Characteristic DNA Methylation Research Articles

Comparative Evaluation of Machine Learning Models for Subtyping Triple-Negative Breast Cancer: A Deep Learning-Based Multi-Omics Data Integration Approach.

Objective: Triple-negative breast cancer (TNBC) poses significant diagnostic challenges due to its aggressive nature. This research develops an innovative deep learning (DL) model based on the latest multi-omics data to enhance the accuracy of TNBC subtype and prognosis prediction. The study focuses on addressing the constraints of prior studies by showcasing a model with substantial advancements in data integration, statistical performance, and algorithmic optimization. Methods: Breast cancer-related molecular characteristic data, including mRNA, miRNA, gene mutations, DNA methylation, and magnetic resonance imaging (MRI) images, were retrieved from the TCGA and TCIA databases. This study not only compared single-omics with multi-omics machine learning models but also applied Bayesian optimization to innovatively optimize the neural network structure of a DL model for multi-omics data. Results: The DL model for multi-omics data significantly outperformed single-omics models in subtype prediction, achieving a 98.0% accuracy in cross-validation, 97.0% in the validation set, and 91.0% in an external test set. Additionally, the MRI radiomics model showed promising performance, especially with the training set; however, a decrease in performance during transfer testing underscored the advantages of the DL model for multi-omics data in data consistency and digital processing. Conclusion: Our multi-omics DL model presents notable innovations in statistical performance and transfer learning capability, bearing significant clinical relevance for TNBC classification and prognosis prediction. While the MRI radiomics model proved effective, it requires further optimization for cross-dataset application to enhance accuracy and consistency. Our findings offer new insights into improving TNBC classification and prognosis through multi-omics data and DL algorithms.

Current Molecular and Clinical Landscape of ATRT - The Link to Future Therapies.

ATRT is a highly aggressive and rare pediatric CNS tumor of very young children. Its genetic hallmark is bi-allelic inactivation of SMARCB1 encoding INI1. Rarely SMARCA4 encoding BRG1 is affected. Up to 30% are associated with constitutional heterozygous pathogenic variants in one of the two genes, giving rise to the Rhabdoid-Tumor-Predisposition-Syndromes (RTPS) 1 and 2. Characteristic DNA methylation profiles distinguish ATRT from other SMARCB1-deficient entities. Three distinct subtypes ATRT-MYC, -TYR, and -SHH are on record. ATRT-SHH may be further divided into the subgroups ATRT-SHH1A, -SHH1B, and -SHH2. The cure of ATRT remains challenging, notwithstanding an increasing understanding of molecular pathomechanisms and genetic background. The implementation of multimodal institutional treatment protocols has improved prognosis. Regardless of treatment approaches, clinical risk factors such as age, metastases, and DNA methylation subtype affect survival probability. We provide a critical appraisal of current conventional multimodal regimens and emerging targeted treatment approaches investigated in clinical trials and entity-specific registries. Intense treatment approaches featuring radiotherapy (RT) and high-dose chemotherapy (HDCT) face the difficulty of balancing tumor control and treatment-related toxicity. Current approaches focus on minimizing radiation fields by proton beam therapy or to withhold RT in HDCT-only approaches. Still, a 40-75% relapse rate upon first-line treatment reveals the need for novel treatment strategies in primary and even more in recurrent/refractory (r/r) disease. Among targeted treatments, immune checkpoint inhibitors and epigenetically active agents appear most promising. Success remains limited in single agent approaches. We hypothesize that mechanism-informed combination therapy will enhance response, as the low mutational burden of ATRT may contribute to acquiring resistance to single targeted agents. As DNA methylation group-specific gene expression profiles appear to influence response to distinct agents, the future treatment of ATRT should respect clinical and biological heterogeneity in risk group adjusted treatment protocols.

Timing is everything: A connection between medulloblastoma prognosis and foetal cerebellar development.

