DNA methylation-based classification of central nervous system tumours.

Accurate pathological diagnosis is crucial for optimal management of patients with cancer. For the approximately 100 known tumour types of the central nervous system, standardization of the diagnostic process has been shown to be particularly challenging-with substantial inter-observer variability in the histopathological diagnosis of many tumour types. Here we present a comprehensive approach for the DNA methylation-based classification of central nervous system tumours across all entities and age groups, and demonstrate its application in a routine diagnostic setting. We show that the availability of this method may have a substantial impact on diagnostic precision compared to standard methods, resulting in a change of diagnosis in up to 12% of prospective cases. For broader accessibility, we have designed a free online classifier tool, the use of which does not require any additional onsite data processing. Our results provide a blueprint for the generation of machine-learning-based tumour classifiers across other cancer entities, with the potential to fundamentally transform tumour pathology.

A complementary study approach unravels novel players in the pathoetiology of Hirschsprung disease.

Hirschsprung disease (HSCR, OMIM 142623) involves congenital intestinal obstruction caused by dysfunction of neural crest cells and their progeny during enteric nervous system (ENS) development. HSCR is a multifactorial disorder; pathogenetic variants accounting for disease phenotype are identified only in a minority of cases, and the identification of novel disease-relevant genes remains challenging. In order to identify and to validate a potential disease-causing relevance of novel HSCR candidate genes, we established a complementary study approach, combining whole exome sequencing (WES) with transcriptome analysis of murine embryonic ENS-related tissues, literature and database searches, in silico network analyses, and functional readouts using candidate gene-specific genome-edited cell clones. WES datasets of two patients with HSCR and their non-affected parents were analysed, and four novel HSCR candidate genes could be identified: ATP7A, SREBF1, ABCD1 and PIAS2. Further rare variants in these genes were identified in additional HSCR patients, suggesting disease relevance. Transcriptomics revealed that these genes are expressed in embryonic and fetal gastrointestinal tissues. Knockout of these genes in neuronal cells demonstrated impaired cell differentiation, proliferation and/or survival. Our approach identified and validated candidate HSCR genes and provided further insight into the underlying pathomechanisms of HSCR.

Tumor DNA-methylome derived epigenetic fingerprint identifies HPV-negative head and neck patients at risk for locoregional recurrence after postoperative radiochemotherapy.

Biomarkers with relevance for loco-regional therapy are needed in human papillomavirus negative aka HPV(-) head and neck squamous cell carcinoma (HNSCC). Based on the premise that DNA methylation pattern is highly conserved, we sought to develop a reliable and robust methylome-based classifier identifying HPV(-) HNSCC patients at risk for loco-regional recurrence (LR) and all-event progression after postoperative radiochemotherapy (PORT-C). The training cohort consisted of HPV-DNA negative HNSCC patients (n = 128) homogeneously treated with PORT-C in frame of the German Cancer Consortium-Radiation Oncology Group (DKTK-ROG) multicenter biomarker trial. DNA Methylation analysis was performed using Illumina 450 K and 850 K-EPIC microarray technology. The performance of the classifier was integrated with a series of biomarkers studied in the training set namely hypoxia-, 5-microRNA (5-miR), stem-cell gene-expression signatures and immunohistochemistry (IHC)-based immunological characterization of tumors (CD3/CD8/PD-L1/PD1). Validation occurred in an independent cohort of HPV(-) HNSCC patients, pooled from two German centers (n = 125). We identified a 38-methylation probe-based HPV(-) Independent Classifier of disease Recurrence (HICR) with high prognostic value for LR, distant metastasis and overall survival (P < 10-9 ). HICR remained significant after multivariate analysis adjusting for anatomical site, lymph node extracapsular extension (ECE) and size (T-stage). HICR high-risk tumors were enriched for younger patients with hypoxic tumors (15-gene signature) and elevated 5-miR score. After adjustment for hypoxia and 5-miR covariates, HICR maintained predicting all endpoints. HICR provides a novel mean for assessing the risk of LR in HPV(-) HNSCC patients treated with PORT-C and opens a new opportunity for biomarker-assisted stratification and therapy adaptation in these patients.

