Profiling amyloid-β peptides as biomarkers for cerebral amyloid angiopathy

Abstract

Brain amyloid-β (Aβ) deposits are key pathological hallmarks of both cerebral amyloid angiopathy (CAA) and Alzheimer's disease (AD). Microvascular deposits in CAA mainly consist of the Aβ₄₀ peptide, whereas Aβ₄₂ is the predominant variant in parenchymal plaques in AD. The relevance in pathogenesis and diagnostic accuracy of various other Aβ isoforms in CAA remain understudied. We aimed to investigate the biomarker potential of various Aβ isoforms in cerebrospinal fluid (CSF) to differentiate CAA from AD pathology. We included 25 patients with probable CAA, 50 subjects with a CSF profile indicative of AD pathology (AD-like), and 23 age- and sex-matched controls. CSF levels of Aβ_1-34, Aβ_1-37, Aβ_1-38, Aβ_1-39, Aβ_1-40, and Aβ_1-42 were quantified by liquid chromatography mass spectrometry. Lower CSF levels of all six Aβ peptides were observed in CAA patients compared with controls (p = 0.0005-0.03). Except for Aβ_1-42 (p = 1.0), all peptides were decreased in CAA compared with AD-like subjects (p = 0.007-0.03). Besides Aβ_1-42, none of the Aβ peptides were decreased in AD-like subjects compared with controls. All Aβ peptides combined differentiated CAA from AD-like subjects better (area under the curve [AUC] 0.84) than individual peptide levels (AUC 0.51-0.75). Without Aβ_1-42 in the model (since decreased Aβ_1-42 served as AD-like selection criterion), the AUC was 0.78 for distinguishing CAA from AD-like subjects. CAA patients and AD-like subjects showed distinct disease-specific CSF Aβ profiles. Peptides shorter than Aβ_1-42 were decreased in CAA patients, but not AD-like subjects, which could suggest different pathological mechanisms between vascular and parenchymal Aβ accumulation. This study supports the potential use of this panel of CSF Aβ peptides to indicate presence of CAA pathology with high accuracy.

Keywords: Alzheimer's disease; amyloid‐β; biomarkers; cerebral amyloid angiopathy; cerebrospinal fluid; mass spectrometry.

Figure 1.

Cerebrospinal fluid levels (pg/ml) of amyloid-β (Aβ) peptides in controls (n=23 for Aβ_1–34, n=22 for other peptides), cerebral amyloid angiopathy (CAA) patients (n=25), and Alzheimer’s disease-like (AD-like) subjects (n=50). Statistical comparison was performed with analysis of variance with Bonferroni’s post hoc test (for Aβ_1–34, Aβ_1–37, Aβ_1–38, Aβ_1–39, and Aβ_1–40), or Kruskal-Wallis with Dunn’s post hoc test (for Aβ_1–42), as appropriate. P values: *p ≤ .05, **p ≤ .01, ***p ≤ .001, ****p ≤ .0001. Median values and interquartile range are indicated.

Figure 2.

Receiver operator characteristics curves for discrimination of (A) cerebral amyloid angiopathy (CAA) patients from controls and (B) CAA patients from Alzheimer’s disease-like (AD-like) subjects. Area under the curve (AUC) was calculated for all individual amyloid-β (Aβ) peptides, and for the combination of all six peptides.

Figure 3.

Correlations of individual amyloid-β (Aβ) peptide levels and Aβ_42/40 ratio with Montreal Cognitive Assessment (MoCA) score and cerebrovascular imaging markers in patients with cerebral amyloid angiopathy (CAA). Spearman rank correlation coefficients are displayed. MoCA was available in a subset (n=21) of CAA patients. Asterisk indicates a significant p value (≤0.05). Abbreviations: CMB, cerebral microbleeds; cSS, cortical superficial siderosis; EPVS, enlarged perivascular spaces; ICH, intracerebral haemorrhage; SVD, small vessel disease; WMH, white matter hyperintensities.