Brain tumor classification methods

Abstract

The disclosure relates to methods that include obtaining cell-free deoxyribonucleic acid (cfDNA) from a sample of cerebrospinal fluid (CSF) from a subject having a brain tumor, profiling DNA methylation of the cfDNA using enzymatic methylation sequencing of the cfDNA to produce sequencing data and classifying the brain tumor of the subject.

Description

[0001] BRAIN TUMOR CLASSIFICATION METHODS

[0002] RELATED APPLICATION

[0003] This application claims the benefit under 35 U.S.C. § 119(e) of U.S. Provisional Application No. 63 / 618,561, filed January 8, 2024, which is incorporated by reference herein in its entirety.

[0004] BACKGROUND

[0005] Molecular diagnosis of brain tumors has rapidly advanced in the last few years with genome-wide DNA methylation profiling emerging as the gold standard. The DNA methylation-based tumor classification algorithm developed by the German Cancer Research Center (DKFZ) has been used widely for diagnosis in clinical settings.

[0006] SUMMARY

[0007] Brain and central nervous system (CNS) tumors constitute the fifth most common form of cancer overall and are the most prevalent type of cancer among children ages 0 to 14 years. Part of brain and CNS tumor diagnosis includes the classification of the brain tumor, which is guided in part by guidelines provided by the World Health Organization that define distinct brain tumor entities, of which there are approximately 100 (see, e.g, Louis el al. Neuro Oncol. 2021. 23(8): 1231-1251). Accurate diagnosis and classification is important for optimal treatment and management of patient care. Profiling of DNA methylation patterns of brain tumors, for example via biopsy of the tumor itself, has become the gold standard of molecular diagnosis. Despite the advantage of DNA methylation profiling, the invasive nature of brain tumor biopsy - especially for deep-brain tumors - as well as the small size and intratumoral heterogeneity of some brain tumors limit the utility of DNA methylation profiling. Moreover, repeated biopsy of brain tumors to determine the response to treatment over time is infeasible.

[0008] Due to the limitations of solid tissue sampling for DNA methylation profiling of brain tumors, liquid biopsy (LB) presents an attractive alternative. LB assays have historically been reported using plasma or serum, though the potential of using cerebral spinal fluid (CSF) has recently been entertained. Despite this possibility, CSF analysis has yet to be adopted for clinical diagnosis and monitoring of brain tumors due to the limitations it presents, the most challenging of which is the collection of CSF in an amount suitable for DNA methylation profiling. It has previously been reported that CSF samples of 200 pL to 500 pL of CSF typically yield less than 50 ng of cell-free deoxyribonucleic acid (cfDNA), which is insufficient for certain assays.

[0009] Presented herein are methods of DNA methylation profiling on low amounts of cfDNA (e.g., less than 5 ng) isolated from small amounts of CSF (e.g., less than 2 mL, milliliters) that result in accurate tumor classification which can be useful for both initial diagnostic processes of brain and CNS tumors, as well as ongoing monitoring of tumors to determine response to treatment.

[0010] In some aspects, the disclosure relates to methods that include obtaining cfDNA from a sample of CSF from a subject having a brain tumor, profiling DNA methylation of the cfDNA using enzymatic methylation sequencing (also referred to as enzymatic methyl sequencing) of the cfDNA to produce sequencing data and classifying the brain tumor of the subject.

[0011] In some embodiments, a sample of CSF has a volume of less than 2 mL. In some embodiments, the sample of CSF has a volume of less than 1 mL.

[0012] In some embodiments, an amount of cfDNA used in a profiling step is less than 5 ng. In some embodiments, an amount of cfDNA used in the profiling is about 0.1 ng to about 1 ng-

[0013] In some embodiments, enzymatic methylation (methyl) sequencing includes: (a) combining the cfDNA with Tet methylcytosine dioxygenase 2 (TET2) (and optionally an oxidation enhancer, for example, T4 phage P-glucosyltransferase (T4-PGT)) to convert 5- methylcytosine (5mC) and 5 -hydroxy methylcytosine (5hmC) of the cfDNA into products that cannot be deaminated by APOBEC3A; and (b) combining the cfDNA with APOBEC3A to convert unmodified cytosines of the cfDNA to uracils.

[0014] In some embodiments, a method further includes converting the sequencing data into a binarized format of methylated / unmethylated for each mapped CpG site.