The childhood brain tumour medulloblastoma is typically classified into multiple discrete molecular subgroups with characteristic DNA methylation and expression patterns. Several of these subgroups are used as, or proposed to be, an effective basis for treatment stratification. Here, we highlight the close connection between the findings described in a recent series of studies which, together, strongly imply a continuous association between survival outcome, the transcriptional profile of a Group3/Group4 (i.e. non-WNT/non-SHH) medulloblastoma and the specific point during early foetal cerebellar development at which initial pathogenic disruption took place. This has important implications for future efforts to model the disease by incorporating driving molecular features into their specific developmental context. This further suggests that instead of relying upon discrete DNA methylation subgroups, using expression biomarkers as the basis of a continuous risk predictor may produce a more effective risk stratification of patients with Group3/Group4 medulloblastoma.

Epigenetic biomarkers to track differentiation of pluripotent stem cells.

Quality control of induced pluripotent stem cells remains a challenge. For validation of the pluripotent state, it is crucial to determine trilineage differentiation potential toward endoderm, mesoderm, and ectoderm. Here, we report GermLayerTracker, a combination of site-specific DNA methylation (DNAm) assays that serve as biomarker for early germ layer specification. CG dinucleotides (CpGs) were identified with characteristic DNAm at pluripotent state and after differentiation into endoderm, mesoderm, and ectoderm. Based on this, a pluripotency score was derived that tracks reprogramming and may indicate differentiation capacity, as well as lineage-specific scores to monitor either directed differentiation or self-organized multilineage differentiation in embryoid bodies. Furthermore, we established pyrosequencing assays for fast and cost-effective analysis. In the future, the GermLayerTracker could be used for quality control of pluripotent cells and to estimate lineage-specific commitment during initial differentiation events.

Comparative epigenomics by machine learning approach for neuroblastoma

BackgroundNeuroblastoma (NB) is the second most common pediatric solid tumor. Because the number of genetic mutations found in tumors are small, even in some patients with unfavorable NB, epigenetic variation is expected to play an important role in NB progression. DNA methylation is a major epigenetic mechanism, and its relationship with NB prognosis has been a concern. One limitation with the analysis of variation in DNA methylation is the lack of a suitable analytical model. Therefore, in this study, we performed a random forest (RF) analysis of the DNA methylome data of NB from multiple databases.ResultsRF is a popular machine learning model owing to its simplicity, intuitiveness, and computational cost. RF analysis identified novel intermediate-risk patient groups with characteristic DNA methylation patterns within the low-risk group. Feature selection analysis based on probe annotation revealed that enhancer-annotated regions had strong predictive power, particularly for MYCN-amplified NBs. We developed a gene-based analytical model to identify candidate genes related to disease progression, such as PRDM8 and FAM13A-AS1. RF analysis revealed sufficient predictive power compared to other machine learning models.ConclusionsRF is a useful tool for DNA methylome analysis in cancer epigenetic studies, and has potential to identify a novel cancer-related genes.

Characteristic DNA methylation profiles of chorionic villi in recurrent miscarriage

Dysregulation of transcriptional programs that are tightly regulated by DNA methylation during placental and fetal development at different gestational stages, may cause recurrent miscarriage. Here, we examined genome-wide DNA methylation in chorionic villi and decidual tissues from patients suffering RM and from healthy women who had undergone artificial abortion (n = 5 each). We found that 13,426 and 5816 CpG sites were differentially methylated in chorionic villi and decidua, respectively. DNA methylation profiles of chorionic villi, but not decidua, in RM patients was clearly distinct from AA controls. Among the differentially methylated genes, the enhancer region of SPATS2L was significantly more highly methylated in RM patients (n = 19) than AA controls (n = 19; mean methylation level, 52.0%-vs.-28.9%, P < 0.001), resulting in reduced expression of SPATS2L protein in the former. Functionally, depletion of SPATS2L in extravillous trophoblast cells decreased their invasion and migration abilities. Our data indicate that particularly the chorionic villi in RM patients exhibit distinct DNA methylation profiles compared with normal pregnancies and that this changed DNA methylation status may impede the progression of embryo development via the altered expression of genes such as SPATS2L in the villi.

