Toward Personalized Interventions for Psoriasis Vulgaris: Molecular Subtyping of Patients by Using a Metabolomics Approach

Abstract

Aim: Psoriasis vulgaris (PV) is a complicated autoimmune disease characterized by erythema of the skin and a lack of available cures. PV is associated with an increased risk of metabolic syndrome and cardiovascular disease, which are both mediated by the interaction between systemic inflammation and aberrant metabolism. However, whether there are differences in the lipid metabolism between different levels of severity of PV remains elusive. Hence, we explored the molecular evidence for the subtyping of PV according to alterations in lipid metabolism using serum metabolomics, with the idea that such subtyping may contribute to the development of personalized treatment. Methods: Patients with PV were recruited at a dermatology clinic and classified based on the presence of metabolic comorbidities and their Psoriasis Area and Severity Index (PASI) from January 2019 to November 2019. Age- and sex-matched healthy controls were recruited from the preventive health department of the same institution for comparison. We performed targeted metabolomic analyses of serum samples and determined the correlation between metabolite composition and PASI scores. Results: A total of 123 participants, 88 patients with PV and 35 healthy subjects, were enrolled in this study. The patients with PV were assigned to a "PVM group" (PV with metabolic comorbidities) or a "PV group" (PV without metabolic comorbidities) and further subdivided into a "mild PV" (MP, PASI <10) and a "severe PV" (SP, PASI ≥10) groups. Compared with the matched healthy controls, levels of 27 metabolites in the MP subgroup and 28 metabolites in the SP subgroup were found to be altered. Among these, SM (d16:0/17:1) and SM (d19:1/20:0) were positively correlated with the PASI in the MP subgroup, while Cer (d18:1/18:0), PC (18:0/22:4), and PC (20:0/22:4) were positively correlated with the PASI in the SP subgroup. In the PVM group, levels of 17 metabolites were increased, especially ceramides and phosphatidylcholine, compared with matched patients from the PV group. In addition, the correlation analysis indicated that Cer (d18:1/18:0) and SM (d16:1/16:1) were not only correlated with PASI but also has strongly positive correlations with biochemical indicators. Conclusion: The results of this study indicate that patients with PV at different severity levels have distinct metabolic profiles, and that metabolic disorders complicate the disease development. These findings will help us understand the pathological progression and establish strategies for the precision treatment of PV.

Keywords: lipid metabolites; metabolic diseases; metabolomics; molecular subtyping; psoriasis vulgaris; severity biomarkers.

FIGURE 1

Comparison of metabolites using PLS-DA models. (A) PLS-DA model of the MP and HC-MP subgroups. (B) PLS-DA model of the SP and HC-SP subgroups. (C) PLS-DA model of the MP and SP subgroups. (D) PLS-DA model of the PVM and PV groups. HC-MP, healthy controls matched to the group with mild psoriasis vulgaris; HC-SP, healthy controls matched to the group with severe psoriasis vulgaris; MP, mild psoriasis vulgaris; PLS-DA, partial least squares discriminant analysis; PV, psoriasis vulgaris without metabolic diseases; PVM, psoriasis vulgaris with metabolic diseases; SP, severe psoriasis vulgaris.

FIGURE 2

Volcano plots illustrating differential metabolites. (A) Volcano plot of differential metabolites between the MP and HC-MP subgroups. (B) Volcano plot of differential metabolites between the SP and HC-SP subgroups. (C) Volcano plot of differential metabolites between the MP and SP subgroups. (D) Volcano plot of differential metabolites between the PVM and PV groups. HC-MP, healthy controls matched to the group with mild psoriasis vulgaris; HC-SP, healthy controls matched to the group with severe psoriasis vulgaris; MP, mild psoriasis vulgaris; SP, severe psoriasis vulgaris; PV, psoriasis vulgaris without metabolic diseases; PVM, psoriasis vulgaris with metabolic diseases.

FIGURE 3

Heatmaps illustrating Spearman’s correlations between the altered metabolites. (A) Correlation heatmap of altered metabolites (MP vs. HC-MP) in the MP subgroup. (B) Correlation heatmap of altered metabolites (SP vs. HC-SP) in the SP subgroup. (C) Correlation heatmap of altered metabolites (MP vs. SP) in the MP subgroup. (D) Correlation heatmap of altered metabolites (MP vs. SP) in the SP subgroup. (E) Correlation heatmap of altered metabolites (PVM vs. PV) in the PVM group. Correlation analysis performed with Spearman’s correlation coefficient, *p < 0.05, **p < 0.01, and ***p < 0.001. HC-MP, healthy controls matched to the group with mild psoriasis vulgaris; HC-SP, healthy controls matched to the group with severe psoriasis vulgaris; MP, mild psoriasis vulgaris; SP, severe psoriasis vulgaris; PV, psoriasis vulgaris without metabolic diseases; PVM, psoriasis vulgaris with metabolic diseases.

FIGURE 4

Spearman’s correlations between altered metabolites and the PASI. (A) Correlation analyses between PASI and altered metabolites (MP vs. HC-MP) in the MP subgroup: SM (d16:1/17:0): R = 0.37, p = 0.041; SM (d19:1/20:0): R = 0.39, p = 0.032. (B) Positive correlations between PASI and altered metabolites (SP vs. HC-SP) in the SP subgroup: Cer (d18:1/18:0): R = 0.42, p = 0.015; PC (18:0/22:4): R = 0.45, p = 0.01; and PC (20:0/22:4): R = 0.39, p = 0.029. PASI, Psoriasis Area and Severity Index; HC-MP, healthy controls matched to the group with mild psoriasis vulgaris; HC-SP, healthy controls matched to the group with severe psoriasis vulgaris; MP, mild psoriasis vulgaris; SP, severe psoriasis vulgaris.

FIGURE 5

Graphical abstract showing the major findings of the study.