Hypermethylated Colorectal Cancer Tumours Present a Myc-Driven Hypermetabolism with a One-Carbon Signature Associated with Worsen Prognosis.

Colorectal cancer (CRC) is the second cause of cancer-related death; the CpG-island methylation pathway (CIMP) is associated with KRAS/BRAF mutations, two oncogenes rewiring cell metabolism, worse prognosis, and resistance to classical chemotherapies. Despite this, the question of a possible metabolic rewiring in CIMPs has never been investigated. Here, we analyse whether metabolic dysregulations are associated with tumour methylation by evaluating the transcriptome of CRC tumours. CIMP-high patients were found to present a hypermetabolism, activating mainly carbohydrates, folates, sphingolipids, and arachidonic acid metabolic pathways. A third of these genes had epigenetic targets of Myc in their proximal promoter, activating carboxylic acid, tetrahydrofolate interconversion, nucleobase, and oxoacid metabolisms. In the Myc signature, the expression of GAPDH, TYMS, DHFR, and TK1 was enough to predict methylation levels, microsatellite instability (MSI), and mutations in the mismatch repair (MMR) machinery, which are strong indicators of responsiveness to immunotherapies. Finally, we discovered that CIMP tumours harboured an increase in genes involved in the one-carbon metabolism, a pathway critical to providing nucleotides for cancer growth and methyl donors for DNA methylation, which is associated with worse prognosis and tumour hypermethylation. Transcriptomics could hence become a tool to help clinicians stratify their patients better.

In Silico Gene Prioritization Highlights the Significance of Bone Morphogenetic Protein 4 (BMP4) Promoter Methylation across All Methylation Clusters in Colorectal Cancer.

The cytosine-phosphate-guanine (CpG) island methylator phenotype (CIMP) represents one of the pathways involved in the development of colorectal cancer, characterized by genome-wide hypermethylation. To identify samples exhibiting hypermethylation, we used unsupervised hierarchical clustering on genome-wide methylation data. This clustering analysis revealed the presence of four distinct subtypes within the tumor samples, namely, CIMP-H, CIMP-L, cluster 3, and cluster 4. These subtypes demonstrated varying levels of methylation, categorized as high, intermediate, and very low. To gain further insights, we mapped significant probes from all clusters to Ensembl Regulatory build 89, with a specific focus on those located within promoter regions or bound regions. By intersecting the methylated promoter and bound regions across all methylation subtypes, we identified a total of 253 genes exhibiting aberrant methylation patterns in the promoter regions across all four subtypes of colorectal cancer. Among these genes, our comprehensive genome-wide analysis highlights bone morphogenic protein 4 (BMP4) as the most prominent candidate. This significant finding was derived through the utilization of various bioinformatics tools, emphasizing the potential role of BMP4 in colorectal cancer development and progression.

DNA hypermethylation driven by DNMT1 and DNMT3A favors tumor immune escape contributing to the aggressiveness of adrenocortical carcinoma.

BACKGROUND: Adrenocortical carcinoma is rare and aggressive endocrine cancer of the adrenal gland. Within adrenocortical carcinoma, a recently described subtype characterized by a CpG island methylator phenotype (CIMP) has been associated with an especially poor prognosis. However, the drivers of CIMP remain unknown. Furthermore, the functional relation between CIMP and poor clinical outcomes of patients with adrenocortical carcinoma stays elusive. RESULTS: Here, we show that CIMP in adrenocortical carcinoma is linked to the increased expression of DNA methyltransferases DNMT1 and DNMT3A driven by a gain of gene copy number and cell hyperproliferation. Importantly, we demonstrate that CIMP contributes to tumor aggressiveness by favoring tumor immune escape. This effect could be at least partially reversed by treatment with the demethylating agent 5-azacytidine. CONCLUSIONS: In sum, our findings suggest that co-treatment with demethylating agents might enhance the efficacy of immunotherapy and could represent a novel therapeutic approach for patients with high CIMP adrenocortical carcinoma.

A Multi-Omics Overview of Colorectal Cancer to Address Mechanisms of Disease, Metastasis, Patient Disparities and Outcomes.

