Inflammatory Proteomic Heterogeneity Beyond Glycemia Status in Severe Obesity

Figures

• 

  Figure 1. Differential expression of circulating inflammatory proteins according to glycemic status. Volcano plots illustrate the differential expression of plasma inflammatory proteins measured via proximity extension assay among glycemic stages. In (A) T2DM vs. noDM, (B) PreDM vs. noDM, and (C) T2DM vs. PreDM, Differential protein expressions using log2 fold change (log2 Fold Change) in NPX values between glycemic groups were plotted against negative log10 p values. The dashed horizontal line indicates the threshold for statistical significance after Benjamini–Hochberg false discovery rate (FDR) correction, which was used as the primary criterion for significance. Each point corresponds to one protein. Dark green points indicate proteins that remain significant after FDR-correction, while light green points represent proteins reaching nominal significance (unadjusted p < 0.05) but not when adjusted through multiple testing correction. Selected proteins with the largest effect sizes and/or strongest statistical support are labeled. Detailed statistics for all proteins are reported in Table S2.

• 

  Figure 2. NPX distributions of CDCP1, Flt3L, and IL-8 among glycemic categories and stratified according to sex. Dot plots showing individual plasma NPX values for the most significantly differentially expressed proteins in plasma samples according to glycemic groups (noDM, PreDM, T2DM). In (A) CDCP1 NPX values, (B) Flt3L NPX values and (C) IL-8 NPX values. Each dot represents the NPX value for the protein in the plasma of each individual. Inside violin plots, horizontal dashed lines indicate median values with interquartile ranges. Panels are stratified by sex to illustrate sex-related variability. Significance was achieved at a p-value less than 0.05. *** p < 0.001; ** p < 0.01 and * p < 0.05; ns, not significant.

• 

  Figure 3. Topological data analysis (TDA) of inflammatory proteomic profiles. TDA network constructed from plasma inflammatory protein expression profiles. Each node represents one or more individuals with similar multidimensional protein expression patterns, and edges represent similar relationships. In principal component analysis (PCA) after TDA, nodes are colored according to the seven distinct clusters (cl1–cl7) identified, reflecting heterogeneous inflammatory phenotypes. (PCA) projection is shown for comparison, highlighting the limited separation of individuals by glycemic status using linear dimensionality reduction.

• 

  Figure 4. Cluster-specific inflammatory protein signatures identified through topological data analysis. Cluster-specific inflammatory protein signatures across TDA-defined clusters (cl1–cl7). Heatmap showing scaled log2 fold-change values for selected inflammatory proteins in the TDA-defined clusters (cl1–cl7). Values represent relative protein expression in each cluster compared with the remainder of the cohort. Red indicates higher relative expression, and blue indicates lower relative expression.

• 

  Figure 5. Distribution of glycemic and metabolic variables across TDA-defined clusters (cl1–cl7). Violin plots in the panels show (A) age, (B) fasting plasma glucose (FPG), (C) glycated hemoglobin (HbA1c), (D) monocyte counts, (E) neutrophil counts, and (F) total white blood cell counts among clusters. Data are presented as individual values with median and interquartile ranges. Median values are indicated by red horizontal lines within each violin plot. Significance was achieved at a p-value less than 0.05. ** p < 0.01 and * p < 0.05.