Case Report: Concordance of cerebrospinal fluid-derived circulating tumor DNA with tumor tissue in spinal glioblastoma associated with Lynch syndrome Authors: Carlen A. Yuen, David O. Kamson, Changrui Xiao, William H. Yong, Jennifer E. Soun, Mark Linskey, Michelle Paff, Caressa Hui, Alexander M. Lopez, Yael Freiberg, Hansen Bow Published: 2026 ABSTRACT BackgroundSpinal glioblastoma (GBM) is an exceptionally rare entity and poses significant diagnostic challenges due to a substantial risk of neurological morbidity with surgical biopsy. Tissue-based analyses are frequently limited by sampling constraints and tumor heterogeneity. Conventional diagnostic modalities, including MRI and cerebrospinal fluid (CSF) cytology, often lack specificity and sensitivity. Commercially available CSF liquid biopsy platforms offer a minimally invasive approach for detecting tumor-derived cell-free DNA (cfDNA) to enable genomic profiling.MethodsWe report a case of a 37-year-old male with spinal GBM who underwent prospective CSF cfDNA analysis using Belay Summit™ and GTC Liquid Trace™, yielding a provisional diagnosis of GBM and raising suspicion for Lynch syndrome.ResultsSubsequent tissue-based molecular profiling with Tempus™ confirmed the diagnosis of GBM, while germline testing via Ambry Genetics™ established Lynch syndrome.ConclusionBased on the findings of mismatch repair deficiency and Lynch syndrome, the decision to proceed with immunotherapy was made. CSF cfDNA is a promising ancillary tool for guiding therapy in difficult-to-biopsy cases. Introduction Genomic profiling is integral to glioblastoma (GBM) management and enables molecular classification and therapeutic stratification. Following the 2021 World Health Organization (WHO) classification of central nervous system (CNS) tumors, GBM diagnosis increasingly relies upon sequencing-defined molecular alterations, including IDH mutation status, EGFR amplification, CDKN2A/B homozygous deletion, TERT promoter mutation, and combined whole chromosome 7 gain/10 loss, which complement or supersede histopathologic criteria (1). Beyond MGMT promoter methylation status, additional predictive biomarkers, including microsatellite instability (MSI) and tumor mutational burden (TMB), carry clinically actionable relevance. Elevated TMB and MSI-high status are associated with enhanced responsiveness to immune checkpoint blockade and underpin the tumor-agnostic FDA approvals of these agents, including nivolumab and pembrolizumab. Identification of mismatch repair (MMR) deficiency is therefore critical for guiding immunotherapeutic strategies in select GBM patients. In the landmark report of two pediatric siblings with constitutional mismatch repair deficiency (CMMRD) and recurrent multifocal GBM, durable response to single-agent nivolumab was achieved following progression with standard therapy (2). This outcome markedly diverges from the typical aggressive clinical course of recurrent GBM patients. CMMRD is characterized by loss of mismatch repair function due to biallelic pathogenic variants of DNA mismatch repair genes MLH1, MSH2, MSH6, or PMS2, resulting in high TMB and markedly elevated neoantigen burden (2). Lynch syndrome is allelic with CMMRD, but affected individuals are heterozygous for germline variants, causing genetic susceptibility to malignancies including GBM (3). Although Lynch syndrome-associated gliomas are recognized, the clinical behavior and survival outcomes are poorly characterized with fewer than five cases reported to date (Table 1). Spinal GBM constitutes an exceptionally rare subset within this already uncommon disease entity. Table 1Lynch syndrome-associated glioblastoma treated with PD-1 blockade.CaseAge/SexInitial treatmentProgressive DiseasePD-1 blockadeOutcomePresent Case36/MCyclophosphamide →TMZ x3 days → NTR → radiation therapy → pembrolizumab–pembrolizumabongoingSherman et al. (15)57/FLeft occipital resection → RT (60 Gy) + TMZ, followed by TTF. Right frontal recurrence with resection → RT + nivolumab–nivolumab concurrently with radiation5 years out from initial diagnosisAnghileri et al. (16)33/FFrontal resection → radiotherapy, and TMZ → re-recurrence + GTR with surgery → nivolumab15 monthssecond surgery and treatment with nivolumabReceiving nivolumab 84 months after surgeryCabezas-Camarero et al. (17)34/MBiopsy → TMZ × 12 cycles → partial resection→ RT (60 Gy) → fotemustine → bevacizumab × 27 cycles → WandS × 12 m → bevacizumab × 13 cycles → tumor progression and biopsy → nivolumab → irinotecan + bevacizumab–Third Surgery and compassionate authorization of nivolumabComfort care, 7 years from diagnosisNTR, near total resection; RT, radiation therapy; TMZ, temozolomide; TTF, tumor-treating fields; GTR, gross total resection. Lynch syndrome-associated glioblastoma treated with PD-1 blockade. NTR, near total resection; RT, radiation therapy; TMZ, temozolomide; TTF, tumor-treating fields; GTR, gross total resection. Liquid biopsy has emerged as a minimally invasive approach for both diagnostic and real-time longitudinal monitoring of tumor burden and molecular status through analysis of circulating tumor DNA (ctDNA) that is shed by the tumor (4). While plasma-based and serum-based ctDNA assays are well established in systemic malignancies, these assays have limited utility in CNS tumors due to the low contribution of tumor DNA into the peripheral blood due to the blood–brain barrier. However, CSF provides direct access to the CNS tumor microenvironment, enabling more sensitive detection of tumor-specific genomic alterations. Although integration of CSF-derived ctDNA into routine clinical workflows for CNS tumors remains investigational, the field is rapidly advancing. CSF platforms, including Belay Summit™ and GTC Liquid Trace™, enable genomic and epigenomic profiling from low-input CSF tumor-derived DNA. These assays also enable the assessment of MSI and TMB. However, data evaluating the concordance between CSF-derived ctDNA and matched tumor tissue remain limited (5). We present a case of provisionally diagnosed Lynch-associated spinal GBM based on CSF ctDNA analysis with Belay Summit™ and GTC Liquid Trace™ diagnostic testing. Tempus™ confirmed GBM, and Ambry™ germline testing confirmed Lynch syndrome. This case underscores the clinical utility of CSF liquid biopsy for CNS tumor profiling and partial concordance to matched tissue-based next-generation sequencing. Methods Patient clinical data were retrospectively collected during the course of routine clinical care and subsequently reviewed for this case report. CSF was collected for ctDNA analysis using the Belay Summit™ and GTC Liquid Trace™ assays. Tissue-based molecular profiling was performed using the Tempus™ next-generation sequencing platform. These clinically validated assays assessed single-nucleotide variants, insertions/deletions, copy number alterations, gene fusions, and TMB in cancer-associated genes according to the manufacturers’ validated protocols. Sequencing results were interpreted in the context of our patient’s clinical history, radiographic findings, and histopathologic features. Confirmatory germline testing was performed using Ambry™ multigene panel testing. Case description A 36-year-old man with no significant past medical history presented to an outside hospital with progressive bilateral lower-extremity weakness, a sensory level below the chest, and urinary incontinence (Figure 1). MRI of the complete spine demonstrated an enhancing intramedullary lesion spanning C6-T2 (Figure 2). His family history was notable for maternally reported Lynch syndrome. Patient