We assessed 173 participants from the Translational Biomarkers of Aging and Dementia (TRIAD) [23] cohort: 23 cognitively unimpaired young adults (CUY), 74 cognitively unimpaired (CU) older adults, 36 individuals with Mild Cognitive Impairment (MCI), 24 individuals with Alzheimer’s clinical syndrome (AD) and 16 participants with other neurodegenerative diseases (OND). All participants had CSF assessments of p-tau181, p-tau217 and p-tau231, from both immunoassays and mass spectrometry. All participants were also evaluated with amyloid-PET with [18F]AZD4694 and tau-PET with [18F]MK6240. Clinical evaluations of participants included a review of their medical history and an interview with the participant and their study partner, a neurological examination by a physician and a neuropsychological examination. Participants were approached consecutively, and data was collected prospectively from October 2017 to August 2021. CU individuals had no objective cognitive impairment and a Clinical Dementia Rating (CDR) score of 0. Cognitively impaired (CI) participants had objective cognitive impairment and a CDR score of 0.5, 1 or 2. Participants were excluded from this study if they had systemic conditions which were not adequately controlled through a stable medication regimen. Other exclusion criteria were active substance abuse, recent head trauma, recent major surgery, or MRI/PET safety contraindications. The study was approved by the Montreal Neurological Institute PET working committee and the Douglas Mental Health University Institute Research Ethics Board. Written informed consent was obtained for all participants. The present study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.

We assessed 394 individuals from the prospective BioFINDER-2 study. This group comprised individuals with mild cognitive impairment (MCI), AD with dementia, various other neurodegenerative conditions, and cognitively unimpaired (CU) individuals. Individuals with AD met the diagnostic criteria outlined in the Diagnostic and Statistical Manual of Mental Disorders [Fifth Edition] [24] in addition to having positive amyloid-beta (Aβ) biomarker results [2]. The inclusion criteria for other neurodegenerative diseases encompassed meeting the criteria for frontotemporal dementia, Parkinson’s disease (PD), PD with dementia, subcortical vascular dementia, progressive supranuclear palsy, multiple system atrophy, or semantic variant primary progressive aphasia, as previously described [25]. CU participants had to not meet criteria for MCI or dementia, showing no history of cognitive decline over time and possessing a CDR score of 0. Recruitment occurred at Skåne University Hospital between April 2017 and September 2019. All participants underwent the Mini-Mental State Examination to assess overall cognition. Ethical approval was granted by the Regional Ethical Committee in Lund, Sweden.

Collection of CSF samples has been reported previously for the TRIAD cohort [15] and BioFINDER-2 cohort [10]. In the TRIAD cohort, CSF concentrations of p-tau181, p-tau217 and p-tau231 were quantified using custom Single molecule array (Simoa; Simoa HD-X instruments, Quanterix, Billerica, MA, USA) assays, as previously described [13, 26]. In the BioFINDER-2 cohort, antibody-based CSF quantification of p-tau181 and p-tau217 were performed at Eli Lilly using the Meso Scale Discovery (MSD) platform, and p-tau231 was quantified by ELISA [10]. For both the TRIAD and BioFINDER-2 cohorts, antibody-free mass spectrometry-based quantification of p-tau was performed using liquid chromatography-mass spectrometry (LC–MS) from a 300 µl sample as described previously [27]. Briefly, CSF samples of 250 µl were spiked with a 10 µl heavy isotope-labeled peptide standards (AQUA peptides, Thermo Scientific) mixture. The spike-in amount of each heavy peptide was modified to yield a light-to heavy peak area ratio of approximately 0.1 – 0.2 in CSF from subjects without AD. The peptide standards were diluted by mixing 10 pmol lyophilized aliquots with 20% acetonitrile. The final 1:10 dilution was performed in 50 mM ammonium bicarbonate to prevent acetonitrile interference during sample preparation. Protein precipitation was performed by adding perchloric acid (15 µl, 60% v/v) to the samples, which then were briefly vortexed and incubated on ice for 15 min. Under such circumstances, a majority of CSF proteins precipitate, though tau does not. The precipitated proteins were then pelleted by centrifugation at 30,000 × g for 10 min at 4 °C, and the supernatants were transferred to a 96-well filter microtitre plate (AcroPrep Advance, 350 µl, 0.45 µm, Supor membrane, Pall Corporation). A vacuum manifold was employed to pass samples through the filter plate and load them on a 96-well SPE plate (Oasis PRiME HLB 96-well µElution Plate, 3 mg Sorbent per Well, Waters). The SPE plate was washed in duplicate with 200 µl 5% methanol (v/v), and peptides were eluted into a microtitre plate with 200 µl 50% acetonitrile, 0.1% trifluoroacetic acid, and the eluates were lyophilized by vacuum centrifugation. Trypsin (Sequencing grade, Promega) was dissolved in the manufacturer diluent and diluted to 2.5 µg/ml in 50 mM ammonium bicarbonate. A 40 µl trypsin solution was added to the dry samples, which were then vortexed and incubated at 37 °C overnight. TFA (1 µl, 10% v/v) was added to the samples to quench additional proteolysis. The samples were then stored at -20 °C until LC–MS analysis. The tryptic peptides measured are described in the supplement. A parallel reaction monitoring (PRM) assay was used on a Hybrid Orbitrap mass spectrometer (Fusion Tribrid, Thermo Scientific). Single-point calibration was performed by adding internal heavy labeled peptides with the same sequence as the targeted peptides at a known concentration. Sample preparation took 2 days, with every sample requiring one hour to be analyzed in the Mass Spectrometer. These mass spectrometry measurements allow for the quantification of concentrations as low as in fmol/ml range, with an absolute precision of 0-2 ppm for quantified peptides. LC–MS data was analyzed using the Skyline v. 21 software package (MacCoss Lab, University of Washington, USA). Mass spectrometry-based measurements of p-tau residues in CSF were performed at the Clinical Neurochemistry Laboratory, University of Gothenburg by scientists blinded to clinical and biomarker information.

