Sparsity and scarcity: Multiomic studies in a low resource setting (A study in archival FFPE cancer tissue)
| dc.contributor.advisor | Gabrielson, Edward | |
| dc.contributor.committeeMember | Sukumar, Saraswati | |
| dc.contributor.committeeMember | Umbricht, Christopher B | |
| dc.contributor.committeeMember | Cope, Leslie M | |
| dc.contributor.committeeMember | Marchionni, Luigi | |
| dc.creator | Cho, Soon Weng | |
| dc.creator.orcid | 0000-0002-1637-2875 | |
| dc.date.accessioned | 2019-04-15T03:59:42Z | |
| dc.date.created | 2017-05 | |
| dc.date.issued | 2016-11-29 | |
| dc.date.submitted | May 2017 | |
| dc.date.updated | 2019-04-15T03:59:42Z | |
| dc.description.abstract | Archival formalin-fixed, paraffin-embedded (FFPE) tissues are an invaluable resource for biomarker discovery due to their vast number in pathology laboratories and availability of high-quality, long-term clinical information. These materials are even more precious for studies on diseases with rare events and long time-to-event intervals. However, these tissues are usually available in limited amounts and formalin treatment of tissue renders recovery of nucleic acids difficult, leading to a setting with low resource. At the time of writing, no robust methodologies have been developed for genomic analysis of FFPE material. To that end, this thesis addresses fundamental questions in genomic profiling in a low resource setting, identified best practice workflows for doing so, and demonstrated their application in three different tumor types. The first section of this dissertation, spanning three chapters, considers the technical challenges and extends the repertoire of methods in performing genomics analysis in low resource settings. We begin by reviewing the nucleic acid modifications and technological advancements for molecular profiling of FFPE tissues. We developed workflows for genomic analysis of FFPE materials and showed application of these findings in a series of microarray experiments. Extending the tools for genomics analysis in a low resource setting, we developed Epicopy, a method to obtain copy number variation (CNV) data from Illumina 450K methylation microarrays and demonstrated its ability to make concordant CNV calls. We also showed comparable, if not better, performance by Epicopy compared to two previously published methods, CHAMP-CNV and CopyNumber450K. In the next section, we next validated the use of these methods in answering biological questions across three tumor types. First, we analyzed a series of ductal carcinoma in situ (DCIS) samples in the context of disease progression into IDC. We observed the presence of global methylation changes in DCIS-adjacent normal tissue and identified the presence of four epitypes of DCIS samples, associated with grade and a CIMP-like phenotype. The CNV profiles of these DCIS samples reflect those of previous studies and differential CNV changes were identified between DCIS that progressed to invasive cancer and DCIS that did not recur. Second, multiomic analysis performed on a series of ER-negative breast cancers identified three functionally relevant clusters of androgen receptor-driven, immune infiltration high, and CNV rich disease. In the clinical context of recurrence in the absence of adjuvant chemotherapy, we discovered a 130-gene panel of markers that was able to predict recurrence in both the institutional ER-negative cohort and validated it in an independent external dataset. Finally, in a total RNA-seq profiling of follicular thyroid cancer (FTC) to study molecular landscapes associated with distant metastasis, we identified a set of biomarkers of distant metastasis, enriched in epithelial-mesenchymal transition genes, and predicted a distant metastasis event in an FTC tumor, 10-years before the fact. These genes were validated in the TCGA thyroid cancer dataset for their ability to predict distant metastasis in follicular variant papillary thyroid cancer (FVPTC), a tumor type which molecularly resembles FTC. Taken together, this work establishes our ability to molecularly profile FFPE tissues using microarray and NGS platforms, unlocking the potential of using archival materials with high quality clinical follow-up information to address important clinical questions. | |
| dc.format.mimetype | application/pdf | |
| dc.identifier.uri | http://jhir.library.jhu.edu/handle/1774.2/60757 | |
| dc.language.iso | en_US | |
| dc.publisher | Johns Hopkins University | |
| dc.publisher.country | USA | |
| dc.subject | genomics | |
| dc.subject | FFPE | |
| dc.subject | breast cancer | |
| dc.subject | thyroid cancer | |
| dc.subject | multiomic analysis | |
| dc.subject | microarray | |
| dc.subject | RNA sequencing | |
| dc.title | Sparsity and scarcity: Multiomic studies in a low resource setting (A study in archival FFPE cancer tissue) | |
| dc.type | Thesis | |
| dc.type.material | text | |
| local.embargo.lift | 2021-05-01 | |
| local.embargo.terms | 2021-05-01 | |
| thesis.degree.department | Cellular and Molecular Medicine | |
| thesis.degree.discipline | Cellular & Molecular Medicine | |
| thesis.degree.grantor | Johns Hopkins University | |
| thesis.degree.grantor | School of Medicine | |
| thesis.degree.level | Doctoral | |
| thesis.degree.name | Ph.D. |