quality of gDNA. All cell lines were verified monthly using the Mycoplasma PCR detection kit (Applied Biological Materials) to ensure lack of Mycoplasma contamination. All cell lines were obtained from ATCC, and were used within ten passages within the time of thawing.

DNA was isolated from cell lines and frozen tumor tissue using the DNeasy Blood and Tissue Kit (Qiagen) as recommended by the manufacturer. DNA was eluted in AE buffer (Qiagen) and stored at −20°C until further processing. DNA concentration and purity were determined using the NanoDrop One (ThermoFisher Scientific).

Circulating nucleic acid was extracted from plasma using the QIAamp Circulating Nucleic Acid Kit (Qiagen) or ExoLution PLUS (Exosome Diagnostics) as per the manufacturer's instructions. cfDNA eluted in 20 μL AVE buffer (QIAamp Circulating Nucleic Acid Kit) or in 20 μL nuclease-free water (ExoLution PLUS Kit) and stored at −20°C until quantification and subsequent ddPCR test.

Because both the C228T and C250T TERT promoter mutations yield the same sequence (Fig. 1B), a single probe was used to detect both mutations. A second probe was also used to recognize the C228 wild-type locus. As described by McEvoy et. al, Locked Nucleic Acid (LNA) modifications were introduced on probes due to the short size of the probe, indicated by “+” (20). The sequences for the probes used are as follows: TERT promoter mutant (5′-FAM/CCC +C+T+T+CCGG/3IABkFQ/) and TERT promoter wild-type (5′-HEX/CCC C+C+T +CCG G/3IABkFQ/). Probes were synthesized by Integrated DNA Technologies (IDT). ddPCR amplification was performed using either 4 μL of cfDNA template or using 100 ng of tumor gDNA, 1× ddPCR Supermix for probes (no dUTP, Bio-Rad), either 1× Q-sol or 200 mmol/L 7-deaza-dGTP (7dG; New England Biolabs), 250 nmol/L of each probe, and 900 nmol/L of each primer (5′-CCTGCCCCTTCACCTTCCAG-3′ and 5′-AGAGCGGAAAGGAAGGGGA-3′) with template (100 ng of tumor tissue or 4 uL of cfDNA) in a total reaction mix of 20 μL. The QX200 manual droplet generator (Bio-Rad) was used to generate droplets. Thermocycling conditions were as follows: 95°C (51% ramp) for 10 minutes, 40 cycles of 94°C (51% ramp) for 30 seconds, and 57°C for 1 minute, followed by 98°C for 10 minutes and held at 4°C until further processing. Droplets were counted and analyzed using the QX200 droplet reader (Bio-Rad) and QuantaSoft analysis (Bio-Rad) was performed to acquire data.

Copies per mL of plasma is calculated by taking copies/20μL (C; provided by QuantaSoft), multiplied by elution volume in μL (EV), divided by the total volume added to the reaction (TV), divided by the plasma volume (PV). In short, Copies/mL of plasma = C*EV/TV/PV.

Gates were constructed for both channel 1 (4–8) and channel 2 (2–6) in increments of 0.05. Combinations of channel 1 gates and channel 2 gates were used to calculate MAF, the number of channel 1–positive droplets divided by the number of channel 2 positive droplets. Thresholds were calculated to fulfill one of the following criteria: maximize specificity, set specificity close to 90%, and to minimize the distance from the ROC curve to the point (0,1). These gating strategies (trained using a discovery cohort) were then used to analyze the data from the multi-institution validation cohort in a blinded fashion. Code available on GitHub Repository koushikmuralidharan/BILL.

Statistical analyses were performed using unpaired two-tailed Student t test in GraphPad Prism 8 software and P < 0.05 was considered statistically significant. Confidence intervals were calculated using exact binomial distributions. The results are presented as the mean ± SD. “***” indicates P value less than or equal to 0.001, and “****” indicates P value less than or equal to 0.0001.