[18F]AZD4694 PET and [18F]MK6240 PET scans were obtained using a Siemens High Resolution Research Tomograph. [18F]AZD4694 PET images were obtained 40–70 min post- injection and reconstructed on a 4-dimensional volume with 3 frames (3 × 600 s), as previously described [28]. [18F]MK6240 PET images were acquired at 90–110 min post-injection and reconstructed on a 4-dimensional volume with 4 frames (4 × 300 s) [29]. MRI acquisition and processing has been described previously [30]. To minimize interference of meningeal spillover, [18F]MK6240 images were meninges-striped before they were blurred, as described previously [31]. [18F]AZD4694 standardized uptake value ratio (SUVR) maps were calculated using the whole cerebellum gray matter as the reference region and [18F]MK6240 SUVR maps were generated using the inferior cerebellar grey matter as a reference region. Spatial smoothing allowed the PET images to achieve an 8-mm full-width at half-maximum resolution. Amyloid-β SUVR from a neocortical region of interest (ROI) for each participant was estimated by averaging the SUVR from the precuneus, prefrontal, orbitofrontal, parietal, temporal, and cingulate cortices [28], with amyloid-β positivity defined as an [18F]AZD4694 above 1.55 [28]. Tau-PET SUVRs were calculate in regions comprising Braak stages I-IV as previously described [30].

Details regarding PET image acquisition and processing in the BioFINDER-2 study have been documented previously [25]. Briefly, amyloid-PET and tau-PET scans were obtained using [18F]flutemetamol and [18F]RO948 radiotracers, respectively. Amyloid-PET binding was quantified using a standardized uptake value ratio (SUVR) with a neocortical meta-region of interest normalized to the cerebellar grey matter. Tau-PET binding was assessed within a meta-ROI covering temporal brain regions. The reference region used for tau-PET quantification was the inferior cerebellar cortex. Furthermore, tau-PET binding was also evaluated in regions comprising Braak stages I-IV as previously described [27].

Statistical analyses were performed in R v4.1.1 and GraphPad Prism v9. CSF biomarkers (immunoassay and mass spectrometry) were compared between indicated groups by parametric t-test with FDR-correction for multiple testing. Effect sizes of group differences between amyloid-PET-positive and -negative individuals were determined using Cohen’s d. We also looked at the mean fold change between amyloid-PET-positive and -negative groups for all p-tau biomarkers. Bland–Altman analyses assessed the agreement between measurements from mass spectrometry and immunoassays. Area under the Receiver Operating Characteristic (ROC) curve values were calculated for all p-tau biomarkers. Nonparametric Spearman correlation coefficients were calculated for PET biomarker associations due to the non-normal distribution of the data. We selected PET biomarkers as the reference standard instead of clinical diagnosis in accordance to the biological definition of AD [2]. Comparison of correlation coefficients was performed using the cocor package in R and Zhou’s test was used to determine 95% CIs of differences [32]. Statistical differences in area under the ROC curves were tested with DeLong’s test using the pROC package in R [33].