Human colorectal cancer (CRC) is one of the most common malignancies in men and women across the globe, albeit CRC incidence and mortality shows a substantial racial and ethnic disparity, with the highest burden in African American patients. Even with effective screening tools such as colonoscopy and diagnostic detection assays, CRC remains a substantial health burden. In addition, primary tumors located in the proximal (right) or distal (left) sides of the colorectum have been shown to be unique tumor types that require unique treatment schema. Distal metastases in the liver and other organ systems are the major causes of mortality in CRC patients. Characterizing genomic, epigenomic, transcriptomic and proteomic (multi-omics) alterations has led to a better understanding of primary tumor biology, resulting in targeted therapeutic advancements. In this regard, molecular-based CRC subgroups have been developed that show correlations with patient outcomes. Molecular characterization of CRC metastases has highlighted similarities and differences between metastases and primary tumors; however, our understanding as to how to improve patient outcomes based on metastasis biology is lagging and remains a major obstacle to improving CRC patient outcomes. In this review, we will summarize the multi-omics features of primary CRC tumors and their metastases across racial and ethnic groups, the differences in proximal and distal tumor biology, molecular-based CRC subgroups, treatment strategies and challenges for improving patient outcomes.

Impact of IDH Mutations, the 1p/19q Co-Deletion and the G-CIMP Status on Alternative Splicing in Diffuse Gliomas.

By generating protein diversity, alternative splicing provides an important oncogenic pathway. Isocitrate dehydrogenase (IDH) 1 and 2 mutations and 1p/19q co-deletion have become crucial for the novel molecular classification of diffuse gliomas, which also incorporates DNA methylation profiling. In this study, we have carried out a bioinformatics analysis to examine the impact of the IDH mutation, as well as the 1p/19q co-deletion and the glioma CpG island methylator phenotype (G-CIMP) status on alternative splicing in a cohort of 662 diffuse gliomas from The Cancer Genome Atlas (TCGA). We identify the biological processes and molecular functions affected by alternative splicing in the various glioma subgroups and provide evidence supporting the important contribution of alternative splicing in modulating epigenetic regulation in diffuse gliomas. Targeting the genes and pathways affected by alternative splicing might provide novel therapeutic opportunities against gliomas.

Recent advances in understanding DNA methylation of prostate cancer.

Epigenetic modifications, such as DNA methylation, is widely studied in cancer. DNA methylation patterns have been shown to distinguish between benign and malignant tumors in various cancers, including prostate cancer. It may also contribute to oncogenesis, as it is frequently associated with downregulation of tumor suppressor genes. Aberrant patterns of DNA methylation, in particular the CpG island hypermethylator phenotype (CIMP), have shown associative evidence with distinct clinical features and outcomes, such as aggressive subtypes, higher Gleason score, prostate-specific antigen (PSA), and overall tumor stage, overall worse prognosis, as well as reduced survival. In prostate cancer, hypermethylation of specific genes is significantly different between tumor and normal tissues. Methylation patterns could distinguish between aggressive subtypes of prostate cancer, including neuroendocrine prostate cancer (NEPC) and castration resistant prostate adenocarcinoma. Further, DNA methylation is detectable in cell-free DNA (cfDNA) and is reflective of clinical outcome, making it a potential biomarker for prostate cancer. This review summarizes recent advances in understanding DNA methylation alterations in cancers with the focus on prostate cancer. We discuss the advanced methodology used for evaluating DNA methylation changes and the molecular regulators behind these changes. We also explore the clinical potential of DNA methylation as prostate cancer biomarkers and its potential for developing targeted treatment of CIMP subtype of prostate cancer.

DNA Methylation and Histone Modification in Low-Grade Gliomas: Current Understanding and Potential Clinical Targets.

