Sarah Al Sharie
Khaled Sawaftah2
Maysa Al-Hussaini- 1Laboratory of Science and Translation in Critical Illness, Vanderbilt University Medical Center, Nashville, TN, United States
- 2Department of Surgery, Jordan Hospital, Amman, Jordan
- 3Department of Pathology and Laboratory Medicine, Jordan University of Science and Technology, Irbid, Jordan
- 4Department of Cell Therapy and Applied Genomics, King Hussein Cancer Center, Amman, Jordan
- 5Department of Pathology and Laboratory Medicine, King Hussein Cancer Center, Amman, Jordan
DNA methylation profiling has emerged as a transformative tool in the diagnosis and classification of central nervous system (CNS) tumors. Traditional approaches like histology, immunohistochemistry, and targeted molecular testing cannot fully capture the biological and clinical diversity of these neoplasms. In contrast, genome-wide methylation analysis provides highly reproducible epigenetic “fingerprints” that reflect both lineage and oncogenic alterations, enabling objective tumor classification, refinement of existing categories, and discovery of novel entities. This comprehensive review summarizes the principles of DNA methylation, available platforms, and the application of methylation-based classifiers across major CNS tumor groups, including diffuse gliomas, medulloblastomas, ependymomas, and meningiomas. We highlight how methylation profiling complements other molecular techniques by simultaneously generating copy number profiles and promoter methylation data, consolidating multiple diagnostic assays into a single platform. Practical considerations such as tissue quality, bioinformatic pipelines, interpretation thresholds, and cost are addressed, as are comparisons with sequencing, RNA expression, and immunohistochemistry. Finally, we explore future directions, including nanopore-based rapid testing, liquid biopsy, and artificial intelligence, which promise to extend the reach and clinical utility of methylation profiling. Collectively, these advances are establishing DNA methylation analysis as a cornerstone of precision neuropathology, aligning diagnostic and prognostic assessment with tumor biology to improve patient care.
Introduction
The diagnosis of central nervous system (CNS) tumors has traditionally relied on microscopic evaluation of histological features, supported by immunohistochemistry and, more recently, targeted molecular assays (1). While these methods remain indispensable, it has become increasingly evident that histology alone cannot fully capture the biological diversity of CNS tumors (2). Tumors that appear identical under the microscope may differ profoundly in clinical behavior, prognosis, and therapeutic response (2). Likewise, morphologically distinct tumors may share a common molecular basis (3). Among the tools now available, DNA methylation profiling has emerged as one of the most transformative (4). DNA methylation is an epigenetic modification that regulates gene expression and reflects both the cell of origin and the genetic alterations driving tumor development (5). Each CNS tumor type, and often each subtype, carries a characteristic “methylation fingerprint” (6). These methylation signatures are highly reproducible, stable across time, and measurable on a genome-wide scale (7). The relevance of this approach extends beyond improving diagnostic accuracy (8). DNA methylation profiling has revealed new subgroups within established tumor categories, clarified the boundaries between overlapping entities, and even uncovered entirely novel tumor types (9). It has also contributed valuable prognostic insights, as certain methylation patterns correlate with clinical outcomes more strongly than histological grade alone (10). In this way, methylation analysis not only strengthens diagnosis but also enhances risk stratification and guides patient management (10). Importantly, methylation profiling does not stand in isolation but forms part of an integrated diagnostic framework (11). Rather than replacing histology, immunohistochemistry, or genetic testing, it complements them, bringing an additional dimension that ties morphological and molecular features together (12). This integrated approach improves diagnostic confidence, reduces inter-observer variability, and provides clinicians with a more reliable basis for treatment decisions (12). As the field evolves, methylation profiling is steadily moving from research into routine diagnostics (13). Its application is particularly impactful in cases where morphology is ambiguous, available tissue is limited, or standard molecular tests are inconclusive (14).
While methylation arrays can generate estimates of MGMT promoter methylation and genome-wide copy-number profiles, cIMPACT-NOW Update 9 emphasizes that these outputs must be interpreted cautiously (15). Array-derived MGMT status can be supportive of clinical decision-making but requires internal laboratory validation against established methods (e.g., pyrosequencing) due to variability in probe performance, bisulfite conversion efficiency, and algorithmic thresholds (16). Similarly, copy-number profiles derived from array intensities provide valuable contextual information—such as broad chromosomal gains/losses or focal events—but should not replace dedicated clinical-grade CNV assays unless validated locally (17).
Both MGMT methylation calls and CNV plots are particularly vulnerable to compromised sample quality (18). Low tumor cell content, extensive necrosis, or high inflammatory background can reduce probe detection rates and distort signal intensities, leading to underestimation of methylation levels or unreliable segmentation of copy-number states (19).
This review provides a comprehensive overview of the role of DNA methylation profiling in neuropathological tumor diagnosis. We will outline the principles underlying this approach, its clinical applications across major CNS tumor categories, comparisons with other molecular diagnostic methods, and the practical considerations involved in its use. Finally, we will discuss current limitations and future directions, highlighting how epigenomic profiling is shaping a new era of precision neuropathology.
Fundamentals of DNA methylation
DNA methylation is the covalent addition of a methyl group (CH3) to cytosine bases, typically within CpG dinucleotides (20). Clusters of CpG sites, or CpG islands, often occur in gene promoter regions where their methylation status strongly influences transcription (21). Promoter hypermethylation is usually linked to gene silencing, while gene-body methylation has more nuanced effects (22). As a normal epigenetic mechanism, methylation regulates embryonic development, X-chromosome inactivation, and genomic imprinting (23). Each cell type carries a distinctive methylation landscape reflecting its lineage and state of differentiation (24).
In cancer, characteristic alterations arise such as the widespread hypomethylation that contributes to genomic instability, or focal promoter hypermethylation which silences tumor suppressor genes (25, 26). Importantly, tumors also retain lineage-specific methylation signatures from their cell of origin (27). The resulting composite profile, part developmental imprint, part oncogenic alteration, is highly robust, even in formalin-fixed paraffin-embedded tissue, making it a reliable diagnostic biomarker (28). For instance, IDH-mutant diffuse gliomas display the CpG island methylator phenotype (G-CIMP), while medulloblastomas, ependymomas, and meningiomas exhibit distinct methylome patterns tied to their biology (29). These tumor-specific epigenetic fingerprints often correlate more closely with behavior and prognosis than morphology alone (30).
In addition to intrinsic tumor and lineage-specific epigenetic features, the methylation profile generated by array-based platforms also reflects the contribution of non-neoplastic cells within the tumor microenvironment (31). Tumors with substantial immune infiltration, such as those enriched with macrophages, microglia, or lymphocytes, often display composite methylation signatures influenced by these cell populations, and this can shape how the classifier algorithm assigns them to a methylation class (32). These microenvironment-derived methylation patterns are particularly relevant in high-grade tumors with robust inflammatory components and underscore the importance of interpreting methylation results in the context of tumor purity and histologic background (33).
Genome-wide methylation profiling in clinical practice is most often performed using microarrays (34). Tumor DNA is bisulfite-treated, converting unmethylated cytosines to uracil while leaving methylated cytosines intact, then hybridized to platforms such as the Illumina HumanMethylation450 or EPIC 850K arrays (35). These measure methylation at hundreds of thousands of CpG sites spanning promoters, gene bodies, and regulatory elements enriched for cancer-relevant loci (36). Data are generated as quantitative β-values and analyzed bioinformatically to generate a tumor-specific methylation profile (37). Alternative methods such as targeted bisulfite sequencing or whole-genome bisulfite sequencing offer higher resolution but remain less practical for routine use (38). Arrays currently represent the best balance of coverage (~1–3% of CpGs) and cost-effectiveness in diagnostics (39). Since the introduction of the Illumina HumanMethylation450 (450K) array, methylation microarray technology has undergone two major updates aimed at expanding genomic coverage and improving assay performance. The first transition replaced the 450K with the MethylationEPIC v1 (EPIC 850K) array, which added >350,000 CpG loci enriched for enhancers and regulatory elements, while maintaining broad backward compatibility with 450K-derived datasets and classification frameworks (40). The most recent iteration, the MethylationEPIC v2 (MEP v2) array, has further refined probe chemistry, redistributed low-performing probes, and improved representation of regulatory elements, while preserving backward compatibility emphasized in cIMPACT-NOW Update 9 (41). This continuity allows laboratories to integrate new data with historical 450K and EPIC v1 datasets and ensures ongoing compatibility with widely used classifiers.
From a practical standpoint, cIMPACT-NOW 9 highlights the importance of adequate DNA input, particularly for FFPE tissue (15). While manufacturers may specify lower minimum amounts, most diagnostic laboratories recommend ≥100 ng of high-quality bisulfite-converted DNA to ensure robust signal performance, sufficient probe detection rates, and reliable copy-number output, especially in samples with variable FFPE preservation (42).
In addition to traditional bisulfite-based methods, several next-generation enzymatic methylation sequencing platforms have recently emerged. New England Biolabs’ Enzymatic Methyl-seq (EM-seq) replaces bisulfite conversion with an enzymatic oxidation/protection strategy that preserves DNA integrity, enabling more accurate and less fragmented genome-wide methylation profiling (43). Illumina’s 5-base sequencing chemistry similarly allows simultaneous detection of methylated cytosines and standard base substitutions during whole-genome sequencing, providing an integrated readout of both epigenetic and genetic alterations (44). These platforms represent an important evolution in methylation technology and are likely to become increasingly relevant as sequencing-based assays gain traction in clinical molecular pathology.
Table 1 summarizes the major platforms currently used for DNA methylation profiling in neuropathology, highlighting their coverage, sample requirements, turnaround times, advantages, and limitations.
Table 1. Current DNA methylation profiling platforms in neuropathology.
Principles of methylation-based tumor classification
The most widely used algorithm for CNS tumor methylation classification is the DKFZ/Heidelberg Molecular Neuropathology (MNP) classifier, originally developed at the German Cancer Research Center (DKFZ) and Heidelberg University. The current clinical implementation of this classifier—now commercialized through Heidelberg Epignostix—has expanded considerably beyond the initial training cohort of approximately 2,800 reference cases, incorporating many thousands of additional well-annotated tumors from both adult and pediatric populations (doi: 10.1101/2025.05.28.25328344). This expanded dataset has further refined class boundaries, improved calibration behavior, and strengthened the classifier’s ability to resolve rare or previously ambiguous entities”.
“Recent advancements have also focused on extending the versatility of methylation-based classification across platforms and biospecimen types. Notably, the MNP-Flex model (doi: 10.1038/s41591-025-03562-5) enables platform-agnostic methylation profiling, allowing reliable class assignment from data generated by different array designs and sequencing-based methylation assays. Additional parallel developments include classifier frameworks established at St. Jude Children’s Research Hospital (doi: 10.1038/s41698-024-00718-3) and the NCI Laboratory of Pathology’s Methylscape environment, both of which provide alternative structures for CNS tumor classification, benchmarking, and visualization. These complementary resources collectively broaden the diagnostic and research applicability of methylation-based tumor profiling.
