Cerebrospinal Fluid Sphingomyelins in Alzheimer's Disease, Neurodegeneration, and Neuroinflammation

Abstract

Background: Sphingomyelin (SM) levels have been associated with Alzheimer's disease (AD), but the association direction has been inconsistent and research on cerebrospinal fluid (CSF) SMs has been limited by sample size, breadth of SMs examined, and diversity of biomarkers available.

Objective: Here, we seek to build on our understanding of the role of SM metabolites in AD by studying a broad range of CSF SMs and biomarkers of AD, neurodegeneration, and neuroinflammation.

Methods: Leveraging two longitudinal AD cohorts with metabolome-wide CSF metabolomics data (n = 502), we analyzed the relationship between the levels of 12 CSF SMs, and AD diagnosis and biomarkers of pathology, neurodegeneration, and neuroinflammation using logistic, linear, and linear mixed effects models.

Results: No SMs were significantly associated with AD diagnosis, mild cognitive impairment, or amyloid biomarkers. Phosphorylated tau, neurofilament light, α-synuclein, neurogranin, soluble triggering receptor expressed on myeloid cells 2, and chitinase-3-like-protein 1 were each significantly, positively associated with at least 5 of the SMs.

Conclusion: The associations between SMs and biomarkers of neurodegeneration and neuroinflammation, but not biomarkers of amyloid or diagnosis of AD, point to SMs as potential biomarkers for neurodegeneration and neuroinflammation that may not be AD-specific.

Keywords: Alzheimer’s disease; biomarkers; cerebrospinal fluid; metabolomics; neurodegeneration; neuroinflammation; sphingolipid; sphingomyelin.

Figure 1.. Pairwise correlations of SMs.

A heatplot showing the correlation between each metabolite pair. Darker colors indicate stronger correlations, while lighter colors indicate weaker correlations. The lowest correlation coefficient was 0.47 and the highest was 0.89.

Figure 2.. Associations between biomarkers, diagnosis, and stearoyl SM (d18:1/18:0).

Results from stearoyl SM (d18:1/18:0) are displayed because this SM was consistently the most significantly associated SM and all other significant associations with SMs were in the same direction. A) Boxplots showing stearoyl SM (d18:1/18:0) levels for individuals with CU, MCI, and dementia-AD diagnoses. A small but statistically insignificant increase in metabolite level can be seen moving from the CU to AD group. None of the SM-diagnosis regressions were statistically significant. B) Scatterplots with each of the four AD-specific outcomes (x-axis) and stearoyl SM (d18:1/18:0) (y-axis). Stearoyl SM was not significantly associated with the measures of amyloid, but was significantly, positively associated with p-tau₁₈₁ after Bonferroni correction (Table 2). C) Scatterplots of each of the three neurodegeneration biomarkers (x-axis) plotted against stearoyl SM (d18:1/18:0) (y-axis). Stearoyl SM was significantly, positively associated with NfL, neurogranin, and α-synuclein after Bonferroni correction (Table 3). D) Scatterplots of each of the three neuroinflammation biomarkers (x-axis) plotted against stearoyl SM (d18:1/18:0) (y-axis). Stearoyl SM was significantly, positively associated with YKL40 and sTREM2 and nominally, negatively associated with IL6 after Bonferroni correction (Table 4). A-D: The units of stearoyl SM (d18:1/18:0) as well as p-tau₁₈₁, p-tau₁₈₁/Aβ₄₂, NfL, and IL6 are standardized by log₁₀-transformation as laid out in the methods section of this paper. B-D: Longitudinal data were used to perform the regressions shown. Best fit lines constructed using linear regression models of form outcome ~ stearoyl SM with 95% confidence intervals are drawn onto each of the scatterplots.