A systematic evaluation of high-dimensional, ensemble-based regression for exploring large model spaces in microbiome analyses.

BACKGROUND: Microbiome studies incorporate next-generation sequencing to obtain profiles of microbial communities. Data generated from these experiments are high-dimensional with a rich correlation structure but modest sample sizes. A statistical model that utilizes these microbiome profiles to explain a clinical or biological endpoint needs to tackle high-dimensionality resulting from the very large space of variable configurations. Ensemble models are a class of approaches that can address high-dimensionality by aggregating information across large model spaces. Although such models are popular in fields as diverse as economics and genetics, their performance on microbiome data has been largely unexplored. RESULTS: We developed a simulation framework that accurately captures the constraints of experimental microbiome data. Using this setup, we systematically evaluated a selection of both frequentist and Bayesian regression modeling ensembles. These are represented by variants of stability selection in conjunction with elastic net and spike-and-slab Bayesian model averaging (BMA), respectively. BMA ensembles that explore a larger space of models relative to stability selection variants performed better and had lower variability across simulations. However, stability selection ensembles were able to match the performance of BMA in scenarios of low sparsity where several variables had large regression coefficients. CONCLUSIONS: Given a microbiome dataset of interest, we present a methodology to generate simulated data that closely mimics its characteristics in a manner that enables meaningful evaluation of analytical strategies. Our evaluation demonstrates that the largest ensembles yield the strongest performance on microbiome data with modest sample sizes and high-dimensional measurements. We also demonstrate the ability of these ensembles to identify microbiome signatures that are associated with opportunistic Candida albicans colonization during antibiotic exposure. As the focus of microbiome research evolves from pilot to translational studies, we anticipate that our strategy will aid investigators in making evaluation-based decisions for selecting appropriate analytical methods.

Sabatolimab in combination with spartalizumab in patients with non-small cell lung cancer or melanoma who received prior treatment with anti-PD-1/PD-L1 therapy: a phase 2 multicentre study.

OBJECTIVE: This study evaluates the safety/efficacy of sabatolimab plus spartalizumab in patients with melanoma or non-small cell lung cancer (NSCLC). DESIGN, SETTING AND PARTICIPANTS: This is a phase 1-1b/2, open-label, multinational, multicentre study of patients with advanced/metastatic melanoma or NSCLC with ≥1 measurable lesion. INTERVENTIONS: Patients were given sabatolimab 800 mg every 4 weeks plus spartalizumab 400 mg every 4 weeks until unacceptable toxicity, disease progression and/or treatment discontinuation. OUTCOME MEASURES: The phase 2 primary outcome measure was overall response rate and secondary objectives included evaluation of the safety, tolerability, efficacy and pharmacokinetics of sabatolimab in combination with spartalizumab. RESULTS: 33 patients (melanoma n=16, NSCLC n=17) received sabatolimab plus spartalizumab. 31 (94%) experienced ≥1 adverse event (AE); 15 (46%) experienced grade 3/4 events. The most frequent grade ≥3 AEs for NSCLC were anaemia, dyspnoea and pneumonia (each n=2, 12%); for patients with melanoma, the most frequent grade ≥3 AEs were physical health deterioration, hypokalaemia, hypophosphataemia, pathological fracture and tumour invasion (each n=1; 6%). One (3%) patient discontinued treatment due to AE. Stable disease was seen in three patients with melanoma (19%) and six patients with NSCLC (35%). Median progression-free survival was 1.8 (90% CI 1.7 to 1.9) and 1.7 (90% CI 1.1 to 3.4) months for patients with melanoma and NSCLC, respectively. Patients with stable disease had higher expression levels of CD8, LAG3, programmed death-ligand 1 and anti-T-cell immunoglobulin and mucin-domain containing-3 at baseline. The pharmacokinetics profile of sabatolimab was consistent with the phase 1 study. CONCLUSIONS: Sabatolimab plus spartalizumab was well tolerated in patients with advanced/metastatic melanoma or NSCLC who had progressed following antiprogrammed death-1/antiprogrammed death-ligand 1 treatment. Limited antitumour activity was observed. The tolerability of sabatolimab administration supports the potential to explore treatment with sabatolimab in various combination regimens and across a spectrum of tumour types. TRIAL REGISTRATION NUMBER: NCT02608268.