Source data for figures are provided within supplementary figures. Code is available on GitHub Repository koushikmuralidharan/BILL.

The most common TERT promoter mutations, C228T and C250T, are heterozygous and mutually exclusive, but both mutations result in the generation of an 11-bp identical sequence, 5′-CCCCTTCCGGG-3′. We used a 10-bp LNA-mutant probe to simultaneously detect both mutations and an LNA wild-type probe complementary to the C228 locus (Fig. 1B). Assay specificity for each mutation was established using U87 (C228T mutant), A431 (C250T mutant), and HBMVEC (TERT WT) gDNA (Fig. 1C). For inputs with mutant allele frequencies (MAF) greater than 10%, 2D amplitude analysis can distinguish between C250T/C228T mutations (Fig. 1D; Supplementary Fig. 1A). TERT promoter mutations are situated in a GC-rich region resistant to amplification that hinders assay performance. To stabilize amplification, we compared Q-sol additive to 7-deaza-dGTP (7dG), a modified nucleotide that inhibits secondary structure formation (17). TERT assay performance with 7dG was superior to Q-Sol (1.8–3.9-fold increase, P = 0.04) with higher absolute mutant detection and comparable WT detection (Fig. 1E and F). We report analytical parameters including limit-of-detections (LODs) of 0.27% MAF (C250T) and 0.42% (C228T) MAF and limit of blanks of 0.02% (C250T) and 0.04% MAF (C228T; Fig. 1G and H; Supplementary Fig. S1). In a comparison of extraction platforms to optimize TERT recovery, a 2-fold increase in TERT WT (P = 0.001) was seen while using ExoLution PLUS in comparison with the QiAmp Circulating Nucleic Acid kit (Qiagen; Fig. 1I). Two milliliters of plasma was determined to be the optimal input for recovery as a function of copies of TERT WT per mL of plasma (Fig. 1J and K).

To standardize gating without operator bias, we sought to mathematically define and automate the gating strategy. Our gating strategy is based on the algorithm used by the R-program ddPCR, which defines empty droplets as those that lie within 7 SDs above the mean amplitude of channel 1 (21). To determine the optimal gating strategy, we gated continuously from 4–8 SDs above the mean amplitude of channel 1 and 2–6 SDs above the mean amplitude of channel 2 in increments of 0.05, with pseudocode provided in Supplementary Fig. S2. Bootstrapping with 1000 replicates was used to determine a threshold, and a list of gating strategies with threshold values was generated to ensure that specificity was greater than or equal to 90%. We present one such gating strategy that maximizes the sum of discovery sensitivity, validation sensitivity, and overall sensitivity. This gating strategy gates at 7.15 SDs above the mean of channel 1 amplitude, 2.95 SDs above the mean of channel 2 amplitude, with a threshold of 0.26% MAF (Supplementary Fig. S2). This is in accordance with prior literature, which suggests that positive droplets lie 7 SDs above the mean amplitude of channel 1 (21). Notably, the threshold calculated by the program, 0.26% MAF, was equivalent to the experimentally determined LOD (MAF) of the TERT Promoter ddPCR Assay using cell line derived gDNA (Fig. 1G).

We selected a cohort (n = 157) of molecularly characterized glioma patients (n = 114), nontumor patients with enhancing lesions on MRI (n = 9), and age matched healthy controls (n = 33). Our patient population spanned tumor diagnosis (20% astrocytoma, 7% oligodendroglioma, 72% glioblastoma, 1% gliosarcoma); grade (6% II; 15% III, 71% IV, 8% not reported) and molecular characteristics: IDH1 mutant (19%), TERT mutant (45% C228T; 15% C250T), 1p19q codeletion (7%), EGFR amplified (20%), MGMT methylated (33%; Table 1; Fig. 2). The study population was randomly assigned to either a discovery cohort (n = 83) or a blinded multi-institutional validation cohort (n = 74). To assess the clinical performance of the assay, tumor tissue and matched plasma samples were analyzed for the presence of the TERT mutations (Supplementary Fig. S3).