DNA-methylome-assisted classification of patients with poor prognostic subventricular zone associated IDH-wildtype glioblastoma.

Glioblastoma (GBM) derived from the "stem cell" rich subventricular zone (SVZ) may constitute a therapy-refractory subgroup of tumors associated with poor prognosis. Risk stratification for these cases is necessary but is curtailed by error prone imaging-based evaluation. Therefore, we aimed to establish a robust DNA methylome-based classification of SVZ GBM and subsequently decipher underlying molecular characteristics. MRI assessment of SVZ association was performed in a retrospective training set of IDH-wildtype GBM patients (n = 54) uniformly treated with postoperative chemoradiotherapy. DNA isolated from FFPE samples was subject to methylome and copy number variation (CNV) analysis using Illumina Platform and cnAnalysis450k package. Deep next-generation sequencing (NGS) of a panel of 130 GBM-related genes was conducted (Agilent SureSelect/Illumina). Methylome, transcriptome, CNV, MRI, and mutational profiles of SVZ GBM were further evaluated in a confirmatory cohort of 132 patients (TCGA/TCIA). A 15 CpG SVZ methylation signature (SVZM) was discovered based on clustering and random forest analysis. One third of CpG in the SVZM were associated with MAB21L2/LRBA. There was a 14.8% (n = 8) discordance between SVZM vs. MRI classification. Re-analysis of these patients favored SVZM classification with a hazard ratio (HR) for OS of 2.48 [95% CI 1.35-4.58], p = 0.004 vs. 1.83 [1.0-3.35], p = 0.049 for MRI classification. In the validation cohort, consensus MRI based assignment was achieved in 62% of patients with an intraclass correlation (ICC) of 0.51 and non-significant HR for OS (2.03 [0.81-5.09], p = 0.133). In contrast, SVZM identified two prognostically distinct subgroups (HR 3.08 [1.24-7.66], p = 0.016). CNV alterations revealed loss of chromosome 10 in SVZM- and gains on chromosome 19 in SVZM- tumors. SVZM- tumors were also enriched for differentially mutated genes (p < 0.001). In summary, SVZM classification provides a novel means for stratifying GBM patients with poor prognosis and deciphering molecular mechanisms governing aggressive tumor phenotypes.

Mutational patterns and regulatory networks in epigenetic subgroups of meningioma.

DNA methylation patterns delineate clinically relevant subgroups of meningioma. We previously established the six meningioma methylation classes (MC) benign 1-3, intermediate A and B, and malignant. Here, we set out to identify subgroup-specific mutational patterns and gene regulation. Whole genome sequencing was performed on 62 samples across all MCs and WHO grades from 62 patients with matched blood control, including 40 sporadic meningiomas and 22 meningiomas arising after radiation (Mrad). RNA sequencing was added for 18 of these cases and chromatin-immunoprecipitation for histone H3 lysine 27 acetylation (H3K27ac) followed by sequencing (ChIP-seq) for 16 samples. Besides the known mutations in meningioma, structural variants were found as the mechanism of NF2 inactivation in a small subset (5%) of sporadic meningiomas, similar to previous reports for Mrad. Aberrations of DMD were found to be enriched in MCs with NF2 mutations, and DMD was among the most differentially upregulated genes in NF2 mutant compared to NF2 wild-type cases. The mutational signature AC3, which has been associated with defects in homologous recombination repair (HRR), was detected in both sporadic meningioma and Mrad, but widely distributed across the genome in sporadic cases and enriched near genomic breakpoints in Mrad. Compared to the other MCs, the number of single nucleotide variants matching the AC3 pattern was significantly higher in the malignant MC, which also exhibited higher genomic instability, determined by the numbers of both large segments affected by copy number alterations and breakpoints between large segments. ChIP-seq analysis for H3K27ac revealed a specific activation of genes regulated by the transcription factor FOXM1 in the malignant MC. This analysis also revealed a super enhancer near the HOXD gene cluster in this MC, which, together with general upregulation of HOX genes in the malignant MC, indicates a role of HOX genes in meningioma aggressiveness. This data elucidates the biological mechanisms rendering different epigenetic subgroups of meningiomas, and suggests leveraging HRR as a novel therapeutic target.