Each CNS tumor type, defined by lineage and key genetic drivers, exhibits a characteristic DNA methylation signature (45). Comparing a tumor’s methylation profile to large reference datasets enables objective assignment of class or subtype (46). The seminal DKFZ classifier assembled a reference library of >2,800 brain tumors spanning 82 methylation classes, covering most recognized entities and several novel ones (6). Unsupervised clustering of these references showed that histologically defined tumors generally segregate into discrete epigenetic clusters (47). Some clusters map one-to-one to WHO entities (Category 1), others reveal molecularly distinct subgroups within single entities (Category 2, e.g., ependymoma or medulloblastoma subtypes), and still others merge previously separate pathologies or define entirely new tumor groups (48).
Operationally, the classifier uses a Random Forest model trained on these reference classes and returns calibrated scores reflecting the probability of membership in each class (49). A single top class typically emerges; scores ≥0.90 are considered confident matches (50). Intermediate scores (≈0.5–0.89) indicate uncertainty or support only a broader assignment to a methylation class family (MCF), for example, families encompassing multiple glioblastoma subclasses or medulloblastoma subtypes, so results can still be informative when sibling subclasses are difficult to distinguish (18, 51). Earlier versions of the classifier (e.g., v11b4 in the 2018 validation study) suggested an exploratory cutoff of 0.84; however, current practice and cIMPACT-NOW Update 9 endorse a ≥0.90 calibrated score as the general benchmark for high-confidence class assignment. Scores below ~0.5 are generally deemed unclassifiable and may reflect either truly novel biology absent from the reference set or suboptimal sample quality (13).
Importantly, cIMPACT-NOW Update 9 emphasizes that calibrated classifier scores exist on a continuum rather than representing a binary ‘match’ or ‘no-match’ outcome (15). Subthreshold scores, particularly those within a relevant methylation class family, may still meaningfully support a diagnosis when histology, immunophenotype, and molecular findings are concordant (15). Thus, classifier results should be interpreted within an integrated framework rather than rejected solely because the calibrated value falls slightly below a numerical threshold.
Different classifier implementations are currently in use, including the DKFZ/Heidelberg classifier (e.g., versions 11b4 through 12.8) and the NCI/Bethesda classifier, each with distinct training sets and class structures (52). Laboratories should specify which classifier version was applied, as updates may refine class boundaries, introduce new entities, or recalibrate probability estimates.
Current clinical practice follows classifier-specific interpretation rules. For the Heidelberg/DKFZ MNP classifier (versions 11b4 through 12.x), a calibrated score ≥0.90 is considered a high-confidence match to a specific methylation class. Scores between 0.50–0.89 support classification only at the methylation class family level, while scores <0.50 are generally considered non-classifiable. In contrast, the NCI/Bethesda Methylscape classifier uses a two-level scoring system requiring both a high super-family score and a sufficiently high class-level score for a definitive match. Specifically, the super-family score must exceed the model-defined threshold (typically ≥0.85), and the class-level score should be ≥0.90 for a high-confidence call. If the super-family score is high but the class score falls below threshold, the result should be reported as a super-family assignment only. Because reporting rules differ between classifiers, laboratories must follow the interpretation guidelines of the specific classifier used and should explicitly document classifier version, score thresholds, and match level (class, family, or super-family) in the final integrated diagnosis.
As an example:
Histopathology: Posterior fossa tumor with classic ependymoma morphology (perivascular pseudorosettes) and retained EMA expression.
Methylation profiling (DKFZ/Heidelberg v12.x classifier):
● Super-family: Posterior fossa ependymoma family calibrated score 0.94
● Best-matching class: PFA calibrated score 0.87 (below class-level match threshold; therefore reported as family-level match only)
● Integrated diagnosis: Posterior fossa ependymoma, PFA (family-level methylation support), WHO CNS5 criteria.
Diagnosis is based primarily on histopathologic findings and supported by methylation profiling and H3K27me3 loss.
Beyond classification, methylation arrays yield genome-wide copy number variation (CNV) profiles derived from probe intensities (53). In a single assay, one can detect hallmark alterations such as the 1p/19q codeletion in oligodendroglioma, EGFR amplification in glioblastoma, MYCN amplification, or CDKN2A/B homozygous deletions (54). This CNV layer complements the class call and adds diagnostic and prognostic value (e.g., confirming an IDH-mutant tumor with 1p/19q codeletion as oligodendroglioma) (55). Arrays can also report promoter methylation at clinically relevant loci, most notably MGMT in glioblastoma, which informs temozolomide responsiveness; although often performed by separate assays, MGMT status can be extracted from array data (56).
cIMPACT-NOW update 9: practical recommendations for diagnostic use
cIMPACT-NOW Update 9 provides detailed guidance on when and how genome-wide DNA methylation profiling should be incorporated into clinical neuropathology (15). The update emphasizes that methylation profiling represents one diagnostic layer within the WHO/ICCR integrated reporting framework, to be interpreted alongside histology, immunophenotype, and targeted molecular testing. Reports should explicitly document the classifier version, calibrated score, and match level (“match,” “match within family,” or “no match”), thereby ensuring transparency and reproducibility across institutions (15).
The update also recommends the use of dimensionality-reduction visualization tools, such as UMAP or t-SNE, particularly for borderline or ambiguous cases (57). These plots allow the queried tumor to be visually compared to reference clusters, helping assess whether a subthreshold result is nonetheless topologically consistent with a recognized class or family (57).
Importantly, cIMPACT-NOW 9 clarifies the relationship between methylation classes and WHO tumor types, noting that some classes map directly to WHO entities while others represent biologically meaningful subgroups or families that require integration with additional data (see update 9 Tables 1–3) (15). Thus, classification should not be used in isolation but incorporated into a layered diagnostic model that clearly communicates the degree of confidence and any limitations.
Table 2. Characteristic methylation signatures of major CNS tumor entities.
Table 3. Diagnostic vs prognostic utility of methylation profiling.
A practical workflow aligned with update 9 includes:
1. When to order profiling:
● When histopathological findings and/or other molecular tests (IHC, targeted sequencing, FISH) are ambiguous, discordant, or insufficient to reach a confident diagnosis.
● When the differential diagnosis includes entities where methylation profiling provides essential diagnostic or prognostic stratification (e.g., PFA vs PFB ependymoma, pediatric-type diffuse high-grade gliomas, rare methylation-defined CNS tumors).
● When conventional molecular biomarkers are negative, equivocal, or fail to explain the observed histologic features.
2. How to integrate results into layered reporting:
● Include classifier version (e.g., DKFZ/Heidelberg v12.8, NCI/Bethesda), calibrated score, match level;
● Add visualization output (UMAP/t-SNE) when results are borderline;
● Clearly state whether the methylation class corresponds to a WHO entity or to a broader class family.
3. Management of low-confidence or discordant results:
1. Reassess technical quality and tumor purity.
Low tumor content, FFPE degradation, necrosis, or poor probe detection can all produce subthreshold classifier scores. Re-examining the selected tissue block and repeating extraction or macrodissection is often necessary.
2. Evaluate biological concordance with histology, immunophenotype, and known driver alterations.
This step is of primary importance. A methylation result—especially a low-confidence or borderline one—must be interpreted in light of the tumor’s morphology, IHC profile (e.g., IDH1 R132H, H3K27me3, SMARCB1), and molecular findings (e.g., IDH mutation, 1p/19q codeletion, ZFTA/YAP1 fusion). Concordance or discordance here strongly determines the validity of the classifier output.
3. Manually review the CNV profile for consistency with the suspected entity and other diagnostic data.
CNV plots derived from methylation arrays must be interpreted manually, as no automated scoring or decision system exists. The goal is to assess whether observed chromosomal alterations (e.g., 1p/19q codeletion, +7/–10, CDKN2A/B homozygous deletion, MYCN amplification) support or contradict the histology and methylation class family. This is a correlation exercise—not an automated validation step.
4. If discordance persists, report the case transparently as low-confidence or uncertain.
Following WHO/ICCR layered reporting and cIMPACT-NOW 9, such cases should be documented with explicit mention of which diagnostic layers are supportive, conflicting, or inconclusive.
Clinical applications in neuropathology
Diffuse gliomas
Diffuse gliomas comprise adult-type astrocytomas and oligodendrogliomas, as well as pediatric-type diffuse gliomas (58). WHO 2021 stratifies them by key alterations such as IDH mutation, 1p/19q codeletion, and histone mutations (1). Methylation profiling provides strong support for this framework (59). Adult-type gliomas segregate into three entities: astrocytoma, IDH-mutant; oligodendroglioma, IDH-mutant and 1p/19q-codeleted; and glioblastoma, IDH-wildtype (60). The classifier distinguishes these with high fidelity, since IDH mutations induce a CpG island methylator phenotype absent in IDH-wildtype glioblastomas (61). Oligodendrogliomas form a separate cluster, reinforced by detection of the 1p/19q codeletion on copy number plots (62). IDH-mutant astrocytomas often display a “G-CIMP-high” profile, correlating with favorable prognosis (63). Table 2 summarizes the characteristic methylation signatures and ancillary genetic hallmarks of major CNS tumor entities currently defined by the 2021 WHO classification.
In glioblastomas, methylation has revealed biologically distinct subgroups such as RTK I, RTK II, and Mesenchymal, which parallel transcriptional subtypes (64). While not yet guiding therapy, these have prognostic trends and may influence future management (65). Methylation profiling is particularly useful for identifying IDH-wildtype tumors that histologically resemble lower-grade astrocytomas but biologically align with glioblastoma. According to WHO CNS5, if methylation profiling and integrated diagnostic data support a glioblastoma-type methylation class or demonstrate characteristic glioblastoma-associated molecular alterations, the tumor should be designated as ‘Glioblastoma, IDH-wildtype, WHO grade 4,’ rather than ‘Diffuse astrocytic glioma, IDH-wildtype, with molecular features of glioblastoma.’ The latter terminology is obsolete and no longer used in contemporary practice (66). The classifier can also act as a surrogate when genetic testing is unavailable, though confirmation of critical mutations remains advisable (66).
In pediatric gliomas, methylation profiling defined entities such as diffuse midline glioma, H3 K27-altered, and diffuse hemispheric glioma, H3 G34-mutant (67). It also contributed to delineating the broader group of pediatric high-grade gliomas, which contain multiple methylation-based subsets under study (68). Even in low-grade pediatric gliomas, methylation aids in clarifying diagnosis (69). For example, distinguishing BRAF fusion positive diffuse low-grade glioma from more aggressive infant-type hemispheric gliomas with RTK fusions (70, 71).
Recent methylation analyses have shown that diffuse midline gliomas, H3K27-altered, segregate into at least two robust methylation subgroups (DMG-K27A and DMG-K27B), with distinct developmental and clinical implications (72). These refinements are not captured by histopathology or single-gene assays, emphasizing the essential role of methylation profiling for accurate subclassification.