Abstract 2491: Molecular and clinical characterization of the new WHO entity ‘Astroblastoma, MN1 altered’ and its molecular subgroups

Abstract Using DNA methylation profiling, already in 2016 we had identified novel molecularly defined CNS tumor entities, including CNS high grade neuroepithelial tumors with MN1 alteration (HGNET-MN1), which have in the recently updated WHO classification of CNS tumors been renamed into Astroblastoma, MN1 altered. However, little is still known about the molecular and clinical characteristics of these tumors. To further molecularly and clinically characterize this unique entity, we collected patient material, clinical information and molecular data from 176 Astroblastoma, MN1 altered tumors, which were all identified based on their characteristic DNA methylation pattern. T-SNE clustering analyses of the DNA methylation data together with our large database of almost 100,000 reference cases showed that this entity forms one main cluster defined by interchromosomal gene fusions of mainly MN1 and BEND2 and a smaller, distinct, but adjacent cluster that mainly showed tumors with MN1:CXXC5 fusions. Comparing molecular data of the BEND2- and CXXC5-fused groups we found that chromosomal copy numbers show distinct patterns, such as 16q loss in one third of cases from the BEND2-fused group, and a gain of chromosome 5 in one third of CXXC5-fused cases. Patients of the BEND2-fused group showed a strong enrichment of female patients (85%), whereas the CXXC5-fused group showed the contrary trend and consisted of 75% male patients. Original histological diagnoses of 110 BEND2-fused cases revealed that most tumors were diagnosed as an Astroblastoma (39%), followed by Ependymoma (20%) and PNET histologies (15%), which is in line with previous findings on Astroblasotma, MN1 altered cases. The CXXC5-fused subset, however, showed a distinct histological distribution and only 1 out of 13 cases (8%) had an Astroblastoma histology in the original report, whereas the other cases were diagnosed as PNET (38%), Ependymoma (23%), HGG (23%), or unknown (8%). Investigations of clinical data revealed that BEND2-fused cases (n=65) show a 5/10 year OS of 97% and 89% and a 5/10 year PFS of 48% and 35%, respectively, which is in line with previous studies in smaller series indicating a favorable overall but unfavorable progression-free survival. The smaller subset of CXXC5-fused patients (n=8) indicated a 5/10 year OS of 83% and a 5/10 year PFS of 60% each. First analyses of BEND2-fused cases indicate that an Astroblastoma histology might be associated with a favorable OS, however further analyses, including a central pathology review, are needed to validate this observation. Citation Format: Felix Schmitt-Hoffner, Johannes Gojo, Monika Mauermann, Katja von Hoff, Martin Sill, Andrey Korshunov, Damian Stichel, Felix Sahm, Natalie Jäger, Stefan Pfister, Marcel Kool. Molecular and clinical characterization of the new WHO entity ‘Astroblastoma, MN1 altered’ and its molecular subgroups [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2491.

Epigenetic lockdown of CDKN1A (p21) and CDKN2A (p16) characterises the neoplastic spindle cell component of giant cell tumours of bone.

Giant cell tumour of bone (GCTB) comprises the eponymous osteoclastic multinucleated giant cells eliciting bone lysis, an H3F3A-mutated neoplastic mononucleated fibroblast-like cell population, and H3F3A wild-type mononucleated stromal cells. In this study, we characterised four new cell lines from GCTB. Furthermore, we compared the genome-wide DNA methylation profile of 13 such tumours and three further cell lines with giant cell-rich lesions comprising three H3F3B-mutated chondroblastomas, three USP6-rearranged aneurysmal bone cysts, three non-ossifying fibromas, two hyperparathyroidism-associated brown tumours as well as mesenchymal stem cells, osteoblasts, and osteoclasts. In an unsupervised analysis, we delineated GCTB and chondroblastomas from the other analysed tumour entities. Using comparative methylation analysis, we demonstrated that the methylation pattern of the cell lines approximately equals that of H3F3A-mutated stromal cells in tissue. These patterns more resemble that of osteoblasts than that of mesenchymal stem cells, which argues for the osteoblast as the cell of origin of giant cell tumours of bone. Using enrichment analysis, we detected distinct hypermethylated clusters containing histone and collagen genes as well as target genes of the tumour suppressor p53. We found that the promotor regions of CDKN1A, CDKN2A, and IGFBP3 are methylated more strongly in GCTB than in the other giant cell-containing lesions, mesenchymal stem cells, osteoblasts, and osteoclasts (p < 0.001). This hypermethylation correlates with the lower gene expression at the mRNA level for these three genes in the cell lines, the lack of p16 and p21 in these cell lines, and the lower expression of p16 and p21 in GCTB. Overall, our analysis reveals characteristic DNA methylation patterns of giant cell tumours of bone and chondroblastomas and shows that cell lines of giant cell tumours of bone are a valid model for further analysis of H3F3A-mutated tumour cells. © 2022 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

