B型肝炎帶原者進展至肝細胞癌之血液全基因體DNA甲基化圖譜分析與生物標記探勘

Blood Genome-Wide DNA Methylation Profiling in Progression from HBV Carrier to Hepatocellular Carcinoma with Impact on Biomarker Discovery

高瑋怡(Wei-Yi Kao)

指導教授 ： 于明暉

國立臺灣大學/公共衛生學院/流行病學與預防醫學研究所/博士(2016年)

https://doi.org/10.6342/NTU201602210

摘要

B型肝炎病毒相關肝細胞癌的平均發病年齡約50歲左右，進而衍生個體過早死亡等重要公共衛生問題，然而如何有效從年經HBV帶原者中早期偵測出具肝細胞癌高危險性者仍充滿著挑戰。B型肝炎病毒相關肝細胞癌的致病機轉包含複雜的病毒-宿主交互作用參與其中，長期進展過程裡逐步累積各種遺傳與表觀遺傳變異。DNA甲基化是一影響基因靜默之主要表觀遺傳機轉，異常甲基化常見於B型肝炎病毒相關肝細胞癌之多階段發展過程之中，可能擔任致癌機轉的重要影響關鍵。本研究透過周邊血液基礎之全表觀基因體相關分析尋找可作為追蹤HBV帶原者未來發展至肝細胞癌之甲基化訊號，連結異常甲基化與病毒/臨床因子之相關性以闡明甲基化變異對病毒-宿主交互作用之塑形。此將有助於更完整系統性地了解甲基化變化與B型肝炎病毒相關肝細胞癌之關係，提供未來發展疾病早期預測甲基化生物指標之指引方向。
本研究以巢式病例對照研究設計針對96個罹病前收集之周邊血液檢體進行進行B型肝炎病毒相關早發型肝細胞癌之全表觀基因體掃描。透過DNA微陣列晶片Illumina Infinium® HumanMethylation450K BeadChip獲取甲基化數據，經質量控制檢驗與數據標準化及校正後，系統性地檢視B型肝炎病毒相關肝細胞癌之全表觀基因體甲基化圖譜，並探討異常甲基化與各種肝細胞癌危險/臨床因子之相關性，釐清DNA甲基化變化與肝細胞癌進展及其相關危險因子之間潛在影響。此外我們觀察HBV慢性感染自然史各階段之甲基化整體變動狀態，探討HBV病毒量高低與全表觀基因體甲基化的變化情形。此將提供新觀點以推論病毒驅動肝細胞癌致癌過程中甲基化重新編程的可能機轉，而這些甲基化訊號未來有機會發展為肝細胞癌危險性評估管理之全新中介指標，特別是針對HBV相關早發型肝細胞癌之疾病進展監控策略。本研究主要分為三個研究主題，包括：

(I) 肝細胞癌之全表觀基因體甲基化圖譜分析
背景與目的：現階段對於如何有效早期偵測出早發型肝細胞癌高危險性者仍充滿著挑戰。已知表觀遺傳變異是B型肝炎病毒相關肝細胞癌發展過程中常見的異常現象，然而對其相關致病機轉目前了解有限。本研究針對B型肝炎病毒相關早發型肝細胞癌進行周邊血液全表觀基因體相關分析，找尋可作為追蹤HBV帶原者未來發展至肝細胞癌之甲基化訊號。
方法：全表觀基因體相關分析使用450K晶片平台，針對48對肝細胞癌病例及配對對照樣本之罹病前周邊血液檢體進行分析。單點CpG位點相關性檢測與區域分析分別以Wilcoxon signed-rank法和bump-hunting法執行。我們建構一甲基化分數評估這些肝細胞癌相關甲基化圖譜之累加疾病作用效應，建構方式係將各相關探針之β值以其線性迴歸模型之迴歸係數加權後加總所得。這些肝細胞癌相關甲基化圖譜之疾病分類能力以數種監督式與非監督式類別演算法進行評估。
結果：肝細胞癌全表觀基因體相關分析指出有38911個CpG位點於配對樣本間呈現甲基化差異，散布於各染色體。其中41.4%的位點在肝細胞癌病例組中呈現低甲基化且主要分布於CpG富含區域。根據甲基化分數估計，肝細胞癌相關甲基化圖譜在不同顯著閾值設定下可解釋多達30.3-54.8%之肝細胞癌變異程度，且反映出疾病特異性。這些甲基化圖譜對於區分肝細胞癌病例與對照組具有相當好的區別能力，六種監督式分類建構法之分類正確率皆達85%以上，非監督式群集演算結果亦觀察到不同病例與對照組別間有明顯之集中歸類趨勢。
結論：本研究系統性地描繪出B型肝炎病毒相關肝細胞癌之血液甲基化變化圖譜，印證在肝細胞癌致病過程中會伴隨多重表基因改變影響疾病發展。這些以周邊血液為基礎所得之甲基化訊息未來可提供肝細胞癌偵測及危險性評估指標發展之方向。