Medulloblastomas
The application of DNA methylation-based classification to medulloblastoma exemplifies how epigenomic profiling can profoundly reshape tumor taxonomy (73). WHO now recognizes four core molecular subgroups: WNT-activated, SHH-activated (TP53-wildtype or mutant), Group 3, and Group 4 (74). Methylation profiling robustly assigns tumors to these subgroups and consistently outperforms immunohistochemical surrogates (75).
WNT-activated tumors, which frequently harbor CTNNB1 mutations and monosomy 6, form a tight epigenetic cluster and have an excellent prognosis, supporting treatment de-escalation strategies (76). SHH-activated tumors, driven by pathway alterations and further subdivided by TP53 status, demonstrate variable outcomes; TP53-mutant pediatric SHH tumors in particular carry a poor prognosis (77). Within the SHH group, additional methylation-defined subclusters (α–δ) capture age-related and genetic differences, although these remain investigational (78).
Group 3 and Group 4 tumors—historically merged under a single category—were first differentiated by methylation profiling, which revealed distinct biological and clinical behaviors (79). Group 3 tumors, often characterized by MYC amplification, have the worst prognosis, whereas Group 4 tumors, frequently associated with features such as isochromosome 17q, are more common and exhibit intermediate outcomes (80).
It is important to note that the foundational subdivision of medulloblastoma into WNT, SHH, Group 3, and Group 4 was initially established through integrative transcriptomic and copy-number analyses, particularly via gene-expression array profiling (81). Subsequent DNA methylation-based classification reproduced these biologically meaningful subgroups with high fidelity and provided a more stable, clinically scalable framework for diagnostic subgroup assignment.
Ependymomas
DNA methylation profiling was central to establishing the modern molecular taxonomy of ependymomas, enabling the separation of posterior fossa tumors into the biologically distinct PFA and PFB groups and redefining several supratentorial categories (82). In the supratentorial compartment, methylation distinguished ZFTA fusion–positive tumors from YAP1 fusion–positive ones, which differ in biology and prognosis (83). In the posterior fossa, methylation studies identified two major groups: PFA, hypermethylated and aggressive with H3K27me3 loss, and PFB, associated with older patients and better outcomes (84). These cannot be separated reliably by histology, making DNA methylation the diagnostic gold standard (84).
It is important to note that the original separation of posterior fossa ependymomas into the PFA and PFB subgroups was first identified through transcriptome-based stratification using expression array profiling (85). Subsequent DNA methylation profiling reproduced these two core groups with high concordance and provided a more robust, reproducible, and clinically applicable framework for classification (85). This alignment between expression-based and DNA methylation–based clustering helped establish PFA and PFB as biologically and clinically distinct disease entities.
Spinal ependymomas usually show NF2 alterations, but methylation identified a distinct, aggressive MYCN-amplified subtype, now a separate WHO entity (86). Importantly, methylation arrays simultaneously detect fusion status, CNVs, and subgroup assignment, often resolving diagnostic ambiguities (87).
DNA methylation profiling is essential for accurately distinguishing PFA from PFB posterior fossa ependymomas, a separation that cannot be reliably achieved by histology, immunohistochemistry, or targeted sequencing alone (88). PFA tumors show a characteristic hypermethylated profile and loss of H3K27me3, whereas PFB tumors cluster separately and maintain H3K27me3 expression (89). However, classifier accuracy is highly dependent on tumor purity: low-cellularity samples, biopsies with abundant reactive tissue, or tumors with extensive necrosis frequently produce subthreshold scores or ambiguous class-family assignments (90). Furthermore, cIMPACT-NOW Update 11 provides updated diagnostic criteria for both posterior fossa ependymoma and pediatric-type diffuse high-grade gliomas (IDH-wt, H3-wt), emphasizing the importance of integrated molecular–methylation analysis for entities previously categorized as “NOS” or “NEC” (91).
Meningiomas
Histologic grading has limited predictive power for meningiomas, as some grade 1 tumors recur aggressively while others remain indolent (92). Methylation studies have proposed biologically driven risk groups, now confirmed across multiple cohorts (93). Classes include benign Merlin-intact tumors with low recurrence risk, immune-enriched tumors with intermediate behavior, and proliferative or hypermetabolic tumors with poor prognosis (94). These classes align partly with genetic features such as NF2, CDKN2A/B, and TERT promoter mutations (95).
Methylation-based stratification predicts recurrence more accurately than histologic grade alone, especially within grade 2 meningiomas, where outcomes vary widely (96). This has major clinical implications for postoperative management, influencing decisions regarding surveillance or adjuvant therapy (97). While not yet formalized in WHO 2021, methylation groups are likely to be incorporated into future classification systems, given their superior prognostic performance and therapeutic relevance (15).
Recent studies further demonstrate that methylation-based risk groups consistently outperform conventional WHO grading in predicting recurrence, even when adjusting for extent of resection (96). These risk groups should be interpreted in conjunction with key molecular alterations, most notably CDKN2A/B homozygous deletion and TERT promoter mutations, both of which designate WHO grade 3 irrespective of histology and strongly correlate with high-risk methylation classes (98).
In addition, cIMPACT-NOW Update 8 reinforces the central role of DNA methylation profiling in meningioma classification and recommends its integration into routine diagnostic practice, particularly for tumors with borderline histologic features or discordant clinical behavior (99).
Rare and ambiguous CNS tumors
Methylation profiling is particularly powerful in rare or diagnostically uncertain tumors (59). Historically vague categories, such as CNS primitive neuro-ectodermal tumors, have been redefined through methylome analysis into distinct entities like CNS neuroblastoma, FOXR2-activated, and CNS tumor with BCOR internal tandem duplication (100). Similarly, astroblastoma, MN1-altered, and several pediatric glioneuronal tumors (e.g., PLNTY, DGONC) were delineated through methylation-based clustering (101).
The approach also refined existing categories, as with atypical teratoid/rhabdoid tumors (AT/RT), where methylation identified TYR, SHH, and MYC subgroups with different clinical associations (102). For diagnostically ambiguous or “NOS” cases, methylation often provides a definitive match, guiding clinical management (103). Even when no exact class match is achieved, methylation data can place tumors within a broader family, offering useful diagnostic direction (104). Importantly, many new CNS tumor types recognized in WHO 2021 originated from such “no match” clusters in methylation space, underscoring the method’s role in discovery as well as diagnosis (6).
Collectively, these examples demonstrate how methylation profiling not only clarifies histologic diagnosis but also provides prognostic and sometimes predictive information across tumor types. Table 3 summarizes the diagnostic and prognostic contributions of methylation profiling across key CNS tumor entities.
Practical considerations in the diagnostic workflow
Implementing DNA methylation profiling in diagnostic neuropathology requires attention to technical, logistical, and interpretive aspects (4). Adequate tissue and DNA quality are fundamental (4). While methylation arrays can be performed on Formalin-Fixed Paraffin-Embedded (FFPE) tissue, the DNA must not be too degraded and should be available in sufficient quantity, typically >250 ng for the Illumina EPIC 850K array (105). Low tumor content poses difficulties, as background signal from non-neoplastic brain or inflammatory cells can obscure the tumor profile (106). Small biopsies or diffuse gliomas with sparse cellularity are especially problematic (68). Macrodissection of slides to enrich tumor tissue is therefore recommended before DNA extraction (107). Figure 1 provides a simplified overview of where DNA methylation profiling fits within the broader diagnostic workflow, illustrating how histopathology, immunohistochemistry, targeted sequencing, and methylation-based classification are integrated into routine practice.
Figure 1. Schematic overview of how DNA methylation profiling is implemented in clinical neuropathology. DNMT, DNA methyltransferase; SAM, S-adenosyl-methionine; MGMT, O-6-methylguanine-DNA methyltransferase; RNA Pol II, RNA polymerase II; Me, methyl group.
The standard workflow includes DNA extraction, bisulfite conversion, amplification and hybridization to the array, scanning and data processing, and classification (108). In practice, this takes 3–7 days, with final reports available in 1–2 weeks (108). Some laboratories batch samples for efficiency, which may introduce minor delays (109). As a result, methylation profiling generally informs the finalized diagnostic report and postoperative planning, rather than intraoperative decisions (110). Efforts are underway to shorten turnaround times using nanopore sequencing, which can generate methylation data within hours or days, though this is not yet routine in clinical laboratories (111).
Cost and availability remain important considerations (112). A single array test may cost several hundred to over one thousand US dollars, and not all hospitals have access to on-site facilities (113). Consequently, many samples are referred to specialized centers (113).
Each array includes probes that monitor bisulfite conversion efficiency and hybridization quality, and laboratories must ensure these falls within acceptable limits (114). Batch effects are another challenge, as running samples at different times or on different array models can introduce minor variations (115). Although the classifier is designed to accommodate these, best practice is to run samples in consistent batches with controls (116). Bioinformatics support may still be needed for data handling and quality control, but the Heidelberg and Methylscape classifiers automatically generate key outputs, including copy-number plots, so manual CNV computation is generally not required (117). Many centers have automated pipelines to standardize analysis and reduce variability (118).
Interpretation of classifier results is essential for determining whether the DNA methylation profile reliably supports a specific CNS tumor class and for integrating this information with morphology, immunohistochemistry, and molecular findings (119, 120). The output from the DKFZ/Heidelberg Molecular Neuropathology (MNP) classifier, currently the most widely used diagnostic tool, is provided as a set of calibrated scores representing the confidence of assignment to each reference methylation class (18). These calibrated scores form a continuum of classifier confidence rather than a strict binary decision, and their interpretation requires attention to class-specific thresholds, match levels, and ancillary data (18). Because subsequent steps in the diagnostic workflow (including reporting of match level and the use of score cutoffs) depend directly on this classifier behavior, clear understanding of the classifier’s output structure is critical (14).
A score ≥0.90 is typically considered a confident assignment, while scores between 0.5 and 0.89 suggest a possible but less certain match, often only at the level of a broader methylation class family (120). Scores below 0.5 are usually regarded as unclassifiable, possibly indicating novel biology or poor sample quality (18). Importantly, the relative distribution of scores matters: a top score of 0.93 with all others near zero provides strong confidence, whereas similar scores across multiple classes indicate uncertainty or a tumor with intermediate features (121).
Despite its robustness, classifier output must always be interpreted within clinical and histopathologic context (122). Even a high score can occasionally be misleading, especially in cases with heavy inflammatory infiltration or unusual morphology (123). The cIMPACT-NOW consortium emphasizes integration, recommending that methylation findings be reported alongside histology, immunohistochemistry, and molecular data (15).
Finally, the integration of methylation profiling requires a multidisciplinary approach. Molecular tumor boards increasingly review methylation findings in conjunction with genomic and histopathologic data (124). Pathologists often present classifier scores, copy number profiles, and clustering results to explain diagnostic changes to oncologists and neurosurgeons (14). This collaborative discussion builds confidence in the method and facilitates clinical translation (14). As experience grows, methylation profiling is transitioning from a research tool to an essential diagnostic pillar, particularly for difficult cases and novel entities (14).
