Deciphering diabetic neuropathy: immune cell causality revealed

This study assessed the causal effects of immune cell phenotypes on diabetic neuropathy (DN) using Mendelian randomization (MR) analysis. Datasets of immune cell phenotypes and DN were collected from the European Bioinformatics Institute and FinnGen. Single nucleotide polymorphisms that met the specified criteria were subjected to a stepwise selection based on the hypotheses of association, independence, and exclusivity. Inverse variance weighted was employed as the main tool for MR analysis. Horizontal pleiotropy, heterogeneity, and robustness of the MR results were evaluated using the MR-Egger intercept, Cochran’s Q, and leave-one-out sensitivity analyses, respectively. MR analysis revealed that CD24+CD27+ %lymphocyte (odds ratio [OR]: 1.043, 95% confidence interval [CI]: 1.007‒1.080, <i>p</i> = 0.018, false discovery rate [FDR] = 0.998), CD24+CD27+ AC (OR: 1.041, 95% CI: 1.004‒1.079, <i>p</i> = 0.030, FDR = 0.998), CD28–CD127–CD25++CD8br %T cell (OR: 1.069, 95% CI: 1.003‒1.140, <i>p</i> = 0.042, FDR = 0.998), CD28–CD25++CD8br AC (OR: 1.095, 95% CI: 1.019‒1.177, <i>p</i> = 0.014, FDR = 0.998), CD33–HLA DR – AC (OR: 1.079, 95% CI: 1.007‒1.156, <i>p</i> = 0.031, FDR = 0.998), CD8 on CM CD8br (OR: 1.135, 95% CI: 1.012‒1.273, <i>p</i> = 0.030, FDR = 0.998), and naïve CD4+ �4+ (OR: 1.119, 95% CI: 1.030‒1.215, <i>p</i> = 0.008, FDR = 0.787) were associated with increased genetic susceptibility to DN. The MR-Egger intercept analysis indicated the absence of horizontal pleiotropy (<i>p</i> ≥ 0.05) and Cochran’s Q test showed that the results were not heterogeneous (<i>p</i> ≥ 0.05). This MR analysis revealed seven immune cell phenotypes associated with increased genetic susceptibility to DN. These findings are preliminary and warrant further experimental validation in different populations to confirm their significance.

本研究采用孟德尔随机化（Mendelian randomization, MR）分析方法，评估了免疫细胞表型对糖尿病周围神经病变（diabetic neuropathy, DN）的因果效应。研究从欧洲生物信息学研究所与FinnGen数据库中获取免疫细胞表型及糖尿病周围神经病变相关数据集，筛选符合特定标准的单核苷酸多态性位点，并基于关联、独立与排他性假说开展逐步筛选。本研究以逆方差加权法作为孟德尔随机化分析的核心工具，分别通过MR-Egger截距项、科克伦Q检验与留一法敏感性分析，对MR分析结果的水平多效性、异质性与稳健性进行评价。孟德尔随机化分析显示，CD24+CD27+ 淋巴细胞占比（比值比[OR]：1.043，95%置信区间[CI]：1.007‒1.080，p=0.018，错误发现率[FDR]=0.998）、CD24+CD27+ 细胞绝对计数（OR：1.041，95%CI：1.004‒1.079，p=0.030，FDR=0.998）、CD28–CD127–CD25++CD8br T细胞占比（OR：1.069，95%CI：1.003‒1.140，p=0.042，FDR=0.998）、CD28–CD25++CD8br T细胞绝对计数（OR：1.095，95%CI：1.019‒1.177，p=0.014，FDR=0.998）、CD33–HLA DR– 细胞绝对计数（OR：1.079，95%CI：1.007‒1.156，p=0.031，FDR=0.998）、CM CD8br记忆样CD8+ T细胞的CD8表达水平（OR：1.135，95%CI：1.012‒1.273，p=0.030，FDR=0.998）以及初始CD4+ T细胞（OR：1.119，95%CI：1.030‒1.215，p=0.008，FDR=0.787）均与糖尿病周围神经病变的遗传易感性升高相关。MR-Egger截距项分析提示未存在水平多效性（p≥0.05），科克伦Q检验结果显示研究无显著异质性（p≥0.05）。本孟德尔随机化分析共发现7种与糖尿病周围神经病变遗传易感性升高相关的免疫细胞表型。本研究结果尚属初步，需在不同人群中开展进一步实验验证以明确其意义。