Establishment of updated biomarker cut‐off values for the second‐generation Elecsys β‐amyloid(1–42), pTau and tTau CSF immunoassays

Background Compared with the first‐generation (Gen1), second‐generation (Gen2) Elecsys ® cerebrospinal fluid (CSF) immunoassays (currently under development, Roche Diagnostics International Ltd) for detection of β‐amyloid(1–42) (Abeta42), phospho‐tau(181P) (pTau) and total‐tau (tTau) have higher thresholds for biotin interference. Abeta42 Gen2 is re‐standardized to Abeta42 certified reference materials ERM‐DA480/IFCC, ERM‐DA481/IFCC and ERM‐DA482/IFCC and has an extended measurement range. Additionally, the newly developed pre‐analytical ‘Routine‐Use’ protocol provides more reliable analyte recovery for Abeta42 against comparators (Hansson et al., Alzheimers Dement. 2020;16(S3):e047394). Here, we establish adjusted cut‐off values for concordance of Abeta42, pTau/Abeta42 and tTau/Abeta42 Gen2 results with amyloid‐PET based on those previously established for Gen1, accounting for different pre‐analytical protocols. Method CSF samples from 26 patients with normal pressure hydrocephalus were collected per the Routine‐Use protocol for fresh CSF (n=3 aliquots/patient) and two clinical trial protocols (BioFINDER and Roche‐Trial) for frozen CSF (<–60°C; n=4 aliquots/patient). Abeta42, pTau and tTau were quantified using Elecsys Gen1/Gen2 immunoassays (cobas e 601 analyzer; Roche Diagnostics International Ltd). Bias at medical decision point was estimated for comparison of Gen1 and Gen2 immunoassays (Roche‐Trial protocol) by weighted‐Deming regression. Original cut‐off values were determined in the BioFINDER cohort using Gen1 immunoassays (Hansson et al.). Adjusted cut‐off values for Gen2 immunoassays with both Routine‐Use and Roche‐Trial protocols were determined. Concentration bias between Routine‐Use and Roche‐Trial protocols was also investigated, measured using Gen2 immunoassays. Mean percentage differences provided estimates of overall bias between protocols and protocol‐assay combinations. Result No clinically meaningful differences between assay generations and pre‐analytical protocols were obtained for pTau or tTau. Bias at medical decision point for Abeta42 Gen2 immunoassay versus Gen1 was –20% (95% confidence interval [CI]: –22%, –18%). Bias between concentrations of Abeta42 measured using BioFINDER/Gen1 versus Routine‐Use/Gen2 was –6.3% (95% CI: –8.7%, –3.9%). Cut‐off values for Routine‐Use/Gen2 were adjusted based on this result (). Bias between Routine‐Use versus Roche‐Trial protocols for Abeta42 Gen2 was –22% (95% CI: –24%, –19%). Cut‐off values for Roche‐Trial/Gen2 were adjusted based on this result. Conclusion Elecsys CSF Abeta42 results were independently affected by immunoassay generation and pre‐analytical protocol, while tau results were unaffected. Appropriate adjusted cut‐off values have been established.