Multi-kingdom gut microbiota characterization in Chinese patients with idiopathic inflammatory myopathies.

All three microbial kingdoms (bacteriome, mycobiome, and virome) were significantly altered in IIM patients compared to healthy controls.

• Shotgun metagenomic sequencing was performed on fecal samples from 34 IIM patients and 37 healthy controls.
• Taxonomic, functional, network, and machine-learning analyses were used to profile gut microbiota across all three kingdoms.
• This was described as the first comprehensive multi-kingdom gut microbiota analysis in IIM.

Several inflammation-associated bacterial taxa were enriched in IIM patients.

• Enriched bacterial taxa in IIM included Rothia mucilaginosa, Streptococcus parasanguinis, and Trueperella pyogenes.
• These taxa are described as inflammation-associated.
• SCFA-producing bacteria were depleted in IIM patients.

Opportunistic fungi, particularly Aspergillus spp., were enriched in IIM patients, while SCFA-producing fungi were depleted.

• Aspergillus spp. were identified as enriched opportunistic fungi in IIM.
• SCFA-producing fungi were among the depleted microbial taxa in IIM patients.
• The mycobiome was significantly altered in IIM as part of the multi-kingdom dysbiosis.

Virome analysis revealed substantial shifts in IIM, with higher abundance of Siphoviridae compared to healthy controls.

• Siphoviridae abundance was higher in IIM patients.
• Altered viral functional gene profiles suggested enhanced phage-mediated genome integration, recombination, and bacterial stress adaptation in IIM.
• The virome demonstrated the strongest discriminatory power among the three microbial kingdoms in machine-learning models.

Multi-kingdom network analysis showed extensive rewiring in IIM characterized by increased network connectivity and a shift toward fungi-centered ecological hubs.

• In IIM, ecological hubs shifted toward fungi-centered networks.
• In healthy controls, networks were bacteria/virus-dominated.
• The increased network connectivity and hub reorganization characterized the dysbiotic state in IIM.

A combined multi-kingdom machine-learning classifier achieved an AUC of 0.997 for discriminating IIM from healthy controls, with viral signatures dominating the model.

• The combined multi-kingdom classifier achieved AUC = 0.997.
• Viral signatures dominated the combined multi-kingdom classifier.
• The virome alone demonstrated the strongest discriminatory power among single-kingdom models.
• Machine-learning analyses were performed as part of the analytical pipeline alongside taxonomic, functional, and network analyses.