Comparisons with other molecular diagnostic approaches
DNA methylation profiling is a powerful diagnostic tool that complements rather than replaces other molecular techniques (59). Sequencing remains essential for detecting specific mutations such as IDH1/2, TERT promoter, BRAF V600E or rare fusions, which drive therapy choices and clinical trial eligibility (125). Methylation arrays do not directly identify sequence changes but reveal the downstream epigenetic effects of genetic events and can reclassify tumors when conventional tests give ambiguous or conflicting results (126). For example, IDH1 immunostaining and 1p/19q FISH may suffice for classic oligodendroglioma, but methylation profiling helps resolve contradictory findings or uncover unexpected entities such as ependymoma profiles in presumed glioblastoma (127).
Compared with RNA expression profiling, methylation analysis is more stable and reproducible, especially in FFPE tissue, and provides additional copy number information (128). While expression data can still be useful for pathway analysis, methylation has largely replaced expression arrays for CNS tumor classification, with integrated approaches that combine methylation, mutation, and expression now under study (6).
Methylation arrays also yield genome-wide copy number plots with high concordance to traditional cytogenetic techniques, reliably detecting hallmark alterations such as 1p/19q codeletion, +7/–10, EGFR amplification, or CDKN2A/B deletion (129). This often reduces the need for separate FISH or chromosomal microarray testing, although focal or cryptic alterations may still require targeted assays (130). Immunohistochemistry, meanwhile, remains indispensable for rapid, inexpensive detection of protein-level changes like IDH1 p.R132H, H3K27M, or SMARCB1 loss, but methylation offers more objective classification and deeper subclassification when staining is equivocal (131).
Emerging single-cell and spatial methods promise unprecedented resolution of intratumoral heterogeneity and epigenetic diversity but are not yet part of routine diagnostics (132). Currently, methylation profiling serves as a practical “broad brush” tool, consolidating classification, copy number analysis, and some immunohistochemical surrogates in a single assay, while sequencing adds precise mutational data (133). Together these technologies are moving the field toward integrated molecular neuropathology reports that unite genotype and epigenotype to guide diagnosis, prognosis, and treatment planning (129, 130, 132). Table 4 compares methylation profiling with sequencing panels, RNA expression profiling, and immunohistochemistry, emphasizing their respective strengths and limitations in clinical neuropathology.
Table 4. Comparison of methylation profiling with other molecular approaches.
Practical summary of best-practice recommendations for diagnostic methylation profiling
1. Ensure adequate tissue and DNA quality.
Optimal performance of methylation arrays requires sufficient material and high-quality DNA (134). For EPIC v1 and v2 platforms, at least 100 ng of well-preserved FFPE DNA is generally recommended, as suboptimal or heavily degraded DNA is associated with low-confidence classifier outputs (41). Tissue regions used for extraction should contain a high proportion of viable tumor cells, ideally 70% or greater, because excessive non-neoplastic tissue, treatment effect, or necrosis reduces the tumor signal and undermines both calibrated scores and CNV profiles (135).
2. Use macrodissection to enhance tumor enrichment.
Guided macrodissection of H&E-marked slides remains a simple but essential step for ensuring tumor purity (107). Removing admixed normal brain, vascular tissue, calcifications, hemorrhage, or necrotic areas helps maintain epigenetic signal quality and prevents dilution of the tumor-specific profile (136). In infiltrative gliomas, selecting regions with cytologic atypia or increased proliferation improves classifier performance and minimizes the risk of underrepresentation of the neoplastic component (136).
3. Apply rigorous pre-analytical quality control.
Pre-array quality control measures, including ΔCt thresholding, assessment of bisulfite conversion efficiency, and evaluation of DNA integrity, are crucial for determining whether a sample is suitable for reliable methylation analysis (134). Poor-quality DNA frequently leads to subthreshold calibrated scores or uninterpretable profiles (137). When QC metrics fall below acceptable levels, repeating the extraction or selecting an alternative block is preferable to proceeding with technically inadequate material (138).
4. Select cases thoughtfully to maximize clinical utility.
Methylation profiling should be applied selectively in cases where it is expected to change diagnostic interpretation (139). It is most valuable in scenarios where morphology and immunohistochemistry are ambiguous or conflicting, where essential molecular biomarkers yield negative or indeterminate results, or where the differential diagnosis includes pediatric-type high-grade gliomas, posterior fossa ependymoma subgroups, or rare and emerging methylation-defined entities such as HGAP, HPAP, or BCOR-altered tumors (89). Conversely, when a diagnosis is straightforward based on morphology, immunophenotype, and targeted sequencing, or when the biopsy is too small or dominated by necrosis, methylation profiling offers limited benefit and may not be appropriate (140).
5. Interpret results within a layered and integrated diagnostic framework.
Methylation classifier outputs should be interpreted within the WHO/ICCR layered reporting structure and understood as a continuum rather than a binary outcome (46). Each report should specify the classifier version used (for example, Heidelberg v12.8 or NCI/Bethesda), the calibrated score, and the level of match achieved, whether “match,” “family match,” or “no-match” (15). In borderline cases or subthreshold matches, dimensionality-reduction tools such as UMAP or t-SNE can help visualize relationships to reference classes (141). Ultimately, the methylation result must be reconciled with morphology, immunohistochemistry, targeted sequencing, and tumor location to achieve a coherent integrated diagnosis (142).
6. Maintain expert neuropathologic oversight.
Despite the increasing diagnostic power of methylation profiling, expert pathology review remains indispensable (143). Array results cannot substitute for comprehensive assessment of histology and ancillary molecular data (143). In cases where the classifier output conflicts with morphologic or genetic findings, the final interpretation should prioritize integrated reasoning rather than a single modality (144). The role of the neuropathologist is therefore central to ensuring that methylation data complements, rather than replaces, traditional diagnostic expertise. As summary, Table 5 presents common pitfalls in methylation profiling and troubleshooting approaches.
Table 5. Common pitfalls in methylation profiling and troubleshooting approaches.
The central role of the neuropathologist in synthesizing histologic, immunophenotypic, molecular, and methylation data is outlined in Figure 2.
Figure 2. Practical diagnostic workflow illustrating how histopathology, immunohistochemistry, targeted molecular testing, and DNA methylation profiling are combined in routine neuropathology practice.
Future directions
The field of methylation profiling in neuropathology is advancing rapidly, promising greater diagnostic precision and broader clinical use (145). As more CNS tumors, including rare and underrepresented types, are profiled worldwide, reference libraries will expand, enabling classifiers to add new tumor categories and refine existing groups into more homogeneous subtypes (146). Larger datasets will also improve discrimination between closely related entities, provided that new array- or sequencing-based platforms maintain backward compatibility with existing data (147).
Emerging sequencing-based methylation technologies are likely to broaden the diagnostic utility of methylation profiling in the near future. Enzymatic Methyl-seq (EM-seq), by avoiding bisulfite-induced DNA damage, offers higher-quality data from FFPE samples and may become a preferred approach for genome-wide methylation profiling (148). As these technologies mature and undergo clinical validation, they may provide a unified workflow for comprehensive molecular diagnostics.
While methylation arrays remain the current standard, sequencing approaches are gaining momentum (149). Whole-genome bisulfite sequencing offers comprehensive coverage but is resource-intensive, whereas reduced-representation or targeted bisulfite sequencing may provide efficient coverage of key CpGs (150). Long-read platforms such as Oxford Nanopore can directly detect methylation without bisulfite conversion, potentially integrating methylation class, copy number, and mutational data within 24–48 hours (151). With falling costs and more mature bioinformatics pipelines, sequencing-based methylation may supplant arrays over the next decade (152).
Liquid biopsy is another frontier (153). Because each tumor class exhibits a distinctive methylation signature, sensitive assays applied to plasma or cerebrospinal fluid could allow diagnosis without surgical biopsy and enable non-invasive monitoring (154). Early studies in gliomas and medulloblastomas show that CSF-derived DNA can be classified successfully, particularly valuable for high-risk lesions such as diffuse midline gliomas, although detecting low-abundance tumor DNA remains challenging (155).
Moreover, Deep-learning models trained on routine H&E slides already predict some mutations and could be extended to infer methylation classes directly from morphology, offering rapid triage or screening (156). Beyond reproducing current classifiers, AI may identify novel prognostic methylation features and integrate them into digital pathology workflows to enhance precision (157).
Finally, methylation-defined subgroups open opportunities for therapeutic stratification (158). Epigenetically defined glioblastoma or medulloblastoma subtypes may respond differently to targeted or immune therapies, just as MGMT promoter methylation predicts temozolomide response (159). The success of CNS classifiers is spurring development in other tumor systems, including pediatric sarcomas, renal tumors, and nerve-sheath tumors, with the longer-term goal of pan-cancer methylation classifiers that can pinpoint tissue of origin or distinguish primary CNS tumors from metastases (6).
Conclusions
DNA methylation profiling has transformed neuropathological diagnosis by providing an objective and reproducible method for classifying CNS tumors. Unique epigenetic signatures now define more than a hundred tumor types; many incorporated into WHO classifications. This approach improves diagnostic accuracy, clarifies ambiguous cases, and enhances prognostic stratification, with clear benefits shown in gliomas, medulloblastomas, ependymomas, and meningiomas. Its greatest strength lies in an integrated framework, where methylation complements histology, immunohistochemistry, and sequencing. Beyond classification, arrays deliver valuable ancillary data such as copy number and promoter methylation, consolidating multiple analyses into one. Methylation profiling occupies a distinct niche: it does not replace mutation testing or rapid immunohistochemistry, but it provides a genome-wide epigenetic context that unifies diverse findings into a coherent diagnosis. Looking forward, advances in sequencing, liquid biopsy, and AI will expand its speed, accessibility, and clinical relevance. Ultimately, methylation profiling is establishing itself as a core pillar of precision neuropathology, aligning diagnosis and prognosis with tumor biology to improve patient care.
Author contributions
SS: Validation, Data curation, Visualization, Conceptualization, Writing – review & editing, Methodology, Software, Writing – original draft. KS: Writing – review & editing, Methodology, Data curation, Software, Conceptualization, Writing – original draft. HQ: Conceptualization, Methodology, Writing – original draft, Writing – review & editing, Investigation. MA-H: Conceptualization, Funding acquisition, Writing – review & editing, Writing – original draft, Visualization, Project administration, Methodology, Resources.
Funding
The author(s) declared that financial support was not received for this work and/or its publication.