MODL-07. DNA methylation-based biobank of murine models for pediatric tumors

Recent advances in molecular profiling methods led to the identification of multiple new molecularly defined tumor types and subtypes, distinguished by distinct molecular markers and characteristic DNA methylation signatures. While the analysis of human methylome using microarrays has become an affordable and a routine in many labs, this technology until recently was not available for murine samples. In the past years, we have successfully generated a variety of mouse models for childhood tumors (e.g brain tumors and sarcomas) using different techniques, most of which faithfully reflect the human tumor counterparts at the histological level. With the recently released Infinium Mouse Methylation BeadChip, we now set out to use our models to generate the first DNA methylation database for murine pediatric tumors. We profiled more than 70 mouse models of pediatric tumors including gliomas, medulloblastomas, ependymomas and sarcomas, as well as 40 normal brain and muscle control tissues. We are currently performing a cross-species comparative analysis of established mouse models and the human counterparts. This will assess, how faithfully each models reflect the human situation and examine the effects of multiple passages of allografting. We will also analyze purified immune cell populations and use the derived methylation signatures to assess the model-specific immune microenvironment. Furthermore, we will investigate the methylomes of multiple putative cells-of-origin, which is hardly possible in the human context due to the lack of purified material. We will correlate these to murine tumor samples and thereby provide novel insights into tumor origins. In summary, our study will generate a validated biobank of murine models for pediatric cancers and provide a valuable resource for future developmental studies and preclinical trials.

Oligosarcomas, IDH-mutant are distinct and aggressive

Oligodendrogliomas are defined at the molecular level by the presence of an IDH mutation and codeletion of chromosomal arms 1p and 19q. In the past, case reports and small studies described gliomas with sarcomatous features arising from oligodendrogliomas, so called oligosarcomas. Here, we report a series of 24 IDH-mutant oligosarcomas from 23 patients forming a distinct methylation class. The tumors were recurrences from prior oligodendrogliomas or developed de novo. Precursor tumors of 12 oligosarcomas were histologically and molecularly indistinguishable from conventional oligodendrogliomas. Oligosarcoma tumor cells were embedded in a dense network of reticulin fibers, frequently showing p53 accumulation, positivity for SMA and CALD1, loss of OLIG2 and gain of H3K27 trimethylation (H3K27me3) as compared to primary lesions. In 5 oligosarcomas no 1p/19q codeletion was detectable, although it was present in the primary lesions. Copy number neutral LOH was determined as underlying mechanism. Oligosarcomas harbored an increased chromosomal copy number variation load with frequent CDKN2A/B deletions. Proteomic profiling demonstrated oligosarcomas to be highly distinct from conventional CNS WHO grade 3 oligodendrogliomas with consistent evidence for a smooth muscle differentiation. Expression of several tumor suppressors was reduced with NF1 being lost frequently. In contrast, oncogenic YAP1 was aberrantly overexpressed in oligosarcomas. Panel sequencing revealed mutations in NF1 and TP53 along with IDH1/2 and TERT promoter mutations. Survival of patients was significantly poorer for oligosarcomas as first recurrence than for grade 3 oligodendrogliomas as first recurrence. These results establish oligosarcomas as a distinct group of IDH-mutant gliomas differing from conventional oligodendrogliomas on the histologic, epigenetic, proteomic, molecular and clinical level. The diagnosis can be based on the combined presence of (a) sarcomatous histology, (b) IDH-mutation and (c) TERT promoter mutation and/or 1p/19q codeletion, or, in unresolved cases, on its characteristic DNA methylation profile.

