Application Of Next-generation Sequencing In Detection Of HBV Variation In HBV-related Liver Cancer Patients

Hepatocellular carcinoma (HCC) is the most common primary liver cancer in mainland of China, ranking the fourth highest morbidity and third highest mortality in all cancers. In our country, 65% of HCC was caused by hepatitis b virus (HBV) infection.There are four overlapping open reading frames (ORFs) in HBV genome: preS/S, preC/C,P and X, encoding proteins including: hepatitis B surface antigen (HBsAg), hepatitis B core antigen (HBcAg), hepatitis B x antigen (HBx) and so on. HBcAg and HBsAg are structure proteins, while HBx is regulatory protein. All these proteins play crucial roles in the occurrence and advancement of HCC, via affecting the proliferation, invasion,differentiation, epigenetic features and chromatin stability of host cells. Due to the lack of proofreading function of HBV reverse transcriptase (RT) and high abundance of HBV population in host, the HBV genome is polytropic which complicate the effect of HBV proteins on advancement of HCC. So it’s vital to study the association between HBV mutations and their implication on HCC occurrence and advancement.Though numerous researches have disclosed the relationship between HBV mutations and advancement of HCC, because of multiple genotypes and geographic heterogeneity of HBV, researches on divergent regions, population and infection stages are still indispensable. Furthermore,studies on association between HBV variation and HCC prognosis were still limited and mainly focused on basal core promoter (BCP)region, so the relationship between HCC prognosis and HBV mutations in other regions,such as preC/C or HBx,is still worthy of study and was probed in order to extension the knowledge on HBV variation and its impact on HCC development in this study.In addition, recent researches implied that HBV exist as complex quasispecies rather than haplotype in host and the viral quasispecies become more sophisticated accompanying the infection progression. While canonical Sanger sequencing could only acquire limited quasispecies information which stem the deep understanding of HBV quasispecies. With high throughput and sensitivity, next generation sequencing (NGS)facilitated the possibility of probing the whole virus population simultaneously. Hence,the preS region of HBV genome of HCC and chronic hepatitis B (CHB) patients were deep sequenced with NGS method for comparison of quasispecies in different infection stages and estimation of the feasibility of NGS data as tumor predictive markers.PART 1. Correlation between HBV PreC/C Region Mutations with Postoperative Prognosis of HBV-related Hepatocellular Carcinoma PatientsThis study is intended to explore the relationship between HBc mutation and the postoperative prognosis of HBV-related HCC patients. 98 patients suffered from HBV-related HCC and treated with surgery were enrolled with 48 months follow-up. The preC/C region of HBV genomes from tumor tissue (TT) and paired adjacent non-tumor tissue (ANTT) of these patients were sequenced and the phylogenetic tree was constructed. The correlations between viral features and evolutionary divergence of preC/C sequences from paired TTs and ANTTs were analyzed. Cox proportional hazard model analysis was applied for postoperative hazard risk evaluation.Phylogenetic analysis revealed that all of the preC/C sequences were ascribed to genotype C. The evolutionary divergence of preC/C sequences from matched TTs and ANTTs was significantly negative correlative with serum and intrahepatic HBV DNA levels. Multivariate analysis revealed that HBc E77 mutation was associated with shorter overall survival and HBc S87 and P156 mutation were independent risk factors of relapse.Furthermore, inconsistent with patients without S87 mutation, no correlation was observed between serum HBV DNA and intrahepatic HBV DNA in HCC patients with S87 mutation. Analysis of intrahepatic preC/C sequence may benefit the understanding of viral status and thus is useful for prognosis prediction for HBV-related HCC patients .Part 2. Correlation between HBV X Region Mutations with Postoperative Prognosis of HBV-related Hepatocellular Carcinoma PatientsThis study is intended to compare the HBx between TTs and ANTTs of HBV-related HCC patients, explore the relationship between HBx mutation and the postoperative prognosis of HBV-related HCC patients and investigate the effect of important HBx mutation on hepatocyte function.90 patients suffered from HBV-related HCC and treated with surgery were enrolled with 48 months follow-up. The X region of HBV genome from TT and paired ANTT of these patients were sequenced and the phylogenetic tree was constructed. Cox proportional hazard model analysis was applied for postoperative hazard risk evaluation and clinical features of patients with or without significant HBx site mutation were compared. Plasmid with HBx mutations were constructed and its effect on cell functions were evaluated.Phylogenetic analysis revealed that 69 HBx sequences were ascribed to genotype C and 