Figure 1

Metabolomic changes in the serum after NF. (a) Study timeline. A window of + /− 1 day was permitted for each sample. (b) Pathway distribution of metabolites detected in 260 serum samples from 36 patients with acute myeloid leukemia. 872 metabolites that were detectable in at least half of the samples were included. (c) Principal components analysis using metabolite concentrations. Pre-NF and post-NF samples in this unsupervised analysis are shown in different colors and shapes. (d) Volcano plot comparing metabolite concentrations between sample groups. Each point represents a metabolite. Points above the q = 0.05 line represent metabolites significantly associated with sample groups. q values were derived from per-metabolite two-sided Welch’s t-tests followed by correction for multiple testing using the Benjamini–Hochberg method. Points to the right (left) of the right (left) vertical line represent metabolites with > twofold higher concentration in post-NF (pre-NF) samples. These points were magnified for better visualization. Detailed results for significant metabolites are provided in Supplementary Data 2. (e) Hierarchical clustering using Euclidean distances and the complete agglomeration method for clustering. NF status was superimposed on the heatmap after the completion of clustering. Significant metabolites (q < 0.05) in the volcano plot (panel c) were used to generate the heatmap. Samples in panels (c–e) were classified into two groups: pre-NF (collected before NF) versus post-NF (collected after NF). NF: neutropenic fever.