PATH-23. OLIGOSARCOMA, IDH-MUTANT IS A DISTINCT AGGRESSIVE TYPE

Abstract Oligodendrogliomas are defined by IDH-mutations and the co-deletion of chromosomal arms 1p and 19q. In the past, case reports and small series described gliomas with sarcomatous features arising from oligodendrogliomas, so called oligosarcomas. Here, we report a group of 23 IDH-mutant oligosarcomas that form a distinct methylation group. The tumors were recurrences from prior oligodendrogliomas or developed de novo. Precursor tumors of 11 oligosarcomas were histologically and molecularly indistinguishable from conventional oligodendrogliomas. Tumor cells were embedded in a reticulin fibers network in all oligosarcomas, frequently showing p53 accumulation, positivity for SMA, and regain of H3K27me3 as compared to primary lesions. In 5 oligosarcomas no 1p/19q co-deletion was detectable, even though it was present in the primary lesions. Oligosarcomas harbored an increased chromosomal copy number variation load with frequent CDKN2A/B deletions. Proteomic profiling demonstrated oligosarcomas to be highly distinct from conventional grade 3 oligodendrogliomas with consistent evidence for smooth muscle differentiation. Expression of several tumor suppressors including NF1 and STAG2 was lost whereas oncogenic YAP1 was aberrantly overexpressed in oligosarcomas. Panel sequencing revealed mutations in TP53 and NF1 along with IDH1/2 and TERT promoter mutations. Survival of patients was significantly poorer for oligosarcomas as for grade 3 oligodendrogliomas but comparable to grade 4 IDH-mutant astrocytomas. These results establish oligosarcoma as a distinct type of IDH-mutant glioma differing from conventional oligodendrogliomas on the histologic, epigenetic, proteomic, molecular and clinical level. Diagnosis can be based on the characteristic DNA methylation profile or the combined evidence of sarcomatous histology, IDH-mutation and TERT promoter mutation with or without 1p/19q co-deletion.

Optimized CRISPR-mediated gene knockin reveals FOXP3-independent maintenance of human Treg identity.

Regulatory Tcell (Treg) therapy is a promising curative approach for a variety of immune-mediated conditions. CRISPR-based genome editing allows precise insertion of transgenes through homology-directed repair, but its use in human Tregs has been limited. We report an optimized protocol for CRISPR-mediated gene knockin in human Tregs with high-yield expansion. To establish a benchmark of human Treg dysfunction, we target the master transcription factor FOXP3 in naive and memory Tregs. Although FOXP3-ablated Tregs upregulate cytokine expression, effects on suppressive capacity invitro manifest slowly and primarily in memory Tregs. Moreover, FOXP3-ablated Tregs retain their characteristic protein, transcriptional, and DNA methylation profile. Instead, FOXP3 maintains DNA methylation at regions enriched for AP-1 binding sites. Thus, although FOXP3 is important for human Treg development, it has a limited role in maintaining mature Treg identity. Optimized gene knockin with human Tregs will enable mechanistic studies and the development of tailored, next-generation Treg cell therapies.

DNA methylation profiling in mummified human remains from the eighteenth-century

Reconstruction of ancient epigenomes by DNA methylation (DNAm) can shed light into the composition of cell types, disease states, and age at death. However, such analysis is hampered by impaired DNA quality and little is known how decomposition affects DNAm. In this study, we determined if EPIC Illumina BeadChip technology is applicable for specimens from mummies of the eighteenth century CE. Overall, the signal intensity on the microarray was extremely low, but for one of two samples we were able to detect characteristic DNAm signals in a subset of CG dinucleotides (CpGs), which were selected with a stringent processing pipeline. Using only these CpGs we could train epigenetic signatures with reference DNAm profiles of multiple tissues and our predictions matched the fact that the specimen was lung tissue from a 28-year-old woman. Thus, we provide proof of principle that Illumina BeadChips are applicable for DNAm profiling in ancient samples.