Gliomas, the most common type of malignant primary brain tumor, were conventionally classified through WHO Grades I-IV (now 1-4), with low-grade gliomas being entities belonging to Grades 1 or 2. While the focus of the WHO Classification for Central Nervous System (CNS) tumors had historically been on histopathological attributes, the recently released fifth edition of the classification (WHO CNS5) characterizes brain tumors, including gliomas, using an integration of histological and molecular features, including their epigenetic changes such as histone methylation, DNA methylation, and histone acetylation, which are increasingly being used for the classification of low-grade gliomas. This review describes the current understanding of the role of DNA methylation, demethylation, and histone modification in pathogenesis, clinical behavior, and outcomes of brain tumors, in particular of low-grade gliomas. The review also highlights potential diagnostic and/or therapeutic targets in associated cellular biomolecules, structures, and processes. Targeting of MGMT promoter methylation, TET-hTDG-BER pathway, association of G-CIMP with key gene mutations, PARP inhibition, IDH and 2-HG-associated processes, TERT mutation and ARL9-associated pathways, DNA Methyltransferase (DNMT) inhibition, Histone Deacetylase (HDAC) inhibition, BET inhibition, CpG site DNA methylation signatures, along with others, present exciting avenues for translational research. This review also summarizes the current clinical trial landscape associated with the therapeutic utility of epigenetics in low-grade gliomas. Much of the evidence currently remains restricted to preclinical studies, warranting further investigation to demonstrate true clinical utility.

MCC Gene Silencing Is a CpG Island Methylator Phenotype-Associated Factor That Predisposes Colon Cancer Cells to Irinotecan and Olaparib.

Chemotherapy is a mainstay of colorectal cancer treatment, and often involves a combination drug regime. CpG island methylator phenotype (CIMP)-positive tumors are potentially more responsive to the topoisomerase-inhibitor irinotecan. The mechanistic basis of the increased sensitivity of CIMP cancers to irinotecan is poorly understood. Mutated in Colorectal Cancer (MCC) is emerging as a multifunctional tumor suppressor gene in colorectal and liver cancers, and has been implicated in drug responsiveness. Here, we found that CIMP tumors undergo MCC loss almost exclusively via promoter hypermethylation rather than copy number variation or mutations. A subset of cancers display hypomethylation which is also associated with low MCC expression, particularly in rectal cancer, where CIMP is rare. MCC knockdown or deletion was found to sensitize cells to SN38 (the active metabolite of irinotecan) or the PARP-inhibitor Olaparib. A synergistic effect on cell death was evident when these drugs were used concurrently. The improved SN38/irinotecan efficacy was accompanied by the down-regulation of DNA repair genes. Thus, differential methylation of MCC is potentially a valuable biomarker to identify colorectal cancers suitable for irinotecan therapy, possibly in combination with PARP inhibitors.

Deciphering the etiology and role in oncogenic transformation of the CpG island methylator phenotype: a pan-cancer analysis.

Numerous cancer types have shown to present hypermethylation of CpG islands, also known as a CpG island methylator phenotype (CIMP), often associated with survival variation. Despite extensive research on CIMP, the etiology of this variability remains elusive, possibly due to lack of consistency in defining CIMP. In this work, we utilize a pan-cancer approach to further explore CIMP, focusing on 26 cancer types profiled in the Cancer Genome Atlas (TCGA). We defined CIMP systematically and agnostically, discarding any effects associated with age, gender or tumor purity. We then clustered samples based on their most variable DNA methylation values and analyzed resulting patient groups. Our results confirmed the existence of CIMP in 19 cancers, including gliomas and colorectal cancer. We further showed that CIMP was associated with survival differences in eight cancer types and, in five, represented a prognostic biomarker independent of clinical factors. By analyzing genetic and transcriptomic data, we further uncovered potential drivers of CIMP and classified them in four categories: mutations in genes directly involved in DNA demethylation; mutations in histone methyltransferases; mutations in genes not involved in methylation turnover, such as KRAS and BRAF; and microsatellite instability. Among the 19 CIMP-positive cancers, very few shared potential driver events, and those drivers were only IDH1 and SETD2 mutations. Finally, we found that CIMP was strongly correlated with tumor microenvironment characteristics, such as lymphocyte infiltration. Overall, our results indicate that CIMP does not exhibit a pan-cancer manifestation; rather, general dysregulation of CpG DNA methylation is caused by heterogeneous mechanisms.

ARID1A loss-of-function induces CpG island methylator phenotype.

