Diagnostic Performance of the α-Synuclein Seed Amplification Assay for Dementia With Lewy Bodies: A Comparison Across 4 Laboratories

Background and objectives: The α-synuclein (α-syn) seed amplification assay (SAA) has shown promising results for diagnosing dementia with Lewy bodies (DLB) using CSF samples. A barrier to implementing α-syn SAA clinically is the use of different protocols for the assay. It is unknown how different protocols perform in comparison with each other. We compared the performance of α-syn SAA across 4 laboratories using CSF samples of patients with DLB. Methods: This was a retrospective cross-sectional study that included data from 4 different European laboratories. We included probable patients with DLB with a positive dopamine transporter (DaT)-SCAN and known amyloid-β status who had mild-to-moderate dementia, along with age-matched and sex-matched controls. The α-syn SAA was run across 4 laboratories using different protocols varying α-syn concentration and plate reader settings. CSF samples were provided by a fifth independent laboratory, which also performed statistical and result analyses. Results: We included 20 patients with DLB (mean age 67 ± 6 years, 60% male) and 10 controls (mean age 67 ± 2 years, 70% male). Neuropathologic confirmation was available for 2 patients with DLB. On average, the 4 laboratories achieved 78.8% sensitivity (minimum 55%, maximum 100%), 77.5% specificity (minimum 60%, maximum 100%), and 78.5% accuracy (minimum 57%, maximum 100%) for discriminating DLB from controls, but our findings show that diagnostic performance of SAA varied across laboratories: Lab A achieved 100% sensitivity (CI 84%-100%) and 100% specificity (CI 72%-100%); Lab B achieved 85% sensitivity (CI 64%-95%) and 90% specificity (CI 59%-99%); Lab C achieved 55% sensitivity (CI 34%-74%) and 60% specificity (CI 31%-83%); and Lab D achieved 75% sensitivity (CI 53%-89%) and 60% specificity (CI 31%-83%). In general, SAA results showed numerically lower sensitivity in β-amyloid (Aβ)-positive patients with DLB (70%) compared with Aβ-negative patients with DLB (87.5%) (nonstatistically significant). A fair agreement of SAA results was obtained across the 4 laboratories (average κ = 0.246). Discussion: This study highlights challenges for the reproducibility of α-syn SAA results across different protocols applied by different laboratories. This finding, together with the methodological variability reported across laboratories, may challenge the clinical implementation of the α-syn SAA. This study provides relevant support for initiating harmonization and standardization of SAA protocols to move the field toward the clinical implementation of SAAs for the biomarker-based diagnosis of DLB. Classification of evidence: This study provides Class III evidence of variations in the accuracy of CSF α-syn SAA across 4 separate laboratories in distinguishing patients with DLB from healthy controls.