disease course. Postcontrast sagittal T1 fat-suppressed sequences demonstrating the tumor (A) pretreatment, (B) postoperative, and (C) post-radiation. (A) Preoperative: heterogeneously enhancing intramedullary tumor spanning the C6-T2 spinal cord. (B) Postoperative: C6-T4 laminectomies and expansion duroplasty for microsurgical resection of the intramedullary tumor. Residual mild intramedullary enhancement. (C) Post-radiation: Increased intramedullary enhancement at the treatment bed. New pseudomeningocele in the posterior soft tissues. Within 2 weeks, he deteriorated to flaccid paraplegia with bowel and bladder dysfunction. Empiric treatment with plasmapheresis, intravenous methylprednisolone, rituximab, and tocilizumab yielded no clinical improvement. CSF studies, including cytology, showed negative results (Table 2). Belay Summit™ 2.0 CSF ctDNA detected a PTEN mutation (variant allele frequency (VAF) 10.6%) and MMR alterations (MLH1 VAF 54.1%, MSH2 VAF 2.2%) with MSI, raising suspicion for GBM in the setting of Lynch syndrome (Table 3). Temozolomide (TMZ) (150 mg/m²) was initiated but discontinued after 3 days due to constipation and abdominal distention. Table 2Patient’s CSF studies.CSF study (#)1 (D0)2 (D51)3 (D66)4 (D100)OP (cm H20)Noted as normalNot obtainedNot obtained22ml2310unknown26RBC (cumm)033356WBC (cumm)41151 Lymphocytes (%)55578131 Neutrophils (%)022012 Monocytes (%)45211958Glucose (mg/dL)106456563Protein (mg/dL)61819758CulturenegativenegativenegativeMeningitis/encephalitis panel (PCR)1–––negativeACE, quantitative (U/L)<565–HSV 1/2Not detectedNot detected––VDRL, qualitativeNon-reactiveNon- reactive––WNV IgM, qualitative, EIA<0.90–––Coccidioides–––No antibodies detectedCryptococcus Antigen–––negativeIgG (mg/dL)5.4––negativeOCGabsentabsent–negativeEncephalitis/Autoimmune/Paraneoplastic Panel2–negativenegativenegativeCytologynegativenegative–negativeFlow Cytometry–––negative-, not obtained. D, day; OP, opening pressure; PCR, polymerase chain reaction; ACE, angiotensin-converting enzyme; HSV, herpes simplex virus; VDRL, venereal disease research laboratory; WNV, West Nile Virus; IgG, immunoglobulin G; OCG, oligoclonal bands. 1 Escherichia coli K1, Haemophilus influenzae, Listeria monocytogenes, Neisseria meningitidis, Streptococcus agalactiae, Streptococcus pneumoniae, Cytomegalovirus, Enterovirus, Herpes simplex virus 1, Herpes simplex virus 2, Human herpesvirus 6, Human parechovirus, Varicella zoster virus, Cryptococcus neoformans/gattii. 2 AGNA-1, Amphiphysin Ab, ANNA-1, ANNA-2, ANNA-3, CRMP-5-IgG, PCA-1, PCA-2, PCA-Tr, GAD65 Ab Assay, NMDA-R Ab CBA, AMPA-R Ab CBA, GABA-B-R Ab CBA, LGI1-IgG CBA, CASPR2-IgG CBA, mGluR1 Ab IFA, GFAP IFA, NIF IFA, Neurochondrin IFA, Septin-7 IFA, DPPX Ab CBA, IgLON5 CBA, PDE10A Ab IFA, TRIM46 Ab IFA. Patient’s CSF studies. -, not obtained. D, day; OP, opening pressure; PCR, polymerase chain reaction; ACE, angiotensin-converting enzyme; HSV, herpes simplex virus; VDRL, venereal disease research laboratory; WNV, West Nile Virus; IgG, immunoglobulin G; OCG, oligoclonal bands. Escherichia coli K1, Haemophilus influenzae, Listeria monocytogenes, Neisseria meningitidis, Streptococcus agalactiae, Streptococcus pneumoniae, Cytomegalovirus, Enterovirus, Herpes simplex virus 1, Herpes simplex virus 2, Human herpesvirus 6, Human parechovirus, Varicella zoster virus, Cryptococcus neoformans/gattii. AGNA-1, Amphiphysin Ab, ANNA-1, ANNA-2, ANNA-3, CRMP-5-IgG, PCA-1, PCA-2, PCA-Tr, GAD65 Ab Assay, NMDA-R Ab CBA, AMPA-R Ab CBA, GABA-B-R Ab CBA, LGI1-IgG CBA, CASPR2-IgG CBA, mGluR1 Ab IFA, GFAP IFA, NIF IFA, Neurochondrin IFA, Septin-7 IFA, DPPX Ab CBA, IgLON5 CBA, PDE10A Ab IFA, TRIM46 Ab IFA. Table 3Comparison of patient’s molecular analyses.GeneCSFTumor tissueSalivaGenomic testing cooperativeliquid traceBelaysummit 