Extended Data Fig. 2: Associations of host characteristics with the gut microbial composition.
(a–d) Species-level and (e–h) pathway-level alpha diversity across participants with different kidney function status. P values for group differences were determined using multivariate linear regression, with adjustments for age, sex, BMI, diabetes, hypertension, cancer, medication use, HDL-c, and LDL-c. The boxes represented the median and interquartile range (IQR) of the distributions, and top and bottom whiskers marked the point at 1.5×IQR. Participants of impaired kidney function, n = 328; Participants of normal kidney function, n = 1,222; Participants of eGFR ≥ 90 mL/min/1.73 m2, n = 187; Participants of eGFR ≥ 60 & < 90 mL/min/1.73 m2, n = 1035; Participants of eGFR < 60 mL/min/1.73 m2, n = 328. (i) The microbial composition among participants with impaired kidney function or normal kidney function at the pathway-level. Group difference was calculated by permutational multivariate analysis of variance (PERMANOVA) using Bray-Curtis dissimilarity. Participants with impaired kidney function, n = 328; Participants with normal kidney function, n = 1,222. (j–l) Proportion of the variance in microbiome taxonomic composition explained by each measured kidney function parameters and other host characteristics. (m–p) Proportion of the variance in microbiome functional composition explained by each measured kidney function parameters and other host characteristics. The explained variance (R2) for (j–p) was calculated by PERMANOVA using Bray-Curtis dissimilarity, and all variables were jointly fitted in one model for each panel. ns, not significant (P > 0.05). eGFR, estimated glomerular filtration rate. BMI, body mass index. Medication Use, whether people have had at least five types of drugs. HDL-c, high-density lipoprotein cholesterol. LDL-c, low-density lipoprotein cholesterol. P values were not adjusted for multiple comparisons. All statistical tests were two-sided.