Conflict of interest
The authors declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
1. Louis DN, Perry A, Wesseling P, Brat DJ, Cree IA, Figarella-Branger D, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro-Oncology. (2021) 23:1231–51. doi: 10.1093/neuonc/noab106
2. Arvanitis CD, Ferraro GB, and Jain RK. The blood-brain barrier and blood-tumour barrier in brain tumours and metastases. Nat Rev Cancer. (2020) 20:26–41. doi: 10.1038/s41568-019-0205-x
3. Li X, Liu D, Wu Z, and Xu Y. Diffuse tumors: Molecular determinants shared by different cancer types. Comput Biol Med. (2024) 178:108703. doi: 10.1016/j.compbiomed.2024.108703
4. Pages M, Uro-Coste E, Colin C, Meyronet D, Gauchotte G, Maurage CA, et al. The Implementation of DNA Methylation Profiling into a Multistep Diagnostic Process in Pediatric Neuropathology: A 2-Year Real-World Experience by the French Neuropathology Network. Cancers (Bsl). (2021) 13:1377. doi: 10.3390/cancers13061377
5. Kim M and Costello J. DNA methylation: an epigenetic mark of cellular memory. Exp Mol Med. (2017) 49:e322–e. doi: 10.1038/emm.2017.10
6. Capper D, Jones DTW, Sill M, Hovestadt V, Schrimpf D, Sturm D, et al. DNA methylation-based classification of central nervous system tumours. Nature. (2018) 555:469–74. doi: 10.1038/nature26000
7. Feinberg AP, Irizarry RA, Fradin D, Aryee MJ, Murakami P, Aspelund T, et al. Personalized epigenomic signatures that are stable over time and covary with body mass index. Sci Transl Med. (2010) 2:49ra67. doi: 10.1126/scitranslmed.3001262
8. Arora J, Nassar M, and Baraka B. Unravelling the potential of plasma DNA methylation in the detection and surveillance of esophageal cancer. World J Gastrointest Oncol. (2025) 17:103333. doi: 10.4251/wjgo.v17.i6.103333
9. Witte T, Plass C, and Gerhauser C. Pan-cancer patterns of DNA methylation. Genome Med. (2014) 6:66. doi: 10.1186/s13073-014-0066-6
10. Lietz CE, Newman ET, Kelly AD, Xiang DH, Zhang Z, Luscko CA, et al. Genome-wide DNA methylation patterns reveal clinically relevant predictive and prognostic subtypes in human osteosarcoma. Commun Biol. (2022) 5:213. doi: 10.1038/s42003-022-03117-1
11. Zhang YW, Zheng Y, Wang JZ, Lu XX, Wang Z, Chen LB, et al. Integrated analysis of DNA methylation and mRNA expression profiling reveals candidate genes associated with cisplatin resistance in non-small cell lung cancer. Epigenetics. (2014) 9:896–909. doi: 10.4161/epi.28601
12. Bertogliat MJ, Morris-Blanco KC, and Vemuganti R. Epigenetic mechanisms of neurodegenerative diseases and acute brain injury. Neurochem Int. (2020) 133:104642. doi: 10.1016/j.neuint.2019.104642
13. Leung ML, Abdullaev Z, Santana-Santos L, Skaugen JM, Moore S, and Ji J. Microarray-Based DNA Methylation Profiling: Validation Considerations for Clinical Testing. J Mol Diagn. (2024) 26:447–55. doi: 10.1016/j.jmoldx.2024.02.001
14. Lee K, Jeon J, Park JW, Yu S, Won JK, Kim K, et al. SNUH methylation classifier for CNS tumors. Clin Epigenet. (2025) 17:47. doi: 10.1186/s13148-025-01824-0
15. Aldape K, Capper D, von Deimling A, Giannini C, Gilbert MR, Hawkins C, et al. cIMPACT-NOW update 9: Recommendations on utilization of genome-wide DNA methylation profiling for central nervous system tumor diagnostics. Neurooncol Adv. (2025) 7:vdae228. doi: 10.1093/noajnl/vdae228
16. Rosas-Alonso R, Colmenarejo-Fernandez J, Pernia O, Rodriguez-Antolín C, Esteban I, Ghanem I, et al. Clinical validation of a novel quantitative assay for the detection of MGMT methylation in glioblastoma patients. Clin Epigenet. (2021) 13:52. doi: 10.1186/s13148-021-01044-2
17. Soong D, Stratford J, Avet-Loiseau H, Bahlis N, Davies F, Dispenzieri A, et al. CNV Radar: an improved method for somatic copy number alteration characterization in oncology. BMC Bioinf. (2020) 21:98. doi: 10.1186/s12859-020-3397-x
18. Capper D, Stichel D, Sahm F, Jones DTW, Schrimpf D, Sill M, et al. Practical implementation of DNA methylation and copy-number-based CNS tumor diagnostics: the Heidelberg experience. Acta Neuropathol. (2018) 136:181–210. doi: 10.1007/s00401-018-1879-y
19. Le J, Dian Y, Zhao D, Guo Z, Luo Z, Chen X, et al. Single-cell multi-omics in cancer immunotherapy: from tumor heterogeneity to personalized precision treatment. Mol Cancer. (2025) 24:221. doi: 10.1186/s12943-025-02426-3
20. Moore LD, Le T, and Fan G. DNA methylation and its basic function. Neuropsychopharmacology. (2013) 38:23–38. doi: 10.1038/npp.2012.112
21. Deaton AM and Bird A. CpG islands and the regulation of transcription. Genes Dev. (2011) 25:1010–22. doi: 10.1101/gad.2037511
22. Yang X, Han H, De Carvalho DD, Lay FD, Jones PA, and Liang G. Gene body methylation can alter gene expression and is a therapeutic target in cancer. Cancer Cell. (2014) 26:577–90. doi: 10.1016/j.ccr.2014.07.028
23. Fedoriw A, Mugford J, and Magnuson T. Genomic imprinting and epigenetic control of development. Cold Spring Harb Perspect Biol. (2012) 4:a008136. doi: 10.1101/cshperspect.a008136
24. Roy R, Ramamoorthy S, Shapiro BD, Kaileh M, Hernandez D, Sarantopoulou D, et al. DNA methylation signatures reveal that distinct combinations of transcription factors specify human immune cell epigenetic identity. Immunity. (2021) 54:2465–80.e5. doi: 10.1016/j.immuni.2021.10.001
25. Sheaffer KL, Elliott EN, and Kaestner KH. DNA Hypomethylation Contributes to Genomic Instability and Intestinal Cancer Initiation. Cancer Prev Res (Phila). (2016) 9:534–46. doi: 10.1158/1940-6207.CAPR-15-0349
26. Chim CS, Liang R, and Kwong YL. Hypermethylation of gene promoters in hematological neoplasia. Hematol Oncol. (2002) 20:167–76. doi: 10.1002/hon.694
27. Qin Q, Zhou Y, Guo J, Chen Q, Tang W, Li Y, et al. Conserved methylation signatures associate with the tumor immune microenvironment and immunotherapy response. Genome Med. (2024) 16:47. doi: 10.1186/s13073-024-01318-3
28. Delpu Y, Cordelier P, Cho WC, and Torrisani J. DNA methylation and cancer diagnosis. Int J Mol Sci. (2013) 14:15029–58. doi: 10.3390/ijms140715029
29. Malta TM, De Souza CF, Sabedot TS, Silva TC, Mosella MS, Kalkanis SN, et al. Glioma CpG island methylator phenotype (G-CIMP): biological and clinical implications. Neuro-Oncology. (2018) 20:608–20. doi: 10.1093/neuonc/nox183
30. Monabbati S, Corredor G, Pathak T, Peacock C, Yang K, Koyfman S, et al. Pathogenomic fingerprinting to identify associations between tumor morphology and epigenetic states. Eur J Cancer. (2025) 221:115429. doi: 10.1016/j.ejca.2025.115429
31. Liang WW, Lu RJ, Jayasinghe RG, Foltz SM, Porta-Pardo E, Geffen Y, et al. Integrative multi-omic cancer profiling reveals DNA methylation patterns associated with therapeutic vulnerability and cell-of-origin. Cancer Cell. (2023) 41:1567–85.e7. doi: 10.1016/j.ccell.2023.07.013
32. Jeschke J, Bizet M, Desmedt C, Calonne E, Dedeurwaerder S, Garaud S, et al. DNA methylation-based immune response signature improves patient diagnosis in multiple cancers. J Clin Invest. (2017) 127:3090–102. doi: 10.1172/JCI91095
33. Saha D, Kanjilal P, Kaur M, Menon SV, Ashraf A, Kumar MR, et al. Transforming Cancer Diagnostics: The Emergence of Liquid Biopsy and Epigenetic Markers. MedComm (2020). (2025) 6:e70388. doi: 10.1002/mco2.70388
34. Montano C and Timp W. Evolution of genome-wide methylation profiling technologies. Genome Res. (2025) 35:572–82. doi: 10.1101/gr.278407.123
35. Huang J and Wang L. Cell-Free DNA Methylation Profiling Analysis-Technologies and Bioinformatics. Cancers (Bsl). (2019) 11:1741. doi: 10.3390/cancers11111741
36. Liu Y, Baggerly KA, Orouji E, Manyam G, Chen H, Lam M, et al. Methylation-eQTL analysis in cancer research. Bioinformatics. (2021) 37:4014–22. doi: 10.1093/bioinformatics/btab443
37. Zhang L, Peng Y, Huang S, and Zhong L. Integrative analysis of DNA methylation and gene expression in skin cutaneous melanoma by bioinformatic approaches. Arch Dermatol Res. (2025) 317:545. doi: 10.1007/s00403-025-03863-2
38. Agius DR, Kapazoglou A, Avramidou E, Baranek M, Carneros E, Caro E, et al. Exploring the crop epigenome: a comparison of DNA methylation profiling techniques. Front Plant Sci. (2023) 14:1181039. doi: 10.3389/fpls.2023.1181039
39. Newman-Toker DE, McDonald KM, and Meltzer DO. How much diagnostic safety can we afford, and how should we decide? A health economics perspective. BMJ Qual Saf. (2013) 22:ii11–20. doi: 10.1136/bmjqs-2012-001616
40. Solomon O, MacIsaac J, Quach H, Tindula G, Kobor MS, Huen K, et al. Comparison of DNA methylation measured by Illumina 450K and EPIC BeadChips in blood of newborns and 14-year-old children. Epigenetics. (2018) 13:655–64. doi: 10.1080/15592294.2018.1497386
41. Poggiali B, Dupont ME, Kampmann ML, Vidaki A, Pereira V, Børsting C, et al. Evaluation of the HumanMethylationEPIC v2.0 Bead Chip Using Low Quality and Quantity DNA Samples. Biol Proced Online. (2025) 27:34. doi: 10.1186/s12575-025-00292-3
42. Cui XL, Nie J, Zhu H, Kowitwanich K, Beadell AV, West-Szymanski DC, et al. LABS: linear amplification-based bisulfite sequencing for ultrasensitive cancer detection from cell-free DNA. Genome Biol. (2024) 25:157. doi: 10.1186/s13059-024-03262-2