In tissue, we demonstrated that our assay and the CLIA certified Solid SNAPSHOT assay (22) used in the MGH Department of Pathology, detected TERT-positive tumors with 100% concordance across 97 tested samples for the presence of the TERT promoter mutation. (Fig. 2A; Supplementary Fig. S3). In one patient, using parallel tumor tissue aliquots, our assay detected the C250T mutation while the SNAPSHOT assay detected the C228T mutation.

Plasma samples (total n = 83; TERT mutant n = 46; TERT WT n = 17; healthy controls n = 20) were analyzed and gated (Supplementary Fig. S2) in two separate discovery sample sets (Fig. 2A). In sample set 1, we report positivity in 16 of 21 (76%) TERT-mutant samples, and in 1 of 14 (7%) WT control samples (Fig. 2B), with a sensitivity of 76.19% (CI, 52.83–91.78), and specificity of 92.86% (CI, 66.3–99.82). In sample set 2 (PCR blinded), we report positivity in 17 of 25 (68%) of TERT-mutant samples, and in 3 of 20 (15%) of WT control samples (Fig. 2C), with a sensitivity of 68.00% (CI, 46.50–85.05), and specificity of 86.96% (CI, 66.41–97.22). Overall, in the discovery cohort, we report positivity in 33 of 46 (72%) TERT-mutant samples, and in 4 of 37 (11%) TERT WT samples (Fig. 2D), with a sensitivity of 71.74% (CI, 56.54–84.01) and specificity of 89.19% (CI, 74.58–96.97).

To validate assay performance in multi-institutional samples, we analyzed a blinded cohort [total n = 74; TERT mutant n = 41 (Henry Ford Hospital n = 12; Washington University n = 2; Massachusetts General Hospital n = 27); TERT WT n = 17; healthy control n = 14, CNS disease nonglioma n = 9] using the parameters previously described (Fig. 3A; Supplementary Fig. S2). The CNS disease nontumor samples are from patients who either had a nonmalignant contrast enhancing mass on MRI, such as a demyelinating lesion or fungal abscess, that was initially suspected as glioma or other nontumor conditions such as normal pressure hydrocephalus. In matched plasma, we report positivity in 22 of 42 (52%) confirmed mutant samples and 3 of 33 (9%) confirmed wild-type with a sensitivity of 52.38% (CI, 36.42%–68.00%), specificity of 90.91% (CI, 75.67%–98.08%; Fig. 3B–D).

The blinded multi-institution validation cohort verifies and validates our assay performance for the detection of the TERT promoter mutation in plasma (23–25). In summary, combining all three cohorts (n = 157), we demonstrate a sensitivity of 62.50% (CI, 51.53%–72.60%), and a specificity of 90% (CI, 80.48%–95.8%; Figs. 2E and 3D). Of patients who were classified positive by plasma analysis but negative by tissue analysis at the MAF threshold selected, the clinical scenario of these false positive samples included 4 of 34 healthy controls, 1 of 9 patients with nonglioma CNS disease, and 2 of 34 TERT WT gliomas.

Within the limits of our cohort, we detected no significant correlation between TERT MAF in plasma and age, duration of symptoms, tumor grade, mutational status (C228T/C250T), tumor volume, contrast enhancement, overall survival and progression-free survival (Supplementary Fig. S4). However, we noted that patients with contrast enhancing tumors (increased breakdown of blood–brain barrier) tended to have higher MAF than patients with non-contrast–enhancing tumors (Supplementary Fig. S1G). Furthermore, patients with MAF above threshold tended to have poorer progression-free survival and overall survival compared to patients with below threshold MAF (Supplementary Fig. S1A and S1B).

We also report perfect concordance in 4 available matched CSF samples from both cohorts compared to a tissue gold standard (TERT mutant n = 3; TERT WT n = 1). We also detect a TERT mutation in the CSF of a patient sample whose plasma TERT MAF was below the defined assay threshold (Supplementary Fig. S5).

