European Real-World Assessment of the Clinical Validity of a CE-IVD Panel for Ultra-Fast Next-Generation Sequencing in Solid Tumors.

Molecular profiling of solid tumors facilitates personalized, targeted therapeutic interventions. The ability to perform next-generation sequencing (NGS), especially from small tissue samples, in a short turnaround time (TAT) is essential to providing results that enable rapid clinical decisions. This multicenter study evaluated the performance of a CE in vitro diagnostic (IVD) assay, the Oncomine Dx Express Test, on the Ion Torrent Genexus System for detecting DNA and RNA variants in solid tumors. Eighty-two archived formalin-fixed paraffin embedded (FFPE) tissue samples from lung, colorectal, central nervous system, melanoma, breast, gastric, thyroid, and soft tissue cancers were used to assess the presence of single nucleotide variants (SNVs), insertions and deletions (indels), copy number variations (CNVs), gene fusions, and splice variants. These clinical samples were previously characterized at the various academic centers using orthogonal methods. The Oncomine Dx Express Test showed high performance with 100% concordance with previous characterization for SNVs, indels, CNVs, gene fusions, and splice variants. SNVs and indels with allele frequencies as low as 5% were correctly identified. The test detected all the expected ALK, RET, NTRK1, and ROS1 fusion isoforms and MET exon 14-skipping splice variants. The average TAT from extracted nucleic acids to the final variant report was 18.3 h. The Oncomine Dx Express Test in combination with the Ion Torrent Genexus System is a CE-IVD-compliant, performant, and multicenter reproducible method for NGS detection of actionable biomarkers from a range of tumor samples, providing results in a short TAT that could support timely decision- making for targeted therapeutic interventions.

Only EBUS-Guided Mediastinal Lymph Node Cryobiopsy Enabled Immunotherapy in a Patient with Non-Small Cell Lung Cancer.

Personalized treatment of metastatic non-squamous non-small cell lung cancer (NSCLC) requires detailed molecular characterization of the tumour including detection of predictive driver mutations and programmed death ligand 1 (PD-L1) expression. Complete detection is influenced by the amount of tumour cells sampled as well as their quality. Different sampling techniques may be necessary to provide sufficient tumour material for comprehensive molecular characterization. Missing the detection of targetable molecular genetic aberrations would have a serious impact on the quality of life and prognosis of a patient. This case report highlights the importance of biopsy technique in a patient with NSCLC. Several procedures-pleural puncture, transthoracic lung biopsy and endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA)-could not provide sufficient tumour material for precise tumour characterization. Only the addition of EBUS-guided transbronchial lymph node cryobiopsy (EBUS-TBLNC) enabled complete immunohistochemical and genetic tumour characterization, demonstrating PD-L1 expression in 100% of the tumour cells in the absence of actionable genetic alterations. Based on these results, immunotherapy was initiated.

Genetic and epigenetic analysis of hepatocellular adenomas with atypical morphological features.

BACKGROUND: Hepatocellular adenoma (HCA) is a rare liver tumour, which can have atypical morphological features such as cytological atypia, pseudoglandular architecture, and altered reticulin framework. Little is known about the genetic and epigenetic alterations of such HCAs and whether they show the alterations classically found in hepatocellular carcinoma (HCC) or in HCA without atypical morphology. METHODS: We analysed five HCAs with atypical morphological features and one HCA with transition to HCC. Every tumour was subtyped by immunohistochemistry, sequenced by a targeted NGS panel, and analysed on a DNA methylation microarray. RESULTS: Subtyping of the five HCAs with atypical features revealed two ß-catenin mutated HCA (b-HCA), two ß-catenin mutated inflammatory HCA (b-IHCA), and one sonic hedgehog activated HCA (shHCA). None of them showed mutations typically found in HCC, such as, e.g. TERT or TP53 mutations. The epigenomic pattern of HCAs with atypical morphological features clustered with reference data for HCAs without atypical morphological features but not with HCC. Similarly, phyloepigenetic trees using the DNA methylation data reproducibly showed that HCAs with morphological atypia are much more similar to nonmalignant samples than to malignant samples. Finally, atypical HCAs showed no relevant copy number variations (CNV). CONCLUSION: In our series, mutational, DNA methylation, as well as CNV analyses, supported a relationship of atypical HCAs with nonatypical HCAs rather than with HCC. Therefore, in cases with difficult differential diagnosis between HCC and HCA, it might be advisable to perform targeted sequencing and/or combined methylation/copy number profiling.