43. Dai Q, Baldwin T, Lyu R, Daniels B, Ye C, Cao C, et al. Ultra-mild bisulfite outperforms existing methods for 5-methylcytosine detection with low input DNA. Nat Commun. (2025) 16:9939. doi: 10.1038/s41467-025-66033-y
44. Crawford R, Charlesworth T, Riffle S, Yardley K, and Osborne R. 5mC and 5hmC methylation sequencing: the power of 6-base sequencing in a multiomic era. Epigenomics. (2025) 12:1–15. doi: 10.1080/17501911.2025.2586452
45. Singh J, Sahu S, Mohan T, Mahajan S, Sharma MC, Sarkar C, et al. Current status of DNA methylation profiling in neuro-oncology as a diagnostic support tool: A review. Neuro-Oncol Pract. (2023) 10:518–26. doi: 10.1093/nop/npad040
46. Park JW, Lee K, Kim EE, Kim SI, and Park SH. Brain Tumor Classification by Methylation Profile. J Kor Med Sci. (2023) 38:e356. doi: 10.3346/jkms.2023.38.e356
47. Lobo J, Tavares NT, Fonseca D, Jerónimo C, Henrique R, Wyvekens N, et al. MicroRNA-371–373 cluster and methylome analysis suggests that a subset of ‘somatic-type’ malignancies arising in germ cell tumors may originate in yolk sac tumor components. J Pathol. (2025) 266:160–76. doi: 10.1002/path.6412
48. Okonechnikov K, Joshi P, Sepp M, Leiss K, Sarropoulos I, Murat F, et al. Mapping pediatric brain tumors to their origins in the developing cerebellum. Neuro-Oncology. (2023) 25:1895–909. doi: 10.1093/neuonc/noad124
49. Dankowski T and Ziegler A. Calibrating random forests for probability estimation. Stat Med. (2016) 35:3949–60. doi: 10.1002/sim.6959
50. Crow M, Paul A, Ballouz S, Huang ZJ, and Gillis J. Characterizing the replicability of cell types defined by single cell RNA-sequencing data using MetaNeighbor. Nat Commun. (2018) 9:884. doi: 10.1038/s41467-018-03282-0
51. Drexler R, Schüller U, Eckhardt A, Filipski K, Hartung TI, Harter PN, et al. DNA methylation subclasses predict the benefit from gross total tumor resection in IDH-wildtype glioblastoma patients. Neuro-Oncology. (2023) 25:315–25. doi: 10.1093/neuonc/noac177
52. Sill M, Schrimpf D, Patel A, Sturm D, Jäger N, Sievers P, et al. Advancing CNS tumor diagnostics with expanded DNA methylation-based classification. medRxiv. (2025). doi: 10.1093/neuonc/noaf201.1012
53. LaFlamme CW, Rastin C, Sengupta S, Pennington HE, Russ-Hall SJ, Schneider AL, et al. Diagnostic utility of DNA methylation analysis in genetically unsolved pediatric epilepsies and CHD2 episignature refinement. Nat Commun. (2024) 15:6524. doi: 10.1038/s41467-024-50159-6
54. Romo CG, Palsgrove DN, Sivakumar A, Elledge CR, Kleinberg LR, Chaichana KL, et al. Widely metastatic IDH1-mutant glioblastoma with oligodendroglial features and atypical molecular findings: a case report and review of current challenges in molecular diagnostics. Diagn Pathol. (2019) 14:16. doi: 10.1186/s13000-019-0793-5
55. Torp SH, Solheim O, and Skjulsvik AJ. The WHO 2021 Classification of Central Nervous System tumours: a practical update on what neurosurgeons need to know-a minireview. Acta Neurochir (Wi). (2022) 164:2453–64. doi: 10.1007/s00701-022-05301-y
56. Butler M, Pongor L, Su YT, Xi L, Raffeld M, Quezado M, et al. MGMT Status as a Clinical Biomarker in Glioblastoma. Trends Cancer. (2020) 6:380–91. doi: 10.1016/j.trecan.2020.02.010
57. Ahmad Z, Alzubaidi M, Al-Thelaya K, Calí C, Boughorbel S, Schneider J, et al. Advancing open-source visual analytics in digital pathology: A systematic review of tools, trends, and clinical applications. J Pathol Inf. (2025) 18:100454. doi: 10.1016/j.jpi.2025.100454
58. Nafe R, Porto L, Samp PF, You SJ, and Hattingen E. Adult-type and Pediatric-type Diffuse Gliomas: What the Neuroradiologist Should Know. Clin Neuroradiol. (2023) 33:611–24. doi: 10.1007/s00062-023-01277-z
59. Pratt D, Sahm F, and Aldape K. DNA methylation profiling as a model for discovery and precision diagnostics in neuro-oncology. Neuro-Oncology. (2021) 23:S16–s29. doi: 10.1093/neuonc/noab143
60. Bader JM, Deigendesch N, Misch M, Mann M, Koch A, and Meissner F. Proteomics separates adult-type diffuse high-grade gliomas in metabolic subgroups independent of 1p/19q codeletion and across IDH mutational status. Cell Rep Med. (2023) 4:100877. doi: 10.1016/j.xcrm.2022.100877
61. Choate KA, Pratt EPS, Jennings MJ, Winn RJ, and Mann PB. IDH Mutations in Glioma: Molecular, Cellular, Diagnostic, and Clinical Implications. Biol (Bsl). (2024) 13:885. doi: 10.3390/biology13110885
62. Mateo-Nouel EJ, Mondragón-Soto MG, Nicolás-Cruz CF, Villanueva-Castro E, Rodriguez Hernandez LA, Reyes-Moreno I, et al. Efficient Detection of Oligodendroglioma With 1p/19q Codeletion Mutation via Methionine PET Imaging: A Promising Diagnostic Approach. Cureus. (2025) 17:e77826. doi: 10.7759/cureus.77826
63. Umphlett M, Bilal KH, Martini ML, Suwala AK, Ahuja S, Rashidipour O, et al. IDH-mutant astrocytoma with EGFR amplification-Genomic profiling in four cases and review of literature. Neurooncol Adv. (2022) 4:vdac067. doi: 10.1093/noajnl/vdac067
64. Verdugo E, Puerto I, and Medina M. An update on the molecular biology of glioblastoma, with clinical implications and progress in its treatment. Cancer Commun (Lond). (2022) 42:1083–111. doi: 10.1002/cac2.12361
65. Lucchini S, Constantinou M, and Marino S. Unravelling the mosaic: Epigenetic diversity in glioblastoma. Mol Oncol. (2024) 18:2871–89. doi: 10.1002/1878-0261.13706
66. Lee B, Hwang S, Bae H, Choi KH, and Suh YL. Diagnostic utility of genetic alterations in distinguishing IDH-wildtype glioblastoma from lower-grade gliomas: Insight from next-generation sequencing analysis of 479 cases. Brain Pathol. (2024) 34:e13234. doi: 10.1111/bpa.13234
67. Vallero SG, Bertero L, Morana G, Sciortino P, Bertin D, Mussano A, et al. Pediatric diffuse midline glioma H3K27- altered: A complex clinical and biological landscape behind a neatly defined tumor type. Front Oncol. (2022) 12:1082062. doi: 10.3389/fonc.2022.1082062
68. Wenger A and Carén H. Methylation Profiling in Diffuse Gliomas: Diagnostic Value and Considerations. Cancers (Bsl). (2022) 14:5679. doi: 10.3390/cancers14225679
69. Ryall S, Tabori U, and Hawkins C. Pediatric low-grade glioma in the era of molecular diagnostics. Acta Neuropathol Commun. (2020) 8:30. doi: 10.1186/s40478-020-00902-z
70. Clarke M, Mackay A, Ismer B, Pickles JC, Tatevossian RG, Newman S, et al. Infant High-Grade Gliomas Comprise Multiple Subgroups Characterized by Novel Targetable Gene Fusions and Favorable Outcomes. Cancer Discov. (2020) 10:942–63. doi: 10.1158/2159-8290.CD-19-1030
71. Roosen M, Odé Z, Bunt J, and Kool M. The oncogenic fusion landscape in pediatric CNS neoplasms. Acta Neuropathol. (2022) 143:427–51. doi: 10.1007/s00401-022-02405-8
72. Stegat L, Eckhardt A, Gocke A, Neyazi S, Pohl L, Schmid S, et al. Integrated analyses reveal two molecularly and clinically distinct subtypes of H3 K27M-mutant diffuse midline gliomas with prognostic significance. Acta Neuropathol. (2024) 148:40. doi: 10.1007/s00401-024-02800-3
73. Alharbi M, Mobark N, Bashawri Y, Abu Safieh L, Alowayn A, Aljelaify R, et al. Methylation Profiling of Medulloblastoma in a Clinical Setting Permits Sub-classification and Reveals New Outcome Predictions. Front Neurol. (2020) 11:167. doi: 10.3389/fneur.2020.00167
74. Brown HG, Kepner JL, Perlman EJ, Friedman HS, Strother DR, Duffner PK, et al. Large cell/anaplastic” medulloblastomas: a Pediatric Oncology Group Study. J Neuropathol Exp Neurol. (2000) 59:857–65. doi: 10.1093/jnen/59.10.857
75. Slocum CC, Park HJ, Baek I, Catalano J, Wells MT, Liechty B, et al. Towards a single-assay approach: a combined DNA/RNA sequencing panel eliminates diagnostic redundancy and detects clinically-relevant fusions in neuropathology. Acta Neuropathol Commun. (2022) 10:167. doi: 10.1186/s40478-022-01466-w
76. Nobre L, Zapotocky M, Khan S, Fukuoka K, Fonseca A, McKeown T, et al. Pattern of Relapse and Treatment Response in WNT-Activated Medulloblastoma. Cell Rep Med. (2020) 1:100038. doi: 10.1016/j.xcrm.2020.100038
77. Menyhárt O and Győrffy B. Principles of tumorigenesis and emerging molecular drivers of SHH-activated medulloblastomas. Ann Clin Transl Neurol. (2019) 6:990–1005. doi: 10.1002/acn3.762
78. Welsh H, Batalha C, Li W, Souza-Pinto NC, Duarte YAO, Naslavsky MS, et al. Age-related changes in DNA methylation in a sample of elderly Brazilians. Clin Epigenet. (2025) 17:17. doi: 10.1186/s13148-025-01821-3
79. Williamson D, Schwalbe EC, Hicks D, Aldinger KA, Lindsey JC, Crosier S, et al. Medulloblastoma group 3 and 4 tumors comprise a clinically and biologically significant expression continuum reflecting human cerebellar development. Cell Rep. (2022) 40:111162. doi: 10.1016/j.celrep.2022.111162
80. Menyhárt O, Giangaspero F, and Győrffy B. Molecular markers and potential therapeutic targets in non-WNT/non-SHH (group 3 and group 4) medulloblastomas. J Hematol Oncol. (2019) 12:29. doi: 10.1186/s13045-019-0712-y
81. Northcott PA, Korshunov A, Witt H, Hielscher T, Eberhart CG, Mack S, et al. Medulloblastoma comprises four distinct molecular variants. J Clin Oncol. (2011) 29:1408–14. doi: 10.1200/JCO.2009.27.4324
82. Pajtler KW, Witt H, Sill M, Jones DT, Hovestadt V, Kratochwil F, et al. Molecular Classification of Ependymal Tumors across All CNS Compartments, Histopathological Grades, and Age Groups. Cancer Cell. (2015) 27:728–43. doi: 10.1016/j.ccell.2015.04.002
