I. A comprehensive genomic pan-cancer classification using The Cancer Genome Atlas gene expression data Tumor classifications based on gene expression data have revealed distinct tumor subtypes and uncovered expression patterns that were associated with clinical outcomes. Landmark studies like those demonstrated that gene expression data can provide valuable information about tumor characteristics which allow targeted options for treatment and for patient care and management. TCGA RNA-seq gene expression data provides a great opportunity to discover unique features that can distinguish multiple different tumor types. Those features may serve as biomarkers for tumor diagnosis and/or potential targets for drug development. Sex differences in cancer susceptibility are one of the most consistent, but least understood, findings in cancer epidemiology. Males are more prone to develop cancer and have worse overall survival than females with the same tumors. For instance, female patients with melanoma tend to exhibit longer survival than male patients. Males have a threefold greater risk for developing bladder cancer than females. Hepatocellular carcinoma is the most common liver cancer that occurs mainly in men. Sex differences in immune response and hormones may play a role. Although additional factors such as sex chromosomes and life style may also contribute, the mechanisms that influence sex differences cancer susceptibility remain largely unknown. Thanks to TCGA, large scale analyses of differences between male and female patients become possible and start to emerge. Knowing when features that can distinguish tumor types differ between males and females might enhance those features utility as biomarkers. We undertook a comprehensive pan-cancer classification of 9,096 tumor samples from 31 tumor types from TCGA using RNA-seq gene expression data.
We aim ed to identify a set of genes whose expression levels can classify all 31 TCGA pan-cancer tumor types. Moreover, we sought to identify, separately in men and in women, analogous sets of genes that can distinguish the 23 sex non-specific tumors types. We hope to gain insight into sexual dimorphism in some tumors from those analyses. We carried out pan-cancer classification of 9,096 TCGA (The Cancer Genome Atlas) tumor samples from 31 tumor types using RNA-seq gene expression data. We randomly assigned 75% of samples into a training set and 25% into a validation/test set, proportionally allocating samples from each tumor type. We were able to correctly classify more than 90% of the test set samples. Classification accuracies were high for all except three of the 31 tumor types. In particular, READ (rectum adenocarcinoma) was largely indistinguishable from COAD (colon adenocarcinoma). Among the genes whose expression best discriminated the 31 tumor types, one third were pseudogenes, suggesting that pseudogenes might better serve as features than their functional counterparts for distinguishing tumor types. Another one third encoded transcription factors and the final third encoded proteins involved in cell adhesion, ion and small molecular transport, protein synthesis and folding, and lung function. We also carried out pan-cancer classification on 23 sex non-specific tumor types (those that are not related to reproductive organs) separately for males and females. Results from these sex-specific pan-cancer classification analyses largely recapitulated our analysis of 33 tumor types when sex was ignored. More than 80% of the 100 most discriminative genes were common to both sexes. Genes that were differentially expressed between males and females in at least one of the 23 tumor types included: BNC1, FAT2, FOXA1, and HOXA11. To our knowledge, our analysis is the first sex-specific pan-cancer classification; the differential discriminative genes we identified might be related to sex differences in tumor incidence and prognosis. II. Putative genomic characteristics of BRAF V600K versus V600E cutaneous melanoma BRAF is a serine/threonine protein kinase that activates the MAP kinase/ERK-signaling pathway. Approximately, 40-60% of cutaneous melanomas harbor activating BRAF mutations Cancer Genome Atlas. The majority of the BRAF mutations constitute a substitution of the valine residue at position 600 by a glutamate (V600E) through mutation of a single nucleotide GTG to GAG. Another prevalent BRAF mutation at the same residue is V600K mutation in which the valine residue is replaced by a lysine through two nucleotide substitutions (GTG to AAG). V600K mutation occurs in 10 to 30% of all BRAF V600 melanomas. The other less common BRAF mutations at V600 residue include V600R and V600D. Patients with V600K and those with V600E mutation seem to have some distinctive clinical features. Patients with V600K tumors appear to be older (over 50) males, and the tumors often occur in the head and neck area (prone to sun damage). Pathologically, V600K tumors appear to be thicker and are more mitotically active than V600E tumors. Clinically, patients with V600K tumors have an increased risk for brain and lung metastases, are at a significantly increased risk of relapse and have a shorter time from diagnosis to metastasis than those with V600E tumors. Despite those differences, a large clinical trial showed that V600K tumors were sensitive to vemurafenib, a BRAF inhibitor, and that patients with V600K or V600E tumors, when treated with vemurafenib, had similar overall survival and progression-free survival outcomes. Although the clinical and histopathological differences between V600K and V600E tumors are well documented, little is known about their genomic differences. To gain insight into those differences, we systematically compared protein expression, RNA-seq gene expression, and miRNA-seq microRNA expression between BRAF V600K and BRAF V600E melanoma tumor samples from TCGA. In addition, we analyzed somatic mutation, copy number alteration, and clinical data between those two BRAF tumor subtypes. We found that c-Kit protein expression and KIT gene expression levels were significantly higher in V600K tumors than in V600E tumors. The increased expression was correlated with down-regulation of mir-222. We also found that elements of energy-metabolism and protein-translation pathways were up-regulated and that pro-apoptotic pathways were down-regulated in V600K tumors compared to V600E tumors. We suggest that up-regulation of c-Kit and increased proliferation and survival signals may be the genomic features that account for the observed relatively aggressive nature of V600K tumors. We also found that several microRNAs known to regulate c-Kit expression, especially mir-222, were down-regulated in V600K tumors compared to V600E tumors. The down-regulation of mir-222 may in part be responsible for the up-regulation of c-Kit in V600K tumors. Finally, we showed that biological terms of growth and metabolism were enriched with genes whose expression was negatively correlated with that of mir-222 were enriched. Our results may provide useful information for clinical management and targeted chemotherapeutical interventions for the two BRAF melanoma subtypes.
Showing the most recent 10 out of 36 publications