[0015] In some embodiments, a classifying step is performed using a DNA methylationbased tumor classification algorithm. In some embodiments, a DNA methylation-based tumor classification algorithm is the DKFZ tumor classification algorithm.

[0016] DETAILED DESCRIPTION

[0017] DNA methylation profile has become the gold standard of molecular diagnosis of brain tumors in recent years. While tumor classification algorithms work well on DNA methylation data generated from tumor tissues, the invasive nature of brain tumor tissue biopsy (e.g., deep in the brain), small size, and intratumoral heterogenicity of certain types of brain tumor are factors limiting the utility of the current approach on tumor tissue. The current approach is also unsuitable for the molecular monitoring of a tumor’s response to ongoing treatment, which requires periodic resampling of the tumor.

[0018] Compared to conventional tissue biopsy, liquid biopsy (LB) is an attractive alternative for DNA methylation-based tumor diagnosis. It has multiple advantages over conventional tissue biopsy, including (1) non-invasiveness and minimal risk; (2) ease of accessibility; (3) less sampling bias for heterogeneous tumors; and (4) potential use in real-time, longitudinal monitoring of disease progression and treatment response. Most of the LB assays have been reported using plasma or serum. Alternatively, cerebrospinal fluid (CSF) is obtained during routine procedure in the care of certain types of brain tumors, for example, intracranial germ cell tumors (iGCTs), such that no additional sampling is required for developing new clinical assays such as those described herein. CSF analysis, however, has been limited due to small sample size and the requirement for threshold DNA amounts when using standard detection methods. For example, methylation profiling using Illumina Infinium MethylationEPIC BeadChip microarray (hereafter “MethylationEPIC”), the standard input for the current German Cancer Research Center (“DKFZ”) tumor classification algorithm, requires amounts of DNA that are not easily obtained from CSF samples. This input amount, 200 ng as recommended by the manufacturer, is the biggest challenge for CSF samples, as published studies usually report obtaining less than 50 ng cfDNA from 200-500 pl of CSF. In addition, certain methyl modifications, such as 5-hydroxymethylcytosine (5hmC), are abundant in brain tissue but appear to be depleted in brain tumor samples (Azizgolshani et al., Clin Epigenetics 23: 176, 2021). Thus, a sensitive method for detecting differential 5hmC patterns in brain tumor patients is needed. Further, it is not clear if the methylation profile of cfDNA from CSF has the same robustness in tumor classification as those with tumor tissue. Therefore, there is an unmet need in the clinical and research community to find alternative approaches to profile cfDNA methylation patterns in the CSF of brain tumor patients for accurate tumor diagnosis using the DKFZ algorithm. Accordingly, provided herein are methods useful for classifying brain tumors in subjects using a small volume of CSF and / or a low amount of cfDNA.

[0019] Subjects, Cerebrospinal Fluid, and Cell-Free DNA

[0020] Some aspects of the disclosure relate to methods that include obtaining cell-free deoxyribonucleic acid (cfDNA) from a sample of cerebrospinal fluid (CSF) from a subject having a brain tumor. CSF is produced by tissue that lines the ventricles of the central nervous system and flows in and around the brain and spinal cord. Collection of cfDNA from CSF is advantageous because it is less invasive than brain tumor biopsy, and CSR sampling (e.g., via lumbar puncture) is a routine procedure in the care of some brain tumors. Any suitable method can be used to obtain (e.g., isolate) cfDNA from a sample of CSF. For example, a sample of CSF can be centrifuged to pellet cellular components and the remaining supernatant subject to a cfDNA isolation kit (e.g., a commercial cfDNA isolation kit; see Example 1). Such methods are known and used in the art.

[0021] Cell-free DNA includes DNA fragments that are freely circulating in the bloodstream, outside of cells. Cell-free DNA typically originates from normal cell turnover, as well as from the breakdown of cells due to tissue injury or disease, such as cancer, for example. Brain and CNS tumors shed cfDNA into the CSF, and methods provided herein were used to demonstrate that cfDNA can be a suitable proxy for detecting methylation patterns within the tumor itself (see Example 1).