Microbiome in human health and disease.

Metagenomic studies have truly revolutionised biology and medicine, and changed the way we study genomics. As genome sequencing becomes cheaper it is being applied to study complex metagenomes. 'Metagenome' is the genetic material recovered directly from an environmental sample or niche. By delivering fast, cheap, and large volumes of data Next Generation Sequencing (NGS) platforms have facilitated a deeper understanding of the fundamentals of genomes, gene functions and regulation. Metagenomics, also referred to as environmental or community genomics, has brought about radical changes in our ability to analyse complex microbial communities by direct sampling of their natural habitat paving the way for the creation of innovative new areas for biomedical research. Many metagenomic studies involving the 'human microbiome'have been undertaken to date. Samples from of a number of diverse habitats including different human body sites have been subject to metagenomic examinations. Huge national and international projects with the purpose of elucidating the biogeography of microbial communities living within and on the human body, are well underway. The analysis of human microbiome data has brought about a paradigm shift in our understanding of the role of resident microflora in human health and disease and brings non-traditional areas such as gut ecology to the forefront of personalised medicine. In this chapter we present an overview of the state-of-the-art in current literature and projects pertaining to human microbiome studies.

MU2A--reconciling the genome and transcriptome to determine the effects of base substitutions.

MOTIVATION: Next-generation sequencing technologies enable the identification of sequence variation in the genome and transcriptome. Differences between the reference genome and transcript libraries complicate the determination of the effect of genomic sequence variants on protein products; similarly, these differences complicate the mapping of sequence variants found in transcripts to their respective genomic position. We have developed MU2A, a publicly available web service for variant annotation that reconciles differences between the genome and transcriptome, enabling the rapid and accurate determination of the effects of genomic variants on protein products, and the mapping of variants detected in transcripts to genomic coordinates. The MU2A web service is available at http://krauthammerlab.med.yale.edu/mu2a. We have released MU2A as open source, available at http://code.google.com/p/mu2a/.

Integrated next-generation sequencing of 16S rDNA and metaproteomics differentiate the healthy urine microbiome from asymptomatic bacteriuria in neuropathic bladder associated with spinal cord injury.

BACKGROUND: Clinical dogma is that healthy urine is sterile and the presence of bacteria with an inflammatory response is indicative of urinary tract infection (UTI). Asymptomatic bacteriuria (ABU) represents the state in which bacteria are present but the inflammatory response is negligible. Differentiating ABU from UTI is diagnostically challenging, but critical because overtreatment of ABU can perpetuate antimicrobial resistance while undertreatment of UTI can result in increased morbidity and mortality. In this study, we describe key characteristics of the healthy and ABU urine microbiomes utilizing 16S rRNA gene (16S rDNA) sequencing and metaproteomics, with the future goal of utilizing this information to personalize the treatment of UTI based on key individual characteristics. METHODS: A cross-sectional study of 26 healthy controls and 27 healthy subjects at risk for ABU due to spinal cord injury-related neuropathic bladder (NB) was conducted. Of the 27 subjects with NB, 8 voided normally, 8 utilized intermittent catheterization, and 11 utilized indwelling Foley urethral catheterization for bladder drainage. Urine was obtained by clean catch in voiders, or directly from the catheter in subjects utilizing catheters. Urinalysis, urine culture and 16S rDNA sequencing were performed on all samples, with metaproteomic analysis performed on a subsample. RESULTS: A total of 589454 quality-filtered 16S rDNA sequence reads were processed through a NextGen 16S rDNA analysis pipeline. Urine microbiomes differ by normal bladder function vs. NB, gender, type of bladder catheter utilized, and duration of NB. The top ten bacterial taxa showing the most relative abundance and change among samples were Lactobacillales, Enterobacteriales, Actinomycetales, Bacillales, Clostridiales, Bacteroidales, Burkholderiales, Pseudomonadales, Bifidobacteriales and Coriobacteriales. Metaproteomics confirmed the 16S rDNA results, and functional human protein-pathogen interactions were noted in subjects where host defenses were initiated. CONCLUSIONS: Counter to clinical belief, healthy urine is not sterile. The healthy urine microbiome is characterized by a preponderance of Lactobacillales in women and Corynebacterium in men. The presence and duration of NB and method of urinary catheterization alter the healthy urine microbiome. An integrated approach of 16S rDNA sequencing with metaproteomics improves our understanding of healthy urine and facilitates a more personalized approach to prevention and treatment of infection.