Glioblastomas with primitive neuronal component harbor a distinct methylation and copy-number profile with inactivation of TP53, PTEN, and RB1

Glioblastoma IDH-wildtype presents with a wide histological spectrum. Some features are so distinctive that they are considered as separate histological variants or patterns for the purpose of classification. However, these usually lack defined (epi-)genetic alterations or profiles correlating with this histology. Here, we describe a molecular subtype with overlap to the unique histological pattern of glioblastoma with primitive neuronal component. Our cohort consists of 63 IDH-wildtype glioblastomas that harbor a characteristic DNA methylation profile. Median age at diagnosis was 59.5 years. Copy-number variations and genetic sequencing revealed frequent alterations in TP53, RB1 and PTEN, with fewer gains of chromosome 7 and homozygous CDKN2A/B deletions than usually described for IDH-wildtype glioblastoma. Gains of chromosome 1 were detected in more than half of the cases. A poorly differentiated phenotype with frequent absence of GFAP expression, high proliferation index and strong staining for p53 and TTF1 often caused misleading histological classification as carcinoma metastasis or primitive neuroectodermal tumor. Clinically, many patients presented with leptomeningeal dissemination and spinal metastasis. Outcome was poor with a median overall survival of only 12 months. Overall, we describe a new molecular subtype of IDH-wildtype glioblastoma with a distinct histological appearance and genetic signature.

The epigenetic pioneer EGR2 initiates DNA demethylation in differentiating monocytes at both stable and transient binding sites

The differentiation of human blood monocytes (MO), the post-mitotic precursors of macrophages (MAC) and dendritic cells (moDC), is accompanied by the active turnover of DNA methylation, but the extent, consequences and mechanisms of DNA methylation changes remain unclear. Here, we profile and compare epigenetic landscapes during IL-4/GM-CSF-driven MO differentiation across the genome and detect several thousand regions that are actively demethylated during culture, both with or without accompanying changes in chromatin accessibility or transcription factor (TF) binding. We further identify TF that are globally associated with DNA demethylation processes. While interferon regulatory factor 4 (IRF4) is found to control hallmark dendritic cell functions with less impact on DNA methylation, early growth response 2 (EGR2) proves essential for MO differentiation as well as DNA methylation turnover at its binding sites. We also show that ERG2 interacts with the 5mC hydroxylase TET2, and its consensus binding sequences show a characteristic DNA methylation footprint at demethylated sites with or without detectable protein binding. Our findings reveal an essential role for EGR2 as epigenetic pioneer in human MO and suggest that active DNA demethylation can be initiated by the TET2-recruiting TF both at stable and transient binding sites.

Blood-Based Detection of Colorectal Cancer Using Cancer-Specific DNA Methylation Markers.

Cancer tissues have characteristic DNA methylation profiles compared with their corresponding normal tissues that can be utilized for cancer diagnosis with liquid biopsy. Using a genome-scale DNA methylation approach, we sought to identify a panel of DNA methylation markers specific for cell-free DNA (cfDNA) from patients with colorectal cancer (CRC). By comparing DNA methylomes between CRC and normal mucosal tissues or blood leukocytes, we identified eight cancer-specific methylated loci (ADGRB1, ANKRD13, FAM123A, GLI3, PCDHG, PPP1R16B, SLIT3, and TMEM90B) and developed a five-marker panel (FAM123A, GLI3, PPP1R16B, SLIT3, and TMEM90B) that detected CRC in liquid biopsies with a high sensitivity and specificity with a droplet digital MethyLight assay. In a set of cfDNA samples from CRC patients (n = 117) and healthy volunteers (n = 60), a panel of five markers on the platform of the droplet digital MethyLight assay detected stages I–III and stage IV CRCs with sensitivities of 45.9% and 95.7%, respectively, and a specificity of 95.0%. The number of detected markers was correlated with the cancer stage, perineural invasion, lymphatic emboli, and venous invasion. Our five-marker panel with the droplet digital MethyLight assay showed a high sensitivity and specificity for the detection of CRC with cfDNA samples from patients with metastatic CRC.