[0022] A sample of CSF used in the methods herein, in some embodiments, has a volume of about 0.1 mL to about 2 mL. For example, a sample of CSF may have a volume of about 0.1 mL to about 1.5 mL, about 0.1 mL to about 1 mL, about 0.1 mL to about 0.5 mL, about 0.5 mL to about 2 mL, about 0.5 mL to about 1.5 mL, about 0.5 mL to about 1 mL, about 0.75 mL to about 1 mL, about 1 mL to about 2 mL, about 1 mL to about 1.5 mL, or about 1.5 mL to about 2 mL. In some embodiments, a sample of CSE has a volume of less than 2 mL. In some embodiments, a sample of CSE has a volume of less than 1 mL. In some embodiments, a sample of CSE has a volume of 0.5 mL. In some embodiments, a sample of CSE has a volume of 1 mL.

[0023] The amount of cfDNA obtained and / or used for DNA methylation profiling, in some embodiments is less than 10 ng. Lor example, the amount of cfDNA may be less than 5 ng, less than 4 ng, less than 3 ng, less than 2 ng, or less than 1 ng. In some embodiments, the amount of cfDNA obtained and / or used for DNA methylation profiling is about 0.1 ng to about 1 ng. Lor example, the amount of cfDNA can be about 0.1 ng to about 0.5 ng, about 0.1 ng to about 0.4 ng, about 0.1 ng to about 0.3 ng, or about 0.1 ng to about 0.2 ng. In some embodiments, the amount of cfDNA is 0.1 ng. In some embodiments, the amount of cfDNA is 0.2 ng. In some embodiments, the amount of cfDNA is 0.3 ng. In some embodiments, the amount of cfDNA is 0.4 ng. In some embodiments, the amount of cfDNA is 0.5 ng. In some embodiments, the amount of cfDNA is 0.6 ng. In some embodiments, the amount of cfDNA is 0.7 ng. In some embodiments, the amount of cfDNA is 0.8 ng. In some embodiments, the amount of cfDNA is 0.9 ng. In some embodiments, the amount of cfDNA is 1.0 ng.

[0024] A subject having a brain tumor may have, for example, a glioma (e.g., astrocytoma, oligodendroglioma, or ependymoma), meningioma, a pituitary tumor, a medulloblastoma, a schwannoma, a pineal gland tumor, or a craniopharyngioma. A subject may have any type of brain tumor, for example any brain tumor type defined by the World Health Organization (see, e.g., Louis et al. Neuro Oncol. 2021.). Other brain tumors are contemplated herein.

[0025] DNA Methylation Profiling

[0026] The methods described herein, in some embodiments, provide unexpected sensitivity for detecting methylation patterns, also known as DNA methylation profiling. DNA methylation is a type of epigenetic modification that is typically associated with gene silencing, and aberrant DNA methylation patterns are a common feature of cancer. The data provided herein demonstrates that methylation profiling of cfDNA from CSF samples can be done using suitable methods, such as a modified version of the Enzymatic Methylation Sequencing (EM-Seq) assay (see Vaisvila R et al. Genome Res. 2021 Jul;31 (7) : 1280-1289, incorporated herein by reference).

[0027] In some embodiments, the methods employ a two-enzyme system, while in other embodiments, the methods employ a three-enzyme system. For example, as discussed further below, the methods can include the use of TET2 and APOBEC3A in some instances, and in other instances an additional enzyme, referred to as an “oxidation enhancer” (e.g., T4 Phage P-glucosyltransferase (T4-PGT)) is also used. The data herein shows that successful cfDNA methylation profiling is achievable using as low as 1 ng cfDNA extracted from the CSF samples, for example. Furthermore, the method described herein resulted in accurate tumor classification from CSF based on the DKFZ algorithm (see Capper D. et al. Nature. 2018 Mar 22;555(7697):469-474, incorporated herein by reference and described elsewhere herein), compared to the classification using matched tumor tissue.

[0028] Any suitable method for detecting methylation patterns can be used. Examples of alternative methods of detecting methylated nucleotides include, but are not limited to, (i) (e.g., as originally described in Frommer et al. PNAS, 1992. 89(5): 1827- 1831); (ii) chemical- assisted pyridine borane sequencing (CAPS); (iii) TET-assisted pyridine borane sequencing (TAPS); and / or (iv) TAPS with PGT protection (TAPSP) (see, e.g., Eiu et al. Nat Commun, 2021. 12( 1):618). In such sequencing methods, methylated nucleotides in DNA (e.g., cfDNA) are converted to dihydrouracil, after which the nucleic acids are amplified and dihydrouracil is converted to thymine. In some embodiments, methylation sequencing comprises bisulfite sequencing. In some embodiments, methylation sequencing comprises CAPS. In some embodiments, methylation sequencing comprises TAPS. In some embodiments, methylation sequencing comprises TAPSp.