To assess the performance of the TERT assay in a longitudinal setting, we analyzed cfDNA from serial plasma samples of five patients with TERT-mutant glioma maintaining the same analytical parameters. TERT-mutant copies over the course of therapy paralleled imaging findings and clinical performance of patients (Fig. 4). Each of the patients had TERT mutant MAF above threshold prior to initial surgical resection and levels returned to baseline postoperatively. Patients P1 and P2 had stable disease with the TERT-mutant levels remaining below baseline over time. On follow up, 3 patients (P3, P4, P5) developed contrast enhancing lesions on MR images after chemoradiation. P4 and P5 had histopathologic confirmation of progression while P3 had clinically diagnosed progression. TERT MAF increased with the development of contrast-enhancing recurrent lesions coincident with clinical deterioration. Plasma from P4 was not available for analysis at the time of suspected disease progression.

This study demonstrates for the first time, in a majority of glioma patients with a known tissue TERT promoter mutation, the detection of the same mutation in glioma patients' plasma, using a novel enhanced ddPCR assay. We report an overall sensitivity of 62.5% and a specificity of 90% of detecting cfDNA TERT promoter mutations in plasma compared with a matched tumor tissue gold standard in 114 patients with glioma. The ability to detect the highly prevalent TERT promoter mutation in the plasma of patients with glioma enhances our ability to diagnose, monitor and assess responses to therapy.

In this report, we overcome several challenges that have prevented detection of cfDNA point mutations in specific genes in the plasma of patients with brain tumor, including for the TERT promoter specifically. In contrast to other tumor types, where cfDNA point mutations are abundant, this is not the case in glioma. Several prior reports have detected specific oncogenic point mutations in cfDNA, including at the TERT promoter locus, in less than 5% of patients (5, 8, 26). This may be related to generally lower levels of CNS-derived circulating tumor DNA (ctDNA) in the blood, thought to be a function of the blood–brain barrier (BBB; ref. 5). While in our patient cohort, we show that tumor size does not correlate with plasma TERT MAF (similar to a prior report of non-specific cfDNA quantitation in glioma; ref. 27), we did note a tendency for patients with contrast-enhancing lesions (indicative of BBB breakdown) to have higher plasma TERT MAF. Taken with the fact that recurrence timepoints (after surgery and chemoradiation) have more edema and higher plasma TERT MAF compared with baseline despite smaller tumor volumes, we hypothesize that plasma TERT MAF is correlated with breakdown of the BBB.

Our approach included several technical optimization strategies to detect low concentrations of TERT promoter mutations in plasma. Both C228T and C250T generate identical sequences, which can be detected with a single probe with LNA enhancements that stabilize probe-template duplexes to improve SNV discrimination. Furthermore, the TERT promoter region has a high GC-content (>80%) which we address by using 7-deaza-dGTP as a ddPCR additive, which interrupts the formation of these secondary structures. In combination, with standardized handling strategies and an unbiased analytic method, we observed a significant improvement in assay sensitivity over previously published reports for TERT promoter detection in plasma of patients with glioma (28).

