Li Jiayu, Chen Xuedi, Min Hang, Du Zouxi, Zhang Leyuan, Hua Wenting, Tian Limin
The First School of Clinical Medicine, Lanzhou University, Lanzhou, Gansu, China.
Department of Endocrinology, Gansu Provincial Hospital, Lanzhou, Gansu, China.
Diabetes Metab Res Rev. 2025 Sep;41(6):e70076. doi: 10.1002/dmrr.70076.
We aimed to explore the gut microbial and serum metabolic disturbances associated with the course of type 1 diabetes mellitus (T1DM), and identify potential biomarkers for discriminating T1DM from normoglycemia individuals by machine learning.
We performed 16s ribosomal RNA gene sequencing and untargeted metabolomics in a cohort of 41 patients with T1DM of varying diabetes duration and 39 healthy controls (HCs) to characterise complex interactions between the gut microbiome and serum metabolome during T1DM progression.
We identified 25 microbial genera that significantly altered in patients with T1DM, eight genera changed as the disease course, and 17 genera changed only in a specific T1DM-course group. Metabolomics analysis revealed that serum glycerophospholipid and amino acids levels were significantly changed in T1DM patients with varying disease course. Notably, we observed significantly higher levels of serum estrone, whereas lower levels of corticosterone and estrone glucuronide as the course of T1DM prolonged; Glycated haemoglobin and fasting blood glucose levels were positively correlated with estrone, and negatively correlated with corticosterone and estrone glucuronide. Furthermore, these notably changes in gut microbiota and serum metabolome were accompanied by functional alterations in sphingolipid, glutathione and taurine and hypotaurine metabolism pathways with T1DM progression. Finally, we successfully selected seven microbial and three metabolic biomarkers to differentiate T1DM from HCs.
Perturbed diabetes course-related gut microbiota was highly correlated with the alternation of metabolic patterns in T1DM, and multi-omics coupled with machine learning algorithms can be used to develop diagnostic models based on selected biomarkers.
我们旨在探索与1型糖尿病（T1DM）病程相关的肠道微生物和血清代谢紊乱情况，并通过机器学习识别区分T1DM患者与血糖正常个体的潜在生物标志物。
我们对41例糖尿病病程各异的T1DM患者和39名健康对照（HCs）进行了16s核糖体RNA基因测序和非靶向代谢组学分析，以表征T1DM进展过程中肠道微生物组与血清代谢组之间的复杂相互作用。
我们鉴定出25个在T1DM患者中显著改变的微生物属，其中8个属随病程变化，17个属仅在特定的T1DM病程组中发生变化。代谢组学分析显示，不同病程的T1DM患者血清甘油磷脂和氨基酸水平发生显著变化。值得注意的是，我们观察到随着T1DM病程延长，血清雌酮水平显著升高，而皮质酮和雌酮葡萄糖醛酸水平降低；糖化血红蛋白和空腹血糖水平与雌酮呈正相关，与皮质酮和雌酮葡萄糖醛酸呈负相关。此外，随着T1DM进展，肠道微生物群和血清代谢组的这些显著变化伴随着鞘脂、谷胱甘肽以及牛磺酸和亚牛磺酸代谢途径的功能改变。最后，我们成功选择了7种微生物和3种代谢生物标志物来区分T1DM患者与HCs。
与糖尿病病程相关的肠道微生物群紊乱与T1DM代谢模式的改变高度相关，多组学结合机器学习算法可用于基于选定的生物标志物开发诊断模型。
