Concordance between amyloid PET and CSF biomarkers in clinical setting: A cross-platform comparison and in-depth analysis of discordant cases

Background: Accurate detection of amyloid pathology is essential for the diagnosis and treatment of Alzheimer’s disease (AD), especially with the growing use of amyloid-targeting therapies that require biomarker confirmation. Amyloid positron emission tomography (PET) and cerebrospinal fluid (CSF) biomarkers both reflect beta-amyloid accumulation but can yield discordant results. This study aimed to assess the concordance between quantitative amyloid PET and various CSF assays, including both standard enzyme-linked immunosorbent assays (ELISA) and the fully automated LUMIPULSE® G600II analyzer, and to analyze discordant cases in a clinical memory clinic setting. Methods: A total of 153 participants from the Czech Brain Aging Study underwent amyloid PET with flutemetamol and lumbar puncture for CSF biomarker analysis. CSF samples were tested using two ELISA kits (Innogenetics and Euroimmun) and the LUMIPULSE® analyzer. PET images were visually interpreted by expert readers and centiloid values were derived from quantitative PET analysis. Receiver operating characteristic (ROC) curves and area under the curve (AUC) values were calculated to compare CSF analytes and ratios against PET status at different centiloid (CL) thresholds (20 and 30 CL). Overall percent agreement (OPA) and logistic regression models were used to explore predictors of PET/CSF discordance. Results: The p-tau181/Aβ42 ratio demonstrated the best overall agreement with PET status across both ELISA platforms (OPA = 87%) and the LUMIPULSE® analyzer (OPA = 92%). Concordance between visual PET assessment and centiloid quantification was excellent (AUC = 0.99), with 100% sensitivity and 97% specificity at a threshold of 20.9 CL. A subgroup of 6% of participants showed consistent PET/CSF discordance not explained by assay variability. These “true discordant” cases were associated with APOE ε4 carrier status and mixed or non-AD pathologies, including progressive supranuclear palsy, vascular dementia and frontotemporal lobar degeneration. Conclusions: This study supports the strong concordance between CSF hybrid ratios and quantitative amyloid PET, as well as the reliability of expert visual PET interpretation. Automated CSF assays enhance early detection and minimize variability. Discordant cases likely reflect underlying biological heterogeneity and warrant careful interpretation, particularly when considering treatment initiation in symptomatic individuals.