[0029] In some embodiments, the methods further comprise profiling DNA methylation of the cfDNA using enzymatic methylation sequencing of the cfDNA to produce sequencing data. Sequencing refers to the process of determining the sequences of nucleotides in a nucleic acid (e.g., the order of thymine, guanine, cytosine, and adenine in a DNA nucleic acid). “Nucleic acid” as used herein is interchangeably with “polynucleotide” and refers to a polymer of nucleotides (e.g. DNA nucleotides or RNA nucleotides). Sequencing data therefore refers to a collection of nucleic acids with known sequences. In some embodiments, sequencing comprises Sanger sequencing. In some embodiments, sequencing comprises nextgeneration sequencing (NGS), including shotgun sequencing, long-read sequencing sequence, whole exome sequencing, and others.

[0030] Profiling DNA methylation involves analyzing the patterns of methylation marks on nucleic acids, e.g., DNA molecules. DNA methylation is a key epigenetic mechanism used by cells to control gene expression. It involves the addition of a methyl group (CH3) to the DNA, typically at the cytosine nucleotide in the context of a cytosine-guanine (CpG) dinucleotide. Enzymatic methylation sequencing detects 5mC and 5hmC using enzymatic reactions. In one reaction, a ten-eleven translocation (TET) enzyme (and optionally an oxidation enhancer, for example, T4-PGT) converts 5mC and 5hmC into products that cannot be deaminated by a deaminase (e.g., activation-induced cytidine deaminase, AID; and apolipoprotein B mRNA editing enzyme, catalytic polypeptide, APOBEC). In some embodiments, a TET enzyme is TET1, TET2, TET3, mTETl, mTET2, or mTET3. In some embodiments, a TET enzyme is TET2. In some embodiments, a deaminase is AID, APOBEC1, APOBEC2, APOBEC3 (e.g., APOBEC3A), or APOBEC4. In some embodiments, a deaminase is APOBEC3.

[0031] An “oxidation enhancer” includes enzymes that specifically transfer a methylprotective sugar, such as the glucose moiety of uridine diphosphoglucose (UDP-Glc), to the 5 -hydroxy methylcytosine (5-hmC) residues in double- stranded DNA, making beta-glucosyl- 5-hydroxymethylcytosine, for example. In some embodiments, the oxidation enhancer is T4- PGT. In some embodiments, the oxidation enhancer is T4 a-glucosyltransferase (T4-aGT) In another reaction, APOBEC3A deaminates unmodified cytosines by converting them to uracils. See Vaisvila R et al. Genome Res. 2021 Jul;31(7): 1280-1289, incorporated herein by reference. Surprisingly, the enzymatic methods described herein, in some embodiments, provide highly sensitive methylation profiling with identification of 5mC and 5hmC using less than 10 ng (and even less than 5 ng or less than 1 ng) of cfDNA obtained from CSF. Thus, in some embodiments, enzymatic methylation sequencing includes: (a) combining cfDNA (e.g., less than 10 ng) with TET2 (and optionally an oxidation enhancer, for example, T4-PGT) to convert 5mC and 5hmC of the cfDNA into products that cannot be deaminated by APOBEC3A; and (b) combining the cfDNA with APOBEC3A to convert unmodified cytosines of the cfDNA to uracils.

[0032] In some embodiments, the methods further comprise converting sequencing data into a binarized format of methylated / unmethylated for each mapped CpG site. Binarization of sequencing data (e.g., a “0” value for an unmethylated mapped CpG site and a “1” value for a methylated mapped CpG site) allows for rapid determination (e.g., using an algorithm) of methylated nucleotides within a given sequence. CpG sites are regions of DNA where a cytosine nucleotide is connected to a guanine nucleotide by a phosphate bond. CpG sites are not evenly distributed across the genome and are preferred sites of DNA methylation. A “mapped CpG site” refers to a CpG site that has been identified in a nucleic acid, e.g., by sequencing the nucleic acid.

[0033] Tumor Classification Algorithms

[0034] In some embodiments, the methods further comprise classifying the brain tumor of the subject. In some embodiments, the classifying is performed using a DNA methylationbased tumor classification algorithm. Any suitable DNA methylation-based tumor classification algorithm can be used (see, e.g., Park et al., J Korean Med Sci. 2023. 38(43):e356). In some embodiments, the DNA methylation-based tumor classification algorithm is the German Cancer Research Center (DKFZ) tumor classification algorithm. The DKFZ tumor classification algorithm, developed by the German Cancer Research Center, is the DNA methylation-based tumor classification algorithm most widely used for diagnosis in the clinic. See Capper D. et al. Nature. 2018 Mar 22;555(7697):469-474, incorporated herein by reference and described elsewhere herein.