Prevalence of Molecular Alterations in a Swiss Cohort of 512 Colorectal Carcinoma Patients by Targeted Next-Generation Sequencing Analysis in Routine Diagnostics.

INTRODUCTION: Colorectal carcinoma (CRC) is among the most common carcinomas in women and men. In the advanced stage, patients are treated based on the RAS status. Recent studies indicate that in the future, in addition to KRAS and NRAS, alterations in other genes, such as PIK3CA or TP53, will be considered for therapy. Therefore, it is important to know the mutational landscape of routinely diagnosed CRC. METHOD: We report the molecular profile of 512 Swiss CRC patients analyzed by targeted next-generation sequencing as part of routine diagnostics at our institute. RESULTS: Pathogenic and likely pathogenic variants were found in 462 (90%) CRC patients. Variants were detected in TP53 (54.3%), KRAS (48.2%), PIK3CA (15.6%), BRAF (13.5%), SMAD4 (10.5%), FBXW7 (7.8%), NRAS (3.5%), PTEN (2.7%), ERBB2 (1.6%), AKT1 (1.5%), and CTNNB1 (0.9%). The remaining pathogenic alterations were found in the genes ATM(n= 1), MAP2K1(n= 1), and IDH2(n= 1). DISCUSSION/CONCLUSIONS: Our analysis revealed the prevalence of potential predictive markers in a large cohort of CRC patients obtained during routine diagnostic analysis. Furthermore, our study is the first of this size to uncover the molecular landscape of CRC in Switzerland.

CB-103: A novel CSL-NICD inhibitor for the treatment of NOTCH-driven T-cell acute lymphoblastic leukemia: A case report of complete clinical response in a patient with relapsed and refractory T-ALL.

Relapsed T cell acute lymphoblastic leukaemia (T-ALL) has a very poor prognosis. A 24-year-old patient with relapsed high-risk T-ALL (PTEN gene deletion; NOTCH1 mutation), was treated with the NOTCH inhibitor CB-103. Within 1 week of starting CB-103, the bone marrow was free of T-ALL blast infiltration (MRD+) and successfully underwent allogeneic hematopoietic stem cell transplantation (allo-HSCT). Sequential samples of ctDNA to monitor the disease after allo-HSCT showed a decrease of circulating Notch1 and PTEN alterations. This is the first T-ALL patient treated with CB-103. The observed clinical response encourages further exploration of CB-103 in ALL.

Fibroblast growth factor receptor (FGFR) inhibitor rogaratinib in patients with advanced pretreated squamous-cell non-small cell lung cancer over-expressing FGFR mRNA: The SAKK 19/18 phase II study.

BACKGROUND: Patients with advanced squamous-cell lung cancer (SQCLC) frequently (46%) exhibit tumor overexpression of fibroblast growth factor receptor (FGFR) messenger ribonucleic acid (mRNA). Rogaratinib is a novel oral pan-FGFR inhibitor with a good safety profile and anti-tumor activity in early clinical trials as a single agent in FGFR pathway-addicted tumors. SAKK 19/18 determined clinical activity of rogaratinib in patients with advanced SQCLC overexpressing FGFR1-3 mRNA. METHODS: Patients with advanced SQCLC failing standard systemic treatment and with FGFR1-3 mRNA tumor overexpression as defined in the protocol received rogaratinib 600 mg BID until disease progression or intolerable toxicity. A 6-months progression-free survival rate (6mPFS) ≤15 % was considered uninteresting (H0), whereas a 6mPFS ≥38 % was considered promising (H1). According to a Simon 2-stage design, 2 out of 10 patients of the first stage were required to be progression-free at 6 months. Comprehensive Genomic Profiling was performedusing the Oncomine Comprehensive Assay Plus (Thermo Fisher Scientific). RESULTS: Between July 2019 and November 2020, 49 patients were screened and 20 were classified FGFR-positive. Among a total of 15 patients, 6mPFS was reached in 1 patient (6.7 %), resulting in trial closure for futility after the first stage. There were 7 (46.7 %) patients with stable disease and 5 (33.3 %) patients with progressive disease. Median PFS was 1.6 (95 % CI 0.9-3.5) months and median overall survival (OS) 3.5 (95 % CI 1.0-5.9) months. Most frequent treatment-related adverse events (TRAEs) included hyperphosphatemia in 8 (53 %), diarrhea in 5 (33 %), stomatitis in 3 (20 %) and nail changes in 3 (20 %) patients. Grade ≥3 TRAEs occurred in 6 (40 %) patients. No associations between mutational profile and treatment outcome were observed. CONCLUSION: Despite preliminary signals of activity, rogaratinib failed to improve PFS in patients with advanced SQCLC overexpressing FGFR mRNA. FGFR inhibitors in SQCLC remain a challenging field, and more in-depth understanding of pathway crosstalks may lead to the development of drug combinations with FGFR inhibitors resulting in improved outcomes.