Phase II, Randomized Study of Spartalizumab (PDR001), an Anti-PD-1 Antibody, versus Chemotherapy in Patients with Recurrent/Metastatic Nasopharyngeal Cancer.

PURPOSE: No standard treatment exists for platinum-refractory, recurrent/metastatic nasopharyngeal cancer (NPC). This phase II study (NCT02605967) evaluated progression-free survival (PFS) of spartalizumab, an antiprogrammed cell death protein-1 (PD-1) monoclonal antibody, versus chemotherapy, in NPC. PATIENTS AND METHODS: Patients with nonkeratinizing recurrent/metastatic NPC who progressed on/after platinum-based chemotherapy were enrolled. Spartalizumab was dosed 400 mg once every 4 weeks, and chemotherapy was received per investigator's choice. RESULTS: Patients were randomized to receive either spartalizumab (82 patients) or chemotherapy (40 patients). The most common spartalizumab treatment-related adverse events were fatigue (10.3%) and pruritus (9.3%). Median PFS in the spartalizumab arm was 1.9 months versus 6.6 months in the chemotherapy arm (P = 0.915). The overall response rate in the spartalizumab arm was 17.1% versus 35.0% in the chemotherapy arm. Median duration of response was 10.2 versus 5.7 months in the spartalizumab versus chemotherapy arms, respectively. Median overall survival was 25.2 and 15.5 months in the spartalizumab and chemotherapy arms, respectively. Tumor RNA sequencing showed a correlation between response to spartalizumab and IFNγ, LAG-3, and TIM-3 gene expression. CONCLUSIONS: Spartalizumab demonstrated a safety profile consistent with other anti-PD-1 antibodies. The primary endpoint of median PFS was not met; however, median overall survival and median duration of response were longer with spartalizumab compared with chemotherapy.

A mutation-specific, single-arm, phase 2 study of dovitinib in patients with advanced malignancies.

BACKGROUND: Receptor tyrosine kinases (RTKs) play key roles in tumorigenesis. The multi-RTK inhibitor dovitinib has demonstrated promising antitumor activity in multiple cancers. PATIENTS AND METHODS: In this phase 2, open-label, single-arm study, patients with advanced malignancies with RTK-pathway genetic aberrations whose disease progressed on/following standard treatment received dovitinib (500 mg/day; 5-days-on/2-days-off). The primary endpoint was clinical benefit rate (CBR; complete response, partial response [PR], or stable disease [SD] for ≥ 16 weeks). RESULTS: Of 80 patients enrolled, common tumors included gastrointestinal stromal tumors (GIST; 20.0%), colorectal cancer (CRC; 18.8%), and ovarian cancer (10.0%). Patients were heavily pretreated (median prior lines = 4; 67.5% had ≥ 3 prior lines). Genetic aberrations included cKIT (28.8%), FGFR3 (15.0%), and RET (15.0%). The CBR was 13.8%; one PR (GIST) and 10 SD (adenoid cystic [n = 3]; ovarian [n = 3]; GIST [n = 2]; CRC [n = 1]; gastroesophageal junction [n = 1]). The most common treatment-related adverse events were fatigue, diarrhea, nausea, and vomiting. CONCLUSIONS: In this heterogeneous patient population, the safety profile was acceptable for dovitinib therapy. A subset of patients with RTK pathway-activated tumors experienced clinical benefit. However, the primary endpoint was not met, suggesting further refinement of predictive biomarkers is required.