(II) 肝細胞癌相關甲基化變化之共甲基化表現網絡與功能訊息路徑分析
背景與目的：已知肝細胞癌相關病毒/環境因子或臨床特徵對白血球DNA甲基化可能產生各種程度與面向之影響，然而其確切影響目標與相關機轉尚待釐清。本研究分析肝細胞癌相關甲基化變化之共甲基化表現網絡，連結共甲基化與各種病毒/臨床特徵因子之相關性，藉此推論DNA甲基化變化與肝細胞癌及其相關危險因子三者之間潛在影響關係，並分析這些共甲基化表現是否共同參與特定之生物功能訊息路徑以推測背後可能隱含的功能和組織型態意義。
方法：我們使用加權基因共表現網絡法分析10360個肝細胞癌相關位點(p<0.01)之共甲基化表現網絡，共甲基化模組與各種肝細胞癌危險因子之相關模式係計算模組驅動基因與各因子之間的皮爾森相關係數做為評估指標。功能訊息路徑富集分析用以探討模組基因之富集特性。
結果：依據位點之共甲基化表現網絡，這些肝細胞癌相關位點可歸類出7個共甲基化特徵模組。連結模組與各種肝細胞癌危險因子之間的相關性，我們發現這些模組各自與病毒量、HBV基因型、ALT、肝細胞癌家族病史、以及慢性肝病史等重要因子之間具有特異相關模式。觀察模組內之基因富集功能路徑特性亦反映出與相關病毒/臨床因子之間合理的生物醫學連結，例如病毒量相關之模組基因富集於免疫相關生物路徑；而ALT相關模組被觀察到富集於發炎反應相關路徑。
結論：本研究顯示甲基化變化可能受到不同的環境、免疫、或病毒相關因子調控，進而誘發肝細胞癌生成，此結果提供病毒宿主交互作用對人類肝細胞癌影響之新觀點，也提供表基因改變和各種臨床徵狀因子之間更多連結資訊。

(III) HBV慢性感染自然史、病毒複製活性、與甲基化變化之關係
背景與目的：已知發生在HBV DNA之表觀遺傳修飾作用可以調控病毒的基因表現以及複製活性；然而針對宿主因素方面，目前對於宿主表觀遺傳修飾機制和HBV感染進程與病毒複製活性調控關係之了解仍相當有限。本研究觀察血液全表觀基因體於HBV慢性感染自然史過程之甲基化分布狀況，並探討HBV病毒量高低與甲基化變化關係，找出HBV相關甲基化變異的可能參與之生物路徑。本研究亦分析肝細胞癌相關甲基化變化與各種肝細胞癌致病因子暴露之關係，協助釐清甲基化變化參與在宿主、病毒、與免疫交互作用可能的角色，以及HBV相關肝細胞癌之致病機轉。
方法：我們分析晶片所有CpG位點在自然史各階段之甲基化分布差異，探討這些自然史相關甲基化變異位點所在基因之生物功能富集特性並與已知免疫相關基因模組進行比對。針對HBV病毒量多寡與全表觀基因體甲基化變化之相關性，分別以單點CpG位點分析和bump-hunting區域分析檢視。我們比對10360個肝細胞癌相關位點(p<0.01)與病毒/臨床相關位點間的重疊狀況，並透過甲基化分數評估這些肝細胞癌相關位點與病毒/臨床因子間之關係。
結果：全染色體共發現17394個CpG位點在不同HBV自然史期別間存在甲基化分布差異。這些期別相關位點被觀察到富集於免疫相關功能路徑，特別是B細胞功能相關基因。病毒量相關之單點CpG位點分析結果顯示有14458個位點之甲基化變化與病毒量相關，其中14.8%亦與肝細胞癌有關，這些甲基化位點被發現富集於免疫和脂質代謝相關等生物路徑。區域分析結果找出12個病毒量差異甲基化區域，且同樣展現出免疫調節相關特性。我們觀察到在肝細胞癌相關甲基化位點中同與病毒或臨床特徵因子相關之比例，相較於其他位點顯著高出2-4倍；包括病毒量與白血球組成比例等因子可解釋這些相關位點之大部分變異來源。
結論：表觀遺傳相關之免疫調控對HBV慢性感染佔重要影響地位，在慢性感染過程中病毒複製活性高低亦可連結至眾多免疫調節基因區域之甲基化變化。此結果提供了新的思考方向以評估在病毒驅動肝細胞癌致癌過程中，甲基化重新編程之可能機轉。