Preanalytical Variables and Sample Quality Control for Clinical Variant Analysis.

Broad molecular profiling by next-generation sequencing of solid tumors has become a critical tool for clinical decision-making in the era of precision oncology. In addition to many already approved targeted therapies, more than half of ongoing oncology-related clinical trials are biomarker-driven. Therefore, accurate and reliable assays are needed to assess the genetic make-up of tumor cells and guide clinicians in the therapy decision process. In order to obtain high-quality NGS data for variant detection, certain preanalytical steps and quality metrics should be followed. These include assessment of sample types, choice of extraction method, library preparation technology, sequencing platform, and finally sequencing quality control. Each of these steps has certain challenges and pitfalls that need to be addressed and overcome, respectively. In this chapter, we address the preanalytical quality control and how each of the involved steps may influence the final result. Following these guidelines and QC metrics may help in obtaining optimal results that will allow the precise and robust assessment of genetic variants in a clinical setting.

T cell subtype profiling measures exhaustion and predicts anti-PD-1 response.

Anti-PD-1 therapy can provide long, durable benefit to a fraction of patients. The on-label PD-L1 test, however, does not accurately predict response. To build a better biomarker, we created a method called T Cell Subtype Profiling (TCSP) that characterizes the abundance of T cell subtypes (TCSs) in FFPE specimens using five RNA models. These TCS RNA models are created using functional methods, and robustly discriminate between naïve, activated, exhausted, effector memory, and central memory TCSs, without the reliance on non-specific, classical markers. TCSP is analytically valid and corroborates associations between TCSs and clinical outcomes. Multianalyte biomarkers based on TCS estimates predicted response to anti-PD-1 therapy in three different cancers and outperformed the indicated PD-L1 test, as well as Tumor Mutational Burden. Given the utility of TCSP, we investigated the abundance of TCSs in TCGA cancers and created a portal to enable researchers to discover other TCSP-based biomarkers.

Validation of a Targeted Next-Generation Sequencing Panel for Tumor Mutation Burden Analysis: Results from the Onconetwork Immuno-Oncology Consortium.

Tumor mutation burden (TMB) is evaluated as a biomarker of response to immunotherapy. We present the efforts of the Onconetwork Immuno-Oncology Consortium to validate a commercial targeted sequencing test for TMB calculation. A three-phase study was designed to validate the Oncomine Tumor Mutational Load (OTML) assay at nine European laboratories. Phase 1 evaluated reproducibility and accuracy on seven control samples. In phase 2, six formalin-fixed, paraffin-embedded samples tested with FoundationOne were reanalyzed with the OTML panel to evaluate concordance and reproducibility. Phase 3 involved analysis of 90 colorectal cancer samples with known microsatellite instability (MSI) status to evaluate TMB and MSI association. High reproducibility of TMB was demonstrated among the sites in the first and second phases. Strong correlation was also detected between mean and expected TMB in phase 1 (r2 = 0.998) and phase 2 (r2 = 0.96). Detection of actionable mutations was also confirmed. In colorectal cancer samples, the expected pattern of MSI-high/high-TMB and microsatellite stability/low-TMB was present, and gene signatures produced by the panel suggested the presence of a POLE mutation in two samples. The OTML panel demonstrated robustness and reproducibility for TMB evaluation. Results also suggest the possibility of using the panel for mutational signatures and variant detection. Collaborative efforts between academia and companies are crucial to accelerate the translation of new biomarkers into clinical research.

