Study Of Biomarkers For Molecular Classification In Hepatocellular Carcinoma And Prostate Cancer

Recently, next generation sequencing （NGS） have accelerated the molecular characterization of cancers and led to the identification of caner-specific signatures and oncogenic targets in cancers. The discovery effort was also aided by bioinformatics integration of disparate molecular data generated from whole-transcriptome, exome, and genome sequencing. These studies have provided genomic and transcriptomic landscapes for cancers. In addition, these investigations have helped researchers understand tumor heterogeneity as well as causative events in tumor initiation, progression, and therapy resistance. They have also aided in the molecular subclassification of cancers. In this study, we showed the successful example of how NGS fully applied in the analysis of genomic landcape, causative events or molecular subsclassification in hepatocellular carcinoma （HCC） and prostate cancer （PCa）.In the first chapter of the first part of our research, we applied whole genome sequencing to three pairs of HBV-positive hepatocellular carcinomas, adjacent non-tumor tissues and normal blood samples. The somatic single-nucleotide variations （SNVs） including single nucleotide polymorphism （SNPs）, and small insertions and deletions （Indels）, structural variations （SVs） and copy number variations （CNVs） in each sample were identified on Complete Genomics platform. We compared all the genomic mutations among the samples and divided them into three pairwise for further analysis:tumor specific mutation, adjacent tissue specific mutation and shared mutation in both tumor and adjacent tissue.To our best knowledge, this is the first time that tumors, adjacent non-tumor tissues and corresponding normal blood samples in HBV related HCC patients were sequenced simultaneously and analyzed comprehensively. The key mutations and pathways identified in tumor and adjacent tissue will furnish the understanding of underlying molecular mechanisms in HBV related HCC.In the second chapter of the first part research, based on the sequencing result, we selected some previously reported and novel SNPs and also SVs which were reported to play important roles in cancers for validation in another cohort of HCC tumors, adjacent non-tumor tissues and normal blood samples. We found tumor suppressor gene KMT2C （Lysine-Specific Methyltransferase 2C）, also known as MLL3, mixed-lineage leukemia 3) C1114R deleterious mutation extremely prevalent in HCCs （97.8% in tumor,96.2% in adjacent and 76.9% in blood）. VCX gene （Variable charge, X-linked）, a L104P frequent mutation was found both in tumor and adjacent （14.6% in tumor and 11.1% in adjacent）. We also found tumor suppressor TP53 （R249S） mutation significantly associated with poor diagnosis and mutated in 7.7% HCC tumors specifically. We also found giant protein AHNAK2 （M1761I） SNP in tumors and adjacent tissues with 100% mutation rate, but only 10% in blood samples. Zinc finger protein ZNF717 （L689H） SNP was detected in tumors, adjacent tissues and blood samples with 34%,23% and 10% mutation rate, respectively. Besides, PABPC3 （Y377H） SNP and HLA-DQB1（S233G） SNP were detected in all HCC tumors, adjacent tissues and blood samples. Regarding of the chromosomal rearrangement, a fusion between TK1 gene in chr.17 and a non-gene related region in chr.8 was detected in both tumors and adjacent, resulting a C-terminal end truncated TK1 protein with 103 amino acids, which will presumably delay the G2/M phase specific degradation and increase cell proliferation of cancer cells.Integrative analysis of specific SNVs on exomes in tumors and adjacent non-tumor tissues and their impact on the biology signaling pathways found the following two pathways were exclusively enriched with significance in tumors:ECM-receptor interaction and Cell adhesion molecules （CAMs）, indicating these microenvironmental pathways may act as HCC drivers. Also olfactory transduction enriched specifically in adjacent appears to have an important role in the initiation of HCC. Meanwhile, olfactory transduction enriched specifically in adjacent appears to have an important role in the initiation of HBV mediated HCC.Our study also shows the great similarity and difference are shared by the initiation and progression process in HBV mediated HCC. The key mutations and pathways identified will provide a new path towards early diagnosis and therapeutic intervention of the deadly disease.In the second part of our research, we applied whole exome sequencing on 74 （50 FFPE tissues （formarlin fixed, paraffin embedded, FFPE） and 24 fresh frozen tissues） primary localized prostate cancer patients （Gleason score<=7） who received radical prostatectomy with 30 patients relapsed and 44 without relapse. Based on the large number of somatic mutations found recurrent and non-recurrent cohorts, we used machine learning algorithm （Random Forest） to identify the best mutation signature to distinguish the two cohorts. In addition, recurrence related genes, somatic mutations and key biological pathways involed were identified.We mainly focused on comparative analysis of the mutational landscape in the recurrent and non-recurrent cohort. The analysis revealed 33 significantly mutated genes in recurrent cohort specifically, top 6 of which is STK31, ALMS1, PCSK5, AHRR and NCOR2. To our best knowledge, most of these genes were reported here for the first time. Recurent non-frameshift insertion occurred in ALMS1 （p.E15delinsEE） and NOCR2 （p.Q78delinsQQ） gene with 46% and 43% frequency, respectively. MAP3K9 harbored CCT three bases nonframeshift deletion on the 113th base loci of the first exon （37%）. KDM6B also gained nonframeshift deletion of ACC three bases on the 796th base loci of the 9th exon （p.252₂53del） with 17% mutation rate. Besides, stopgain on the 402nd base loci of the 5th exon of IDI2（p.Y134X） was identified in 20% of the recurrent patients. All these mutation may have great impact on the function of the corresponding proteins, which may also affect the key biological signaling pathways involved in the progression and metastasis of prostate cancer.In order to find the best mutation signiture to predict recurrence, we used machine learning algorithm （Random Forest） and identified 22 mutation panel to distinguish the indolent and non-indolent prostate cancer patients, which might be the potential biomarker candidates for protstate cancer recurrence.Integrative analysis of all the somatic mutations identified in recurrent and non-recurrent cohort revealed that METS affect on macrophage differentiation is significantly enriched in recurrent cohort exclusively, key genes of which including NCOR2, HDAC2 and METS. All the 3 genes were significantly mutated in the cohort with relapse. Mutations of these genes will presumably suppress growth of tumor associated macrophage （TAMs）, increase their differentiation and accelerate tumor invasiveness and surrounding inflammation consequently.Collectively, we uncovered the mutational landscape of recurrent and non-recurrent PCa cohorts and found key mutations、genes and biological pathways associated with recurrence. Besides, we defined a 22 mutation panel to predict biochemical recurrence which has the potential to improve the clinical management of prostate cancer.