Array-based DNA-methylation profiling in sarcomas with small blue round cell histology provides valuable diagnostic information.

Undifferentiated solid tumors with small blue round cell histology and expression of CD99 mostly resemble Ewing sarcoma. However, they also may include other tumors such as mesenchymal chondrosarcoma, synovial sarcoma, or small cell osteosarcoma. Definitive classification usually requires detection of entity-specific mutations. While this approach identifies the majority of Ewing sarcomas, a subset of lesions remains unclassified and, therefore, has been termed "Ewing-like sarcomas" or small blue round cell tumors not otherwise specified. We developed an approach for further characterization of small blue round cell tumors not otherwise specified using an array-based DNA-methylation profiling approach. Data were analyzed by unsupervised clustering and t-distributed stochastic neighbor embedding analysis and compared with a reference methylation data set of 460 well-characterized prototypical sarcomas encompassing 18 subtypes. Verification was performed by additional FISH analyses, RNA sequencing from formalin-fixed paraffin-embedded material or immunohistochemical marker analyses. In a cohort of more than 1,000 tumors assumed to represent Ewing sarcomas, 30 failed to exhibit the typical EWS translocation. These tumors were subjected to methylation profiling and could be assigned to Ewing sarcoma in 14 (47%), to small blue round cell tumors with CIC alteration in 6 (20%), to small blue round cell tumors with BCOR alteration in 4 (13%), to synovial sarcoma and to malignant rhabdoid tumor in 2 cases each. One single case each was allotted to mesenchymal chondrosarcoma and adamantinoma. 12/14 tumors classified as Ewing sarcoma could be verified by demonstrating either a canonical EWS translocation evading initial testing, by identifying rare breakpoints or fusion partners. The methylation-based assignment of the remaining small blue round cell tumors not otherwise specified also could be verified by entity-specific molecular alterations in 13/16 cases. In conclusion, array-based DNA-methylation analysis of undifferentiated tumors with small blue round cell histology is a powerful tool for precisely classifying this diagnostically challenging tumor group.

Novel, improved grading system(s) for IDH-mutant astrocytic gliomas.

According to the 2016 World Health Organization Classification of Tumors of the Central Nervous System (2016 CNS WHO), IDH-mutant astrocytic gliomas comprised WHO grade II diffuse astrocytoma, IDH-mutant (AIIIDHmut), WHO grade III anaplastic astrocytoma, IDH-mutant (AAIIIIDHmut), and WHO grade IV glioblastoma, IDH-mutant (GBMIDHmut). Notably, IDH gene status has been made the major criterion for classification while the manner of grading has remained unchanged: it is based on histological criteria that arose from studies which antedated knowledge of the importance of IDH status in diffuse astrocytic tumor prognostic assessment. Several studies have now demonstrated that the anticipated differences in survival between the newly defined AIIIDHmut and AAIIIIDHmut have lost their significance. In contrast, GBMIDHmut still exhibits a significantly worse outcome than its lower grade IDH-mutant counterparts. To address the problem of establishing prognostically significant grading for IDH-mutant astrocytic gliomas in the IDH era, we undertook a comprehensive study that included assessment of histological and genetic approaches to prognosis in these tumors. A discovery cohort of 211 IDH-mutant astrocytic gliomas with an extended observation was subjected to histological review, image analysis, and DNA methylation studies. Tumor group-specific methylation profiles and copy number variation (CNV) profiles were established for all gliomas. Algorithms for automated CNV analysis were developed. All tumors exhibiting 1p/19q codeletion were excluded from the series. We developed algorithms for grading, based on molecular, morphological and clinical data. Performance of these algorithms was compared with that of WHO grading. Three independent cohorts of 108, 154 and 224 IDH-mutant astrocytic gliomas were used to validate this approach. In the discovery cohort several molecular and clinical parameters were of prognostic relevance. Most relevant for overall survival (OS) was CDKN2A/B homozygous deletion. Other parameters with major influence were necrosis and the total number of CNV. Proliferation as assessed by mitotic count, which is a key parameter in 2016 CNS WHO grading, was of only minor influence. Employing the parameters most relevant for OS in our discovery set, we developed two models for grading these tumors. These models performed significantly better than WHO grading in both the discovery and the validation sets. Our novel algorithms for grading IDH-mutant astrocytic gliomas overcome the challenges caused by introduction of IDH status into the WHO classification of diffuse astrocytic tumors. We propose that these revised approaches be used for grading of these tumors and incorporated into future WHO criteria.