Robust assessment of tumor mutational burden in cytological specimens from lung cancer patients.

OBJECTIVES: Tumor mutational burden (TMB) has emerged as a promising predictive biomarker for immune checkpoint inhibitor therapy. While the feasibility of TMB analysis on formalin-fixed paraffin-embedded (FFPE) samples has been thoroughly evaluated, only limited analyses have been performed on cytological samples, and no dedicated study has investigated concordance of TMB between different sample types. Here, we assessed TMB on matched histological and cytological samples from lung cancer patients and evaluated the accuracy of TMB estimation in these sample types. MATERIALS AND METHODS: We analyzed mutations and resulting TMB in FFPE samples and matched ethanol-fixed cytological smears (n = 12 matched pairs) by using a targeted next-generation sequencing assay (Oncomine™ Tumor Mutational Load). Two different variant allele frequency (VAF) thresholds were used to estimate TMB (VAF = 5% or 10%). RESULTS: At 5% VAF threshold, 73% (107/147) of mutations were concordantly detected in matched histological and cytological samples. Discordant variants were mainly unique to FFPE samples (34/40 discordant variants) and mostly C:G > T:A transitions with low allelic frequency, likely indicating formalin fixation artifacts. Increasing the VAF threshold to 10% clearly increased the number of concordantly detected mutations in matched histological and cytological samples to 96% (100/106 mutations), and drastically reduced the number of FFPE-only mutations (from 34 to 4 mutations). In contrast, cytological samples showed consistent mutation count and TMB values at both VAF thresholds. Using FFPE samples, 2 out of 12 patients were classified as TMB-high at VAF cutoff of 5% but TMB-low at 10%, whereas cytological specimens allowed consistent patient classification independently from VAF cutoff. CONCLUSION: Our results show that cytological smears provide more consistent TMB values due to high DNA quality and lack of formalin-fixation induced artifacts. Therefore, cytological samples should be the preferred sample type for robust TMB estimation.

Tumor mutational burden assessed by targeted NGS predicts clinical benefit from immune checkpoint inhibitors in non-small cell lung cancer.

In non-small cell lung cancer (NSCLC), immune checkpoint inhibitors (ICIs) significantly improve overall survival (OS). Tumor mutational burden (TMB) has emerged as a predictive biomarker for patients treated with ICIs. Here, we evaluated the predictive power of TMB measured by the Oncomine™ Tumor Mutational Load targeted sequencing assay in 76 NSCLC patients treated with ICIs. TMB was assessed retrospectively in 76 NSCLC patients receiving ICI therapy. Clinical data (RECIST 1.1) were collected and patients were classified as having either durable clinical benefit (DCB) or no durable benefit (NDB). Additionally, genetic alterations and PD-L1 expression were assessed and compared with TMB and response rate. TMB was significantly higher in patients with DCB than in patients with NDB (median TMB = 8.5 versus 6.0 mutations/Mb, Mann-Whitney p = 0.0244). 64% of patients with high TMB (cut-off = third tertile, TMB ≥ 9) were responders (DCB) compared to 33% and 29% of patients with intermediate and low TMB, respectively (cut-off = second and first tertile, TMB = 5-9 and TMB ≤ 4, respectively). TMB-high patients showed significantly longer progression-free survival (PFS) and OS (log-rank test p = 0.0014 for PFS and 0.0197 for OS). While identifying different subgroups of patients, combining PD-L1 expression and TMB increased the predictive power (from AUC 0.63 to AUC 0.65). Our results show that the TML panel is an effective tool to stratify patients for ICI treatment. A combination of biomarkers might maximize the predictive precision for patient stratification. Our study supports TMB evaluation through targeted NGS in NSCLC patient samples as a tool to predict response to ICI therapy. We offer recommendations for a reliable and cost-effective assessment of TMB in a routine diagnostic setting. © 2019 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of Pathological Society of Great Britain and Ireland.

[Current methods in molecular pathology]. / Aktuelle molekularpathologische Methoden.

