Our laboratory group focuses its efforts on the molecular genetics of gastrointestinal cancers and premalignant lesions, as well as on translational research to improve early detection, prognostic evaluation, and treatment of these conditions. Below, some examples of this work are described.

       carcinoma is among the top ten causes of cancer death worldwide. This disease is usually detected at advanced stages, when available treatments are not very effective. Earlier detection of this cancer has been shown to make a significant impact on patient outcome. Our studies are yielding biomarkers capable of diagnosing synchronous cancer based on gene expression patterns in normal or premalignant tissues. We are extending these studies to confirm that these biomarkers can predict future (metachronous) esophageal cancer risk. Our ultimate goal is the translation of these early detection biomarkers into clinical validation within 5 years. Specifically, we are pursuing the following Aims: Aim 1. In pilot cohort Phase I discovery studies using cDNA microarrays, to develop biomarkers distinguishing between normal or metaplastic tissues of patients with vs. without cancer. Aim 2.In a pilot cohort Phase II validation study, to confirm expression panels and individual genes identified in Aim 1 with quantitative RT-PCR (Q-PCR). Aim 3. In larger cohorts, to perform Phase III cross-sectional retrospective longitudinal validation studies using the Q-PCR method analytically validated in Aim 2. Significant productivity has resulted in our ability to distinguish the normal esophageal mucosae of patients with esophageal cancer from normal mucosae of patients without cancer. These findings offer the potential to screen patients for the presence of cancer or diagnose cancer earlier by assaying normal mucosa. We have also characterized a panel of biomarkers that distinguish between responders and nonresponders to radiation therapy. Plans will consist of extending our esophageal genomics studies to continue to evaluate the predictive value of tissue genomic and epigenomic biomarkers in the early detection and treatment stratification realms. By branching off a tangent from our setasides study and taking it to the next level of validation, we hope to accelerate these biomarkers into the clinical arena.

 

 

 

We have found that freely circulating, abnormally methylated DNA occurs in the plasma of patients with esophageal cancer. Specifically, these data show that methylated alleles of the APC, HPP1 and p16 genes can be detected in the bloodstream of esophageal cancer patients. Furthermore, our data show decreased survival in patients with high plasma levels of methylated APC DNA, which also parallel relapse of malignancy. Thus, circulating methylated nucleic acids are potential biomarkers for the prognostication and monitoring of esophageal cancer patients. The overall scope of this arm of our research is to move these circulating biomarkers from the laboratory into the clinic. In an initial exploratory phase, precise assays of gene-specific methylated plasma DNA are being developed and refined using quantitative real-time methylation-specific PCR, with the established methylation targets p16 and APC serving as templates. In addition, during this exploratory phase, additional novel methylation targets are being identified in neoplastic esophageal tissues and plasma. In a second, developmental phase, novel methylation targets identified in the exploratory phase are clinically validated on a larger, independent cohort by performing clinical correlations with tissue and plasma methylation levels.