DNA methylation-based classification and grading system for meningioma: a multicentre, retrospective analysis.

BACKGROUND: The WHO classification of brain tumours describes 15 subtypes of meningioma. Nine of these subtypes are allotted to WHO grade I, and three each to grade II and grade III. Grading is based solely on histology, with an absence of molecular markers. Although the existing classification and grading approach is of prognostic value, it harbours shortcomings such as ill-defined parameters for subtypes and grading criteria prone to arbitrary judgment. In this study, we aimed for a comprehensive characterisation of the entire molecular genetic landscape of meningioma to identify biologically and clinically relevant subgroups. METHODS: In this multicentre, retrospective analysis, we investigated genome-wide DNA methylation patterns of meningiomas from ten European academic neuro-oncology centres to identify distinct methylation classes of meningiomas. The methylation classes were further characterised by DNA copy number analysis, mutational profiling, and RNA sequencing. Methylation classes were analysed for progression-free survival outcomes by the Kaplan-Meier method. The DNA methylation-based and WHO classification schema were compared using the Brier prediction score, analysed in an independent cohort with WHO grading, progression-free survival, and disease-specific survival data available, collected at the Medical University Vienna (Vienna, Austria), assessing methylation patterns with an alternative methylation chip. FINDINGS: We retrospectively collected 497 meningiomas along with 309 samples of other extra-axial skull tumours that might histologically mimic meningioma variants. Unsupervised clustering of DNA methylation data clearly segregated all meningiomas from other skull tumours. We generated genome-wide DNA methylation profiles from all 497 meningioma samples. DNA methylation profiling distinguished six distinct clinically relevant methylation classes associated with typical mutational, cytogenetic, and gene expression patterns. Compared with WHO grading, classification by individual and combined methylation classes more accurately identifies patients at high risk of disease progression in tumours with WHO grade I histology, and patients at lower risk of recurrence among WHO grade II tumours (p=0·0096) from the Brier prediction test). We validated this finding in our independent cohort of 140 patients with meningioma. INTERPRETATION: DNA methylation-based meningioma classification captures clinically more homogenous groups and has a higher power for predicting tumour recurrence and prognosis than the WHO classification. The approach presented here is potentially very useful for stratifying meningioma patients to observation-only or adjuvant treatment groups. We consider methylation-based tumour classification highly relevant for the future diagnosis and treatment of meningioma. FUNDING: German Cancer Aid, Else Kröner-Fresenius Foundation, and DKFZ/Heidelberg Institute of Personalized Oncology/Precision Oncology Program.

Type 2 diabetes and leucocyte DNA methylation: an epigenome-wide association study in over 1,500 older adults.