Current methods in molecular pathology Abstract. Macroscopy and microscopy were the cornerstones of tumor analysis in pathology for many years. Recently, we have witnessed an enormous increase in knowledge of genetic and epigenetic alterations occurring in tumors. The detection of these alterations is becoming increasingly important during pathological work-up because they have an important impact on the diagnosis, prognosis, therapy, and prevention of tumors. It is therefore crucial to have appropriate methods available to detect these genetic alterations, such as mutations, translocations or changes in the methylation profile. In the following review article, we will present current methods that are being applied in molecular pathology.

Cell-free DNA analysis in healthy individuals by next-generation sequencing: a proof of concept and technical validation study.

Pre-symptomatic screening of genetic alterations might help identify subpopulations of individuals that could enter into early access prevention programs. Since liquid biopsy is minimally invasive it can be used for longitudinal studies in healthy volunteers to monitor events of progression from normal tissue to pre-cancerous and cancerous condition. Yet, cell-free DNA (cfDNA) analysis in healthy individuals comes with substantial challenges such as the lack of large cohort studies addressing the impact of mutations in healthy individuals or the low abundance of cfDNA in plasma. In this study, we aimed to investigate the technical feasibility of cfDNA analysis in a collection of 114 clinically healthy individuals. We first addressed the impact of pre-analytical factors such as cfDNA yield and quality on sequencing performance and compared healthy to cancer donor samples. We then confirmed the validity of our testing strategy by evaluating the mutational status concordance in matched tissue and plasma specimens collected from cancer patients. Finally, we screened our group of healthy donors for genetic alterations, comparing individuals who did not develop any tumor to patients who developed either a benign neoplasm or cancer during 1-10 years of follow-up time. To conclude, we have established a rapid and reliable liquid biopsy workflow that allowed us to study genomic alterations with a limit of detection as low as 0.08% of variant allelic frequency in healthy individuals. We detected pathogenic cancer mutations in four healthy donors that later developed a benign neoplasm or invasive breast cancer up to 10 years after blood collection. Even though larger prospective studies are needed to address the specificity and sensitivity of liquid biopsy as a clinical tool for early cancer detection, systematic screening of healthy individuals will help understanding early events of tumor formation.

PipeIT: A Singularity Container for Molecular Diagnostic Somatic Variant Calling on the Ion Torrent Next-Generation Sequencing Platform.

The accurate identification of somatic mutations has become a pivotal component of tumor profiling and precision medicine. In molecular diagnostics laboratories, somatic mutation analyses on the Ion Torrent sequencing platform are typically performed on the Ion Reporter platform, which requires extensive manual review of the results and lacks optimized analysis workflows for custom targeted sequencing panels. Alternative solutions that involve custom bioinformatics pipelines involve the sequential execution of software tools with numerous parameters, leading to poor reproducibility and portability. We describe PipeIT, a stand-alone Singularity container of a somatic mutation calling and filtering pipeline for matched tumor-normal Ion Torrent sequencing data. PipeIT is able to identify pathogenic variants in BRAF, KRAS, PIK3CA, CTNNB1, TP53, and other cancer genes that the clinical-grade Oncomine workflow identified. In addition, PipeIT analysis of tumor-normal paired data generated on a custom targeted sequencing panel achieved 100% positive predictive value and 99% sensitivity compared with the 68% to 80% positive predictive value and 92% to 96% sensitivity using the default tumor-normal paired Ion Reporter workflow, substantially reducing the need for manual curation of the results. PipeIT can be rapidly deployed to and ensures reproducible results in any laboratory and can be executed with a single command with minimal input files from the users.

A Variant of a Killer Cell Immunoglobulin-like Receptor Is Associated with Resistance to PD-1 Blockade in Lung Cancer.