Primary mismatch repair deficient IDH-mutant astrocytoma (PMMRDIA) is a distinct type with a poor prognosis

Diffuse IDH-mutant astrocytoma mostly occurs in adults and carries a favorable prognosis compared to IDH-wildtype malignant gliomas. Acquired mismatch repair deficiency is known to occur in recurrent IDH-mutant gliomas as resistance mechanism towards alkylating chemotherapy. In this multi-institutional study, we report a novel epigenetic group of 32 IDH-mutant gliomas with proven or suspected hereditary mismatch repair deficiency. None of the tumors exhibited a combined 1p/19q deletion. These primary mismatch repair-deficient IDH-mutant astrocytomas (PMMRDIA) were histologically high-grade and were mainly found in children, adolescents and young adults (median age 14 years). Mismatch repair deficiency syndromes (Lynch or Constitutional Mismatch Repair Deficiency Syndrom (CMMRD)) were clinically diagnosed and/or germline mutations in DNA mismatch repair genes (MLH1, MSH6, MSH2) were found in all cases, except one case with a family and personal history of colon cancer and another case with MSH6-deficiency available only as recurrent tumor. Loss of at least one of the mismatch repair proteins was detected via immunohistochemistry in all, but one case analyzed. Tumors displayed a hypermutant genotype and microsatellite instability was present in more than half of the sequenced cases. Integrated somatic mutational and chromosomal copy number analyses showed frequent inactivation of TP53, RB1 and activation of RTK/PI3K/AKT pathways. In contrast to the majority of IDH-mutant gliomas, more than 60% of the samples in our cohort presented with an unmethylated MGMT promoter. While the rate of immuno-histochemical ATRX loss was reduced, variants of unknown significance were more frequently detected possibly indicating a higher frequency of ATRX inactivation by protein malfunction. Compared to reference cohorts of other IDH-mutant gliomas, primary mismatch repair-deficient IDH-mutant astrocytomas have by far the worst clinical outcome with a median survival of only 15 months irrespective of histological or molecular features. The findings reveal a so far unknown entity of IDH-mutant astrocytoma with high prognostic relevance. Diagnosis can be established by aligning with the characteristic DNA methylation profile, by DNA-sequencing-based proof of mismatch repair deficiency or immunohistochemically demonstrating loss-of-mismatch repair proteins.

Abstract 1276: Analyses of cell-free DNA alterations in physiological conditions and colon diseases for screening, diagnostics and therapy follow-up