20 ones were ascribed to genotype B. Mutation frequency in TTs was higher than counterpart in ANTTs. Multivariate analysis revealed that mutations in HBx nt1479(codon 36) and nt 1799 (synonymous mutation) were associated with postoperative relapse and HBx codon 119 mutation was independent risk factors of shorter OS. Poorer prognosis was observed in HBV genotype B related HCC patients and significant divergence was disclosed between HBx genotype B and C. Furthermore, 69 HBV genotype C related HCC patients were analyzed especially. No divergence was found in distribution and frequency of HBx genotype C mutations between TTs and ANTTs.Multivariate analysis revealed that HBx genotype C nt1479 mutation were associated with postoperative relapse and HCC patients with HBx 1479C displayed larger tumor size,higher γ-glutamyltransferase and more severe clinical manifestations. Plasmid with HBx, HBx P36T and K118T mutation were constructed. Transfection of plasmid with HBx and HBx mutants promoted cell proliferation and invasion while no significant difference was observed in cell functions between cells transfected with wild or mutated HBx plasmid. Analysis of HBx sequence and HBV genotyping could benefit the understanding of viral status and thus is useful for prognosis prediction for HBV-related HCC patients.Part 3. Establishment and Evaluation of Predictive Models for HCC Based on Next Generation Sequencing of HBV PreS Region.This study aimed to compare the HBV preS sequence features and deletion patterns from matched serum, TT and ANTT samples, to compare the HBV preS sequence features and deletion patterns between HBV genotype B and C and between CHB and HCC patients and to establish predictive models for HCC based on sequence features,quasispecies complexity and deletion patterns of HBV preS region. PreS region of HBV genome in 45 CHB and 94 HBV related HCC patients were sequenced by NGS and Manhattan distance based on sequence k-mer frequency and cosine distance based on nucleotide deletion percentage of preS region were calculated. Hierarchical clustering and multidimensional scaling were used for comparison k-mer features and deletion patterns among matched serum, TT and ANTT samples. Hierarchical clustering and multidimensional scaling were also used for comparison k-mer features and deletion patterns between different HBV genotypes and between CHB and HCC patients.Furthermore, support vector machine (SVM) models for HCC prediction based on k-mer features or deletion patterns were established. Besides, elastic net logistic regression predictive model for HCC based on quasispecies complexity was fitted. The performance of these models for HCC prediction were evaluated with 5 fold cross validation (5-CV)and independent cohort validation.No significant divergence was found in k-mer features and deletion patterns of preS among matched serum, TT and ANTT samples. There were more similar deletion patterns between matched serum and ANTT samples, compared to the similarity between matched serum and TT samples. Prominent divergence was found in k-mer features and delletion patterns between HBV preS genotype B and C. Compared to CHB patients,more significant difference was found in deletion pattern between preS genotype B and C infected HCC patients. While k-mer features were more divergent between preS genotype B and C infected CHB patients than in HCC patients. There were significant difference in k-mer features and deletion pattern in preS between CHB and HCC patients and HCC patients with different preS deletion patterns displayed diverse clinical features.With 5-CV, the average area under ROC (AUROC) of the predictive SVM models trained on k-mer features and nucleotide deletion ratio of preS region were 0.956 and 0.729 respectively and the average AUROC of these models in independent cohort validation were 0.624 and 0.694. The highest average AUROC of 5-CV and IV for elastic net logistic regression model trained on nucleotide quasispecies complexity of preS region were 0.98 and 0.840. With NGS method, no significant divergence was found in k-mer features and deletion pattern of preS among matched serum, TT and ANTT samples, and there were significant diversity in k-mer features and deletion patterns between preS genotype B and C and between CHB and HCC patients. The machine learning models trained on k-mer features, nucleotide deletion ratio and quasispecies complexity of preS region could used for the prediction of HCC and have the potential to become HCC markers in future.In summary, in present study, mutations in HBc and HBx were significantly associated with HCC prognosis and may applied for HCC patients precise therapy and prognosis prediction after large size cohort validation and functional validation.Quantitative analysis on preS region with NGS method provide a new approach for studies on HBV quasispecies structure and implication of HBV variation on HCC advancement. With advanced machine learning method, the predictive model for HCC trained on NGS data introduce new ideas and methods for the HCC diagnosis and therapy.