關鍵字

DNA甲基化 ； 全表觀基因體相關 ； B型肝炎 ； 肝細胞癌 ； 周邊血液白血球

並列摘要

The average age at onset of hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC) is around 50 years, which causes important public health problems such as premature mortality. However, it still remains challenging to identify and detect young HBV carriers who are at high risk of HCC at an early stage. The pathogenesis of HBV-related HCC involves complicated virus-host interactions with the accumulation of genetic and epigenetic abnormalities. Aberrant DNA methylation is an epigenetic mechanism of gene silencing that is frequently observed during the progression from a precancerous condition to HCC, suggesting a mechanism for epigenetic tumorigenesis towards HBV-mediated HCC. By using epigenome-wide mapping technique, we aimed to identify methylation signatures of peripheral leukocytes that allow us to track HCC development in HBV carriers, and by determining the correlation between methylation signatures and viral/clinical factors, the role of epigenetic variation in shaping virus-host interactions can be elucidated. These studies may be helpful for systematically understanding aberrant methylation in HBV-related hepatocarcinogenesis, which may open new paths for epigenetic biomarkers development for HCC early detection.
In this study, we performed prospective nested case-control study design to examine genome-wide methylation signals for HBV-related early-onset HCC using leukocyte DNA from 96 prospectively collected prediagnostic blood samples. Signatures of global methylation profiles were obtained from Illumina Infinium® HumanMethylation450K BeadChip by passing quality control checks, data normalization and correction procedures. We sought to systematically establish a global DNA methylation profile in HBV-related HCC and to identify DNA methylation changes associated with major risk factors and clinical correlates, thereby, clarifying the link between methylation signatures, HCC progression, and etiological risk factors. We also characterized the methylation patterns across different phases of HBV infection and identified methylation signatures associated with HBV viral load. These are important processes for providing new insights into the mechanisms underlying epigenetic reprogramming during HBV-induced hepatocarcinogenesis. Furthermore, the methylation signatures may act as novel intermediate biomarkers for the management of HCC risk with great potential in clinical use, especially for early-onset HCC. This research is divided in three parts, including:
(I) Epigenome-wide DNA Methylation Profiles and Early-onset Hepatocellular Carcinoma in Hepatitis B Carriers
Background Aims It remains challenging to identify persons at high risk of HCC for early detection. The epigenetic disruption is frequently observed in HBV-related HCC progression, but there is only limited knowledge regarding its participation in HBV-related HCC pathogenesis. We aimed to identify blood-based genome-wide aberrant DNA methylation profiles in HBV-related early-onset HCCs that allows us to track HCC progress in HBV carriers.
Methods We conducted an epigenome-wide association analysis of prospectively collected leukocyte DNA using 450K array on 48 case-control pairs for HBV-related early-onset HCC. The site-specific and region-level analyses were performed using Wilcoxon signed-rank test and bump-hunting method, respectively. A methylation profile score (MPS) that reflect additive effect of HCC methylation profiles was constructed by computing a sum of the β-values for each associated probes weighed by the regression coefficient derived through a linear regression. To evaluate the discrimination ability of these methylation profiles, both supervised and unsupervised prediction algorithms were applied.
Results Significant differential methylation was observed between HCC cases and matched controls at 38911 loci across the genome, wherein 41.4% were hypomethylated in HCCs and a large part of these loci were resided in CpG-rich regions. These HCC methylation profiles collectively accounted for a substantial proportion of variance (30.3-54.8%, evaluated from MPS according to different significance thresholds) with good specificity and had high quality of discrimination performances. The classification accuracy of six supervised class prediction algorithms reached above 85%, while apparent clustering were observed for HCCs and controls separately by unsupervised methods.
Conclusions The present study demonstrates the significance of blood-based aberrant methylation profiles in HBV-related HCC tumorigenesis, which proves that the HCC pathogenesis is accompanied by multitudes of epigenetic alterations. The blood-based methylation signature holds promise for HCC detection and risk prediction.

