Untargeted serum metabonomics study of psoriasis vulgaris based on ultra-performance liquid chromatography coupled to mass spectrometry

Abstract

Psoriasis is a common, chronic, systemic inflammatory skin disease, the etiology and pathogenesis is unclear. An untargeted high-throughput metabonomics method based on liquid chromatography coupled to mass spectrometry was applied to study the serum metabolic changes in psoriasis vulgaris patients, and to discover serum potential biomarkers for identification, diagnosis and exploring pathogenesis of psoriasis. The serum metabolic profiles from 150 subjects (75 psoriasis patients and 75 healthy controls) were acquired, the raw spectrometric data were processed by multivariate statistical analysis, and 44 potential biomarkers were screened out and identified. The potential biomarkers were mainly involved in glycerophospholipid metabolism, sphingolipid metabolism, arachidonic acid metabolism, bile acid biosynthesis, indicated the pathogenesis of psoriasis may be related to the disturbed metabolic pathways.

Keywords: LC-MS; biomarker; high-throughput; metabonomics; psoriasis.

Figure 1. Typical base peak chromatograms (BPC) of serum sample

The typical base peak chromatograms based on UPLC/Q-TOF MS in positive ion mode of psoriasis patient (A) and healthy control (B) was shown respectively.

Figure 2. Multivariate data analysis and permutation test of psoriasis and healthy controls

(A) PCA scores plots, (B) PLS-DA, (C) OPLS-DA(● PV patients, ● HC), (D) Permutation test of PLS-DA, (E) Permutation test of OPLS-DA, (F) S-plots: The variables of the maximus VIP values were marked red. The normalized data were imported to Simca-P for multivariate statistical analysis. These models showed excellent predictive ability, and validated to be no over fitting by permutation test with 200 iterations. The red marked plots in the S-plot of PV and HC were significant different variables, and considered as the biomarker candidates.

Figure 3. Multivariate data analysis and permutation test of psoriasis in varying degrees of severity

(A) PLS-DA scores plots, (B) OPLS-DA, (C) Permutation test of PLS-DA, (D) Permutation test of OPLS-DA (● mild PV patients, ● moderate to severe PV patients, ● HC). The severity of psoriasis is defined by PASI score, PASI ≦10 being considered mild, PASI >10 being considered moderate to severe, compared with the healthy control, statistical pattern recognition results were shown.

Figure 4. Typical metabolites with significant alterations in psoriasis vs. control

The peaks intensities of potential biomarkers in serum of the two groups were shown.

Figure 5. Heat map of the 44 differential metabolites

(Psoriasis patients are labelled P... and healthy controls are labelled C...). The ratio of metabolite in the subject samples to the average of those in the healthy control samples was calculated, and then the metabolic alteration was demonstrated as log10 (ratio), the major metabolic alterations in psoriasis were visualized in a heat map plot.

Figure 6. Disturbed metabolic pathways showed various metabolism changes when comparing psoriasis and control

The metabolic pathway analysis was an integrating enrichment analysis based on the human metabolic pathways. The different color and size of the symbol means the different level of significance, there were more potential biomarkers in the data were involved in the pathway, the color was darker or the size was larger. Glycerophospholipid metabolism, sphingolipid metabolism, arachidonic acid metabolism was most relevant pathways in psoriasis metabolism.

Figure 7. Pearson correlation network of metabolites

Pearson correlation coefficients between the 44 metabolites were calculated on the basis of the average normalized quantities of metabolites. Highly correlated metabolites with r> 0.8 were connected with solid lines, which metabolite with r< -0.8 were connected with dotted lines in the network. Orange plots indicated up-regulated metabolites and green plots indicated down-regulated metabolites in PV.

Figure 8. The typical ROC curve plots of potential biomarkers

(A~E) Typical ROC curves and AUC values of potential biomarkers with high-performance prediction; (F) Multivariate exploratory ROC analysis overview.