利用中介分析探討B型肝炎病毒基因型C型對於慢性B型肝炎患者產生肝癌的影響

Abstract

背景：慢性B型肝炎病毒（HBV）是導致肝細胞癌（HCC）的主要因子之一。與B型基因型相比，C型基因型與HCC風險增加有關。儘管如此，相關機轉仍不清楚。目前已知C型基因型與e抗原（HBeAg）延遲清除、基礎核心啟動子（BCP）A1762T/G1764A突變發生和肝硬化之間存在關聯，這些因素可能解釋了HCC風險的增加。我們旨在量化這些潛在中介因子對HBV基因型與HCC關係的相對貢獻。 方法：本研究使用了基於醫院的ERADICATE-B和基於社區的REVEAL-HBV這兩個世代研究的數據，進行二元單中介和串聯多重中介分析。所有模型均將C型基因型（相對於B）定義為暴露因素，HCC發病（相對於無HCC）定義為結果變量。測試的潛在中介變量包括HBeAg清除延遲或無清除、BCP A1762T/G1764A突變狀態和肝硬化。此外還調整了性別和年齡。單中介模型使用邏輯回歸將勝算比轉換為風險比（RR），而多重中介模型使用Probit回歸計算風險差（RD）。 結果：兩個群體具有相似的HBV基因型、肝硬化和HCC數據的樣本大小（ERADICATE-B為2547人，REVEAL-HBV為2630人）。相比之下，REVEAL-HBV的追蹤時間略短且HCC事件總數較少（12.4年對比15.3年，140例對比207例）。在所有模型中，HBV基因型與HCC顯著相關。在單中介模型中，所有測試的中介因素均顯示出統計顯著的中介效果。在兩群中，肝硬化表現出最大的中介效果（ERADICATE-B中的RR調整值=2.19 [95% CI 1.99–2.45]，中介比例=66%；REVEAL-HBV中的RR調整值=1.48 [1.34–1.65]，中介比例=48%），其次是BCP突變（RR調整值=1.35 [1.20–1.52] 和1.22 [1.07–1.40]，中介比例=23%和27%）以及HBeAg清除延遲或無清除（RR調整值=1.22 [1.12–1.33] 和1.15 [1.14–1.17]，中介比例=12% 和13%）。在HBeAg清除延遲或無清除接著肝硬化的串聯兩中介模型中， C型基因型主要通過單獨的肝硬化增加HCC風險（ERADICATE-B中的RD調整值=4.95%，REVEAL-HBV中的RD調整值=1.28%；中介比例=58% 和48%），雖然在敏感性分析中HBeAg清除延遲或無清除和肝硬化共同途徑也是顯著的。在BCP突變接著肝硬化的串聯中介模型中，所有三個間接途徑都是顯著的。單獨的肝硬化解釋了ERADICATE-B和REVEAL-HBV中效果的56%（RD調整值=3.14% 和1.21%），單獨的BCP突變解釋了ERADICATE-B的11% 和REVEAL-HBV的17%（RD調整值=0.64% 和0.37%），而BCP突變接著肝硬化解釋了12%和8%（RD調整值=0.66% 和0.18%）。 結論：本研究是首次利用中介分析探究HBV基因型相關致癌機制，為慢性HBV感染的自然史提供了洞見，並為研究其他疾病提供了框架。研究結果支持先前的假設，即HBeAg延遲清除主要通過肝硬化發揮作用，而BCP突變在增加C型基因型感染者的HCC風險時既有進展又有無進展至肝硬化兩種機制。與C型基因型相關的增加HCC風險可能涉及多種機制，而肝硬化的發生起著關鍵作用。

Background: Chronic hepatitis B virus (HBV) infection is a major cause of hepatocellular carcinoma (HCC). HBV genotype C is associated with a higher risk of HCC compared to genotype B, but the mechanisms remain unclear. Previous studies have linked genotype C with delayed hepatitis B e antigen (HBeAg) seroclearance, basal core promoter (BCP) A1762T/G1764A mutation onset, and cirrhosis, all of which may contribute to the increased HCC risk. This study aimed to quantify the relative contributions of these potential mediators to the relationship between HBV genotype and HCC. Methods: Binary single mediation and serial multiple mediation analyses were conducted using data from two Taiwanese chronic HBV cohorts: the hospital-based ERADICATE-B cohort and community-based REVEAL-HBV cohort. HBV genotype C (vs B) was considered the exposure, and incident HCC (vs no HCC) was the outcome. The mediators tested were delayed or no HBeAg seroclearance, BCP A1762T/G1764A mutation status, and cirrhosis. Sex and age were adjusted for. Odds ratios were converted to risk ratios (RRs) for the single mediation models which used logistic regression, while risk differences (RDs) were calculated for the multiple mediation models which used probit regression. Results: The cohorts had similar sample sizes of individuals with HBV genotype, cirrhosis, and HCC data (2547 and 2630 in ERADICATE-B and REVEAL-HBV respectively). REVEAL-HBV had a slightly shorter follow-up time and total number of HCC events compared to ERADICATE-B (12.4 vs 15.3 years, 140 vs 207 events). HBV genotype was significantly associated with HCC in all models. In single mediation models, all tested mediators yielded statistically significant mediated effects. In both cohorts, cirrhosis demonstrated the largest mediated effect (RRadj. = 2.19 [95% CI 1.99–2.45], proportion mediated = 66% in ERADICATE-B and RRadj. =1.48 [1.34–1.65], prop. mediated = 48% in REVEAL-HBV), followed by BCP mutation (RRadj. = 1.35 [1.20–1.52] and 1.22 [1.07–1.40], prop. mediated = 23% and 27%) and delayed or no HBeAg seroclearance (RRadj. = 1.22 [1.12–1.33] and 1.15 [1.14–1.17], prop. mediated = 12% and 13%). In the serial mediation model with delayed or no HBeAg seroclearance followed by cirrhosis, HBV genotype C increased HCC risk primarily through cirrhosis alone (RDadj. = 4.95% in ERADICATE-B and 1.28% in REVEAL-HBV; prop. mediated = 58% and 48%), though the pathway with both delayed or no HBeAg seroclearance and cirrhosis was also significant in sensitivity analyses. In the serial mediation model with BCP mutation followed by cirrhosis, all three indirect pathways were significant. Cirrhosis alone explained 56% of the effect in both ERADICATE-B and REVEAL-HBV (RDadj. = 3.14% and 1.21%), BCP mutation alone explained 11% in ERADICATE-B and 17% in REVEAL-HBV (RDadj. = 0.64% and 0.37%), and BCP mutation followed by cirrhosis explained 12% and 8% (RDadj. = 0.66% and 0.18%). Conclusions: This study, the first to investigate the mechanisms underlying HBV genotype-related carcinogenesis using mediation analysis, provides insights into the natural history of chronic HBV infection and offers a framework for studying other diseases. The findings support previous hypotheses that delayed HBeAg seroclearance primarily acts through cirrhosis, while BCP mutation acts both with and without progression to cirrhosis in increasing HCC risk in individuals with genotype C infection. The increased HCC risk associated with genotype C involves multiple mechanisms, with cirrhosis onset playing a key role.