Background

Blood-based biomarkers (BBMs) are emerging as accessible, non-invasive and cost-effective tools for the diagnosis and prognosis of neurodegenerative diseases. However, the susceptibility of various protein biomarkers to variation when re-sampled within the same individual is not well understood. In this study we assessed the test-retest reliability of plasma measurements analysed on the NULISAseq CNS disease panel from two serial samples taken from the same individuals.

 

Methods

Paired plasma samples collected within 3 months from 93 participants enrolled in studies at the Dementia Research Centre, UCL, were tested in singlicate using the NULISAseq CNS kit (Alamar). Both cognitively impaired and unimpaired participants were included. 41 sample pairs were tested using a previous panel version lacking seven analytes (BD-MAPT, BD-ptau181, BD-ptau217, BD-ptau231, DDC, UBB and YWHAG). Test-retest reliability was assessed using Spearman’s correlation coefficient (ρ). Agreement and subsequent disagreement reflecting potential systematic or proportional bias were evaluated using Passing–Bablok regression, with 95% confidence intervals estimated using a bootstrap (quantile) method.

 

Results

Participants had a mean age of 52 years (SD = 15.7); 43 were female. The cohort comprised 22 individuals with Alzheimer’s disease (AD), 8 with other dementias, 27 presymptomatic autosomal dominant AD mutation carriers, and 36 controls. The median inter-sampling interval was 85 mins (range: 30mins-92 days). Strong correlations (ρ ≥ 0.6) were observed for 104 of 131 analytes, moderate correlations (ρ = 0.4–0.6) for 20 analytes, and low correlations (ρ ≤ 0.4) for 7 analytes (Figure 1). Passing-Bablok regression demonstrated no evidence of disagreement or bias for 120 analytes. Core AD biomarkers showed high correlations (ρ ≥ 0.8); however, NfL and GFAP demonstrated evidence of systematic differences between timepoints despite strong correlation (Table 1, Figure 2).

 

Conclusion

Core AD biomarkers measured using the NULISAseq CNS disease panel demonstrated excellent short-interval test-retest reliability. However, NfL and GFAP showed slight disagreement across timepoints, highlighting the importance of assessing agreement in addition to correlation. Several analytes, including YWHAG, had low test-retest correlation, suggesting greater susceptibility to biological variability, pre-analytical handling, or individual characteristics, requiring further study. These findings emphasise the importance of reliability assessments for highly multiplex proteomic platforms.