Genomic Profiling of Metastatic Uveal Melanoma and Clinical Results of a Phase I Study of the Protein Kinase C Inhibitor AEB071.

Up to 50% of patients with uveal melanoma (UM) develop metastatic disease, for which there is no effective systemic treatment. This study aimed to evaluate the safety and efficacy of the orally available protein kinase C inhibitor, AEB071, in patients with metastatic UM, and to perform genomic profiling of metastatic tumor samples, with the aim to propose combination therapies. Patients with metastatic UM (n = 153) were treated with AEB071 in a phase I, single-arm study. Patients received total daily doses of AEB071 ranging from 450 to 1,400 mg. First-cycle dose-limiting toxicities were observed in 13 patients (13%). These were most commonly gastrointestinal system toxicities and were dose related, occurring at doses ≥700 mg/day. Preliminary clinical activity was observed, with 3% of patients achieving a partial response and 50% with stable disease (median duration 15 weeks). High-depth, targeted next-generation DNA sequencing was performed on 89 metastatic tumor biopsy samples. Mutations previously identified in UM were observed, including mutations in GNAQ, GNA11, BAP1, SF3B1, PLCB4, and amplification of chromosome arm 8q. GNAQ/GNA11 mutations were observed at a similar frequency (93%) as previously reported, confirming a therapeutic window for inhibition of the downstream effector PKC in metastatic UM.In conclusion, the protein kinase C inhibitor AEB071 was well tolerated, and modest clinical activity was observed in metastatic UM. The genomic findings were consistent with previous reports in primary UM. Together, our data allow envisaging combination therapies of protein kinase C inhibitors with other compounds in metastatic UM.

PD-1 Blockade in Anaplastic Thyroid Carcinoma.

PURPOSE: Anaplastic thyroid carcinoma is an aggressive malignancy that is almost always fatal and lacks effective systemic treatment options for patients with BRAF-wild type disease. As part of a phase I/II study in patients with advanced/metastatic solid tumors, patients with anaplastic thyroid carcinoma were treated with spartalizumab, a humanized monoclonal antibody against the programmed death-1 (PD-1) receptor. METHODS: We enrolled patients with locally advanced and/or metastatic anaplastic thyroid carcinoma in a phase II cohort of the study. Patients received 400 mg spartalizumab intravenously, once every 4 weeks. The overall response rate was determined according to RECIST v1.1. RESULTS: Forty-two patients were enrolled. Adverse events were consistent with those previously observed with PD-1 blockade. Most common treatment-related adverse events were diarrhea (12%), pruritus (12%), fatigue (7%), and pyrexia (7%). The overall response rate was 19%, including three patients with a complete response and five with a partial response. Most patients had baseline tumor biopsies positive for PD-L1 expression (n = 28/40 evaluable), and response rates were higher in PD-L1-positive (8/28; 29%) versus PD-L1-negative (0/12; 0%) patients. The highest rate of response was observed in the subset of patients with PD-L1 ≥ 50% (6/17; 35%). Responses were seen in both BRAF-nonmutant and BRAF-mutant patients and were durable, with a 1-year survival of 52.1% in the PD-L1-positive population. CONCLUSION: To our knowledge, this is the first clinical trial to show responsiveness of anaplastic thyroid carcinoma to PD-1 blockade.

Efficacy and safety of buparlisib, a PI3K inhibitor, in patients with malignancies harboring a PI3K pathway activation: a phase 2, open-label, single-arm study.