AIMS/HYPOTHESIS: Development of type 2 diabetes depends on environmental and genetic factors. We investigated the epigenome-wide association of prevalent diabetes with DNA methylation (DNAm) in peripheral blood. METHODS: DNAm was measured in whole blood with the Illumina Infinium HumanMethylation450 BeadChip in two subsamples of participants from the ESTHER cohort study. Cohort 1 included 988 participants, who were consecutively recruited between July and October 2000 and cohort 2 included 527 randomly selected participants. The association of DNAm with prevalent type 2 diabetes at recruitment was estimated using median regression analysis adjusting for sex, age, BMI, smoking behaviour, cell composition and batch at 361,922 CpG sites. RESULTS: Type 2 diabetes was prevalent in 16% of the participants, and diabetes was poorly controlled in 45% of the diabetic patients. In cohort 1 (discovery) DNAm at 39 CpGs was significantly associated with prevalent diabetes after correction for multiple testing. In cohort 2 (replication) at one of these CpGs, DNAm was still significantly associated. Decreasing methylation levels at cg19693031 with increasing fasting glucose and HbA1c concentrations were observed using restricted cubic spline analysis. In diabetic patients with poorly controlled diabetes, the decrease in estimated DNAm levels was approximately 5% in comparison with participants free of diagnosed diabetes. CONCLUSIONS/INTERPRETATION: Cg19693031, which is located within the 3'-untranslated region of TXNIP, might play a role in the pathophysiology of type 2 diabetes. This result appears biologically plausible given that thioredoxin-interacting protein is overexpressed in diabetic animals and humans and 3'-untranslated regions are known to play a regulatory role in gene expression.

Cross-sectional and longitudinal changes in DNA methylation with age: an epigenome-wide analysis revealing over 60 novel age-associated CpG sites.

Understanding the role of epigenetic modifications, e.g. DNA methylation, in the process of aging requires the characterization of methylation patterns in large cohorts. We analysed >480 000 CpG sites using Infinium HumanMethylation450 BeadChip (Illumina) in whole blood DNA of 965 participants of a population-based cohort study aged between 50 and 75 years. In an exploratory analysis in 400 individuals, 200 CpG sites with the highest Spearman correlation coefficients for the association between methylation and age were identified. Of these 200 CpGs, 162 were significantly associated with age, which was verified in an independent cohort of 498 individuals using mixed linear regression models adjusted for gender, smoking behaviour, age-related diseases and random batch effect and corrected for multiple testing by Bonferroni. In another independent cohort of 67 individuals without history of major age-related diseases and with a follow-up of 8 years, we observed a gain in methylation at 96% (52%, significant) of the positively age-associated CpGs and a loss at all (89%, significant) of the negatively age-associated CpGs in each individual while getting 8 years older. A regression model for age prediction based on 17 CpGs as predicting variables explained 71% of the variance in age with an average accuracy of 2.6 years. In comparison with cord blood samples obtained from the Ulm Birth Cohort Study, we observed a more than 2-fold change in mean methylation levels from birth to older age at 86 CpGs. We were able to identify 65 novel CpG sites with significant association of methylation with age.

Methylation-based classification of benign and malignant peripheral nerve sheath tumors.

The vast majority of peripheral nerve sheath tumors derive from the Schwann cell lineage and comprise diverse histological entities ranging from benign schwannomas and neurofibromas to high-grade malignant peripheral nerve sheath tumors (MPNST), each with several variants. There is increasing evidence for methylation profiling being able to delineate biologically relevant tumor groups even within the same cellular lineage. Therefore, we used DNA methylation arrays for methylome- and chromosomal profile-based characterization of 171 peripheral nerve sheath tumors. We analyzed 28 conventional high-grade MPNST, three malignant Triton tumors, six low-grade MPNST, four epithelioid MPNST, 33 neurofibromas (15 dermal, 8 intraneural, 10 plexiform), six atypical neurofibromas, 43 schwannomas (including 5 NF2 and 5 schwannomatosis associated cases), 11 cellular schwannomas, 10 melanotic schwannomas, 7 neurofibroma/schwannoma hybrid tumors, 10 nerve sheath myxomas and 10 ganglioneuromas. Schwannomas formed different epigenomic subgroups including a vestibular schwannoma subgroup. Cellular schwannomas were not distinct from conventional schwannomas. Nerve sheath myxomas and neurofibroma/schwannoma hybrid tumors were most similar to schwannomas. Dermal, intraneural and plexiform neurofibromas as well as ganglioneuromas all showed distinct methylation profiles. Atypical neurofibromas and low-grade MPNST were indistinguishable with a common methylation profile and frequent losses of CDKN2A. Epigenomic analysis finds two groups of conventional high-grade MPNST sharing a frequent loss of neurofibromin. The larger of the two groups shows an additional loss of trimethylation of histone H3 at lysine 27 (H3K27me3). The smaller one retains H3K27me3 and is found in spinal locations. Sporadic MPNST with retained neurofibromin expression did not form an epigenetic group and most cases could be reclassified as cellular schwannomas or soft tissue sarcomas. Widespread immunohistochemical loss of H3K27me3 was exclusively seen in MPNST of the main methylation cluster, which defines it as an additional useful marker for the differentiation of cellular schwannoma and MPNST.