[0035] The DKFZ tumor classification algorithm was developed using genomic information including mutation profiles and copy number variations (CNVs), transcriptomic data from RNA sequencing to assess gene expression patterns, and epigenomic data like DNA methylation, which is, as articulated elsewhere, the gold standard for distinguishing tumor subtypes and a key component of the DKFZ algorithm. The DNA methylation-based tumor classification algorithm developed by DKFZ has improved diagnostic precision and revealed previously unrecognized tumor subtypes. Initially, tumors are categorized by histological types, such as gliomas, medulloblastomas, etc., and hematological malignancies like leukemias and lymphomas. Thereafter, the DKFZ algorithm narrows these classifications by identifying molecular subtypes based on cancer-associated genetic mutations (e.g., EGFR, IDH1 / 2, TP53), gene expression patterns (e.g., luminal A / B, basal-like, HER2-enriched subtypes in breast cancer), and chromosomal alterations like copy number variations or structural rearrangements. DNA methylation patterns are also to further refine the tumor classification, and can additionally be used to categorize tumors according to prognostic groups based on aggressiveness, metastatic potential, and response to therapy. Prognostic biomarkers (e.g., certain methylation patterns, such as MGMT methylation in glioblastomas) also guide treatment decisions.

[0036] DKFZ has created an open-access platform known as the Molecular Neuropathology (MNP) Classifier, a web-based tool designed for classifying brain tumors using methylation data, as well as other platforms that integrate multiple -omics datasets for comprehensive tumor analysis. Accordingly, DNA methylation-based tumor classification according to the DKFZ algorithm is available for use as a component of the methods provided herein.

[0037] Additional Embodiments

[0038] The present disclosure further provides embodiments encompassed by the following numbered paragraphs:

[0039] Paragraph 1. A method, comprising: obtaining cell- free deoxyribonucleic acid (cfDNA) from a sample of cerebrospinal fluid (CSF) from a subject having a brain tumor; profiling DNA methylation of the cfDNA using enzymatic methylation sequencing of the cfDNA to produce sequencing data; and classifying the brain tumor of the subject.

[0040] Paragraph 2. The method of Paragraph 1, wherein the sample of CSF has a volume of less than 2 mL.

[0041] Paragraph 3. The method of Paragraph 2, wherein the sample of CSF has a volume of less than 1 mL.

[0042] Paragraph 4. The method of any one of the preceding Paragraphs, wherein the amount of cfDNA used in the profiling is less than 5 ng.

[0043] Paragraph 5. The method of Paragraph 4, wherein the amount of cfDNA used in the profiling is about 0.1 ng to about 1 ng. Paragraph 6. The method of any one of the preceding Paragraphs, wherein the enzymatic methylation sequencing comprises: (a) combining the cfDNA with Tet methylcytosine dioxygenase 2 (TET2) (and optionally an oxidation enhancer, for example, T4 phage P-glucosyltransferase (T4-BGT)) to convert 5-methylcytosine (5mC) and 5- hydroxymethylcytosine (5hmC) of the cfDNA into products that cannot be deaminated by AP0BEC3A; and (b) combining the cfDNA with AP0BEC3A to convert unmodified cytosines of the cfDNA to uracils.

[0044] Paragraph 7. The method of Paragraph 6, wherein the cfDNA is combined with TET2 and the oxidation enhancer, wherein the oxidation enhancer is T4 phage P-glucosyltransferase (T4-BGT).

[0045] Paragraph 8. The method of any one of the preceding Paragraphs further comprising converting the sequencing data into a binarized format of methylated / unmethylated for each mapped CpG site.

[0046] Paragraph 9. The method of any one of the preceding Paragraphs, wherein the classifying is performed using a DNA methylation-based tumor classification algorithm.

[0047] Paragraph 10. The method of Paragraph 9, wherein the DNA methylation-based tumor classification algorithm is the DKFZ tumor classification algorithm.