Specifically, we are pursuing the following Aims: Aim 1. To develop and validate accurate, robust, standardizable, scalable quantitative real-time plasma DNA methylation assays using the established tissue and plasma methylation targets p16 and APC. Assays are analytically validated with known standards of plasma from normal subjects spiked with known levels of genomic DNA, as well as known but blinded positive and negative patient blood samples. Aim 2. To identify and measure 20 novel methylation events in 50 primary esophageal carcinomas and 50 normal esophageal tissues. A panel of candidate genes with known cancer and outcome relevance and reported frequent methylation in gastrointestinal and other human tumors is evaluated using real-time quantitative methylation-specific PCR (MSP). Genes methylated in at least 20% of tumors, but in less than 5% of normal specimens, are further pursued in Aim 3. Aim 3. To study the same pilot group of 50 patients for plasma methylation of target genes identified in tissues during Aim 2. Genes methylated in the plasma of greater than 10% of these patients constitute plasma targets for clinical validation during the developmental phase. The goals of the developmental study phase of the study are first to develop a prediction model, based on tissue and plasma levels of methylation targets identified during the exploratory phase, on a training cohort of 168 patients. This prediction model is then tested on an independent cohort of 88 patients to establish its clinical validity in prognostic, treatment response, and recurrence risk prediction. Samples and clinical data collected and stored in a clinically well-characterized tissue bank and database are being used to establish the clinical validity of these biomarkers during the developmental phase. Plasma and tissue gene-specific methylation levels are correlated with clinical parameters, including demographics, initial stage, histologic grade, overall and disease-specific survival, interval to tumor recurrence or progression (disease-free and progression-free intervals), type, dosage and timing of treatments received, and additional features (e.g., degree of clinical response to radio/chemotherapy, downstaging after neoadjuvant therapy. Aim 4. To construct a prediction model based on tissue and plasma methylation markers identified during the exploratory phase vs. clinical outcome data corresponding to 168 patients. Single-marker as well as multiple-marker panels are being tested as predictive biomarkers using the Cox proportional hazards model, generalized additive models, artificial neural networks, linear discriminant analysis, and additional methods. Internal validation is accomplished by performing cross-validation. Aim 5. To externally validate the prediction models developed in Aim 4. The tissue and/or plasma markers identified by prediction modeling in Aim 4 are being blindly measured in an independent 88-patient cohort, clinical data are collected and entered into a database, and statistical models are tested using the same correlative analyses used in the original prediction models.   

 

esophagus (BE) is a premalignant condition that predisposes patients to the development of esophageal adenocarcinoma (ADCA). Because of this increase in cancer risk, patients with BE undergo recommended endoscopic surveillance (EGD) at regular intervals indefinitely, every two to three years, sometimes submitting to as many as 10 EGDs per lifetime.

           Because the incidence of ADCA in BE is rare (less than 1 per 100 patients per year), most surveillance EGDs in BE will not uncover cancer. The currently accepted marker for cancer risk is histologic dysplasia, with high-grade dysplasia (HGD) being considered a much more accurate and higher risk factor than low-grade dysplasia (LGD). However, better tissue-based markers capable of predicting progression to HGD or ADCA are needed. For the past several years, we have studied the role of DNA methylation in esophageal ADCA origin and progression. We showed that key tumor suppressor genes (TSGs) that become methylated in BE (metaplasia) actually function as biomarkers in the process, predicting whether patients with BE will or will not progress to develop HGD or ADCA. We have now taken this finding a step further: We have incorporated the three tumor suppressor genes involved in the earlier model (p16, HPP1, RUNX3, and their combined methylation index) plus 4 relevant clinical parameters (age, sex, length of BE, and histologic presence or absence of LGD) to construct 2 models using linear discriminant analysis (LDA). We will research a 3-tier stratification model whereby patients will be stratified into high risk (HR), intermediate risk (IR), and low risk (LR) groups. HR patients will be endoscoped more frequently than usual (once per year), IR patients will be endoscoped at the customary interval (once every 2 years), and LR patients will be endoscoped less often than usual (once per 4 years).  We will accomplish the following Specific Aims: Specific Aim 1: In a blinded pilot study in Year 1, we will analyze methylation profiles and perform risk analysis using the 3-layer model outlined in our Preliminary Data on the samples from our consortia and headquarters sites; Specific Aim 2: If acceptable accuracy results and Milestones are met (see below), in years 2 and 3 we will expand the number of consortia and broaden our study to achieve greater assay automation, enlarge assay generalizability, make the assay adherent to CLIA and FDA standards, perform exploratory research to fine-tune the assay, add or subtract clinical or biomolecular markers to sharpen the current panel, and develop statistical and bioinformatics tools to ramp up power on the bioanalytic side.

