B型肝炎病毒全基因突變圖譜、人類白血球抗原遺傳多型性、與肝細胞癌之關係

Abstract

B型肝炎病毒相關肝細胞癌的致病機轉與宿主-病毒間複雜的交互作用高度相關。B型肝炎病毒因為使用缺乏校正功能的反轉錄酶複製，導致不同突變在慢性感染中因免疫選擇壓力而逐漸累積。B型肝炎病毒具有基因多樣性，以8%全基因體核苷酸序列差異為標準，至今已演化出10種基因型。基因型對於肝細胞癌的影響已有相當多的討論。即使在相同基因型，多個核苷酸改變或突變依然會在慢性B型肝炎感染中產生。本研究將探討B型肝炎病毒全基因體核苷酸序列突變圖譜與從健康B型肝炎帶原者到肝細胞癌發生過程的相關。另外，本研究也將偵測第二型人類白血球抗原對病毒變異的影響，與兩者間交互作用對肝細胞癌發生的影響。所有分析資料利用巢式病例對照研究，個案來自4841名男性B型肝炎表面抗原帶原者追蹤22年的世代。本研究共分為三個主題，包括: (I) HBV全基因體單點核苷酸變異與肝細胞癌發展之關係 本研究第一部分為調查B型肝炎病毒變異與肝細胞發展之時序相關。使用巢式病例對照研究中117位肝細胞癌與118位對照個案，其基線血液檢體(距診斷肝細胞癌時間前<1-20年)的病毒全基因體序列資料。 我們發現31個病毒單核苷多型性(SNP)與肝細胞癌發展有相關。接著利用2015年11月30日對GenBank資料庫系統性收集的B型肝炎病毒全基因體序列進行驗證。共有11個與肝細胞癌發展相關的病毒SNPs在二個資料庫中皆有一致的結果。接著將11個病毒變異加總組成突變分數，發現突變分數會隨著肝細胞癌診斷時間間距越近而上升，突變分數1、2、3、≥4相對於沒有任何病毒變異其調整基因型後的相對危險性分別為2.61、4.97、7.38和23.42。在肝細胞癌發病前<4.5年內，突變分數對肝細胞癌罹病風險的鑑別力優於病毒量、病毒基因型和其他各種危險因子的組合(B基因型或/及C基因型的接收者特徵曲線下面積=0.83–0.89；敏感度=72.7%–94.1%；特異度=58.3%–70.5%)。 本研究是第一篇前瞻性地評估B型肝炎全基因體突變圖譜與肝細胞癌的相關，並發展一組突變圖譜可早期預測肝細胞癌發生，以應用於監測慢性B型感染到肝細胞癌發生的進程。 (II) HBV全基因體單點核苷酸變異之共變網絡與肝細胞癌發展之關係 第一部分研究視每個核苷酸變異與病毒上的其它變異各自獨立，然而，許多變異會因為互補變異或生物功能而互有連結。本論文的第二部分研究接續上述巢式病例對照研究，對B型肝炎病毒全基因體單點核苷酸變異使用共變網絡分析探討變異之間的演化在不同基因型、慢性B型肝炎感染自然史階段(免疫耐受期、免疫廓清期、不活化殘餘期、e抗原陰性肝炎期)和肝細胞癌疾病狀態的關係。我們特別針對核苷酸成對間的交互作用(共變)，包含其個數和強度(共變分數)探討分析。 在病毒全基因體569個變異比率>4%的核苷酸中，B基因型共產生248個(43.6%)共變位置，C基因型產生222個(39.0%)。不管在哪個基因型，平均共變分數顯著在HBeAg陰轉後上升(p<0.0001)。不活化殘餘期的共變網絡為軸輻式網絡結構，網絡異質性為0.972-1.076，其擁有最多連結的樞紐核苷酸位置與免疫耐受期和免疫廓清期明顯不同。整體來說，大部分(B基因型:69.8%，C基因型:55.4%)形成共變的核苷酸為基因型特異的核苷酸。然而，在C基因型非基因型特異的核苷酸共變對中，肝細胞癌相較於對照組有較高的共變分數(共變分數S: 2.14 vs. 1.14，p<0.0001)。另外，發現一個只在B基因型對照組出現、由X基因與precore基因上12個核苷酸組成的共變次網絡，網絡中8個變異與低病毒量顯著相關。 共變網絡分析可提供肝細胞癌發展過程中病毒演化研究的另一種角度。而與其它核苷酸變異高度連結的樞紐核苷酸可能可以做為抗病毒藥物的目標。 (III) HLA-DRB1多型性、HBV病毒變異、與肝細胞癌發展之關係 B型肝炎病毒變異的產生來自於逃脫宿主免疫壓力，而病毒與宿主免疫之間的互相作用影響著肝臟疾病的發展。過去許多研究發現人類白血球抗原多型性與B型肝炎進程和病毒清除有關，但是是與哪個特定人類白血球抗原基因座相關仍有不同的看法。本研究的第三部分針對34種第二型人類白血球抗原DRB1基因4位編碼的基因型，探討人類白血球抗原多型性與肝細胞癌的相關。考慮相關背後的可能機轉，再進一步分析DRB1對偶基因對病毒突變的影響，特別針對抗原決定位與上述兩者對肝細胞癌的交互作用討論。 在巢式病例對照研究資料庫156位肝細胞癌與233位對照個案中，發現2種DRB1基因型與肝細胞癌發生有關，DRB1*08032 (相對危險性=2.11，p=0.0148)和DRB1*1401(相對危險性=0.40，p=0.0216)。而DRB1*15011則在病毒量>4 log copies/mL族群中與肝細胞癌發生正相關(相對危險性=2.44，p=0.0295)。再利用同時有病毒序列及HLA-DRB1對偶基因分型資料的105位肝細胞癌與89位對照個案，針對10種DRB1基因分型頻率>5%進行分析，總共發現210個DRB1和病毒變異的顯著相關，包含182個核苷酸位置。高達83.3%篩選出的核苷酸為適應型，與特定人類白血球抗原基因相關，這些核苷酸大多位於已知抗原決定位上和基因無重疊之區域中。無觀察到DRB1基因分型與病毒核肝酸變異對肝細胞癌發生的交互作用。 本研究顯示帶有特定DRB1對偶基因分型會影響肝細胞癌發生。被DRB1對偶基因篩選出的病毒變異其生物機轉需再進一步探討。

The pathogenesis of hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC) involves complicated host-virus interaction. The reverse transcriptase of the virus lacks proofreading function, leading to different mutations accumulating in the viral genome during long-term persistent infection under different immune selection pressures. HBV is genetically diverse and has evolved into ten genotypes defined by more than 8% of divergence in the entire nucleotide sequence. Their impacts on the risk of HCC are well documented. Even within the same genotype; however, multiple nucleotide substitutions or mutations can arise throughout the viral genome during chronic HBV infection. This