[0048] EXAMPLES

[0049] Example 1

[0050] This Example relates to the development of a protocol for DNA methylation-based brain tumor classification using cerebrospinal fluid (CSF). The protocol includes (i) extraction of cell-free DNA (cfDNA) from 1 mL or less CSF sample from patients with brain tumors using Quick-cfDNA™ Serum & Plasma Kit (Zymo Research) or QIAamp® Circulating Nucleic Acid Kit (Qiagen®); (ii) Enzymatic Methyl -Sequencing (EM-Seq) profiling of the extracted cfDNA using a modified version of the NEBNext® Enzymatic Methyl-seq Kit v2 kit and Unique Dual Index Primer pairs (New England Diolabs); and (iii) Conversion of sequencing data into binarized format compatible with that of DNA methylation data generated on microarray; (iv) tumor classification based on the EM-Seq data from cfDNA using DKFZ tumor classification algorithm. As the current gold standard, DNA methylation-based brain tumor classification is based on the Infinium MethylationEPIC BeadChip microarray (Illumina) data from tumor tissues which were collected by surgical procedures. The protocol presented herein has multiple advantages over the gold standard, as outlined below. Firstly, DNA methylation profiling of cfDNA extracted from CSF was used for tumor classification. CSF is collected by less invasive means than surgical procedures which reduces the risk for the patients. CSF also potentially allows tumor monitoring in time series. Secondly, as low as 0.1 ng of cfDNA for EM-Seq was used, compared to a minimal 10 ng input (or 200 ng based on manufacturer’s recommendation) for the Infinium MethylationEPIC BeadChip microarray (hereafter, “MethylationEPIC”) assay. The EM-seq method requires much lower DNA input than the “MethylationEPIC” assay. Thirdly, two (instead of three) enzymes can be used for the identification of 5mC and 5hmC with enzymatic methyl sequencing. Fourthly, while the current tumor classification algorithm can use only MethylationEPIC data as input for tumor classification, the EM-Seq data herein was converted into a format that can be put into the current tumor classification algorithm for classification.

[0051] Results:

[0052] To find an alternative approach of Methylation EPIC array on tumor tissues for brain tumor classification, a platform to perform DNA methylation profiling on cfDNA extracted from CSF samples from brain tumor patients was developed and demonstrated that, as a proof-of-concept study, the resultant cfDNA methylation profiling results can be used for tumor classification.

[0053] CSF samples of two patients with medulloblastoma (MB 11 and MB45) and one patient with high grade glioma (HGG47) were obtained from Connecticut Children’s Medical Center (CCMC) biorepository. cfDNA were extracted from 0.5 or 1 mL CSF and followed by quantification using DNA high sensitivity Qubit assay. The total DNA yield of MB 11, MB45 and HGG47 were 10.85, 2.69 and 5.13 ng, respectively (see Table 1).

[0054] Table 1. Clinicopathological characteristics of patients

[0055] Methylation profiling the cfDNA extracted from CSF was done using the method described herein which is a modified version of EM-Seq kit - NEBNext® Enzymatic Methyl- seq Kit v2 kit. To test the performance of the described method with different cfDNA input amounts, the EM-Seq libraries were prepared with 0.1, 1 and 10 ng input cfDNA. In addition, recommended amount of sheared unmethylated lambda DNA and methylated pUC19 DNA were added to each cfDNA sample as negative and positive controls, respectively. Oxidation of 5-methylcytosines and 5-hydroxymethylcytosines by TET2 and deamination of cytosines by APOBEC were carried out . Sequencing was done on Illumina Novaseq X Plus platform generating paired end reads of 150bp with targeted number of reads of 750 million reads per library.

[0056] DNA methylation profiling of CSF samples from MB 11 and HGG47 yielded molecular classification that is in 100% agreement with classification based on MethylationEPIC data generated from matched tumor tissue. CSF results of MB 45 yielded a correct diagnosis of medulloblastoma but with lower confidence level. Upon further investigation, MethylationEPIC diagnosis actually gave a diagnosis of suboptimal confidence.

[0057] Materials and Methods:

[0058] Patient samples

[0059] De-identified archival CSF samples were obtained from Connecticut Children’s Medical Center (CCMC) biorepository. The study was approved by both CCMC and The Jackson laboratory Institutional Review Board. cfDNA extraction cfDNA was extracted from 500-1,000 pl CSF using Quick-cfDNA Serum & Plasma Kit (Zymo Research) or QIAamp Circulating Nucleic Acid Kit (Qiagen) according to manufacturers’ procedures. The resultant cfDNA was quantified by DNA high sensitivity Qubit assay (Thermo Fisher Scientific) and Cell-free DNA ScreenTape assay (Agilent Technologies).