Next-generation sequencing in routine brain tumor diagnostics enables an integrated diagnosis and identifies actionable targets.

With the number of prognostic and predictive genetic markers in neuro-oncology steadily growing, the need for comprehensive molecular analysis of neuropathology samples has vastly increased. We therefore developed a customized enrichment/hybrid-capture-based next-generation sequencing (NGS) gene panel comprising the entire coding and selected intronic and promoter regions of 130 genes recurrently altered in brain tumors, allowing for the detection of single nucleotide variations, fusions, and copy number aberrations. Optimization of probe design, library generation and sequencing conditions on 150 samples resulted in a 5-workday routine workflow from the formalin-fixed paraffin-embedded sample to neuropathological report. This protocol was applied to 79 retrospective cases with established molecular aberrations for validation and 71 prospective cases for discovery of potential therapeutic targets. Concordance of NGS compared to established, single biomarker methods was 98.0 %, with discrepancies resulting from one case where a TERT promoter mutation was not called by NGS and three ATRX mutations not being detected by Sanger sequencing. Importantly, in samples with low tumor cell content, NGS was able to identify mutant alleles that were not detectable by traditional methods. Information derived from NGS data identified potential targets for experimental therapy in 37/47 (79 %) glioblastomas, 9/10 (90 %) pilocytic astrocytomas, and 5/14 (36 %) medulloblastomas in the prospective target discovery cohort. In conclusion, we present the settings for high-throughput, adaptive next-generation sequencing in routine neuropathology diagnostics. Such an approach will likely become highly valuable in the near future for treatment decision making, as more therapeutic targets emerge and genetic information enters the classification of brain tumors.

Discovery of a novel epigenetic cancer marker related to the oxidative status of human blood.

Long-lasting oxidative stress exposure may lead to relatively stable epigenetic modifications of the DNA in order to activate anti-oxidative defence mechanisms. Oxidative stress related DNA methylation may therefore be associated (causally or as a by-product) with cancer. We measured derivatives of reactive oxygen metabolites (D-ROM), total thiol levels (TTL) and DNA methylation with the Illumina Infinium 450K BeadChip in three samples of German individuals aged ≥50 years: n = 1,000 ESTHER study baseline participants (DNA methylation only), n = 99 ESTHER eight-year follow-up participants and n = 142 participants of the BLITZ study. The correlation coefficient of methylation at cg10342304 and D-ROM in the ESTHER 8-year follow-up sample (r = -0.427; P = 1 × 10(-5)) was replicated with a P-value indicating statistical significance after correction for multiple testing in the BLITZ sample (r = -0.192; P = 0.022). The association was robust to adjusting for potential confounders. In the ESTHER baseline sample, the hazard ratio for cancer development in 11 years of follow-up comparing bottom and top quartile of DNA methylation at cg10342304 was 1.86 (95%-confidence-interval 1.01-3.43). In summary, this first epigenome-wide screening and replication study with oxidative status markers observed a negative correlation of D-ROM levels and DNA methylation at cg10342304 in two independent cohorts. This CpG site is located in the body region of the nucleoredoxin gene. The nucleoredoxin protein is a redox-dependent inhibitor of the Wnt/ß-catenin signaling pathway, a well-characterized cancer pathway. If the observed CpG-cancer association can be successfully replicated by other studies, this epigenetic marker could be an interesting biomarker of cancer risk.

Adult IDH wild type astrocytomas biologically and clinically resolve into other tumor entities.