The CpG island methylator phenotype (CIMP) is associated with prognosis and drug sensitivity in multiple cancer types. In gastric cancer, the CIMP is closely associated with Epstein-Barr virus (EBV) infection and AT-rich interactive domain 1A (ARID1A) mutations, a component of the SWI/SNF chromatin remodeling complex. However, the involvement of SWI/SNF defects in CIMP induction has been unclear. In this study, we demonstrate a causal role of ARID1A loss-of-function in CIMP induction. Mutations of SWI/SNF components, especially ARID1A, was associated with the CIMP, as well as EBV infection, in gastric cancers, and also in uterine endometrial and colorectal cancers, which are not affected by EBV infection. Genome-wide DNA methylation analysis showed that ARID1A knockout (KO) in cultured 293FT cells and gastric epithelial cells, GES1, induced aberrant DNA methylation of a substantial number of CpG sites. DNA methylation was induced at genomic regions with high levels of pre-existing histone H3 lysine 27 trimethylation (H3K27me3) and those with acquired H3K27me3 by ARID1A KO. These results showed that the ARID1A mutation induced aberrant DNA methylation, and this is likely to be one of the potential mechanisms of CIMP induction.

CpG Island Methylator Phenotype-A Hope for the Future or a Road to Nowhere?

The CpG island methylator phenotype (CIMP) can be regarded as the most notable emanation of epigenetic instability in cancer. Since its discovery in the late 1990s, CIMP has been extensively studied, mainly in colorectal cancers (CRC) and gliomas. Consequently, knowledge on molecular and pathological characteristics of CIMP in CRC and other tumour types has rapidly expanded. Concordant and widespread hypermethylation of multiple CpG islands observed in CIMP in multiple cancers raised hopes for future epigenetically based diagnostics and treatments of solid tumours. However, studies on CIMP in solid tumours were hampered by a lack of generalisability and reproducibility of epigenetic markers. Moreover, CIMP was not a satisfactory marker in predicting clinical outcomes. The idea of targeting epigenetic abnormalities such as CIMP for cancer therapy has not been implemented for solid tumours, either. Twenty-one years after its discovery, we aim to cover both the fundamental and new aspects of CIMP and its future application as a diagnostic marker and target in anticancer therapies.

PODNL1 Methylation Serves as a Prognostic Biomarker and Associates with Immune Cell Infiltration and Immune Checkpoint Blockade Response in Lower-Grade Glioma.

Lower-grade glioma (LGG) is a diffuse infiltrative tumor of the central nervous system, which lacks targeted therapy. We investigated the role of Podocan-like 1 (PODNL1) methylation in LGG clinical outcomes using the TCGA-LGG transcriptomics dataset. We identified four PODNL1 CpG sites, cg07425555, cg26969888, cg18547299, and cg24354933, which were associated with unfavorable overall survival (OS) and disease-free survival (DFS) in univariate and multivariate analysis after adjusting for age, gender, tumor-grade, and IDH1-mutation. In multivariate analysis, the OS and DFS hazard ratios ranged from 0.44 to 0.58 (p < 0.001) and 0.62 to 0.72 (p < 0.001), respectively, for the four PODNL1 CpGs. Enrichment analysis of differential gene and protein expression and analysis of 24 infiltrating immune cell types showed significantly increased infiltration in LGGs and its histological subtypes with low-methylation levels of the PODNL1 CpGs. High PODNL1 expression and low-methylation subgroups of the PODNL1 CpG sites were associated with significantly increased PD-L1, PD-1, and CTLA4 expressions. PODNL1 methylation may thus be a potential indicator of immune checkpoint blockade response, and serve as a biomarker for determining prognosis and immune subtypes in LGG.

Interaction between Microsatellite Instability (MSI) and Tumor DNA Methylation in the Pathogenesis of Colorectal Carcinoma.

In colorectal cancer (CRC), the role of microsatellite instability (MSI) is well known. In a genome-wide scale, for the first time, we explored whether differential methylation is associated with MSI. We analyzed 250 paired samples from 125 CRC patients (m = 72, f = 53) at different stages. Of them, 101 had left-sided CRC, 30 had MSI, 34 had somatic mutation in KRAS proto-oncogene (KRAS), and 6 had B-Raf proto-oncogene (BRAF) exon 15p.V600E mutation. MSI was more frequent in right-sided tumors (54% vs. 17%, p = 0.003). Among the microsatellite stable (MSS) CRC, a paired comparison revealed 1641 differentially methylated loci (DML) covering 686 genes at FDR 0.001 with delta beta ≥ 20%. Similar analysis in MSI revealed 6209 DML covering 2316 genes. ANOVA model including interaction (Tumor*MSI) revealed 23,322 loci, where the delta beta was different among MSI and MSS patients. Our study shows an association between MSI and tumor DNA methylation in the pathogenesis of CRC. Given the interaction seen in this study, it may be worth considering the MSI status while looking for methylation markers in CRC. The study also indicates an opportunity for potential use of certain immune checkpoint inhibitors (CTLA4 and HAVCR2 inhibitors) in CRC with MSI.