Background: Phosphatidylinositol 3-kinase (PI3K) pathway activation plays a key role in tumorigenesis and has been associated with poor prognosis and resistance to multiple therapies in various cancers. Results: There were 146 patients enrolled; common tumor types were colorectal, sarcoma, and ovarian. Tumors had PI3K pathway alterations and a median of four mutations with tissue-specific patterns of mutation burden (lowest: sarcoma [2.5]; highest: esophagus, germ cell tumor, skin non-melanoma, vaginal [7]). The number of prior therapies did not correlate with the number of genetic alterations (Pearson r = -0.037). The clinical benefit rate was 15.1% (n = 22). An additional patient had an unconfirmed complete response. The most common adverse events were fatigue, nausea, hyperglycemia, decreased appetite, and diarrhea. Patient and Methods: In this phase 2, open-label, single-arm study, patients with solid or hematologic malignancies with PI3K pathway activation and progression on or after standard treatment received buparlisib (100 mg once daily). The primary endpoint was clinical benefit rate per local investigator assessment (response or stable disease at ≥16 weeks). Conclusions: Buparlisib was well tolerated, however efficacy was limited despite selection of PI3K pathway aberrations. Future studies may provide insight into buparlisib efficacy by refining the molecular selection of different tumor types.

Phase 2 Study of the HSP-90 Inhibitor AUY922 in Previously Treated and Molecularly Defined Patients with Advanced Non-Small Cell Lung Cancer.

INTRODUCTION: In this phase 2 study, we evaluated the activity of AUY922 in pretreated patients with stage IV NSCLC. METHODS: Patients with advanced NSCLC were divided into molecularly defined strata based on mutations in the EGFR gene, the ALK receptor tyrosine kinase gene (ALK), the KRAS gene, or the wild type of all three. All patients must have received more than two prior lines of therapy, except for those in a fifth stratum for a less pretreated EGFR cohort (EGFR<2). In the EGFR-mutant and ALK-rearranged strata, prior platinum therapy was not required. Patients with EGFR mutation must have received an EGFR tyrosine kinase inhibitor unless they had de novo resistance (e.g., T790M or exon 20 insertions). Eligible patients received weekly intravenous AUY922, 70 mg/m2. The primary objective was to estimate efficacy (complete or partial response, or in the absence of complete or partial response, stable disease) at 18 weeks, by the Response Criteria in Solid Tumors. RESULTS: A total of 153 patients from 21 global centers were enrolled from October 2010 to November 2014. The investigator-assessed overall response rate and stable disease rate at 18 weeks were 31.8% and 9.1% in the ALK-rearranged stratum, 17.1% and 8.6% in EGFR-mutant stratum, 9.7% and 22.6% in the EGFR<2 stratum, 0% and 7.1% in KRAS-mutant stratum, and 8.8% and 8.8% in wild-type stratum. Biomarker data showed activity of AUY922 in EGFR-mutant patients with exon 19 deletion, T790M mutation, and exon 20 insertion. The most common (≥40%) all-causality adverse events were diarrhea, nausea, and decreased appetite. Visual-related disorders were reported in 79.7% of patients (most were grade 1/2). Thirty-five patients (22.9%) reported night blindness. CONCLUSION: AUY922 is active in patients with NSCLC, particularly among patients with ALK rearrangements and EGFR mutations.

Using Bayesian modelling to investigate factors governing antibiotic-induced Candida albicans colonization of the GI tract.

Receipt of broad-spectrum antibiotics enhances Candida albicans colonization of the GI tract, a risk factor for haematogenously-disseminated candidiasis. To understand how antibiotics influence C. albicans colonization, we treated mice orally with vancomycin or a combination of penicillin, streptomycin, and gentamicin (PSG) and then inoculated them with C. albicans by gavage. Only PSG treatment resulted in sustained, high-level GI colonization with C. albicans. Furthermore, PSG reduced bacterial diversity in the colon much more than vancomycin. Both antibiotic regimens significantly reduced IL-17A, IL-21, IL-22 and IFN-γ mRNA levels in the terminal ileum but had limited effect on the GI fungal microbiome. Through a series of models that employed Bayesian model averaging, we investigated the associations between antibiotic treatment, GI microbiota, and host immune response and their collective impact on C. albicans colonization. Our analysis revealed that bacterial genera were typically associated with either C. albicans colonization or altered cytokine expression but not with both. The only exception was Veillonella, which was associated with both increased C. albicans colonization and reduced IL-21 expression. Overall, antibiotic-induced changes in the bacterial microbiome were much more consistent determinants of C. albicans colonization than either the GI fungal microbiota or the GI immune response.

