For patients started on anti-PD-1 therapy, fresh fecal samples were prospectively collected before and after ICI therapy. For patients with irAEs, fresh fecal samples were collected before irAE treatment, after irAE treatment, and after resuming ICIs. All fecal samples were stored in the Clinical Biobank, Medical Research Center, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences.
Fecal samples were subjected to microbiota analysis by 16S rDNA amplicon sequencing on the Illumina MiSeq (PE300) sequencing platform (Illumina, San Diego, CA) (19 (link)). Operational taxonomic units (OTUs) were detected with QIIME2 (20 (link)) and grouped according to phylum and genus. Differences in microbiota composition were compared according to relative abundance of these two levels. α-diversity was assessed according to the observed species, Shannon index, and Chao1 index, while β-diversity was assessed by Bray-Curtis and weighted UniFrac distances. Bray-Curtis distances were also used for ordination by principal coordinate analysis (PCoA), and differences in composition structure were assessed by Adonis and analysis of molecular variance (AMOVA) (21 (link)). The multiple response permutation procedure (MRPP) (22 (link)) was based on OTUs. Species with statistically significant differences between groups were evaluated by linear discriminant analysis effect size (LEfSe) (23 (link)). Functional prediction of microbiota differences was performed using Tax4fun (24 (link)) and STAMP (Statistical Analysis of Metagenomic Profiles) (25 (link)) using functional inferences from the Kyoto Encyclopedia of Gene and Genomes (KEGG) database. The abundances of the main butyrate-producing bacteria (Faecalibacterium, Agathobacter, Roseburia, Subdoligranulum, Ruminococcus_gnavus_group, Megasphaera, Phascolarctobacterium, Flavonifractor, Eubacterium_ruminantium_group, Coprococcus, Eubacterium_hallii_group, Oscillibacter, Butyricicoccus, Butyricimonas, Anaerostipes, Odoribacter, Porphyromonas, Eubacterium_ventriosum_group, Oscillospira, and Butyrivibrio) were compared.
Fecal samples were subjected to microbiota analysis by 16S rDNA amplicon sequencing on the Illumina MiSeq (PE300) sequencing platform (Illumina, San Diego, CA) (19 (link)). Operational taxonomic units (OTUs) were detected with QIIME2 (20 (link)) and grouped according to phylum and genus. Differences in microbiota composition were compared according to relative abundance of these two levels. α-diversity was assessed according to the observed species, Shannon index, and Chao1 index, while β-diversity was assessed by Bray-Curtis and weighted UniFrac distances. Bray-Curtis distances were also used for ordination by principal coordinate analysis (PCoA), and differences in composition structure were assessed by Adonis and analysis of molecular variance (AMOVA) (21 (link)). The multiple response permutation procedure (MRPP) (22 (link)) was based on OTUs. Species with statistically significant differences between groups were evaluated by linear discriminant analysis effect size (LEfSe) (23 (link)). Functional prediction of microbiota differences was performed using Tax4fun (24 (link)) and STAMP (Statistical Analysis of Metagenomic Profiles) (25 (link)) using functional inferences from the Kyoto Encyclopedia of Gene and Genomes (KEGG) database. The abundances of the main butyrate-producing bacteria (Faecalibacterium, Agathobacter, Roseburia, Subdoligranulum, Ruminococcus_gnavus_group, Megasphaera, Phascolarctobacterium, Flavonifractor, Eubacterium_ruminantium_group, Coprococcus, Eubacterium_hallii_group, Oscillibacter, Butyricicoccus, Butyricimonas, Anaerostipes, Odoribacter, Porphyromonas, Eubacterium_ventriosum_group, Oscillospira, and Butyrivibrio) were compared.
Free full text: Click here
