Gut microbiota biomarkers of chronic kidney disease progression identified by 16S rDNA sequencing and machine...

Chronic kidney disease (CKD) is a global health concern characterized by high prevalence and mortality rates, yet its underlying pathogenesis remains inadequately understood. This study aimed to investigate microbial biomarkers associated with CKD progression across various stages by employing 16S rDNA sequencing, complemented by machine learning techniques including Lasso regression, the Boruta algorithm, and K-fold cross-validation, alongside microbial network analysis. Fecal samples were collected from a cohort consisting of six patients with stage II CKD, five with stage III CKD, seven with stage IV CKD, and nine healthy controls. Following quality control of the sequencing data, we analyzed microbial composition, richness, and diversity, revealing significant differences among the groups. Ten differential microbial taxa were identified, with Actinobacteriota and Bifidobacterium showing the highest relative abundances. Machine learning methods highlighted six microbial biomarkers, including Eubacterium eligens and Lactococcus, all exhibiting AUC values exceeding 0.7, indicating their potential in distinguishing CKD patients from healthy individuals. Furthermore, species driving force analysis uncovered 26, 27, and 39 microbial interaction relationships between stages II, III, IV CKD, and controls, respectively. In conclusion, our integrative analysis elucidates stage-specific gut microbial biomarkers and functional pathways that are implicated in CKD progression, thereby supporting the mechanistic role of the gut-kidney axis and underscoring the potential for microbiota-based interventions and early diagnostic approaches in CKD.