Figure 1.

Partial Least Squares Discriminant Analysis (PLS-DA) based on serum metabolites determined by GC-MS. (A) Two-dimensional score plot showing clustering and separation between healthy CTRL serum profiles (orange) and endometrial cancer-affected patients’ profiles (blue) from the n = 90 training set. (B) Metabolites showing a VIP score > 2.0 in the PLS-DA analysis. (C) Permutation test results based on 2000 iterations. (D) PLS-DA classification performance using increasing number of latent variables. The red star indicates that the best model was achieved using only 1 variable. (E) Volcano plot reporting metabolite concentration fold changes and their statistical significance comparing CTRL vs. EC subjects among the second validation cohort. 1. Glycine, 2. phenyl pyruvic acid, 3. serine, 4. valine, 5. urea, 6. oxyproline, 7. phenylalanine, 8. glyceraldehyde 3 phosphate, 9. gluconic acid 10. glycerol, 11. 3-hydroxybutyric acid, 12. stearic acid.