Non-alcoholic fatty liver disease (NAFLD) is characterized by changes in cell composition that occur throughout disease pathogenesis, which includes the development of fibrosis in a subset of patients. DNA methylation (DNAm) is a plausible mechanism underlying these shifts, considering that DNAm profiles differ across tissues and cell types, and DNAm may play a role in cell-type differentiation. Previous work investigating the relationship between DNAm and fibrosis in NAFLD has been limited by sample size and the number of CpG sites interrogated. Here, we performed an epigenome-wide analysis using Infinium MethylationEPIC array data from 325 individuals with NAFLD, including 119 with severe fibrosis and 206 with no histological evidence of fibrosis. After adjustment for latent confounders, we identified 7 CpG sites whose DNAm associated with fibrosis (p < 5.96 × 10–8). Analysis of RNA-seq data collected from a subset of individuals (N = 56) revealed that gene expression at 288 genes associated with DNAm at one or more of the 7 fibrosis-related CpGs. DNAm-based estimates of cell-type proportions showed that estimated proportions of natural killer cells increased, while epithelial cell proportions decreased with disease stage. Finally, we used an elastic net regression model to assess DNAm as a biomarker of fibrotic stage and found that our model predicted fibrosis with a sensitivity of 0.93 and provided information beyond a model based solely on cell-type proportions. These findings are consistent with DNAm as a mechanism underpinning or marking fibrosis-related shifts in cell composition and demonstrate the potential of DNAm as a possible biomarker of NAFLD fibrosis.

中文翻译：

---

随着 NAFLD 纤维化阶段的进展，差异 DNA 甲基化和细胞类型比例的变化

非酒精性脂肪性肝病 (NAFLD) 的特征是在整个疾病发病机制中发生的细胞组成变化，其中包括部分患者出现纤维化。DNA 甲基化 (DNAm) 是这些转变背后的一个合理机制，考虑到 DNAm 谱在组织和细胞类型之间存在差异，并且 DNAm 可能在细胞类型分化中发挥作用。先前研究 NAFLD 中 DNAm 与纤维化之间关系的工作受到样本量和所询问的 CpG 位点数量的限制。在这里，我们使用来自 325 名 NAFLD 患者的 Infinium MethylationEPIC 阵列数据进行了表观基因组范围的分析，其中包括 119 名患有严重纤维化的患者和 206 名没有纤维化组织学证据的患者。在对潜在混杂因素进行调整后，我们确定了 7 个 CpG 位点，其 DNAm 与纤维化相关（p < 5.96 × 10-8）。对从一部分个体 (N = 56) 收集的 RNA-seq 数据的分析显示，在 7 个与纤维化相关的 CpG 中的一个或多个上，288 个基因的基因表达与 DNAm 相关。基于 DNAm 的细胞类型比例估计表明，自然杀伤细胞的估计比例增加，而上皮细胞比例随着疾病阶段的增加而降低。最后，我们使用弹性网络回归模型评估 DNAm 作为纤维化阶段的生物标志物，发现我们的模型预测纤维化的敏感性为 0.93，并提供了超出仅基于细胞类型比例的模型的信息。