IDH wild type (IDHwt) anaplastic astrocytomas WHO grade III (AA III) are associated with poor outcome. To address the possibilities of molecular subsets among astrocytoma or of diagnostic reclassification, we analyzed a series of 160 adult IDHwt tumors comprising 120 AA III and 40 diffuse astrocytomas WHO grade II (A II) for molecular hallmark alterations and established methylation and copy number profiles. Based on molecular profiles and hallmark alterations the tumors could be grouped into four major sets. 124/160 (78 %) tumors were diagnosed as the molecular equivalent of conventional glioblastoma (GBM), and 15/160 (9 %) as GBM-H3F3A mutated (GBM-H3). 13/160 (8 %) exhibited a distinct methylation profile that was most similar to GBM-H3-K27, however, lacked the H3F3A mutation. This group was enriched for tumors of infratentorial and midline localization and showed a trend towards a more favorable prognosis. All but one of the 120 IDHwt AA III could be assigned to these three groups. 7 tumors recruited from the 40 A II, comprised a variety of molecular signatures and all but one were reclassified into distinct WHO entities of lower grades. Interestingly, TERT mutations were exclusively restricted to the molecular GBM (78 %) and associated with poor clinical outcome. However, the GBM-H3 group lacking TERT mutations appeared to fare even worse. Our data demonstrate that most of the tumors diagnosed as IDHwt astrocytomas can be allocated to other tumor entities on a molecular basis. The diagnosis of IDHwt diffuse astrocytoma or anaplastic astrocytoma should be used with caution.

CpG-biomarkers in tumor tissue and prediction models for the survival of colorectal cancer: A systematic review and external validation study.

The research aimed to identify previously published CpG-methylation-based prognostic biomarkers and prediction models for colorectal cancer (CRC) prognosis and validate them in a large external cohort. A systematic search was conducted, analyzing 298 unique CpGs and 12 CpG-based prognostic models from 28 studies. After adjustment for clinical variables, 48 CpGs and five prognostic models were confirmed to be associated with survival. However, the discrimination ability of the models was insufficient, with area under the receiver operating characteristic curves ranging from 0.53 to 0.62. Calibration accuracy was mostly poor, and no significant added prognostic value beyond traditional clinical variables was observed. All prognostic models were rated at high risk of bias. While a fraction of CpGs showed potential clinical utility and generalizability, the CpG-based prognostic models performed poorly and lacked clinical relevance.

Large-scale external validation and meta-analysis of gene methylation biomarkers in tumor tissue for colorectal cancer prognosis.

BACKGROUND: DNA methylation biomarkers in colorectal cancer (CRC) tissue hold potential as prognostic indicators. However, individual studies have yielded heterogeneous results, and external validation is largely absent. We conducted a comprehensive external validation and meta-analysis of previously suggested gene methylation biomarkers for CRC prognosis. METHODS: We performed a systematic search to identify relevant studies investigating gene methylation biomarkers for CRC prognosis until March 2024. Our external validation cohort with long-term follow-up included 2303 patients with CRC from 22 hospitals in southwest Germany. We used Cox regression analyses to assess associations between previously suggested gene methylation biomarkers and prognosis, adjusting for clinical variables. We calculated pooled hazard ratios (HRs) and their 95% confidence intervals (CIs) using random-effects models. FINDINGS: Of 151 single gene and 29 multiple gene methylation biomarkers identified from 121 studies, 37 single gene and seven multiple gene biomarkers were significantly associated with CRC prognosis after adjustment for clinical variables. Moreover, the directions of these associations with prognosis remained consistent between the original studies and our validation analyses. Seven single biomarkers and two multi-biomarker signatures were significantly associated with CRC prognosis in the meta-analysis, with a relatively strong level of evidence for CDKN2A, WNT5A, MLH1, and EVL. INTERPRETATION: In a comprehensive evaluation of the so far identified gene methylation biomarkers for CRC prognosis, we identified candidates with potential clinical relevance for further investigation. FUNDING: The German Research Council, the Interdisciplinary Research Program of the National Center for Tumor Diseases (NCT), Germany, the German Federal Ministry of Education and Research.