A Metabolomic Approach to Unexplained Syncope

Figures

• 

  Figure 1. PLDS-DA analysis between the CTRL and each TLC group. PLDS-DA analysis based on the metabolites pattern quantified in the CTRL and CS groups (A); the plasma levels of Phe, Phe/Tyr, Arg, Cit, Orn, C18, and C18OH quantified in the comparison between the CS and CTRL groups (B); cumulative ROC curve based on the significantly modulated metabolites depicted in (B) (CS vs. CTRL) and the associated Confusion Matrix resulting from the predicted class probability across the 100 cross-validation (C); PLDS-DA analysis based on the metabolites pattern quantified in the CTRL and OH groups (D); the serum levels of Phe and Pro quantified in the comparison between the OH and CTRL groups (E); cumulative ROC curve based on the two significantly modulated metabolites depicted in panel E (OH vs. CTRL) and the associated Confusion Matrix resulting from the predicted class probability across the 100 cross-validation (F); PLDS-DA analysis based on the metabolites pattern quantified in the CTRL and US groups (G); the serum levels of Phe and Phe/Tyr in the comparison between the US and CTRL groups (H) and the related cumulative ROC and associated Confusion Matrix resulting from the predicted class…

• 

  Figure 2. PLDS-DA analysis and heatmaps based on the comparison between each TLC group. PLDS-DA analysis based on the metabolite pattern quantified in the TLC groups, in particular, CS in green, NMS in dark blue, OH in yellow, and US in light blue (A); heatmaps based on serum metabolite levels quantified in CS (B), NMS (C), and OH (D) for evaluation of the US group cluster analysis.

• 

  Figure 3. The logistic regression model with C24 and C22:0-LPC. Dot-plot showing plasma levels of C24 (A) and C22:0-LPC (B) in the comparison between the OH and NMS groups; cumulative ROC built by using C24 and C22:0-LPC simultaneously based on a logistic regression model for the comparison between OH and NMS (C); re-classification of US patients based on the application of the logistic regression algorithm described in Table 2 (D). * p-value < 0.05.

• 

  Figure 4. The logistic regression model with ARG and ASA-Total. Dot-plot showing plasma levels of ARG (A) and Asa-Total (B) in the comparison between the OH and CS groups; re-classification of US patients using ARG and ASA-Total simultaneously based on the application of the logistic regression algorithm described in Table S9 (C). * p-value < 0.05.

• 

  Figure 5. The logistic regression models with GLN/LYS and C22:0-LPC. Dot-plot showing plasma levels of GLN/LYS (A) and C22:0-LPC (B) in the comparison between the CS and NMS groups; re-classification of US patients using GLN/LYS plasma based on the application of the logistic regression algorithm described in Table S10 (C); re-classification of US patients using C22:0-LPC plasma based on applying the logistic regression algorithm described in Table S11 (D). * p-value < 0.05.