(II) Co-methylation network and pathway analysis of HCC-associated methylation signatures
Background Aims Aberrant methylation of leukocyte DNA is associated with various environmental agents/clinical features known to be risk factors for HCC, but the precise targets and underlying mechanisms have not been elucidated. We aimed to examine the co-methylation networks for HCC-associated differentially methylated probes (HCC-DMPs) and correlate the identified co-methylation modules to the viral/clinical features, thereby, clarifying the link between etiological risk factors, HCC progression, and methylation signatures. We also performed pathway analysis to explore the underlying functional organization of these co-methylation changes.
Methods We applied a weighted correlation network analysis to identify the co-methylation modules for 10360 HCC-DMPs (p<0.01). The correlations between modules and the viral/clinical features were evaluated by calculating the Pearson correlation coefficients of module eigengenes and each factor. Gene set enrichment analysis was performed for these modules to test for potential overlaps with biological processes and pathways.
Results We identified a total of 7 co-methylation modules, and each of them was significantly correlated with specific viral/clinical features including viral load, HBV genotype, alanine transaminase (ALT), history of chronic liver disease, and first-degree family history of HCC. The enrichment of functional pathways in each of the modules may further reflect their biological concordance in relation to viral/clinical factors, such as, the viral load-related module genes are enriched in the immune-related pathways, while the ALT-related module is associated with inflammatory-related pathways.
Conclusions The present study showed that these HCC-DMPs exhibit specific methylation signatures associated with viral/clinical factors. The findings provided not only new insights into the virus-host interaction underlying HBV-related HCC, but also an informative link between etiological risk factors and methylation changes is established.

(III) Natural history of HBV chronic infection, viral replication activity, and DNA methylation
Background Aims HBV-DNA methylation changes are observed in human tissues and may be crucial for HBV replication. However, regarding host DNA, little is known about epigenetic changes in human blood DNA in relation to HBV infection. We aimed to identify blood-based genome-wide aberrant DNA methylation signatures associated with the natural history of HBV infection and virus replication, elucidating the biological pathways enriched in these epigenetic changes. We also investigated the association between HCC-DMPs and key components of etiology for HCC, which allows us to clarify the virus-host-immune interaction underlying HBV-related HCC pathogenesis.
Methods We characterized the global methylation patterns across four phases of HBV infection and explored the biologic relevance of these alterations by linking to predefined immune-related gene modules. The epigenome-wide association study of HBV viral load was conducted using both site-specific and region-level bump-hunting analyses. We cross matched the 10360 HCC-DMPs (p<0.01) to viral/clinical factor-related probes. The methylation profile score was constructed to evaluate the link between HCC-DMPs and viral/clinical factors.
Results A total of 17394 differentially methylated loci associated with phases of HBV-infection were identified. These phase-related genes showed enrichment for immune-related pathways, with abundance of B-cell function-related genes. From site-specific analysis of HBV viral load, we obtained 14458 associated loci, with 14.8% of these loci being related to HCC and were enriched in immune and lipid metabolism-related pathways. Region-level analysis detected 12 viral load-associated methylation regions that also exhibit immune-related properties. We found a two-to-four fold enrichment of the HCC-DMPs with viral/clinical factors-related probes as compared to the HCC-unrelated probes. A substantial proportion of variance in HCC-DMPs was attributed to viral load and leukocyte subtype composition.
Conclusions These data highlight a role of DNA methylation changes in the natural course of HBV-infection and a role in the interplay between virus replication and host immune response. Therefore, offering insights into the mechanisms underlying epigenetic reprogramming during HBV-induced hepatocarcinogenesis.

並列關鍵字

DNA methylation ； epigenome-wide association ； hepatitis B ； hepatocellular carcinoma ； peripheral blood leukocytes

參考文獻

1. Venook AP, Papandreou C, Furuse J, et al. The incidence and epidemiology of hepatocellular carcinoma: a global and regional perspective. Oncologist 2010;15 Suppl 4:5-13.

2. Bosch FX, Ribes J, Borras J. Epidemiology of primary liver cancer. Semin Liver Dis 1999;19(3):271-85.

3. Lavanchy D. Hepatitis B virus epidemiology, disease burden, treatment, and current and emerging prevention and control measures. J Viral Hepat 2004;11(2):97-107.

4. Yu MW, Chen CJ. Hepatitis B and C viruses in the development of hepatocellular carcinoma. Crit Rev Oncol Hematol 1994;17(2):71-91.

5. Liaw YF, Chu CM. Hepatitis B virus infection. Lancet 2009;373(9663):582-92.