83. Lim KY, Lee KH, Phi JH, Yun H, Won JK, Choi SH, et al. ZFTA-YAP1 fusion-positive ependymoma can occur in the spinal cord: Letter to the editor. Brain Pathol. (2022) 32:e13020. doi: 10.1111/bpa.13020
84. Pun M, Pratt D, Nano PR, Joshi PK, Jiang L, Englinger B, et al. Common molecular features of H3K27M DMGs and PFA ependymomas map to hindbrain developmental pathways. Acta Neuropathol Commun. (2023) 11:25. doi: 10.1186/s40478-023-01514-z
85. Witt H, Mack SC, Ryzhova M, Bender S, Sill M, Isserlin R, et al. Delineation of two clinically and molecularly distinct subgroups of posterior fossa ependymoma. Cancer Cell. (2011) 20:143–57. doi: 10.1016/j.ccr.2011.07.007
86. Ghasemi DR, Sill M, Okonechnikov K, Korshunov A, Yip S, Schutz PW, et al. MYCN amplification drives an aggressive form of spinal ependymoma. Acta Neuropathol. (2019) 138:1075–89. doi: 10.1007/s00401-019-02056-2
87. Galbraith K, Serrano J, Shen G, Tran I, Slocum CC, Ketchum C, et al. Impact of Rare and Multiple Concurrent Gene Fusions on Diagnostic DNA Methylation Classifier in Brain Tumors. Mol Cancer Res. (2024) 22:21–8. doi: 10.1158/1541-7786.MCR-23-0627
88. Larrew T, Saway BF, Lowe SR, and Olar A. Molecular Classification and Therapeutic Targets in Ependymoma. Cancers (Bsl). (2021) 13:6218. doi: 10.3390/cancers13246218
89. Gödicke S, Kresbach C, Ehlert M, Obrecht D, Altendorf L, Hack K, et al. Clinically relevant molecular hallmarks of PFA ependymomas display intratumoral heterogeneity and correlate with tumor morphology. Acta Neuropathol. (2024) 147:23. doi: 10.1007/s00401-023-02682-x
90. Menzel M, Endris V, Schwab C, Kluck K, Neumann O, Beck S, et al. Accurate tumor purity determination is critical for the analysis of homologous recombination deficiency (HRD). Transl Oncol. (2023) 35:101706. doi: 10.1016/j.tranon.2023.101706
91. Wesseling P, Capper D, Reifenberger G, Sarkar C, Hawkins C, Perry A, et al. cIMPACT-NOW update 11: Proposal on adaptation of diagnostic criteria for IDH- and H3-wildtype diffuse high-grade gliomas and for posterior fossa ependymal tumors. Brain Pathol. (2025) 36:e70035. doi: 10.1111/bpa.70035
92. Roehrkasse AM, Peterson JEG, Fung KM, Pelargos PE, and Dunn IF. The Discrepancy Between Standard Histologic WHO Grading of Meningioma and Molecular Profile: A Single Institution Series. Front Oncol. (2022) 12:846232. doi: 10.3389/fonc.2022.846232
93. Zhang H, Hao D, Gong Y, Xu Y, Ding C, Wang J, et al. Associations of twelve DNA methylation algorithms of aging with mortality. hLife. (2025). doi: 10.1016/j.hlife.2025.05.010
94. Choudhury A, Chen WC, Lucas CG, Bayley JC, Harmanci AS, Maas SLN, et al. Hypermitotic meningiomas harbor DNA methylation subgroups with distinct biological and clinical features. Neuro-Oncology. (2023) 25:520–30. doi: 10.1093/neuonc/noac224
95. Maas SLN, Stichel D, Hielscher T, Sievers P, Berghoff AS, Schrimpf D, et al. Integrated Molecular-Morphologic Meningioma Classification: A Multicenter Retrospective Analysis, Retrospectively and Prospectively Validated. J Clin Oncol. (2021) 39:3839–52. doi: 10.1200/JCO.21.00784
96. Landry AP, Wang JZ, Patil V, Gui C, Mamatjan Y, Patel Z, et al. Validation and next-generation update of a DNA methylation-based recurrence predictor for meningioma: A multicenter prospective study. Neuro-Oncology. (2025) 27:1004–16. doi: 10.1093/neuonc/noae236
97. Ehret F, Perez E, Teichmann D, Meier S, Geiler C, Zeus C, et al. Clinical implications of DNA methylation-based integrated classification of histologically defined grade 2 meningiomas. Acta Neuropathol Commun. (2024) 12:74. doi: 10.1186/s40478-024-01739-6
98. Lee SH, Kim TG, Ryu KH, Kim SH, and Kim YZ. CDKN2A Homozygous Deletion Is a Stronger Predictor of Outcome than IDH1/2-Mutation in CNS WHO Grade 4 Gliomas. Biomedicines. (2024) 12:2256. doi: 10.3390/biomedicines12102256
99. Sahm F, Aldape KD, Brastianos PK, Brat DJ, Dahiya S, von Deimling A, et al. cIMPACT-NOW update 8: Clarifications on molecular risk parameters and recommendations for WHO grading of meningiomas. Neuro-Oncology. (2025) 27:319–30. doi: 10.1093/neuonc/noae170
100. Sturm D, Orr BA, Toprak UH, Hovestadt V, Jones DTW, Capper D, et al. New Brain Tumor Entities Emerge from Molecular Classification of CNS-PNETs. Cell. (2016) 164:1060–72. doi: 10.1016/j.cell.2016.01.015
101. d’Amati A, Bargiacchi L, Rossi S, Carai A, Bertero L, Barresi V, et al. Pediatric CNS tumors and 2021 WHO classification: what do oncologists need from pathologists? Front Mol Neurosci. (2024) 17:1268038. doi: 10.3389/fnmol.2024.1268038
102. Ho B, Johann PD, Grabovska Y, De Dieu Andrianteranagna MJ, Yao F, Frühwald M, et al. Molecular subgrouping of atypical teratoid/rhabdoid tumors-a reinvestigation and current consensus. Neuro-Oncology. (2020) 22:613–24. doi: 10.1093/neuonc/noz235
103. Wu Z, Abdullaev Z, Pratt D, Chung HJ, Skarshaug S, Zgonc V, et al. Impact of the methylation classifier and ancillary methods on CNS tumor diagnostics. Neuro-Oncology. (2022) 24:571–81. doi: 10.1093/neuonc/noab227
104. Ma X, Wang YW, Zhang MQ, and Gazdar AF. DNA methylation data analysis and its application to cancer research. Epigenomics. (2013) 5:301–16. doi: 10.2217/epi.13.26
105. Christiansen SN, Andersen JD, Kampmann ML, Liu J, Andersen MM, Tfelt-Hansen J, et al. Reproducibility of the Infinium methylationEPIC BeadChip assay using low DNA amounts. Epigenetics. (2022) 17:1636–45. doi: 10.1080/15592294.2022.2051861
106. Izadi N, Solár P, Hašanová K, Zamani A, Akbar MS, Mrázová K, et al. Breaking boundaries: role of the brain barriers in metastatic process. Fl Brr CNS. (2025) 22:3. doi: 10.1186/s12987-025-00618-z
107. Wisner L, Larsen B, and Maguire A. Enhancing Tumor Content through Tumor Macrodissection. J Vis Exp. (2022). doi: 10.3791/62961-v
108. Jeddi F, Faghfuri E, Mehranfar S, and Soozangar N. The common bisulfite-conversion-based techniques to analyze DNA methylation in human cancers. Cancer Cell Int. (2024) 24:240. doi: 10.1186/s12935-024-03405-2
109. Apsley AT, Ye Q, Etzel L, Wolf S, Hastings WJ, Mattern BC, et al. Biological stability of DNA methylation measurements over varying intervals of time and in the presence of acute stress. Epigenetics. (2023) 18:2230686. doi: 10.1080/15592294.2023.2230686
110. Khan TA, Loftus TJ, Filiberto AC, Ozrazgat-Baslanti T, Ruppert MM, Bandyopadhyay S, et al. Metabolomic Profiling for Diagnosis and Prognostication in Surgery: A Scoping Review. Ann Surg. (2021) 273:258–68. doi: 10.1097/SLA.0000000000003935
111. Zheng P, Zhou C, Ding Y, Liu B, Lu L, Zhu F, et al. Nanopore sequencing technology and its applications. MedComm (2020). (2023) 4:e316. doi: 10.1002/mco2.316
112. Guanzon D, Ross JP, Ma C, Berry O, and Liew YJ. Comparing methylation levels assayed in GC-rich regions with current and emerging methods. BMC Genomics. (2024) 25:741. doi: 10.1186/s12864-024-10605-7
113. Feldman LS, Shihab HM, Thiemann D, Yeh H-C, Ardolino M, Mandell S, et al. Impact of Providing Fee Data on Laboratory Test Ordering: A Controlled Clinical Trial. JAMA Internal Med. (2013) 173:903–8. doi: 10.1001/jamainternmed.2013.232
114. Hodges E, Smith AD, Kendall J, Xuan Z, Ravi K, Rooks M, et al. High definition profiling of mammalian DNA methylation by array capture and single molecule bisulfite sequencing. Genome Res. (2009) 19:1593–605. doi: 10.1101/gr.095190.109
115. Zindler T, Frieling H, Neyazi A, Bleich S, and Friedel E. Simulating ComBat: how batch correction can lead to the systematic introduction of false positive results in DNA methylation microarray studies. BMC Bioinf. (2020) 21:271. doi: 10.1186/s12859-020-03559-6
116. Phua SX, Lim KP, and Goh WW. Perspectives for better batch effect correction in mass-spectrometry-based proteomics. Comput Struct Biotechnol J. (2022) 20:4369–75. doi: 10.1016/j.csbj.2022.08.022
117. Clark AJ and Lillard JW Jr. A Comprehensive Review of Bioinformatics Tools for Genomic Biomarker Discovery Driving Precision Oncology. Genes (Bsl). (2024) 15:1036. doi: 10.3390/genes15081036
118. Leung WY, Marschall T, Paudel Y, Falquet L, Mei H, Schönhuth A, et al. SV-AUTOPILOT: optimized, automated construction of structural variation discovery and benchmarking pipelines. BMC Genomics. (2015) 16:238. doi: 10.1186/s12864-015-1376-9
119. Chen W, Zhuang J, Wang PP, Jiang J, Lin C, Zeng P, et al. DNA methylation-based classification and identification of renal cell carcinoma prognosis-subgroups. Cancer Cell Int. (2019) 19:185. doi: 10.1186/s12935-019-0900-4
120. Priesterbach-Ackley LP, Boldt HB, Petersen JK, Bervoets N, Scheie D, Ulhøi BP, et al. Brain tumour diagnostics using a DNA methylation-based classifier as a diagnostic support tool. Neuropathol Appl Neurobiol. (2020) 46:478–92. doi: 10.1111/nan.12610
121. Wei X and Li KC. Exploring the within- and between-class correlation distributions for tumor classification. Proc Natl Acad Sci U.S.A. (2010) 107:6737–42. doi: 10.1073/pnas.0910140107
122. Komura D, Ochi M, and Ishikawa S. Machine learning methods for histopathological image analysis: Updates in 2024. Comput Struct Biotechnol J. (2025) 27:383–400. doi: 10.1016/j.csbj.2024.12.033