PURPOSE: PD-(L)1-blocking antibodies have clinical activity in metastatic non-small cell lung cancer (NSCLC) and mediate durable tumor remissions. However, the majority of patients are resistant to PD-(L)1 blockade. Understanding mechanisms of primary resistance may allow prediction of clinical response and identification of new targetable pathways. EXPERIMENTAL DESIGN: Peripheral blood mononuclear cells were collected from 35 patients with NSCLC receiving nivolumab monotherapy. Cellular changes, cytokine levels, gene expression, and polymorphisms were compared between responders and nonresponders to treatment. Findings were confirmed in additional cohorts of patients with NSCLC receiving immune checkpoint blockade. RESULTS: We identified a genetic variant of a killer cell immunoglobulin-like receptor (KIR) KIR3DS1 that is associated with primary resistance to PD-1 blockade in patients with NSCLC. This association could be confirmed in independent cohorts of patients with NSCLC. In a multivariate analysis of the pooled cohort of 135 patients, the progression-free survival was significantly associated with presence of the KIR3DS1 allele (HR, 1.72; 95% confidence interval, 1.10-2.68; P = 0.017). No relationship was seen in cohorts of patients with NSCLC who did not receive immunotherapy. Cellular assays from patients before and during PD-1 blockade showed that resistance may be due to NK-cell dysfunction. CONCLUSIONS: We identified an association of the KIR3DS1 allelic variant with response to PD-1-targeted immunotherapy in patients with NSCLC. This finding links NK cells with response to PD-1 therapy. Although the findings are interesting, a larger analysis in a randomized trial will be needed to confirm KIRs as predictive markers for response to PD-1-targeted immunotherapy.

A scalable solution for tumor mutational burden from formalin-fixed, paraffin-embedded samples using the Oncomine Tumor Mutation Load Assay.

BACKGROUND: Tumor mutational burden (TMB) is an increasingly important biomarker for immune checkpoint inhibitors. Recent publications have described strong association between high TMB and objective response to mono- and combination immunotherapies in several cancer types. Existing methods to estimate TMB require large amount of input DNA, which may not always be available. METHODS: In this study, we develop a method to estimate TMB using the Oncomine Tumor Mutation Load (TML) Assay with 20 ng of DNA, and we characterize the performance of this method on various formalin-fixed, paraffin-embedded (FFPE) research samples of several cancer types. We measure the analytical performance of TML workflow through comparison with control samples with known truth, and we compare performance with an orthogonal method which uses matched normal sample to remove germline variants. We perform whole exome sequencing (WES) on a batch of FFPE samples and compare the WES TMB values with TMB estimates by the TML assay. RESULTS: In-silico analyses demonstrated the Oncomine TML panel has sufficient genomic coverage to estimate somatic mutations with a strong correlation (r2=0.986) to WES. Further, in silico prediction using WES data from three separate cohorts and comparing with a subset of the WES overlapping with the TML panel, confirmed the ability to stratify responders and non-responders to immune checkpoint inhibitors with high statistical significance. We found the rate of somatic mutations with the TML assay on cell lines and control samples were similar to the known truth. We verified the performance of germline filtering using only a tumor sample in comparison to a matched tumor-normal experimental design to remove germline variants. We compared TMB estimates by the TML assay with that from WES on a batch of FFPE research samples and found high correlation (r2=0.83). We found biologically interesting tumorigenesis signatures on FFPE research samples of colorectal cancer (CRC), lung, and melanoma origin. Further, we assessed TMB on a cohort of FFPE research samples including lung, colon, and melanoma tumors to discover the biologically relevant range of TMB values. CONCLUSIONS: These results show that the TML assay targeting a 1.7-Mb genomic footprint can accurately predict TMB values that are comparable to the WES. The TML assay workflow incorporates a simple workflow using the Ion GeneStudio S5 System. Further, the AmpliSeq chemistry allows the use of low input DNA to estimate mutational burden from FFPE samples. This TMB assay enables scalable, robust research into immuno-oncology biomarkers with scarce samples.

Short sequences can efficiently recruit histone H3 lysine 27 trimethylation in the absence of enhancer activity and DNA methylation.

Trimethylation of histone H3 at lysine 27 (H3K27me3) is a chromatin mark associated with Polycomb-mediated gene repression. Despite its critical role in development, it remains largely unclear how this mark is targeted to defined loci in mammalian cells. Here, we use iterative genome editing to identify small DNA sequences capable of autonomously recruiting Polycomb. We inserted 28 DNA elements at a defined chromosomal position in mouse embryonic stem cells and assessed their ability to promote H3K27me3 deposition. Combined with deletion analysis, we identified DNA elements as short as 220 nucleotides that correctly recapitulate endogenous H3K27me3 patterns. Functional Polycomb recruiter sequences are invariably CpG-rich but require protection against DNA methylation. Alternatively, their activity can be blocked by placement of an active promoter-enhancer pair in cis. Taken together, these data support the model whereby PRC2 recruitment at specific targets in mammals is positively regulated by local CpG density yet obstructed by transcriptional activity or DNA methylation.