Clin Chem Lab Med 2022; 60(2): 207–219 Johan Gobom*, Lucilla Parnetti, Pedro Rosa-Neto, Martin Vyhnalek, Serge Gauthier, Samuela Cataldi, Ondrej Lerch, Jan Laczo, Katerina Cechova, Marcus Clarin, Andrea L. Benet, Tharick A. Pascoal, Neserine Rahmouni, Manu Vandijck, Else Huyck, Nathalie Le Bastard, Jenna Stevenson, Mira Chamoun, Daniel Alcolea, Alberto Lleó, Ulf Andreasson, Marcel M. Verbeek, Giovanni Bellomo, Roberta Rinaldi, Nicholas J. Ashton, Henrik Zetterberg, Katerina Sheardova, Jakub Hort and Kaj Blennow Validation of the LUMIPULSE automated immunoassay for the measurement of core AD biomarkers in cerebrospinal fluid https://doi.org/10.1515/cclm-2021-0651 Abstract Received January 1, 2021; accepted November 2, 2021; published online November 15, 2021 Objectives: The core cerebrospinal fluid (CSF) bio- markers; total tau (tTau), phospho-tau (pTau), amyloid *Corresponding author: Johan Gobom, Department of Psychiatry and Manu Vandijck, Else Huyck and Nathalie Le Bastard, Fujirebio Europe Neurochemistry, Institute of Neuroscience and Physiology, The N.V., Ghent, Belgium Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; Daniel Alcolea and Alberto Lleó, Department of Neurology, Memory and Clinical Neurochemistry Laboratory, Sahlgrenska University Unit, Hospital de la Santa Creu i Sant Pau- Biomedical Research Hospital, Mölndal, Sweden, E-mail: johan.gobom@neuro.gu.se. Institute Sant Pau-Universitat Autònoma de Barcelona, Barcelona, https://orcid.org/0000-0001-6193-6193 Spain; and Centro de Investigación Biomédica en Red Enfermedades Lucilla Parnetti, Samuela Cataldi, Giovanni Bellomo and Roberta Neurodegenerativas (CIBERNED), Madrid, Spain Rinaldi, Laboratory of Clinical Neurochemistry, Section of Neurology, Marcel M. Verbeek, Department of Laboratory Medicine, Radboud University of Perugia, Perugia, Italy University Medical Center, Nijmegen, The Netherlands; and Pedro Rosa-Neto and Serge Gauthier, Department of Neurology and Department of Neurology, Radboud Alzheimer Centre, Radboud Neurosurgery, McGill University Research Centre for Studies in Aging, University Medical Center, Donders Institute for Brain, Cognition and Douglas Research Institute, Le Centre intégré universitaire de santé et Behaviour, Nijmegen, The Netherlands de services sociaux (CIUSSS) de l’Ouest-de-l’Île-de-Montréal, Nicholas J. Ashton, Department of Psychiatry and Neurochemistry, Psychiatry and Pharmacology and Therapeutics, McGill University, Institute of Neuroscience and Physiology, The Sahlgrenska Academy, Montreal, QC, Canada; and Montreal Neurological Institute, Montreal, University of Gothenburg, Gothenburg, Sweden; Wallenberg Centre QC, Canada for Molecular and Translational Medicine, University of Gothenburg, Martin Vyhnalek, Ondrej Lerch, Jan Laczo and Katerina Cechova, Gothenburg, Sweden; King’s College London, Institute of Psychiatry, Department of Neurology, Second Medical Faculty, Charles University, Psychology & Neuroscience, Maurice Wohl Clinical Neuroscience Prague, Czech Republic; Motol University Hospital, Prague, Czech Institute, London, UK; and NIHR Biomedical Research Centre for Republic; and International Clinical Research Center, St. Anne’s Mental Health & Biomedical Research Unit for Dementia at South University Hospital, Brno, Czech Republic. London & Maudsley NHS Foundation, London, UK https://orcid.org/0000-0002-5845-1854 (K. Cechova) Henrik Zetterberg, Department of Psychiatry and Neurochemistry, Marcus Clarin, Ulf Andreasson and Kaj Blennow, Department of Institute of Neuroscience and Physiology, The Sahlgrenska Academy, Psychiatry and Neurochemistry, Institute of