study was designed to detect HBV whole-genome mutation profile in association with progression from healthy HBV carrier state to HCC. We also aimed to determine the effect of human leukocyte antigen (HLA) class II alleles on the generation of viral mutations, and their interactions on the development of HCC. All analyses were conducted using data from a case-control study nested within a population-based cohort consisting of 4841 male hepatitis B surface antigen (HBsAg) carriers with 22-years of follow-up. This dissertation is divided into three parts. (I) The associations between HBV-single nucleotide polymorphisms (SNPs) among the whole-genome and HCC development In part 1, we investigated the temporality of different HBV mutations in blood and the development of HCC. Using HBV whole genome sequence data obtained from<1 to 20 years before HCC diagnosis from 117 HCC cases and 118 controls in a nested case-control study. We identified 31 HCC-associated HBV-SNPs. Then we conducted a systemic search of GenBank database through November 30, 2015 to identify HBV full-length sequences to seek supportive evidence. Eleven of which could be obtained similar results with the GenBank database in accordance with results obtained from the nested case-control study. Furthermore, we found that a simple-count mutation score using the 11 HBV-SNPs increased with time approaching HCC diagnosis. The multivariate-adjusted odds ratios (ORs) associated with HCC were 2.61, 4.97, 7.38 and 23.42, respectively, for the mutation score of 1, 2, 3, and ≥4, as compared with the group without any of the HBV-SNPs. Within time to diagnosis of less than 4.5 years, the mutation score for HCC risk prediction outperformed HBV DNA, genotype, and other combination of risk factors ( area under the receiver operating characteristic curve=0.83-0.89, sensitivity=72.7%-94.1%, specificity=58.3%-70.5% for predominant genotype B and/or C in Taiwan). Our study is the first to prospectively evaluate the association between HBV mutation profiling and HCC risk, and developed mutation score can be used as a short-term risk score for monitoring of hepatitis B progression to HCC. (II) The associations between genome-wide HBV nucleotide covariance networks and HCC development The first part of study treated variation at each nucleotide position as being independent from all others in the viral genome. However, different mutations may interact due to compensatory variation or functional linkage. In part 2 of the dissertation, we conducted genome-wide covariance network analysis of HBV nucleotide sequence to assess coordinated evolution by HBV genotype, distinct clinical phases (i.e., immune-tolerant, immune-clearance, non-/low- replicative, and hepatitis B e antigen (HBeAg)-negative hepatitis phase) of chronic HBV infection, and HCC status in the aforementioned nested case-control study. We addressed the pairwise interactions (covariances), including their number and strength (covariance score). Among 569 nucleotide positions