Abstract Cell-free DNA (cfDNA) level increases in several physiological conditions, such as pregnancy, aging or high physical activity. Furthermore, alterations of cfDNA concentration and genetic or epigenetic (e.g. DNA methylation) patterns can also be observed in pathological processes, as inflammatory diseases or different cancer types including colorectal cancer (CRC). However, studies about quality changes of cfDNA aimed to analyze its fragment length profile and DNA methylation pattern are lacking. We aimed to define the quantity and quality changes of cfDNA, including concentration, fragment length distribution and global DNA methylation pattern during high physical exercise, and in colorectal adenoma and cancer patients. Moreover, we aimed to investigate the treatment response in plasma of metastatic CRC (mCRC) patients analyzing the above-mentioned parameters and the DNA methylation level of SFRP2 gene during therapy. Plasma fraction was separated from blood samples of 6 healthy athletes before, during and after physical training, and from healthy (n=16), adenoma (n=13), IBD (n=19), and CRC (n=16) patients. Moreover, plasma longitudinal assessment was performed in the case of 20 mCRC patients treated with chemotherapy. CfDNA level was quantified with Qubit 1.0 (Thermo Fisher Scientific), fragment length distribution was examined by 2100 Bioanalyzer instrument (Agilent), and global DNA methylation level was measured with bisulfite pyrosequencing of long interspersed nuclear element-1 (LINE-1) (Qiagen) in all samples. Besides, DNA methylation status of SFRP2 gene was determined with droplet digital PCR (Bio-Rad) in mCRC plasma samples. Increased cfDNA amount was observed during physical exercise (198.5±90ng), in comparison with control phase (86.3±40.6ng), and then in restitution period, lower cfDNA level (155.6±119.3ng) was detected. Elevated cfDNA amounts have been found in plasma of adenoma (72.1±37.5ng), IBD (78.0±50.9ng) and CRC (84.5±70.1ng) patients compared to controls (36.2±11.3ng). CfDNA fragment length distribution showed different patterns in each sample group, longer fragments (200-600bp) enriched in AD and CRC samples in addition to short (&amp;lt;200bp) fragments. Slightly lower level of global DNA methylation was noticed in AD (79,5%) and CRC (79,2%) compared to normal (83,8%) samples. Interestingly, the amount and global methylation level alteration of cfDNA revealed negative correlation in CRC patients during therapy and progression. Furthermore, the amount of methylated SFRP2 copies (5-500copies) was increased in the case of progressive disease, while were undetectable in patients achieved complete remission. Characteristic cfDNA level, fragment length, and global DNA methylation level were identified in the different sample groups. The above-mentioned parameters can be applied for the follow-up of mCRC patients during chemotherapy and could be integrated into current follow-up methods. Citation Format: Barbara K. Bartak, Krisztina Andrea Szigeti, Alexandra Kalmar, Orsolya Galamb, Zsófia Brigitta Nagy, Sara Zsigrai, Zsolt Tulassay, Peter Igaz, Magdolna Dank, Bela Molnar. Analyses of cell-free DNA alterations in physiological conditions and colon diseases for screening, diagnostics and therapy follow-up [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1276.

Characterisation of DNA methylation changes in EBF3 and TBC1D16 associated with tumour progression and metastasis in multiple cancer types

BackgroundCharacteristic DNA methylation differences have been identified between primary and metastatic melanomas at EBF3 and/or TBC1D16 gene loci. To further evaluate whether these epigenetic changes may act more generally as drivers of tumour onset and metastasis, we have investigated DNA methylation changes involving EBF3 and TBC1D16 in additional publicly available data of multiple different tumour types.ResultsPromoter hypermethylation and gene body hypomethylation of EBF3 were observed in a number of metastatic tumour types, when compared to normal or primary tumour tissues, as well as in tumour vs normal tissues and in a colorectal primary/metastasis pair, although not all tumour samples or primary/metastasis cancer pairs exhibited altered patterns of EBF3 methylation. In addition, hypomethylation of TBC1D16 was observed in multiple tumours, including a breast cancer primary/metastasis pair, and to a lesser degree in melanoma, although again not all tumours or cancer primary/metastasis pairs exhibited altered patterns of methylation.ConclusionsThese findings suggest characteristic DNA methylation changes in EBF3 and TBC1D16 are relatively common tumour-associated epigenetic events in multiple tumour types, which is consistent with a potential role as more general drivers of tumour progression.

Glioma epigenetics: From subclassification to novel treatment options.

Gliomas are the most common malignant primary brain tumors, of which glioblastoma is the most malignant form (WHO grade IV), and notorious for treatment resistance. Over the last decade mutations in epigenetic regulator genes have been identified as key drivers of subtypes of gliomas with distinct clinical features. Most characteristic are mutations in IDH1 or IDH2 in lower grade gliomas, and histone 3 mutations in pediatric high grade gliomas that are also associated with characteristic DNA methylation patterns. Furthermore, in adult glioblastoma patients epigenetic silencing of the DNA repair gene MGMT by promoter methylation is predictive for benefit from alkylating agent therapy. These epigenetic alterations are used as biomarkers and play a central role for classification of gliomas (WHO 2016) and treatment decisions. Here we review the pivotal role of epigenetic alterations in the etiology and biology of gliomas. We summarize the complex interactions between "driver" mutations, DNA methylation, histone post-translational modifications, and overall chromatin organization, and how they inform current efforts of testing epigenetic compounds and combinations in preclinical and clinical studies.