Thimet oligopeptidase as a potential CSF biomarker for Alzheimer's disease: A cross-platform validation study

Abstract

Introduction: Our previous antibody-based cerebrospinal fluid (CSF) proteomics study showed that Thimet oligopeptidase (THOP1), an amyloid beta (Aβ) neuropeptidase, was increased in mild cognitive impairment with amyloid pathology (MCI-Aβ+) and Alzheimer's disease (AD) dementia compared with controls and dementia with Lewy bodies (DLB), highlighting the potential of CSF THOP1 as an early specific biomarker for AD. We aimed to develop THOP1 immunoassays for large-scale analysis and validate our proteomics findings in two independent cohorts.

Methods: We developed in-house CSF THOP1 immunoassays on automated Ella and Simoa platforms. The performance of the different assays were compared using Passing-Bablok regression analysis in a subset of CSF samples from the discovery cohort (n = 72). Clinical validation was performed in two independent cohorts (cohort 1: n = 200; cohort 2: n = 165) using the Ella platform.

Results: THOP1 concentrations moderately correlated between proteomics analysis and our novel assays (Rho > 0.580). In both validation cohorts, CSF THOP1 was increased in MCI-Aβ+ (>1.3-fold) and AD (>1.2-fold) compared with controls; and between MCI-Aβ+ and DLB (>1.2-fold). Higher THOP1 concentrations were detected in AD compared with DLB only when both cohorts were analyzed together. In both cohorts, THOP1 correlated with CSF total tau (t-tau), phosphorylated tau (p-tau), and Aβ40 (Rho > 0.540) but not Aβ42.

Discussion: Validation of our proteomics findings underpins the potential of CSF THOP1 as an early specific biomarker associated with AD pathology. The use of antibody-based platforms in both the discovery and validation phases facilitated the translation of proteomics findings, providing an additional workflow that may accelerate the development of biofluid-based biomarkers.

Keywords: Alzheimer's disease; CSF; THOP1; biomarkers.

FIGURE 1

THOP1 assay was developed and analytically validated on the Ella platform. (A) The CV% of CSF samples with triplicate or duplicate measurements is plotted against the average THOP1 concentration. Precision plots show that all samples had a CV% <20 and all samples were above the LLOD of 1.78 pg/mL. (B) THOP1 Ella assay showed the THOP1 signal in CSF samples (log‐transformed relative fluorescence unit; RFU signal) following a two‐fold serial dilution (reciprocal relative dilution, log‐transformed) was parallel to the signal obtained from the standard curve. (C) RFU signal of CSF samples measured in a serial dilution shows that the mean %linearity of the assay is within range (85%–115%). Graph is plotted with a log‐transformed y‐axis. (D) % Recovery of low, medium, and high spiked CSF samples measured on the Ella platform detects that only the medium and high spiked CSF samples were within range. (E) THOP1 concentrations in CSF samples were normalized to reference condition with zero freeze and thaw (f/t) cycles and presented in % as the mean of three samples, which showed stable protein concentrations up to seven f/t cycles. Dashed lines show the acceptance range of 85% to 115%. Error bars in D and E represent the standard deviation of the four and three CSF samples measured, respectively. Abbrevations: CSF, cerebrospinal fluid; CV, coefficient of variation; LLOD, lower limit of detection; THOP1, thimet oligopeptidase.

FIGURE 2

THOP1 assay was developed and analytically validated on the Simoa platform. (A) The CV% of CSF samples with duplicate measurements are plotted against the average THOP1 concentration. Precision plots show that three samples had a CV% >20 and all samples were above the LLOD of 5.2 pg/mL. (B) THOP1 concentrations on the Simoa platform show that levels in CSF samples following a two‐fold serial dilution are parallel to the signal obtained from the standard curve. (C) AEB signals of CSF samples measured in a serial dilution show a matrix effect in dilutions 1–4. Upon further diluting, the % linearity is within the acceptable range. Graph is plotted with a log‐transformed y‐axis. (D) % Recovery of low, medium, and high spiked CSF samples measured on the Simoa platform show that all samples are within range. (E) THOP1 concentrations are stable up to two freeze and thaw (f/t) cycles. Dashed lines show the acceptance range of 85% to 115%. Error bars in D and E represent the standard deviation of the four or three CSF samples measured, respectively. Abbrevations: AEB, average enzymes per bead; CSF, cerebrospinal fluid; CV, coefficient of variation; LLOD, lower limit of detection; THOP1, thimet oligopeptidase.

FIGURE 3

THOP1 concentrations are translatable across different platforms. THOP1 concentrations in CSF are increased in patients with AD compared to controls and patients with DLB on three different platforms; antibody‐based proteomics (A), Ella (B), and Simoa (C). THOP1 concentrations strongly correlated between Ella and antibody‐based proteomics (D) and moderately between Simoa and antibody‐based proteomics (E). The novel THOP1 assays correlated strongly between the automated Ella and Simoa platforms (F). Abbrevations: AD, Alzheimer's disease; CSF, cerebrospinal fluid; DLB, dementia with Lewy bodies; THOP1, thimet oligopeptidase. ^***indicates p < 0.001.

FIGURE 4

THOP1 concentrations are increased in MCI‐Aβ+ and AD and associated with Aβ40, p‐tau, and t‐tau in both validation cohorts. THOP1 concentrations were measured on the Ella platform, which showed increased THOP1 concentrations in AD compared to MCI‐Aβ+ compared to controls and patients with DLB in validation cohort 1 (A) and validation cohort 2 (C). The correlation matrix heatmap represents Spearman's correlation coefficient of THOP1 with the classical AD CSF biomarkers and MMSE scores. The blue color depicts a positive correlation coefficient, whereas red depicts a negative correlation coefficient. Significant correlations between THOP1 and Aβ40, p‐tau, and t‐tau in validation cohort 1 (B) and validation cohort 2 (D) were observed. Abbrevations: AD, Alzheimer's disease; Aβ40, amyloid beta 1‐40; Aβ42, amyloid beta 1‐42; CSF, cerebrospinal fluid; DLB, dementia with Lewy bodies; MCI‐Aβ+, mild cognitive impairment with amyloid pathology; MMSE, Mini‐Mental State Examination; p‐tau, phosphorylated tau; THOP1, thimet oligopeptidase; t‐tau, total tau. Statistical significance is indicated as: ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001.