In-depth cerebrovascular lipidomics profiling for discovering novel biomarkers and mechanisms in moyamoya and intracranial atherosclerotic disease

Abstract

Background: Despite considerable research efforts, the precise etiology and underlying pathways contributing to moyamoya disease (MMD) remain poorly understood. Moreover, the overlapping vascular pathologies shared between MMD and intracranial atherosclerotic disease (ICAD) pose challenges in clinical differentiation, even with gold-standard cerebral angiography. An in-depth exploration of lipidomic alterations in cerebral intracranial MMD vessels could offer valuable insights into the pathogenesis of MMD-related mechanisms, encompassing MMD and ICAD, and unveil novel biomarkers and potential therapeutic targets. However, to date, comprehensive lipidomic profiling has been lacking.

Materials and methods: To discover novel biomarkers and unravel the pathophysiological mechanisms underlying MMD, we conducted a lipidomics analysis to characterize various lipid species in matched human extracranial and intracranial artery tissues from patients diagnosed with MMD ( n =99) and ICAD ( n =12).

Results: Our analysis identified 569 lipid species and delineated a robust panel of lipidomic biomarkers capable of effectively distinguishing MMD from ICAD (area under curve=0.98), as determined by receiver operating characteristic curve analysis. Notably, we observed a significantly more pronounced positive correlation of diacylglycerols and a negative association of triglycerides in intracranial artery tissues of MMD patients compared to those with ICAD, suggesting a potential role of dysregulated diacylglycerol-induced signaling in MMD pathogenesis. Furthermore, our investigation into the correlations of critical differential intracranial artery vessel lipid species between MMD and ICAD and clinical parameters revealed negative associations with plasma iron levels, implying a potential link between plasma iron metabolism and artery lipid homeostasis during MMD pathogenesis.

Conclusion: These findings offer promising prospects for advancing clinical diagnosis for enhanced differentiation between the two disease conditions. Additionally, they shed light on the fundamental mechanisms implicated in MMD pathogenesis and suggest potential therapeutic avenues through targeting artery vessel lipids or plasma iron levels.

Figure 1

Lipidomics analysis of recipient and donor cerebral vessel tissues in patients with MMS and ICAD. (A) Three-dimensional plot of sparse PLS discriminant analysis of lipidome of the MMD-R, MMD-D, ICAD-R, and ICAD-D groups. (B) Heatmaps showed differences in lipid class levels among tissues of MMD-R, MMD-D, ICAD-R, and ICAD-D. Clinical parameters were plotted corresponding to disease subtypes, including age and sex. (C) WGCNA analysis of lipid profiles generated four lipid modules, including the turquoise module (M1), blue module (M2), brown module (M3), and gray module (M4). (D) The pairwise correlation heatmap between four modules and disease statuses.

Figure 2

Association of differential lipid species of recipient and donor cerebral vessel tissues with clinical parameters. The top 30 lipid species ranged by VIP (variable importance on projection) score, differentiating (A) MMD-R from MMD-D and (C) ICAD-R from ICAD-D, applying stringent criteria (P<0.05, fold change >2, and VIP score >1). Receiver operating characteristic curves generated by four machine learning algorithms, including logistic regression, Gaussian NB, random forests, and MLP classifier differentiating (B) MMD-R from MMD-D and (D) ICAD-R from ICAD-D. The shade indicates the results of 10-fold cross-validation. Association of clinical parameters with differential lipid species between (E) ICAD-R and ICAD-D and (F) MMD-R and MMD-D.

Figure 3

Association of differential lipid species between intracranial MMD and ICAD recipient artery tissues with clinical parameters. The top 30 lipid species ranged by VIP (variable importance on projection) score, differentiating (A) MMD-D from ICAD-D and (C) MMD-R from ICAD-R. Ten lipid species were selected based on stringent criteria (P<0.1, fold change >2, and VIP score >1) to distinguish (B) MMD-D from ICAD-D, including MG (16:0), MG (18:3), MG (18:0), NAE (22:2), NAE (22:3), MG (20:3), MG (20:0), NAE (20:0), TG (52:1|16:0_18:0_18:1), and PE (34:2), and (C) MMD-R from ICAD-R, including NAE (22:3), MG (16:0), MG (18:3), MG (20:3), NAE (22:2), MG (18:0), MG (20:0), DGTS (32:0|16:0_16:0), NAE (20:0), and TG (52:1|16:0_18:0_18:1), by receiver operating characteristic curves generated by four machine learning algorithms, including logistic regression, Gaussian NB, random forests, and MLP classifier differentiating. (E) Association of clinical parameters with differential lipid species between MMD-R and ICAD-R. (F and G) Linear regression analysis of TIBC and IRON abundance (μmol/l) and differential lipid species.