The challenge in detecting plasma cfDNA from patients with brain tumor highlights a missed opportunity for the advantages of the liquid biopsy approach in a tumor type where initial or repeat direct tissue sampling may pose substantial neurologic risk. While tissue biopsy will likely remain the gold standard for molecular diagnosis of glioma, the ability to detect oncogenic point mutations in the blood may have specific near-term application in two scenarios: (i) the described TERT promoter mutation assay may be useful for upfront diagnosis. Because some malignant tumors are located in deep or inaccessible locations and would be poor surgical candidates for direct tissue biopsy or resection, the combination of a characteristic set of MRI imaging findings in addition to a liquid biopsy that shows TERT promoter mutation may be sufficient to establish a high positive predictive value for the clinical diagnosis malignant glioma and allow adjunctive treatment with radiation and chemotherapy to proceed with confidence. (ii) In another scenario, because of the surgical risk of repeated brain biopsies, a liquid biopsy approach to TERT promoter mutation detection in the blood may be particularly suitable for therapeutic monitoring of disease burden, as exemplified conceptually in the longitudinal cases shown in this report. In our preliminary analysis, levels of TERT promoter mutants, as detected by this assay, do correlate with disease recurrence or progression. This can aid in differentiation between radiation necrosis, pseudoprogression, and true progression, thus minimizing the need for further invasive workup and improving overall quality of care. Furthermore, with emerging insights into intratumoral heterogeneity, the validity of a single, localized tissue biopsy as a true gold standard has begun to be debated (25), raising the possibility that a liquid biopsy, which effectively samples multiple tumor regions as blood perfuses the solid tumor mass, may offer greater sensitivity for TERT promoter mutation than a single, focal, tissue biopsy, in some patients.

Future studies offer the opportunity for further extension of this particular assay and the general approach of liquid biopsy for brain tumors. The false positive rate of our assay at the chosen MAF threshold was 10%, which is similar to the false positive rate of multimodality imaging analysis to infer diagnosis of gliomas. However, this also included assessment of patients with a very low predicted risk of glioma (e.g., healthy controls with no known MRI abnormalities and patients with fungal abscess/normal pressure hydrocephalus). In practical use, the TERT promoter mutation assay would be preferentially used in situations where disease prevalence is estimated to be high, for example, in patients with abnormal MRI scans consistent with glioma, To confirm the presence of glioma with a high positive predictive value. Within our cohort, 0 of 7 patients with an abnormal MRI and nonglioma CNS disease had TERT MAF above threshold. However, every patient with abnormal MRI (FLAIR and/or contrast abnormality) and a plasma TERT MAF above the threshold had a confirmed glioma diagnosis (n = 55 tissue TERT Mutant; n = 2 tissue TERT WT). This is especially true for patients with contrast-enhancing lesions on MRI and plasma TERT MAF above the threshold, as every patient in this subset also had a confirmed glioma diagnosis (n = 42 tissue TERT mutant; n = 2 tissue TERT WT). Thus, while our assay's sensitivity compared with a tissue gold standard is encouraging, future studies should assess the positive predictive value of the assay for the overall diagnosis of glial malignancy in the context of abnormal MRI findings. Furthermore, the combination of this plasma assay with advances in MRI imaging techniques that permit differentiation of various TERT promoter mutant glial tumors with different prognoses (e.g., glioblastoma vs. anaplastic oligodendroglioma) is also anticipated to be useful for clinical prognostication.

While our MAF threshold (0.26%) is optimized based on ROC analysis, from a clinical perspective, it is important to minimize the false positive rate of our assay (10%) below that of multimodal imaging. To this end, clinical studies across larger patient cohorts will be required to validate and standardize the gating parameters, plasma volumes, and clinical performance of the assay in heterogeneous populations with more extensive longitudinal sampling. These expanded studies should provide further insight on factors which produce false-negative and false positive results. Such factors may include assay MAF threshold performance in low prevalence populations (with potential false positives) or sampling of tumors with heterogeneous amounts of BBB breakdown (affecting cfDNA release), that manifest as false negative results in non-contrast–enhancing TERT⁺ gliomas. Expanded studies will also allow further study of samples from multiple institutions; in the current report, the samples were heavily weighted (92%) from a single institution, potentially introducing bias in preanalysis sample handling with concomitant effects on assay performance.

The demonstration herein that the plasma detection of TERT promoter mutations in a majority of patients with glioma who harbor these tumor mutations opens the door to extending this to a multi-gene approach to further classify gliomas using a liquid biopsy approach, mimicking tissue-based molecular classification. This could be accomplished by careful optimization of assays for common glioma point mutations in the blood such as IDH1 R132H mutations, EGFRvIII mutations, HK23M mutations and others. We also anticipate that this assay will also offer a sensitive method for detecting mutations in plasma cfDNA in other TERT-positive tumors.

