Azithromycin (AZM) treatment and relapse are associated with changes in cell energy metabolism pathways. (A-B) Metabolomic analyses in (A) plasma and (B) dried cell pellet samples. Volcano plots and dot plots illustrate metabolite individual changes and pathway enrichment in patients treated with AZM compared with placebo (PLA). Volcano plot P values were calculated by means of 2-sided Wilcoxon signed rank test; enrichment P values were computed by means of hypergeometric test. (C) Omics integration method overview. Ninety-eight and 87 patients with mass cytometry, plasma and cell metabolomics samples in PLA and AZM cohorts were studied, respectively. Variables identified as statistically different between the 2 groups were used to reduce dimensionality with principal component analysis (PCA). Number of components to study was defined by a threshold of 80% cumulative percentage of variance explained. Patient coordinates in a 29-dimensional space were extracted from PCA results. (D) Forest plot representing the hazard ratio of relapse with the corresponding 95% confidence interval of a multivariate Fine-Gray model including each component and treatment group as covariates with death as the competing event. Blue dots indicate principal components statistically associated with relapse. (E) Stacked bar plot depicting percentage of contribution of variables from each omics layers in the 3 components associated with relapse. Top contributing individual variables are shown in Extended Data Figure 1. (F) Chord diagram showing statistically significant inter-omic correlations. (G) Correlation networks of variables included in the multi-omics analyses. Node coordinates were calculated by means of a multidimensional scaling algorithm, and edges are drawn between correlated variables. Variables (nodes) that significantly contributed to relapse are illustrated by the blue color, and size is correlated with the corresponding sum of contribution in significant dimensions. Areas where nodes overlaid are zoomed and highlighted in yellow for visualization purposes. P value and correlation and coefficients were computed according to Spearman rank correlation. Correlations were considered statistically significant if adjusted P values with false discovery rate were <.05 and Spearman’s absolute rho value >0.3. PBMC, peripheral mononuclear blood cell.
Figure 4.

Azithromycin (AZM) treatment and relapse are associated with changes in cell energy metabolism pathways. (A-B) Metabolomic analyses in (A) plasma and (B) dried cell pellet samples. Volcano plots and dot plots illustrate metabolite individual changes and pathway enrichment in patients treated with AZM compared with placebo (PLA). Volcano plot P values were calculated by means of 2-sided Wilcoxon signed rank test; enrichment P values were computed by means of hypergeometric test. (C) Omics integration method overview. Ninety-eight and 87 patients with mass cytometry, plasma and cell metabolomics samples in PLA and AZM cohorts were studied, respectively. Variables identified as statistically different between the 2 groups were used to reduce dimensionality with principal component analysis (PCA). Number of components to study was defined by a threshold of 80% cumulative percentage of variance explained. Patient coordinates in a 29-dimensional space were extracted from PCA results. (D) Forest plot representing the hazard ratio of relapse with the corresponding 95% confidence interval of a multivariate Fine-Gray model including each component and treatment group as covariates with death as the competing event. Blue dots indicate principal components statistically associated with relapse. (E) Stacked bar plot depicting percentage of contribution of variables from each omics layers in the 3 components associated with relapse. Top contributing individual variables are shown in Extended Data Figure 1. (F) Chord diagram showing statistically significant inter-omic correlations. (G) Correlation networks of variables included in the multi-omics analyses. Node coordinates were calculated by means of a multidimensional scaling algorithm, and edges are drawn between correlated variables. Variables (nodes) that significantly contributed to relapse are illustrated by the blue color, and size is correlated with the corresponding sum of contribution in significant dimensions. Areas where nodes overlaid are zoomed and highlighted in yellow for visualization purposes. P value and correlation and coefficients were computed according to Spearman rank correlation. Correlations were considered statistically significant if adjusted P values with false discovery rate were <.05 and Spearman’s absolute rho value >0.3. PBMC, peripheral mononuclear blood cell.

This Feature Is Available To Subscribers Only

or Create an Account

Close Modal
Close Modal