123. Alturkustani M. Unraveling morphology, methylation profiling, and diagnostic challenges in BRAF-Mutant pediatric glial and glioneuronal tumors. Neurosci (Riyadh). (2024) 29:168–76. doi: 10.17712/nsj.2024.3.20230108
124. Galbraith K, Vasudevaraja V, Serrano J, Shen G, Tran I, Abdallat N, et al. Clinical utility of whole-genome DNA methylation profiling as a primary molecular diagnostic assay for central nervous system tumors-A prospective study and guidelines for clinical testing. Neurooncol Adv. (2023) 5:vdad076. doi: 10.1093/noajnl/vdad076
125. Kim T, Lee A, Ahn S, Park JS, Jeun SS, and Lee YS. Comprehensive Molecular Genetic Analysis in Glioma Patients by Next Generation Sequencing. Brain Tum Res Treat. (2024) 12:23–39. doi: 10.14791/btrt.2023.0036
126. Hetzel S, Hodis E, Torlai Triglia E, Kovacsovics A, Steinmann K, Gnirke A, et al. Direct genetic transformation bypasses tumor-associated DNA methylation alterations. Genome Biol. (2025) 26:212. doi: 10.1186/s13059-025-03650-2
127. Reuss DE. Updates on the WHO diagnosis of IDH-mutant glioma. J Neurooncol. (2023) 162:461–9. doi: 10.1007/s11060-023-04250-5
128. Tyekucheva S, Martin NE, Stack EC, Wei W, Vathipadiekal V, Waldron L, et al. Comparing Platforms for Messenger RNA Expression Profiling of Archival Formalin-Fixed, Paraffin-Embedded Tissues. J Mol Diagn. (2015) 17:374–81. doi: 10.1016/j.jmoldx.2015.02.002
129. Emiliani FE, Ismail AAO, Hughes EG, Tsongalis GJ, Zanazzi GJ, and Lin CC. Nanopore-based random genomic sampling for intraoperative molecular diagnosis. Genome Med. (2025) 17:6. doi: 10.1186/s13073-025-01427-7
130. Rosenfeld JA and Patel A. Chromosomal Microarrays: Understanding Genetics of Neurodevelopmental Disorders and Congenital Anomalies. J Pediatr Genet. (2017) 6:42–50. doi: 10.1055/s-0036-1584306
131. Onishi S, Yamasaki F, Amatya VJ, Yonezawa U, Taguchi A, Ozono I, et al. Prognostic value of immunohistochemical staining for H3K27me3 and EZH2 in astrocytoma, IDH-mutant. J Neurooncol. (2025) 172:185–94. doi: 10.1007/s11060-024-04897-8
132. Pearson HCL, Hunt KV, Trahair TN, Lock RB, Lee HJ, and De Bock CE. The Promise of Single-cell Technology in Providing New Insights Into the Molecular Heterogeneity and Management of Acute Lymphoblastic Leukemia. Hemasphere. (2022) 6:e734. doi: 10.1097/HS9.0000000000000734
133. Gerami P, Jewell SS, Morrison LE, Blondin B, Schulz J, Ruffalo T, et al. Fluorescence in situ hybridization (FISH) as an ancillary diagnostic tool in the diagnosis of melanoma. Am J Surg Pathol. (2009) 33:1146–56. doi: 10.1097/PAS.0b013e3181a1ef36
134. Kurdyukov S and Bullock M. DNA Methylation Analysis: Choosing the Right Method. Biol (Bsl). (2016) 5:3. doi: 10.3390/biology5010003
135. Cho M, Ahn S, Hong M, Bang H, Van Vrancken M, Kim S, et al. Tissue recommendations for precision cancer therapy using next generation sequencing: a comprehensive single cancer center’s experiences. Oncotarget. (2017) 8:42478–86. doi: 10.18632/oncotarget.17199
136. García-Heredia A, Guerra-Núñez L, Martín-Climent P, Rojas E, López-Domínguez R, Alcántara-Domínguez C, et al. Transcriptomics and epigenomics datasets of primary brain cancers in formalin-fixed paraffin embedded format. Sci Data. (2025) 12:273. doi: 10.1038/s41597-025-04597-6
137. Hellicar AD, Rahman A, Smith DV, and Henshall JM. Machine learning approach for pooled DNA sample calibration. BMC Bioinf. (2015) 16:214. doi: 10.1186/s12859-015-0593-1
138. Haake J and Steenpass L. Chromosomal quality control in hPSCs: A practical guide to SNP array analysis with GenomeStudio. Front Cell Dev Biol. (2025) 13:1599923. doi: 10.3389/fcell.2025.1599923
139. Barros-Silva D, Marques CJ, Henrique R, and Jerónimo C. Profiling DNA Methylation Based on Next-Generation Sequencing Approaches: New Insights and Clinical Applications. Genes (Bsl). (2018) 9:429. doi: 10.3390/genes9090429
140. Uhl C, Ayoub N, Faust K, Vajkoczy P, Schweizer L, Radke J, et al. DNA methylation-based profiling is an effective asset for identification of tumors in suspected, yet immunohistochemically, unspecified neuro-oncological cases. Brain Spine. (2025) 5:104256. doi: 10.1016/j.bas.2025.104256
141. Dorrity MW, Saunders LM, Queitsch C, Fields S, and Trapnell C. Dimensionality reduction by UMAP to visualize physical and genetic interactions. Nat Commun. (2020) 11:1537. doi: 10.1038/s41467-020-15351-4
142. Handoko H, Susanto E, Aman RA, Aninditha T, Wibowo H, Permata TBM, et al. Conserved methylation signature accurately predicts heavily irradiated CNS tumour with perplexing histopathology: A case report. BioMed Rep. (2025) 23:165. doi: 10.3892/br.2025.2043
143. Aref-Eshghi E, Abadi AB, Farhadieh ME, Hooshmand A, Ghasemi F, Youssefian L, et al. DNA methylation and machine learning: challenges and perspective toward enhanced clinical diagnostics. Clin Epigenet. (2025) 17:170. doi: 10.1186/s13148-025-01967-0
144. Abugabah A, Shukla PK, Shukla PK, and Pandey A. An intelligent healthcare system for rare disease diagnosis utilizing electronic health records based on a knowledge-guided multimodal transformer framework. BioDat Min. (2025) 18:70. doi: 10.1186/s13040-025-00487-0
145. Bertero L, Mangherini L, Ricci AA, Cassoni P, and Sahm F. Molecular neuropathology: an essential and evolving toolbox for the diagnosis and clinical management of central nervous system tumors. Virchows Arch. (2024) 484:181–94. doi: 10.1007/s00428-023-03632-4
146. Mittelbronn M. Neurooncology: 2023 update. Free Neuropathol. (2023) 4:4. doi: 10.17879/freeneuropathology-2023-4692
147. Yassi M, Chatterjee A, and Parry M. Application of deep learning in cancer epigenetics through DNA methylation analysis. Brief Bioinform. (2023) 24:bbad411. doi: 10.1093/bib/bbad411
148. Yang Y, Wen X, and Wang L. Advancements in DNA methylation technologies and their application in cancer diagnosis. Epigenetics. (2025) 20:2539995. doi: 10.1080/15592294.2025.2539995
149. Sahoo K and Sundararajan V. Methods in DNA methylation array dataset analysis: A review. Comput Struct Biotechnol J. (2024) 23:2304–25. doi: 10.1016/j.csbj.2024.05.015
150. Masser DR, Stanford DR, Hadad N, Giles CB, Wren JD, Sonntag WE, et al. Bisulfite oligonucleotide-capture sequencing for targeted base- and strand-specific absolute 5-methylcytosine quantitation. Ag (Dordr). (2016) 38:49. doi: 10.1007/s11357-016-9914-1
151. Liu Y, Rosikiewicz W, Pan Z, Jillette N, Wang P, Taghbalout A, et al. DNA methylation-calling tools for Oxford Nanopore sequencing: a survey and human epigenome-wide evaluation. Genome Biol. (2021) 22:295. doi: 10.1186/s13059-021-02510-z
152. De Abreu AR, Ibrahim J, Lemonidis V, Mateiu L, Van Camp G, and Op De Beeck K. Comparison of current methods for genome-wide DNA methylation profiling. Epigenet Chromat. (2025) 18:57. doi: 10.1186/s13072-025-00616-3
153. Ma L, Guo H, Zhao Y, Liu Z, Wang C, Bu J, et al. Liquid biopsy in cancer: current status, challenges and future prospects. Signal Transduct Targ Ther. (2024) 9:336. doi: 10.1038/s41392-024-02021-w
154. Zhou H, Zhu L, Song J, Wang G, Li P, Li W, et al. Liquid biopsy at the frontier of detection, prognosis and progression monitoring in colorectal cancer. Mol Cancer. (2022) 21:86. doi: 10.1186/s12943-022-01556-2
155. Xiao F, Lv S, Zong Z, Wu L, Tang X, Kuang W, et al. Cerebrospinal fluid biomarkers for brain tumor detection: clinical roles and current progress. Am J Transl Res. (2020) 12:1379–96.
156. Frascarelli C, Venetis K, Marra A, Mane E, Ivanova M, Cursano G, et al. Deep learning algorithm on H&E whole slide images to characterize TP53 alterations frequency and spatial distribution in breast cancer. Comput Struct Biotechnol J. (2024) 23:4252–9. doi: 10.1016/j.csbj.2024.11.037
157. Sahoo K, Lingasamy P, Khatun M, Sudhakaran SL, Salumets A, Sundararajan V, et al. Artificial Intelligence in cancer epigenomics: a review on advances in pan-cancer detection and precision medicine. Epigenet Chromat. (2025) 18:35. doi: 10.1186/s13072-025-00595-5
158. Rombaut D, Sandmann S, Tekath T, Crouch S, de Graaf AO, Smith A, et al. Somatic mutations and DNA methylation identify a subgroup of poor prognosis within lower-risk myelodysplastic syndromes. Hemasphere. (2025) 9:e70073. doi: 10.1002/hem3.70073
Keywords: central nervous system neoplasms, DNA methylation profiling, ependymoma, epigenomics, glioma, medulloblastoma, meningioma, neuropathology
Citation: Al Sharie S, Sawaftah K, Qasim H and Al-Hussaini M (2025) Methylation profiling in neuropathological tumors diagnosis: a comprehensive review. Front. Oncol. 15:1720458. doi: 10.3389/fonc.2025.1720458
Received: 08 October 2025; Accepted: 08 December 2025; Revised: 30 November 2025;
Published: 19 December 2025.
Edited by:
Alessia Pellerino, University Hospital of the City of Health and Science of Turin, ItalyReviewed by:
Luca Bertero, University of Turin, ItalyAntonio d’Amati, University of Bari Aldo Moro, Italy
Aaron Taylor, Jackson Laboratory for Genomic Medicine, United States
Copyright © 2025 Al Sharie, Sawaftah, Qasim and Al-Hussaini. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Maysa Al-Hussaini, bWh1c3NhaW5pQGtoY2Muam8=