In conclusion, we demonstrate a novel and highly sensitive ddPCR based TERT promoter mutation assay, utilizing high affinity LNA enhanced probes and the additive 7dG to reduce the formation of secondary structures, for detection and monitoring of TERT promoter mutations (C228T, C250T) in tumor tissue and cfDNA of matched plasma of patients with glioma with an overall sensitivity of 62.5% and a specificity of 90% in combined discovery and blinded validation cohorts of 157 samples. The ability to detect TERT mutations, which are highly prevalent in glioma patients, in the plasma enhances our ability to diagnose, monitor and assess response to therapy. Liquid biopsy-based monitoring can significantly impact clinical care by guiding patient stratification for clinical trials, offering new opportunities for the development of targeted therapies ultimately improving patient care.

K. Muralidharan reports a patent for TERT Promoter Droplet Digital PCR Assay For the Diagnosis of Malignant Cancer pending. A. Yekula reports a patent for TERT Promoter Droplet Digital PCR Assay for the Diagnosis of Malignant Cancers pending. S. Lau reports grants from United States Government during the conduct of the study. H. Lee reports grants from NCI and grants from DOD during the conduct of the study. C. Bettegowda reports personal fees from Depuy-Synthes and personal fees from Bionaut Labs outside the submitted work. M.R. Chicoine reports grants from IMRIS, Inc., The Head for the Cure Foundation, and Mrs. Carol Rossfeld and The Alex & Alice Aboussie Family Charitable Foundation during the conduct of the study. D.P. Cahill reports personal fees from Merck, Lilly, and Boston Pharmaceuticals outside the submitted work; in addition, D.P. Cahill has a patent 20170369939 pending and a patent 20190160139 pending. L. Balaj reports a patent for TERT Promoter droplet digital PCR Assay for the Diagnosis of Malignant Cancers pending. B.S. Carter reports a patent pending related to plasma-based detection of TERT mutations as described in this report. No disclosures were reported by the other authors.

K. Muralidharan: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. A. Yekula: Data curation, formal analysis, validation, investigation, visualization, methodology, writing-review and editing. J.L. Small: Investigation, writing-review and editing. Z.S. Rosh: Investigation, writing-review and editing. K.M. Kang: Investigation, writing-review and editing. L. Wang: Investigation, writing-review and editing. S. Lau: Investigation. H. Zheng: Formal analysis. H. Lee: Formal analysis, visualization. C. Bettegowda: Formal analysis, validation. M.R. Chicoine: Formal analysis, validation. S.N. Kalkanis: Formal analysis, validation. G.M. Shankar: Formal analysis, writing-review and editing. B.V. Nahed: Formal analysis, writing-review and editing. W.T. Curry: Writing-review and editing. P.S. Jones: Formal analysis, writing-review and editing. D.P. Cahill: Writing-review and editing. L. Balaj: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. B.S. Carter: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing.

This work is supported by grants U01 CA230697 (to B.S. Carter, L. Balaj), UH3 TR000931 (to B.S. Carter), P01 CA069246 (to B.S. Carter), and R01 CA239078, CA237500 (to B.S. Carter, L. Balaj, H. Lee). The funding sources had no role in the writing of the manuscript or decision to submit the manuscript for publication. The authors have not been paid to write this article by any entity. The corresponding author has full access and assumes final responsibility for the decision to submit for publication. The authors thank all of the MGH Neurosurgery clinicians and staff who assisted with the collection of samples. They also thank the patients and their families for their participation in this study. They thank Alexandra Beecroft and Abigail Taylor for editorial help. They also thank Alona Muzikansky for help with the statistical analysis and Gur Rotkop and Sharmistha Chakrabortty for help with computational analysis.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.