Cross-sectional multi-omics analysis reveals stage-associated differences in gut microbiota–fecal metabolite–peripheral immune modules across the Alzheimer’s disease continuum

Alzheimer’s disease (AD) is increasingly viewed as a disorder involving systemic immune and metabolic changes, but how gut microbial composition, fecal metabolites, and peripheral immune proteins vary across the AD clinical continuum remains unclear. We conducted a cross-sectional multi-omics study of 78 participants, including cognitively healthy controls (CH), individuals with mild cognitive impairment (MCI), and patients with AD. Fecal 16 S rRNA gene sequencing, fecal untargeted metabolomics, selected serum immune proteins, and AD-related blood biomarkers were analyzed using group comparisons and exploratory integrative analyses.

Cognitive scores were consistent with the predefined CH, MCI, and AD classifications. Blood biomarker and peripheral immune protein analyses showed group-dependent patterns, including higher Aβ1–42 and p-tau-181 in cognitively impaired groups than in CH, the highest NfL and MCP-1 levels in AD, higher sTREM2 and PGRN in cognitively impaired groups, and lower CX3CL1 in MCI and AD. Microbiome beta-diversity analyses did not support robust categorical separation among groups, whereas selected genus-level analyses highlighted lower relative abundance of Faecalibacterium and Agathobacter and higher relative abundance of Faecalitalea in cognitively impaired groups.

Fecal metabolomics highlighted alterations involving lipid/fatty acyl-, bile acid-, and amino acid/peptide-related metabolism, including 3-methyladipic acid, linoleoyl ethanolamide, lauroyl-l-carnitine, cholic acid, 7-ketodeoxycholic acid, and α-l-Glu-Gly. Integrative analyses suggested fecal metabolites as possible statistical links between gut microbial features and peripheral immune or cognitive readouts. These findings provide hypothesis-generating evidence for gut microbiota–fecal metabolite–peripheral immune associations across the AD continuum and require validation in larger longitudinal cohorts.

Linear discriminant analysis effect size Orthogonal partial least-squares discriminant analysis Soluble triggering receptor expressed on myeloid cells 2 The authors thank all participants and their families for their contribution to this study. Use of artificial intelligence tools. During manuscript preparation, an AI-based language model was used to assist with language polishing and formatting.

The authors reviewed and edited all AI-assisted content and take full responsibility for the final manuscript. This work was supported by the National Natural Science Foundation of China (Grant No. 82675973), the Natural Science Foundation of Zhejiang Province (Grant No. LMS25H270002), and the Lishui Science and Technology Bureau Priority Research and Development Project (Grant No. 2022ZDYF21).

The funders had no role in the study design, data collection, data analysis, data interpretation, manuscript preparation, or decision to submit the manuscript for publication. Zhe Wei and Jinjing Wu contributed equally to this work. School of Medicine, Lishui University, No. 1 Lishui University Road, Lishui, 323000, Zhejiang, China Zhe Wei, Xinlin Wang, Jinghua Yi & Hang Zhang Department of Fundamental Medicine, Wuxi School of Medicine, Jiangnan University, Wuxi, 214000, Jiangsu, China Hangzhou Normal University, Hangzhou, 311121, Zhejiang, China The Second Affiliated Hospital, Heilongjiang University of Chinese Medicine, Heilongjiang, 150000, China The authors declare no competing interests.