Point of care measurement of blood tau in Alzheimer’s dementia

Abstract Background Several assays reliably measure cerebrospinal fluid (CSF) total tau (t‐tau) protein for Alzheimer’s diseases (AD) diagnosis. However, CSF sampling is invasive and hampers availability of biomarker testing in AD. Therefore, less invasive blood t‐tau assays are needed, preferably on different platforms to enable implementation in different clinical settings. Point‐of‐care (POC) platforms enable the development of practical and easily implemented assays. Here, we aim to develop and compare two blood t‐tau POC‐assays, (1)on the Ella microfluidics platform and (2)using upconversion‐nanoparticle based lateral flow (UCNP‐LF), for use in AD diagnosis. Method The prototype assays were developed with the same pair of antibodies (capture‐mAb: ADx205 and detector‐mAb: ADx204 having epitopes in respectively the regions aa 194‐204 and 1‐20) and were calibrated using a recombinant protein (2N4R tau). For the Ella‐assay, we validated the analytical sensitivity, precision (4 different runs) and parallelism. The acceptance range was set to <15% coefficient of variation (CV) for precision and 85‐115% of slope‐accuracy for parallelism. Detectability of t‐tau was assessed using 20 plasma samples, including 10 AD‐dementia patients (age: 64±9y, 60%F) and 10 controls (age: 67±8y, 50%F). For the UCNP‐LF assay, spiked buffer and serum were used for preliminary performance testing. Result For the Ella assay, we determined a limit of detection (LoD) of 3 pg/mL based on mean signal of 16 blanks plus 10x SD (figure1). The assay quantified t‐tau in the clinical samples above assay blank with an average intra‐assay CV of 5.3%. Average intra‐ and inter‐assay CV for QC plasma samples were 8% and 11.4%. The assay showed robust mean parallelism response of (95%) (figure 2). Median t‐tau concentration in AD samples was 36.9 pg/mL vs 8 pg/mL in controls. The prototype UCNP‐LF assay showed a LoD of 92 pg/mL, based upon spiked buffer with the recombinant protein, and 102 pg/mL for spiked serum (figure3). The UCNP‐LF assay has a read‐out time of 30 minutes. Conclusion The prototype assays showed promising preliminary performance. Next, a technology comparison using Bland‐Altman graphs on samples of at least two different clinical groups, would be measured using a reference immunoassay platform, such as Simoa.