Distinct Cerebrospinal Fluid Lipid Signature in Patients with Subarachnoid Hemorrhage-Induced Hydrocephalus

Abstract

Patients with subarachnoid hemorrhage (SAH) may develop posthemorrhagic hydrocephalus (PHH), which is treated with surgical cerebrospinal fluid (CSF) diversion. This diversion is associated with risk of infection and shunt failure. Biomarkers for PHH etiology, CSF dynamics disturbances, and potentially subsequent shunt dependency are therefore in demand. With the recent demonstration of lipid-mediated CSF hypersecretion contributing to PHH, exploration of the CSF lipid signature in relation to brain pathology is of interest. Despite being a relatively new addition to the omic's landscape, lipidomics are increasingly recognized as a tool for biomarker identification, as they provide a comprehensive overview of lipid profiles in biological systems. We here employ an untargeted mass spectroscopy-based platform and reveal the complete lipid profile of cisternal CSF from healthy control subjects and demonstrate its bimodal fluctuation with age. Various classes of lipids, in addition to select individual lipids, were elevated in the ventricular CSF obtained from patients with SAH during placement of an external ventricular drain. The lipidomic signature of the CSF in the patients with SAH suggests dysregulation of the lipids in the CSF in this patient group. Our data thereby reveal possible biomarkers present in a brain pathology with a hemorrhagic event, some of which could be potential future biomarkers for hypersecretion contributing to ventriculomegaly and thus pharmacological targets for pathologies involving disturbed CSF dynamics.

Keywords: SAH; cerebrospinal fluid; lipidomics; mass spectrometry; posthemorrhagic hydrocephalus.

Figure 1

CSF lipidomics in control subjects. (A) Detected CSF lipids (244 compounds) were distributed according to classes (17 classes), based on enrichment analysis. Fourteen groups contained less than four lipids (<1% of the total detected compounds) and were assembled into one group (‘small collection group’). (B) Lipids were normalized based on the geomean of the quality control samples (n = 15) to obtain the total lipid abundance for each group, and an enrichment plot was generated by summing the normalized values of lipids in each class upon which the mean of all controls was calculated. Data are based on cisternal CSF from control subjects (n = 11).

Figure 2

Subject background and CSF lipid distribution in control subjects. (A) The normalized total lipid values (with a descriptive power above 2.5) were grouped for age initially divided at the median age (light gray spheres: ≤60 years; dark gray spheres: >60 years) or sex (B) (blue: male; pink: female) and plotted with respect to their first and second principal components (PC1 and PC2). (C) Total lipid abundance fluctuation was generated based on normalized mean abundance for each lipid group as a function of the age of the control subjects and plotted as a line plot from the youngest (40 years old) to the oldest (77 years old) control subject. Data are based on cisternal CSF from control subjects (n = 11; note that the spheres indicating ages 57 and 77 years are based on two subjects each).

Figure 3

CSF lipid profile in patients with SAH. (A) The total abundance for each CSF lipid class was obtained with normalization based on the geomean of the quality control samples (n = 15). An enrichment plot was generated based on the summing of normalized values of lipids in each class from patients with SAH. (B) The normalized lipid values (with a descriptive power above 2.5) were grouped for age (light gray spheres: ≤60 years; dark gray spheres: >60 years) or sex (C) (blue: male; pink: female) and plotted with respect to their first and second principal components (PC1 and PC2). Data are based on ventricular CSF from patients with SAH (n = 13).

Figure 4

Dysregulated CSF lipid levels in patients with SAH. (A) The normalized total lipid values (with a descriptive power above 2.5) were grouped for control subjects and patients with SAH (white spheres: control subjects; red: SAH) and plotted with respect to their first and fourth principal components (PC1 and PC4). (B) The total abundance of each lipid class in CSF from control subjects and patients with SAH was plotted as bars using normalized data. (C) Volcano plot of normalized individual CSF lipids identified with the fold change (log2 transformed) between control subjects and patients with SAH. Upper dashed line: adjusted p value; indicates cut-off for significance. Lower dashed line: p value < 0.05. A cluster of lipids that were detected as highly elevated in patients with SAH, although not reaching the adjusted significance level, are highlighted by a dashed line box. Statistical evaluation with Welch’s t-test followed by the Benjamini−Hochberg method (with an adjusted p value < 0.1 (false discovery rate, FDR, of 10%). * p < 0.05. FC: fold change.

Figure 5

Isolated lipid class analysis. The lipid content within each lipid class in isolation between control subjects and patients with SAH are plotted as bars using normalized data (without outliers; two-sided Smirnov−Grubbs test, α = 0.05). Lipids with a descriptive power of less than 2.5 were excluded from the analysis. FC: fold change.