[0060] Enzymatic methyl-sequencing (EM-Seq) library preparation and sequencing

[0061] Library preparation was performed using a modified version of the NEBNext® Enzymatic Methyl-seq Kit and Unique Dual Index Primer pairs (New England Biolabs). Enzymatic methylation identification was performed using TET2 and APOBEC enzymes. Sheared unmethylated lambda DNA and methylated pUC19 DNA were added to each cfDNA sample as negative and positive controls, respectively. 1:500 control DNA was used for 0.1 ng input sample, while 1:50 control DNA was used for 1 and 10 ng sample. A total of 14, 11 and 8 amplification cycles were used for 0.1 ng, 1 ng and >1 ng of input DNA, respectively. Libraries were stored at -20C prior to sequencing. Quantification of libraries was performed using real-time qPCR Viia7 (Thermo Fisher Scientific). Sequencing was performed on Illumina Novaseq X Plus platform generating paired end reads of 150bp.

[0062] Data processing and tumor classification

[0063] Following confirmation of libraries and sequencing meeting QC criteria, sequencing data was binarized to a format of methylated / unmethylated for each mapped CpG sites to mimic that of the array data. These reformatted data are then inputted into the DKFZ algorithm for tumor classification.

[0064] All references, patents and patent applications disclosed herein are incorporated by reference with respect to the subject matter for which each is cited, which in some cases may encompass the entirety of the document.

[0065] The indefinite articles “a” and “an,” as used herein in the specification and in the claims, unless clearly indicated to the contrary, should be understood to mean “at least one.”

[0066] It should also be understood that, unless clearly indicated to the contrary, in any methods claimed herein that include more than one step or act, the order of the steps or acts of the method is not necessarily limited to the order in which the steps or acts of the method are recited.

[0067] In the claims, as well as in the specification above, all transitional phrases such as “comprising,” “including,” “carrying,” “having,” “containing,” “involving,” “holding,” “composed of,” and the like are to be understood to be open-ended, i.e., to mean including but not limited to. Only the transitional phrases “consisting of’ and “consisting essentially of’ shall be closed or semi-closed transitional phrases, respectively, as set forth in the United States Patent Office Manual of Patent Examining Procedures, Section 2111.03.

[0068] The terms “about” and “substantially” preceding a numerical value mean ±10% of the recited numerical value.

[0069] Where a range of values is provided, each value between and including the upper and lower ends of the range are specifically contemplated and described herein.

Claims

What is claimed is:CLAIMS1. A method, comprising: obtaining cell-free deoxyribonucleic acid (cfDNA) from a sample of cerebrospinal fluid (CSF) from a subject having a brain tumor; profiling DNA methylation of the cfDNA using enzymatic methylation sequencing of the cfDNA to produce sequencing data; and classifying the brain tumor of the subject.

2. The method of claim 1, wherein the amount of cfDNA used in the profiling is less than 5 ng.

3. The method of claim 2, wherein the amount of cfDNA used in the profiling is about 0.1 ng to about 1 ng.

4. The method of any one of the preceding claims, wherein the sample of CSF has a volume of less than 2 mL.

5. The method of claim 4, wherein the sample of CSF has a volume of less than 1 mL.

6. The method of any one of the preceding claims, wherein the enzymatic methylation sequencing comprises: (a) combining the cfDNA with Tet methylcytosine dioxygenase 2 (TET2) and optionally an oxidation enhancer to convert 5-methylcytosine (5mC) and 5- hydroxymethylcytosine (5hmC) of the cfDNA into products that cannot be deaminated by APOBEC3A; and (b) combining the cfDNA with APOBEC3A to convert unmodified cytosines of the cfDNA to uracils.

7. The method of claim 6, wherein the cfDNA is combined with TET2 and the oxidation enhancer, wherein the oxidation enhancer is T4 phage P-glucosyltransferase (T4-BGT).

8. The method of any one of the preceding claims further comprising converting the sequencing data into a binarized format of methylated / unmethylated for each mapped CpG site.

9. The method of any one of the preceding claims, wherein the classifying is performed using a DNA methylation-based tumor classification algorithm.

10. The method of claim 9, wherein the DNA methylation-based tumor classification algorithm is the DKFZ tumor classification algorithm.

Images