Best ROC curves of 2- and 4-year prediction models. A: For the 2-year prediction model, the best AUROC (0.818916) was obtained with 3 parameters: age, segment length, and M.I. B: For the 4-year model, the best AUROC (0.807844) was acquired using 3 parameters: pathology (non-neoplastic BE vs. LGD), and M.I.

 

instability (MSI), caused by defective DNA mismatch repair, occurs frequently in colorectal carcinomas. MSI is categorized as high (MSI-H), low, (MSI-L), or negative (MSS, or microsatellite-stable), according to the frequency of microsatellite alterations at anonymous (noncoding) loci. Tumors with high-frequency MSI (MSI-H tumors) have clinical behavior that distinguishes them from MSS and MSI-L tumors. In addition, evidence is mounting that tumors with low-frequency MSI (MSI-L) tumors have unique features. Nevertheless, our understanding of both MSI-H and MSI-L tumors remains incomplete, and the existence of MSI-L tumors as a distinct subgroup has been questioned. Hypothesis: MSI-H, MSI-L, and MSS gastrointestinal tumors are phenotypically unique. These distinct biologies can be better defined and understood through comprehensive genomic approaches, including instabilotyping and microarray-based bioinformatics. Moreover, valuable insights into molecular pathways underlying these entities can be gained by identifying and studying candidate genes. This hypothesis will be pursued via the following Specific Aims: 1.To broaden and extend instabilotyping of MSI-H colorectal cancers and cell lines, identifying additional genes targeted by frameshift mutation; 2.To examine functional consequences of mutations in coding region targets of microsatellite instability; 2.a. To demonstrate biallelic inactivation of genes showing frequent frameshift mutation by analyzing for loss of heterozygosity, point mutation, and altered expression; 2.b. To determine functional differences between WT and mutant candidate proteins, ascertaining the effect(s) of mutant proteins on cell biology and behavior by transfecting WT candidate genes into mutated cells and ascertaining effect(s): 2.b.i. To assess cell proliferation, anchorage independent growth, invasion, mobility, apoptosis, and differentiation; 2.b.ii.To evaluate effects of WT-transfected ACTR2 and other candidate genes on protein expression and signal transduction, including phosphorylated and total SMAD2, total SMAD4, caspase 1, and TTK; 2.b.iii. Using cDNA microarrays, to compare colon cancer cells before and after transfection with WT ACTR2, TTK, HDCMA18, CASP1, and as-yet unidentified genes containing frequently mutated microsatellites; 3.To increase our understanding of MSI-H, MSI-L, and MSS colorectal carcinomas by comparing the transcriptomes of these cancers, using cDNA microarrays and bioinformatics strategies; 3.a.To generate global gene expression data from MSI-H, MSI-L, and MSS colorectal tumors: 3.a.i.To produce and probe cDNA microarrays with RNAs from MSI-H and MSS cells; 3.a.ii.To hybridize microarrays to MSI-H, MSI-L, and MSS colorectal tumors; 3.b.To determine whether MSI-L tumors comprise a biologically distinct subgroup: 3.b.i.To apply bioinformatics strategies to confirm the existence and provide clues to the biology of a distinct MSI-L tumor subgroup; 3.c. To identify genes defining molecular genetic pathways underlying MSI-H, MSS, and MSI-L tumors: 3.c.i. To use principal components analysis (PCA) to find genes segregating with MSI status, and significance analysis of microarray data (SAM) to identify genes that are differentially expressed among these three tumor groups.