Next generation sequencing to define prokaryotic and fungal diversity in the bovine rumen.

A combination of Sanger and 454 sequences of small subunit rRNA loci were used to interrogate microbial diversity in the bovine rumen of 12 cows consuming a forage diet. Observed bacterial species richness, based on the V1-V3 region of the 16S rRNA gene, was between 1,903 to 2,432 species-level operational taxonomic units (OTUs) when 5,520 reads were sampled per animal. Eighty percent of species-level OTUs were dominated by members of the order Clostridiales, Bacteroidales, Erysipelotrichales and unclassified TM7. Abundance of Prevotella species varied widely among the 12 animals. Archaeal species richness, also based on 16S rRNA, was between 8 and 13 OTUs, representing 5 genera. The majority of archaeal OTUs (84%) found in this study were previously observed in public databases with only two new OTUs discovered. Observed rumen fungal species richness, based on the 18S rRNA gene, was between 21 and 40 OTUs with 98.4-99.9% of OTUs represented by more than one read, using Good's coverage. Examination of the fungal community identified numerous novel groups. Prevotella and Tannerella were overrepresented in the liquid fraction of the rumen while Butyrivibrio and Blautia were significantly overrepresented in the solid fraction of the rumen. No statistical difference was observed between the liquid and solid fractions in biodiversity of archaea and fungi. The survey of microbial communities and analysis of cross-domain correlations suggested there is a far greater extent of microbial diversity in the bovine rumen than previously appreciated, and that next generation sequencing technologies promise to reveal novel species, interactions and pathways that can be studied further in order to better understand how rumen microbial community structure and function affects ruminant feed efficiency, biofuel production, and environmental impact.

Sequence-specific biosensors report drug-induced changes in epigenetic silencing in living cells.

Treatment with demethylating drugs can induce demethylation and reactivation of abnormally silenced tumor suppressor genes in cancer cells, but it can also induce potentially deleterious loss of methylation of repetitive elements. To enable the observation of unwanted drug effects related to loss of methylation of repetitive DNA, we have developed a novel biosensor capable of reporting changes in DNA accessibility via luminescence, in living cells. The biosensor design comprises two independent modules, each with a polydactyl zinc finger domain fused to a half intein and to a split-luciferase domain that can be joined by conditional protein splicing after binding to adjacent DNA targets. We show that an artificial zinc finger design specifically targeting DNA sequences near the promoter region of the L1PA2 subfamily of Line-1 retroelements is able to generate luminescent signals, reporting loss of epigenetic silencing and increased DNA accessibility of retroelements in human cells treated with the demethylating drugs decitabine or 5-azacytidine.

Phosphoproteomic screen identifies potential therapeutic targets in melanoma.

Therapies directed against receptor tyrosine kinases are effective in many cancer subtypes, including lung and breast cancer. We used a phosphoproteomic platform to identify active receptor tyrosine kinases that might represent therapeutic targets in a panel of 25 melanoma cell strains. We detected activated receptors including TYRO3, AXL, MERTK, EPHB2, MET, IGF1R, EGFR, KIT, HER3, and HER4. Statistical analysis of receptor tyrosine kinase activation as well as ligand and receptor expression indicates that some receptors, such as FGFR3, may be activated via autocrine circuits. Short hairpin RNA knockdown targeting three of the active kinases identified in the screen, AXL, HER3, and IGF1R, inhibited the proliferation of melanoma cells and knockdown of active AXL also reduced melanoma cell migration. The changes in cellular phenotype observed on AXL knockdown seem to be modulated via the STAT3 signaling pathway, whereas the IGF1R-dependent alterations seem to be regulated by the AKT signaling pathway. Ultimately, this study identifies several novel targets for therapeutic intervention in melanoma.