Global Methylome Scores Correlate with Histological Subtypes of Colorectal Carcinoma and Show Different Associations with Common Clinical and Molecular Features.

BACKGROUND: The typical methylation patterns associated with cancer are hypermethylation at gene promoters and global genome hypomethylation. Aberrant CpG island hypermethylation at promoter regions and global genome hypomethylation have not been associated with histological colorectal carcinomas (CRC) subsets. Using Illumina's 450 k Infinium Human Methylation beadchip, the methylome of 82 CRCs were analyzed, comprising different histological subtypes: 40 serrated adenocarcinomas (SAC), 32 conventional carcinomas (CC) and 10 CRCs showing histological and molecular features of microsatellite instability (hmMSI-H), and, additionally, 35 normal adjacent mucosae. Scores reflecting the overall methylation at 250 bp, 1 kb and 2 kb from the transcription starting site (TSS) were studied. RESULTS: SAC has an intermediate methylation pattern between CC and hmMSI-H for the three genome locations. In addition, the shift from promoter hypermethylation to genomic hypomethylation occurs at a small sequence between 250 bp and 1 Kb from the gene TSS, and an asymmetric distribution of methylation was observed between both sides of the CpG islands (N vs. S shores). CONCLUSION: These findings show that different histological subtypes of CRC have a particular global methylation pattern depending on sequence distance to TSS and highlight the so far underestimated importance of CpGs aberrantly hypomethylated in the clinical phenotype of CRCs.

Tetranucleotide and Low Microsatellite Instability Are Inversely Associated with the CpG Island Methylator Phenotype in Colorectal Cancer.

MSH3 gene or protein deficiency or loss-of-function in colorectal cancer can cause a DNA mismatch repair defect known as "elevated microsatellite alterations at selected tetranucleotide repeats" (EMAST). A high percentage of MSI-H tumors exhibit EMAST, while MSI-L is also linked with EMAST. However, the distribution of CpG island methylator phenotype (CIMP) within the EMAST spectrum is not known. Five tetranucleotide repeat and five MSI markers were used to classify 100 sporadic colorectal tumours for EMAST, MSI-H and MSI-L according to the number of unstable markers detected. Promoter methylation was determined using methylation-specific PCR for MSH3, MCC, CDKN2A (p16) and five CIMP marker genes. EMAST was found in 55% of sporadic colorectal carcinomas. Carcinomas with only one positive marker (EMAST-1/5, 26%) were associated with advanced tumour stage, increased lymph node metastasis, MSI-L and lack of CIMP-H. EMAST-2/5 (16%) carcinomas displayed some methylation but MSI was rare. Carcinomas with ≥3 positive EMAST markers (13%) were more likely to have a proximal colon location and be MSI-H and CIMP-H. Our study suggests that EMAST/MSI-L is a valuable prognostic and predictive marker for colorectal carcinomas that do not display the high methylation phenotype CIMP-H.

Comprehensive immunomolecular profiling of endometrial carcinoma: A tertiary retrospective study.