   

 

 

 

cancer is the most prevalent malignancy and the leading cause of death in the digestive system in the United States. To improve clinical care and early detection of this disease, novel molecular biomarkers and a comprehensive understanding of molecular pathology are valuable. Aberrant gene silencing by hypermethylation at CpG islands covering the promoter region is a major hallmark of human cancer of multiple organs, and promoter hypermethylation of some genes is of diagnostic value. In this context, we are conducting global scanning for novel genes with aberrant promoter methylation in colon cancer using a differential expression pattern-oriented approach. We have conducted cDNA microarray analyses of normal and tumorous colon tissues and identified 370 of 14,000 genes to be down-regulated in tumors by Significance Analysis of Microarray Data (SAM). We have prioritized 50 of these 370 genes as candidate targets of tumor-specific methylation using the following criteria: a) presence of CpG islands overlapping the 5’UTR region; b) putative functional linkage to cancer or known genetic alterations in cancer; and c) upregulation of mRNA expression by 5-aza-2’-deoxycytidine (5-aza-dC) treatment, an established in vitro DNA demethylating method. Microarray experiments on colon cancer cell lines with and without 5-aza-dC treatment have been performed to obtain this information. Candidate target genes are being further pre-screened using 14 colon cancer cell lines and normal colon mucosae for the presence of tumor-specific CpG island methylation by methylation specific PCR (MSP). Correlation between the presence of a methylated allele and downregulated mRNA expression is confirmed by real-time quantitative RT-PCR at this stage. Genes showing tumor-specific CpG island methylation correlating with downregulated mRNA expression in cell lines are further analyzed on primary tumors. Finally, we conduct real-time quantitative MSP on 57 primary colon tumors for genes selected by pre-screening cell lines. Correlations between promoter hypermethylation and clinical features for the cases are evaluated as well. We are uncovering previously understudied genes involved in colon carcinogenesis through promoter hypermethylation-mediated gene silencing. These findings constitute novel biomarkers for early cancer detection, tailored treatment, and molecular therapies.

 

Chronic idiopathic inflammatory bowel disease (IBD) predisposes to the development of colorectal carcinoma. The molecular basis of this predisposition has been studied for many years, but much remains to be discovered. For example, we know that unique global gene expression patterns occur early in IBD-associated neoplasias (IBDNs), and that hypermethylation of certain promoter regions is a mechanism of gene inactivation in these lesions. But at which neoplastic stage do these alterations occur during IBD-associated carcinogenesis? Can individual genes be identified from global genomic screens of expression, methylation, or change in copy number? Which global patterns or individual gene alterations predict early neoplastic transformation or progression? The current research project will answer these questions by developing the following unifying hypothesis: The study of IBDNs at all stages of evolution will benefit from global, comprehensive genomic approaches that will illuminate molecular genetic carcinogenetic pathways while simultaneously discovering clinically valuable neoplastic progression biomarkers. This hypothesis will be developed by pursuing the following Aims: 1.To perform a genome-wide characterization of the epigenetic signature of IBD-associated neoplasias (IBDNs), focusing on known as well as novel CpG islands in the promoter or upstream portions of genes. a. Known methylation targets will be analyzed, including E-cadherin(CDH1), p16, p15, p14-ARF, death-associated protein kinase (DAPK), O6-methylguanine DNA methyltransferase (MGMT), human mutL homolog 1 (hMLH1), adenomatous polyposis coli (APC), RASSF1A, deleted in colon carcinoma (DCC), and 14-3-3-F. b. Searches for novel targets of methylation in IBDNs will be performed using CpG island microarrays. 2. To comprehensively scan the genome for alterations in gene copy number at each stage of IBD-neoplasia. a. To probe cDNA microarrays with genomic DNA in order to identify specific genes involved by DNA amplification and  deletion in IBDNs. 3. To perform global gene expression studies of IBDNs using cDNA microarrays. a. To produce cDNA microarrays and probe them with RNAs from IBDNs at all stages of neoplasia. b. To use hierarchical clustering, significance analysis of microarrays (SAM) and artificial neural networks (ANNs) to identify global expression patterns and specific genes at each stage of IBD-associated neoplastic progression. 4. To perform clinical correlations with molecular data. a. Bioinformatics algorithms will be used to define gene expression patterns associated with neoplastic progression in IBDN. b.Clinical parameters will be correlated with gene expression, methylation and copy number data to delineate specific genes potentially relevant to neoplastic progression in IBD.