2.0TempusAmbry geneticsGenomic variantVAF (%)Genomic variantVAF (%)Genomic variantVAF (%)Pathogenic – germline implications BRCA1 Not detectedp.E23fs c.68_69del3.8p.E23fs - LOF22.4Not detected MLH1 c.589-2A>G 43.56 c.589-2A>G 54.1 c.589-2A>G Splice region variant - LOF 71.7 c.589-2A->G (confirmed germline) MSH2 Not detectedp.I883fs c.2647du2.2p.I883fs - LOF13.5Not detectedGlioblastoma (somatic) MGMTp Unmethylated Unmethylated CDKN2A Not detected CDKN2A p.R58* c.172C>T4.4p.R58* Stop gain - LOF52.7 EGFRLow level gain in autosomal chromosomes Not detected Not detected PTEN Not detectedp.N323fs c.968del10.6p.N323fs - LOF30.8 TP53 Not detectedp.G245S c.733G>A1.9 Pathogenic - actionable MTOR Not detected Not detected p.L2431P Missense variant - GOF6.6 NF1 Not detected Not detected p.I679fs - LOF53.7 PIK3R1 Not detectedc.1746-2A>G3.9c.1746-2A>G Splice variant - LOF11.8Pathogenic - non-actionable AJUBA Not tested Not tested p.A244fs - LOF25.6 ARID1A Not detectedp.Y815* c.2445_2446d7.6p.Y815fs - LOF26.2 BCORL1 Not detected P1681fs*** p.1755fs - LOF61.4 KMT2D Not detectedNot detected p.R2915* Stop gain - LOF32.2 MRE11 Not detectedp.T481fs c.1441dup6.7p.T481fs - LOF28.3 NOTCH2c.74C>A22.22 Not detected TCF7L2 Not tested Not detected p.C469fs - GOF31.7Other actionable TMBLow (1mut/Mb)Not detected 13.7mut/MB (92nd %ile) MSINot testedHigh HighThe * and *** notations are a part of the allele nomenclature.The bold values indicate concordance among the CSF, tissue, and saliva assays. Comparison of patient’s molecular analyses. The * and *** notations are a part of the allele nomenclature. The bold values indicate concordance among the CSF, tissue, and saliva assays. Upon transfer to our institution, examination revealed American Spinal Injury Association (ASIA) A T1 complete paraplegia with preserved hand function (partial C7-C8 involvement). Repeat CSF ctDNA with GTC Liquid Trace™ confirmed MLH1 alteration with low TMB and low-level EGFR gain. Given the discordant results from the Belay Summit™ and GTC Liquid Trace™ platforms, which were obtained at different time points in the disease course and that paraplegia was already established, a tissue-based diagnosis was pursued. The patient underwent C6-T3 laminectomy with subtotal resection of the T1–3 intramedullary tumor and expansion duroplasty, with superior resection limited to preserving hand function. Postoperatively, neurological status was unchanged. Histopathology demonstrated a hypercellular astrocytic neoplasm with palisading necrosis and microvascular proliferation, immunopositive for Olig2, ATRX, and GFAP and negative for SOX10, EMA, IDH1 R132H, and BRAF V600E. Ki-67 showed a score of 5%-10% (Figure 3). (A) H&E-fibrillary astrocytes with palisading necrosis (200×). (B) H&E-microvascular proliferation (200×). (C) GFAP-fibrillary and gemistocytic astrocytes (200×). (D) Olig2-extensive nuclear immunopositivity (400×). Tumor next-generation sequencing with Tempus™ confirmed MSI with MLH1 c.589-2A>G splice-site mutation (VAF 71.7%), PTEN and MSH2 frameshift loss-of-function variants, and unmethylated MGMT. Additional alterations included BCORL1, NF1, and CDKN2A loss-of-function mutations, and alterations in KMT2D, MRE11, ARID1A, AJUBA, and TCF7L2. No canonical IDH1/2, H3K27M, and TERT promoter mutations or EGFR amplification was identified. TMB was high in contrast to GTC Liquid Trace™ findings. The tumor was classified as Glioblastoma, IDH-wildtype, CNS WHO grade 4. However, in addition to the MLH1 mutation, the high frequency of BCORL1 alteration among many other alterations is distinct from the usual GBM. Methylation profiling was not possible given the limited tissue. Saliva-based Ambry panel testing confirmed a germline MLH1 c.589-2A>G pathogenic mutation. Adjuvant radiation therapy to 40 Gy in 15 fractions was