OBJECTIVE: Combined immunohistochemical and molecular classification using the Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE) independently predicts prognosis in endometrial carcinoma (EC). As next-generation sequencing (NGS) is entering clinical practice, we evaluated whether more comprehensive immunomolecular profiling (CIMP), including NGS and extended immunohistochemical analysis, could further refine the current ProMisE classification. METHODS: A series of 120 consecutive ECs, classified according to ProMisE, was stained immunohistochemically for CD3, CD8, PD-L1, beta-catenin and L1CAM. An in-house 96 gene NGS panel was performed on a subset of 44 ECs, representing the 4 ProMisE subgroups (DNA polymerase epsilon catalytic subunit exonuclease domain mutated (POLEmut), mismatch repair deficient (MMRd), p53 abnormal (p53 abn) and no specific molecular profile (NSMP) ECs). Cases harboring non-hotspot POLE variants were analyzed with Illumina TruSight Oncology 500 NGS panel (TSO500) as a surrogate for whole-exome sequencing. RESULTS: Eight cases harbored POLE variants, half of which were hotspots. Using TSO500, non-hotspot POLE variants were classified as pathogenic (3) or variant of unknown significance (1). POLEmut and MMRd ECs typically showed higher numbers of CD3+/CD8+ tumor-infiltrating lymphocytes and higher PD-L1 expression in tumor-infiltrating immune cells. p53 abn ECs showed significantly higher L1CAM immunoreactivity and frequently harbored gene amplifications including HER2 (25%), but typically lacked ARID1A or PTEN variants. Beta-catenin-positivity and FGFR2 variants were predominantly found in NSMP ECs. CONCLUSIONS: Our data show that CIMP adds significant value to EC characterization and may help to determine pathogenicity of non-hotspot POLE variants, encountered more frequently than expected in our series. In addition, CIMP may reveal ECs benefitting from immune checkpoint inhibition and allows upfront identification of targetable alterations, such as HER2 amplification in p53 abn ECs.

Highly recurrent CBS epimutations in gastric cancer CpG island methylator phenotypes and inflammation.

BACKGROUND: CIMP (CpG island methylator phenotype) is an epigenetic molecular subtype, observed in multiple malignancies and associated with the epigenetic silencing of tumor suppressors. Currently, for most cancers including gastric cancer (GC), mechanisms underlying CIMP remain poorly understood. We sought to discover molecular contributors to CIMP in GC, by performing global DNA methylation, gene expression, and proteomics profiling across 14 gastric cell lines, followed by similar integrative analysis in 50 GC cell lines and 467 primary GCs. RESULTS: We identify the cystathionine beta-synthase enzyme (CBS) as a highly recurrent target of epigenetic silencing in CIMP GC. Likewise, we show that CBS epimutations are significantly associated with CIMP in various other cancers, occurring even in premalignant gastroesophageal conditions and longitudinally linked to clinical persistence. Of note, CRISPR deletion of CBS in normal gastric epithelial cells induces widespread DNA methylation changes that overlap with primary GC CIMP patterns. Reflecting its metabolic role as a gatekeeper interlinking the methionine and homocysteine cycles, CBS loss in vitro also causes reductions in the anti-inflammatory gasotransmitter hydrogen sulfide (H2S), with concomitant increase in NF-κB activity. In a murine genetic model of CBS deficiency, preliminary data indicate upregulated immune-mediated transcriptional signatures in the stomach. CONCLUSIONS: Our results implicate CBS as a bi-faceted modifier of aberrant DNA methylation and inflammation in GC and highlights H2S donors as a potential new therapy for CBS-silenced lesions.

Expression and regulatory roles of lncRNAs in G-CIMP-low vs G-CIMP-high Glioma: an in-silico analysis.

BACKGROUND: Clinically relevant glioma subtypes, such as the glioma-CpG island methylator phenotype (G-CIMP), have been defined by epigenetics. In this study, the role of long non-coding RNAs in association with the poor-prognosis G-CMIP-low phenotype and the good-prognosis G-CMIP-high phenotype was investigated. Functional associations of lncRNAs with mRNAs and miRNAs were examined to hypothesize influencing factors of the aggressive phenotype. METHODS: RNA-seq data on 250 samples from TCGA's Pan-Glioma study, quantified for lncRNA and mRNAs (GENCODE v28), were analyzed for differential expression between G-CIMP-low and G-CIMP-high phenotypes. Functional interpretation of the differential lncRNAs was performed by Ingenuity Pathway Analysis. Spearman rank order correlation estimates between lncRNA, miRNA, and mRNA nominated differential lncRNA with a likely miRNA sponge function. RESULTS: We identified 4371 differentially expressed features (mRNA = 3705; lncRNA = 666; FDR ≤ 5%). From these, the protein-coding gene TP53 was identified as an upstream regulator of differential lncRNAs PANDAR and PVT1 (p = 0.0237) and enrichment was detected in the "development of carcinoma" (p = 0.0176). Two lncRNAs (HCG11, PART1) were positively correlated with 342 mRNAs, and their correlation estimates diminish after adjusting for either of the target miRNAs: hsa-miR-490-3p, hsa-miR-129-5p. This suggests a likely sponge function for HCG11 and PART1. CONCLUSIONS: These findings identify differential lncRNAs with oncogenic features that are associated with G-CIMP phenotypes. Further investigation with controlled experiments is needed to confirm the molecular relationships.

modelBuildR: an R package for model building and feature selection with erroneous classifications.