with frequencies >4%, genotype B had 248 (43.6%) covariant positions, and 222 (39.0%) were found for genotype C. Regardless of HBV genotype, the average covariance score increased after HBeAg seroconversion (p<0.0001). For non-/low- replicative phase, the covariance network had a hub-and-spoke architecture, with heterogeneity ranging about 0.972 to 1.076 and top hubs that differed from those observed in immune-tolerant or immune-clearance phase. Overall, the majority (69.8% for genotype B; 55.4% for genotype C) of the nucleotide positions that formed the covarying pairs were genotype-dependent. However, HCC cases had higher average covariance score for genotype-independent covarying positions than controls in genotype C subjects (S value: 2.14 vs. 1.14, p<0.0001). Besides, a sub-covariance network composed of 12 HBV-SNPs in X or precore gene was found in genotype B controls, there were significant negative associations between 8 of which and HBV DNA. The covariance network analysis may provide insights into viral evolution during HCC development, and the highly connected hubs that have a large number of functional interactions with other nucleotides positions may be valuable antiviral drug targets. (III) HLA-DRB1 polymorphism, HBV mutation, and HCC development Mutations in HBV genome contribute to its escape from host immune response and the interplay of the virus and the host immune response is likely to result in hepatic disease progression. Polymorphisms of HLA have been associated with hepatitis B progression and viral clearance, although the evidence for a link with specific HLA loci is controversial. In part 3 of the study, we first examined the association between HLA polymorphism and HBV-related HCC risk, focusing in particular on HLA-DRB1 alleles, for which 34 alleles were classified by four-digit allele genotyping. We then considered potential mechanisms underlying the association, and evaluated the impact of HLA-DRB1 alleles on viral mutations, especially specific epitopes, and their interactions in HCC. In a nested case-control study of 156 HCC cases and 233 controls, HCC risk was associated with two DRB1 alleles, DRB1*08032 (OR=2.11, p=0.0148) and DRB1*1401 (OR=0.40, p=0.0216) in overall, while DRB1*15011 was associated with HCC in subjects with HBV DNA >4 log copies/mL (OR=2.44, p=0.0295). Next, using HBV sequence and HLA-DRB1 allelic data from 105 HCC cases and 89 controls, 210 associations between the HLA-DRB1 type and HBV-SNPs among 10 DRB1 alleles with frequencies >5% were identified at 182 sites in the HBV genome. 83.3% of HBV-SNPs were adapted form associated with a given HLA allele, and most of them were located in known HBV epitopes and non-overlapping regions of genes. No interactions between HLA-DRB1 alleles and HBV-SNPs on HCC risk were observed. The present study showed that specific HLA-DRB1 alleles could influence the susceptibility of HCC. The biological function of HBV-SNPs selected by DRB1 alleles needs to be further investigated.