Cancers are heterogeneous and mainly classified depending on their origin of tissues/organ, observed clinical/histology and biological features. The intra-tumor heterogeneity, diverse morphology and lack of definitive clinical diagnostic markers for most of the cancer subtypes make the diagnosis and classification of these diseases challenging and hinder their investigation and optimal treatment. At the same time, histological classification suffers from high intra- and inter-observer variability and does not modestly predict clinical outcomes (van den Bent, 2010). Therefore, clinicians and researchers rely on genetic/molecular aberrations to classify and predict clinical outcome. Notably, available cancer genetic studies to date have shown that the malignancy has a remarkably diverse molecular profile. This intricate heterogeneity has been documented in wide spectrum of cancers. Recent genomics studies provided more detailed driver mutational signature and molecular subtype classifications of cancers, which brings us to a possibility to optimally stratify based on novel molecular aberrations. Meanwhile, growing research society (NCCN recommendations) support cancers molecular profiling in combinations with histology and clinical key features to anchor cancers type/subtype and decision for targeted therapy.
Technological advancement (High Throughput Sequencing (HTS) technology) played critical role in finding different features of cancers and its correlation with survival outcome. The last century of cancer biology research focused on reductionist approach that aim to study each regulatory layer in isolation. This pileup of gathered HTS data at an individual (DNA, RNA, epigenomics etc.) level has created potential opportunities to integrate and stratify cancer subtypes at a molecular level, biomarker discovery and new therapeutics. Integrative cancer genomics is built on elementary principle that any cellular and biological process assembled on many molecular phenomena. This assembly is critical to understand the interplay between and within multi-layered cancer genomics datasets to fully understand with phenotypic features, biomarker and drug discovery. In this study, I propose to integrate G2P (Genotype to Phenotype) multi-layered tumor dataset to stratify tumor subtypes at molecular level with clinical features for biomarker discovery, in Neuroendocrine Tumors (NETs).
Neuroendocrine cells are specialized cells that respond to chemical signals by releasing hormones into the blood (like Insulin produce by beta cells of pancreas). Any tumor originates from hormone producing/diffuse neuroendocrine cells are called Neuroendocrine tumors (NETs). These tumors are mainly classified according to the tissue/organ of origin. The majority of NETs are found in the gastrointestinal (GI) tract (stomach, ileum, appendix, colon, rectum, and pancreas) and in the lungs, although they also occur in other (rare) sites – thymus, adrenal glands, thyroid, skin, ovaries, and salivary glands [Vinik A.I et al., 2010]. NETs are rare and slow growing type of tumors with vague symptoms and easily misunderstood with other condition. Each year, ~8000 people in US are diagnosed with a NET and 5-year survival rate for these tumors varies and depends on several factors, primarily the location of tumor. However, due to increased use of new diagnostics modalities (imaging and endoscopy) and an increased awareness, the incidence and prevalence have increased over the last 30 years [Gustafsson BI et al., 2008]. These tumors are slow growing but often metastasize locally and/or different tissue. Many types of NETs are classically called “Carcinoids (cancer-like)” or “Carcinoid tumors” but these terms do not accurately explain their biology, histopathological differences and secretory capabilities.
Pancreatic neuroendocrine tumors (PanNETs) are relatively rare but second most common group of clinically relevant NETs (~7% of all NETs) (Sadaria MR et al., 2013). Genomics analysis of PanNETs identified driver gene mutations in chromatin remodeling pathways: 44% of the tumors with somatic mutation in MEN1, 43% had somatic mutation in either one of the two subunits of transcription/chromatin remodeling complex (ATRX and DAXX subunit) [Jiao Y, et al science 2011]. Jiao Y et al., also found better prognosis for tumor with mutations in MEN1 and ATRX/DAXX. Using this data, PanNETs can be divided into two main genetically different classes: Mutant PanNETs (Mutation in MEN1 and ATRX/DAXX) and Wild-type PanNETs (no mutation in MEN1 and ATRX/DAXX). Interestingly, not much is known in terms of their mutational specific gene signature, pathways, epigenomic profiles and clinical biomarkers. Questions like, which oncogenic pathways are activated in Mutants vs Wild-type PanNETs? Also within mutant PanNETs, dissecting pathways for MEN1 vs ATRX/DAXX tumors resulting in same phenotype.
Small Intestinal neuroendocrine tumor (SI-NET) arising from enterochromaffin cells is the most common type of gastrointestinal endocrine tumors. Hemizygous loss of chromosome 18 is one of the most frequent (~80%) genetic characteristics of SI-NETs [Francis JM et al., 2013]. A frame shift deletion in CDKN1B gene is recurrent (~8%) somatic deletion and proposed to acts as haploinsufficient tumor suppressor gene in SI-NETs [Francis JM et al., 2013]. Using observed genetic events in SI-NETs, the possibility of two or more genetically distinct subtypes in SI-NETs are suggested with different clinical outcome. Integrative copy number, gene expression and methylation profiling of SI-NETs at the molecular level is needed to identify genetic changes for tumor initiation, progression, stratification, targetable genes and survival analysis.
Lung NETs are groups of heterogeneous group of neoplasms. According to WHO 2015(Travis et al., 2015), Lung NETs represents ~ 20 to 25% of primary lung neoplasm and classified as four histological subtypes with considerably different biological characteristics: 1) Small Cell carcinoma (SCLC) (~20% prevalence), 2) Large Cell NE carcinoma (LCNEC) (~3%), 3) Atypical carcinoids (AC) (~0.2%), 4) Typical Carcinoids (TC) (~2%). The key features of this classification rely on morphology, mitotic index per 2 mm2 and necrosis assessment [Travis W 2004]. The reproducibility of this classification and its prognostic efficacy was disputed with high inter-observer variability [Travis et al., 1998]), especially for typical and Atypical carcinoids (Swarts et al). Crushed artifacts in histological sections are major pitfall for diagnosis and management of lung carcinoids. It has been also reported that, TC and AC tumors are overdiagnosed as SCLC on biopsy specimen [Pelosi G et al., 2005]. Strikingly over the last 30 years, the incidence rate of carcinoids tumors has increased [Gustafsson BI et al., 2008] in United States. But our understanding of pathogenesis at molecular level is still in “grey zone”. Only few studies have been done using recent high-thorough put sequencing technologies to understand the underlying molecular alterations for pathogenesis apart from few low-frequency alterations, like MEN1 mutations [Ref Swarts D et al., 2012]. Recent genomics study by Frenandez-Cuesta et al., 2014 on 74 pulmonary carcinoids (includes TCs and ACs) revealed 0.4 rate of somatic mutation per megabase. This rate is much lower than observed somatic mutation rate for other lung cancers. They found genes with somatic mutations are enriched for chromatin-remodeling genes [Frenandez-Cuesta et al., 2014 nature comm] with recurrent alteration in MEN1, PSIP1 and ARID1A. Contrariwise, no significant mutations and focal copy alterations were observed for frequent mutated genes in lung cancers (KRAS, TP53, EGFR etc). This unique mutational profile suggests different cellular and biological mechanisms for LUCA from those of high-grade NETs (LCNEC and SCLC) and lung cancers. But at the same time, all these studies are based on exploring single layer cancer data and comprehensive integrative genomics analysis is lacking for such kind of tumors for better diagnosis, molecular stratifications and targetable predictions.
Here, I propose to study Genotype to Phenotype integration of neuroendocrine tumors. I plan to elucidate molecular alterations, transcriptomics signature and distinct methylation profiles at a single layer and integrate them to find molecular subtypes for optimal stratifications and prognosis, novel biomarkers for IHC and targetable genes.