Cindy Boer

The Gut Microbiome of Childen and Adults | 215 5.1 α-diversity was negatively associated with BMI after adjusting for age, sex and technical covariates in both cohorts (beta=−0.019, SE=2.3x10 -03 , p-value=1.1*10 -03 for GenR and beta=−0.015, SE=2.8x10 -03 , p-value=1.8x10 -06 for RS). We compared this as- sociation with observations reported in similar cohorts of Northern European origin (LLD and FGFP). The negative correlation of α-diversity with BMI was in line with find- ings reported previously[40]. At genus-level, we identified 27 BMI-associated genera in GenR and 33 BMI-associated genera in RS ( Table 3, Supplementary Table 2 ), of which 14 genera overlapped across both cohorts. In the RS cohort, we confirmed 6 genera found to be associated with BMI in the LLD and FGFP cohorts ( Table 3 ). In GenR we only confirmed two genera ( Alistipes and Barnesiella ) to be associated with BMI in RS, LLD and FGFP cohorts. As previously reported40,48 we also observed the association between increased abundance of genus Akkermansia and lower BMI. Comparison of RS and GenR stool microbiome diversities, compositions and functions To analyze differences in gut microbiome compositions between children and adults, both datasets were combined after exclusion of the non-Northern European samples. The combined dataset included 3,082 samples: 1,371 from RS and 1,712 from GenR and the final OTU table contained 173 genera. Although Shannon α-diversity was not associated with age in each cohort separately (linear model; GenR: beta=−0.003, p-val- ue=0.91; RS: beta=0.001, p-value=0.63), Shannon α-diversity was significantly higher in the RS cohort than in the GenR cohort in the combined dataset (p-value<2.2x10 -16 , mean=5.9, sd=0.74 for RS andmean=5.58, sd=0.82 for GenR). Also, the overall compo- sitions (Bray Curtis distances) differed significantly between both groups ( Figure 3B ; PERMANOVA; R 2  =0.06; p-value=0.001). We observed a 2 to 3 times lower abundance of phylum Firmicutes in the GenR cohort as compared with RS (p-value<2.2x10 -16 ). Furthermore, we observed higher abundances of the gram-negative classes (Bacteroid- ia, Negativicutes and some classes from phylum Proteobacteria) in children. Regres- sion analysis, in MaAsLin, on the relative abundances of the individual genera showed higher relative abundances of 59 genera in RS compared to GenR and higher relative abundances of 20 genera in GenR compared to the RS cohort (q-value<0.05; Supple- mentary Table 3 ). The largest differences were observed for genera from family Lach- nospiraceae including Blautia, Lachnospiraceae, Anaerostipes, Dorea, Fusicatenibacter, Coprococcus, Roseburia, Ruminoclosteridium, Butyricicoccus and Lachnoclostridium that were more abundant in RS, and genera from families Ruminococcaceae and Bacteroi- daceae including Bateroides, Faecalibacterium, Alistipes, Barnesiella, Parabacteroides, Bifidobacterium and Odoribacter that were more abundant in GenR ( Figure 3C ). The most abundant genera were Bacteroides in GenR (children) cohort and Blautia in the RS (adults) cohort ( Figure 3D ).