MicroRNA signatures differentiate melanoma subtypes.

Melanoma is an aggressive cancer that is highly resistance to therapies once metastasized. We studied microRNA (miRNA) expression in clinical melanoma subtypes and evaluated different miRNA signatures in the background of gain of function somatic and inherited mutations associated with melanoma. Total RNA from 42 patient derived primary melanoma cell lines and three independent normal primary melanocyte cell cultures was evaluated by miRNA array. MiRNA expression was then analyzed comparing subtypes and additional clinicopathologic criteria including somatic mutations. The prevalence and association of an inherited variant in a miRNA binding site in the 3'UTR of the KRAS oncogene, referred to as the KRAS-variant, was also evaluated. We show that seven miRNAs, miR-142-3p, miR-486, miR-214, miR-218, miR-362, miR-650 and miR-31, were significantly correlated with acral as compared to non-acral melanomas (p < 0.04). In addition, we discovered that the KRAS-variant was enriched in non-acral melanoma (25%), and that miR-137 under expression was significantly associated with melanomas with the KRAS-variant. Our findings indicate that miRNAs are differentially expressed in melanoma subtypes and that their misregulation can be impacted by inherited gene variants, supporting the hypothesis that miRNA misregulation reflects biological differences in melanoma.

Using semantic web technologies to annotate and align microarray designs.

In this paper, we annotate and align two different gene expression microarray designs using the Genomic ELement Ontology (GELO). GELO is a new ontology that leverages an existing community resource, Sequence Ontology (SO), to create views of genomically-aligned data in a semantic web environment. We start the process by mapping array probes to genomic coordinates. The coordinates represent an implicit link between the probes and multiple genomic elements, such as genes, transcripts, miRNA, and repetitive elements, which are represented using concepts in SO. We then use the RDF Query Language (SPARQL) to create explicit links between the probes and the elements. We show how the approach allows us to easily determine the element coverage and genomic overlap of the two array designs. We believe that the method will ultimately be useful for integration of cancer data across multiple omic studies. The ontology and other materials described in this paper are available at http://krauthammerlab.med.yale.edu/wiki/Gelo.

Loss of epigenetic silencing in tumors preferentially affects primate-specific retroelements.

Close to 50% of the human genome harbors repetitive sequences originally derived from mobile DNA elements, and in normal cells, this sequence compartment is tightly regulated by epigenetic silencing mechanisms involving chromatin-mediated repression. In cancer cells, repetitive DNA elements suffer abnormal demethylation, with potential loss of silencing. We used a genome-wide microarray approach to measure DNA methylation changes in cancers of the head and neck and to compare these changes to alterations found in adjacent non-tumor tissues. We observed specific alterations at thousands of small clusters of CpG dinucleotides associated with DNA repeats. Among the 257,599 repetitive elements probed, 5% to 8% showed disease-related DNA methylation alterations. In dysplasia, a large number of local events of loss of methylation appear in apparently stochastic fashion. Loss of DNA methylation is most pronounced for certain members of the SVA, HERV, LINE-1P, AluY, and MaLR families. The methylation levels of retrotransposons are discretely stratified, with younger elements being highly methylated in healthy tissues, while in tumors, these young elements suffer the most dramatic loss of methylation. Wilcoxon test statistics reveals that a subset of primate LINE-1 elements is demethylated preferentially in tumors, as compared to non-tumoral adjacent tissue. Sequence analysis of these strongly demethylated elements reveals genomic loci harboring full length, as opposed to truncated elements, while possible enrichment for functional LINE-1 ORFs is weaker. Our analysis suggests that, in non-tumor adjacent tissues, there is generalized and highly variable disruption of epigenetic control across the repetitive DNA compartment, while in tumor cells, a specific subset of LINE-1 retrotransposons that arose during primate evolution suffers the most dramatic DNA methylation alterations.