Fig. 4: Changes in throat composition are subtle and mirror to a certain extent changes in sputum microbiome.
a Throat microbiome composition shifts from baseline (PERMANOVA on BC-dissimilarities, stratified by donor, adjusted for sex and age: Months from ETI start R2 = 1.6%, p = 0.018). PCoA shows first two dimensions; arrows connect each sample to its baseline. Colors indicate months from ETI treatment start: black=baseline (N = 21), yellow=3 months (N = 29), blue=6–12 months (N = 80), red=15–18 months (N = 43), green=21-24 months (N = 40). Geometric markers show mean shift from baseline. b Microbiome variance (R2) explained between sampling time points (PERMANOVA on BC-dissimilarities, stratified by donor). Circle size reflects R2; color indicates p-value, details in Supplementary Data 17. c Each point represents a single participant, where the x-axis shows the BC-dissimilarities of throat-throat pairs, and the y-axis displays the BC-dissimilarities of sputum-sputum pairs for the same sampling interval. Point colors represent the sampling interval duration (in days), with purple indicating shorter intervals and yellow indicating longer intervals. Line shows linear fit; shaded areas indicate 95% confidence intervals. LME, controlled for ID-pairs and sample time point-pairs as random effects, revealed a strong positive association between throat-throat and sputum-sputum BC dissimilarities (LME Estimate = 0.51, p = 1.81e-05), suggesting that greater changes in the throat microbiome parallel greater shifts in the sputum microbiome. Additionally, sampling interval duration showed a weak but significant effect (LME Estimate = 2.12e-04, p = 0.03), indicating that longer intervals contribute to increased microbial dissimilarity.