BACKGROUND: Model building is a crucial part of omics based biomedical research to transfer classifications and obtain insights into underlying mechanisms. Feature selection is often based on minimizing error between model predictions and given classification (maximizing accuracy). Human ratings/classifications, however, might be error prone, with discordance rates between experts of 5-15%. We therefore evaluate if a feature pre-filtering step might improve identification of features associated with true underlying groups. METHODS: Data was simulated for up to 100 samples and up to 10,000 features, 10% of which were associated with the ground truth comprising 2-10 normally distributed populations. Binary and semi-quantitative ratings with varying error probabilities were used as classification. For feature preselection standard cross-validation (V2) was compared to a novel heuristic (V1) applying univariate testing, multiplicity adjustment and cross-validation on switched dependent (classification) and independent (features) variables. Preselected features were used to train logistic regression/linear models (backward selection, AIC). Predictions were compared against the ground truth (ROC, multiclass-ROC). As use case, multiple feature selection/classification methods were benchmarked against the novel heuristic to identify prognostically different G-CIMP negative glioblastoma tumors from the TCGA-GBM 450 k methylation array data cohort, starting from a fuzzy umap based rough and erroneous separation. RESULTS: V1 yielded higher median AUC ranks for two true groups (ground truth), with smaller differences for true graduated differences (3-10 groups). Lower fractions of models were successfully fit with V1. Median AUCs for binary classification and two true groups were 0.91 (range: 0.54-1.00) for V1 (Benjamini-Hochberg) and 0.70 (0.28-1.00) for V2, 13% (n = 616) of V2 models showed AUCs < = 50% for 25 samples and 100 features. For larger numbers of features and samples, median AUCs were 0.75 (range 0.59-1.00) for V1 and 0.54 (range 0.32-0.75) for V2. In the TCGA-GBM data, modelBuildR allowed best prognostic separation of patients with highest median overall survival difference (7.51 months) followed a difference of 6.04 months for a random forest based method. CONCLUSIONS: The proposed heuristic is beneficial for the retrieval of features associated with two true groups classified with errors. We provide the R package modelBuildR to simplify (comparative) evaluation/application of the proposed heuristic (http://github.com/mknoll/modelBuildR).

Effects of Somatic Methylation in Colonic Polyps on Risk of Developing Metachronous Advanced Colorectal Lesions.

The utility of molecular markers for predicting the risk of metachronous advanced colorectal lesions (MACLs) remains poorly investigated. We examined the relationship between somatic hypermethylation in polyps at baseline and the risk of developing MACL. This retrospective cohort study included 281 consecutive patients with colonic polyps who were enrolled between 2007 and 2009 and followed-up until 2014. MACLs were defined as adenomas of ≥10 mm, high-grade dysplasia, or with a villous component; and serrated lesions of ≥10 mm or with dysplasia. In total, 595 polyps were removed at baseline colonoscopy and analyzed for pathological characteristics and CpG island methylator phenotype (CIMP) using the MS-MLPA (Methylation-Specific -- Multiplex Ligation-dependent Probe Amplification) technique. Forty-five patients (16.0%) showed at least one CIMP+ polyp. MACL risk was higher in patients with CIMP+ polyps (odds ratio (OR), 4.50; 95% CI, 1.78-11.4; p = 0.002). Patients with CIMP+ polyps also exhibited shorter time to MACL development (33.8 months vs. 50.1 months; p < 0.001), even with adjustment for polyp size and number (OR, 2.40; 95% CI, 1.33-4.34). Adding CIMP analysis improved the sensitivity (57.0% to 70.9%), negative predictive value (71.1% to 77.3%), and overall accuracy (49.8% to 52.0%) for MACL risk estimation. These results highlight that CIMP may be a useful marker for endoscopic surveillance.