initiated but discontinued at 24 Gy due to wound dehiscence and hematoma as a result of cachexia and anticoagulation for DVT. A wound washout and re-closure were performed without complication. Given MSI-high status, MMR deficiency, and elevated TMB, the decision was made to proceed with pembrolizumab 200 mg every 3 weeks. Discussion Spinal GBM can be diagnostically challenging. Surgical biopsy carries substantial risk of neurological morbidity due to the infiltrative nature of GBM and the eloquent anatomy of the spinal cord. Moreover, limited tissue frequently restricts molecular characterization. Conventional minimally invasive diagnostic modalities, including magnetic resonance imaging (MRI) and CSF cytology, are suboptimal for detecting GBM, underscoring the need for more adequate diagnostic tools. To our knowledge, this is the first reported case of a provisionally diagnosed spinal GBM utilizing CSF ctDNA. Our case corroborates the established low sensitivity of conventional diagnostics, including CSF cytology (6), which failed to detect malignant cells on two different samples. CSF ctDNA offers a minimally invasive approach for early diagnosis and treatment guidance, eliminating neurosurgical risk. Moreover, unmethylated MGMT results from Belay Vantage™ and Tempus™ informed the omission of TMZ, while detection of MSI, high TMB, and DNA mismatch repair alterations suggested Lynch syndrome and guided the decision for immunotherapy. Although high ctDNA levels are typically observed in GBM with Ki-67 >20% (7), CSF ctDNA analysis was successful even at a lower proliferative rate (Ki-67=5%-10%) in our case. Belay Summit™ showed superior concordance with the Tempus™ assay on the tumor compared to GTC Liquid Trace™, with presurgical CSF ctDNA findings of PTEN and MLH1/MSH2 mutations, which were suggestive of GBM and Lynch syndrome. While PTEN is not a diagnostic criterion for GBM, PTEN mutations are found in 30-40% of GBMs (8) and raised the suspicion for GBM. Our case illustrates the complementary potential diagnostic value of CSF ctDNA, demonstrating its capacity to identify genomic alterations not detected by tissue-based NGS in the diagnosis of GBM. A low-level EGFR gain was identified in our patient’s tumor only by GTC Liquid Trace™ and not by Belay Summit™ or Tempus™. EGFR amplification is present in ~50% of GBMs and constitutes a diagnostic criterion for GBM (9). This finding raises the possibility of either a false positive or a subclonal event missed by tissue sampling due to low VAF (10). Moreover, ctDNA analyses may include non-tumor-derived alterations, which predominantly occurs after cytotoxic therapy. In our case, Belay Summit™ was performed prior to cyclophosphamide and GTC Liquid Trace™ was performed after cyclophosphamide. Although notable discrepancies were observed between Belay Summit™ and GTC Liquid Trace™, this case was not intended as a head-to-head analytical comparison of these two platforms. A critical confounder is the difference in sampling timepoints. Belay Summit™ was performed pre-chemotherapy, whereas both GTC Liquid Trace™ and tumor tissue profiling via Tempus™ were obtained following cyclophosphamide and TMZ. It remains unclear whether chemotherapy-induced cytotoxicity affected the performance characteristics of GTC Liquid Trace™, potentially through treatment-related alterations resulting in ctDNA shedding or fragmentation. While cytotoxic therapy may theoretically increase ctDNA release through tumor cell lysis, it may also reduce overall tumor burden or alter the integrity of the DNA, thereby impacting the sensitivity of the assay. Furthermore, our patient’s tumor was MGMT unmethylated, with decreased response to alkylating agents and he only received 3 days of TMZ. Lastly, though GTC Liquid Trace™ and Tempus™ were performed at the same treatment timepoint, discordance between these two assays was observed. This discordance may reflect low or absent ctDNA shedding from specific tumor subclones. Another discordance was observed with the TP53 mutation (VAF 1.9%) detected by Belay Summit™ but absent in the matched tumor. This discrepancy may reflect the technical limitations of tissue NGS. GBMs are heterogeneous tumors, and tissue NGS relies upon the sample of tissue provided to and selected by the neuro-pathologist for processing. Conventional NGS is constrained by error rates that limit reliable detection of variants at ≤1% VAF noise (11), whereas ctDNA integrates signal from spatially and genetically distinct tumor subclones. Consistent with this, Pan et al. demonstrated that CSF ctDNA can identify mutations not detected in matched tumor specimens in brainstem gliomas (4). Taken together, these findings support a key advantage of ctDNA in its ability to capture low-frequency and spatially heterogeneous alterations that may be missed by single-site tissue biopsy (4, 12, 13). Lastly, the discordant TMB (not detected on Belay Summit™ and high TMB observed on Tempus™) may be representative of the ongoing genomic evolution of our patient’s tumor under selective pressure given that Belay Summit™ was performed early in his disease course and Tempus™ was performed later in his disease course. Concordant alterations were also observed. A BRCA1 mutation was identified in both CSF and tumor. BRCA1 dysfunction contributes to genomic instability and therapeutic resistance in GBM, with potential implications for treatment with PARP inhibition. Similarly, CDKN2A/B homozygous deletion, a common GBM alteration, was detected by Belay Summit™ and Tempus™, with variable VAFs, reflecting the differences between CSF ctDNA and tissue sampling profiling. Collectively, this case highlights that CSF ctDNA may complement tissue NGS by uncovering clinically relevant mutations missed due to sampling bias and may provide a more comprehensive view of GBM genomic heterogeneity. Concordance between CSF ctDNA and tissue NGS in GBM is not well established, with limited paired tissue data. Our case adds to this limited body of knowledge and emphasizes the potential of for CSF ctDNA in CNS tumors to enable timely, precision-guided care. CSF ctDNA represents a promising tool for the diagnosis of CNS tumors, particularly in cases where lesions are difficult or high-risk for biopsy (14). CSF ctDNA carries future promise to guide targeted therapies, longitudinally monitoring tumor evolution, and improving real-time diagnostic accuracy. Limitations We acknowledge limitations in this study. First, a single case is not generalizable to all spinal glioblastomas predisposed by Lynch syndrome. Second, assessment of the longitudinal response to immunotherapy in our patient is constrained by the limited duration of follow-up. Third, the CSF assays were collected at different timepoints during our patient’s disease course and the disparate results may reflect differences in disease burden and the ongoing genetic evolution of GBM. Lastly, potential differences in specimen volumes, storage, and timing of processing may have also attributed to the mismatch between the test results. Conclusion Cerebrospinal fluid circulating tumor DNA is a promising ancillary tool to enable patient-specific genomic stratification and targeted therapy in difficult-to-biopsy cases. We report a case of a 37-year-old man in whom prospective CSF ctDNA analysis (Belay Summit™ and GTC Liquid Trace™) supported a provisional spinal GBM diagnosis with suspected Lynch syndrome. Paired tumor-based sequencing (Tempus™) confirmed glioblastoma, and germline testing (Ambry™) established Lynch syndrome, thereby guiding therapy for immunotherapy.