Neuroscience and University of Gothenburg, Gothenburg, Sweden; Clinical Physiology, The Sahlgrenska Academy, University of Gothenburg, Neurochemistry Laboratory, Sahlgrenska University Hospital, Gothenburg, Sweden; and Clinical Neurochemistry Laboratory, Mölndal, Sweden; Department of Neurodegenerative Disease, UCL Sahlgrenska University Hospital, Mölndal, Sweden Institute of Neurology, London, UK; and UK Dementia Research Andrea L. Benet, Department of Psychiatry and Neurochemistry, Institute at UCL, London, UK Institute of Neuroscience and Physiology, The Sahlgrenska Academy, Katerina Sheardova, Department of Neurology, Second Medical University of Gothenburg, Gothenburg, Sweden; and Translational Faculty, Charles University, Prague, Czech Republic; and First Neuroimaging Laboratory, McGill Centre for Department of Neurology, Faculty of Medicine, Masaryk University and Studies in Aging, McGill University, Montreal, St. Anne’s University Hospital, Brno, Czech Republic QC, Canada Jakub Hort, Department of Neurology, Second Medical Faculty, Tharick A. Pascoal, Neserine Rahmouni, Jenna Stevenson and Mira Charles University, Prague, Czech Republic; Motol University Hospital, Chamoun, Translational Neuroimaging Laboratory, McGill Centre for Prague, Czech Republic; and First Department of Neurology, Faculty of Studies in Aging, McGill University, Montreal, Medicine, Masaryk University and St. Anne’s University Hospital, QC, Canada Brno, Czech Republic Open Access. © 2021 Johan Gobom et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 International License. 208 Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF β 1-42 (Aβ 1-42), and the Aβ 1-42/Aβ 1-40 ratio have trans- amyloid precursor protein [5]. Tau in its native form, i.e. formed Alzheimer’s disease (AD) research and are today without abnormal phosphorylation, is a protein found increasingly used in clinical routine laboratories as diag- intracellularly and is involved in the stabilization of the nostic tools. Fully automated immunoassay instruments microtubules [6]. with ready-to-use assay kits and calibrators has simplified In AD, the typical biomarker pattern in cerebrospinal their analysis and improved reproducibility of measure- fluid (CSF) is a combination of decreased levels of Aβ 1-42 ments. We evaluated the analytical performance of the and increased levels of pTau and total tau (tTau) [4]. CSF fully automated immunoassay instrument LUMIPULSE G tTau is measured using assays based on antibodies that (Fujirebio) for measurement of the four core AD CSF bio- detect all tau isoforms, independently of phoysphorylation markers and determined cutpoints for AD diagnosis. status [4]. These biomarkers are used extensively in Methods: Comparison of the LUMIPULSE G assays was research but have not yet become included in the clinical performed with the established INNOTEST ELISAs (Fujir- diagnostic criteria for AD, although clinicians use them ebio) for hTau Ag, pTau 181, β-amyloid 1-42, and with already now as support for the diagnosis [7]. The Aβ 1-42 V-PLEX Plus Aβ Peptide Panel 1 (6E10) (Meso Scale Dis- concentration may be normalized to the concentration of covery) for Aβ 1-42/Aβ 1-40, as well as with a LC-MS refer- the 40-amino acid-long form of beta-amyloid (Aβ 1-40) to ence method for Aβ 1-42. Intra- and inter-laboratory obtain the Aβ 1-42/Aβ 1-40 ratio, which has proved to yield reproducibility was evaluated for all assays. Clinical cut- improved diagnostic performance compared to Aβ 1-42 points for Aβ 1-42, tTau, and pTau was determined by alone [8]. While the ratio decreases the impact of pre- analysis of three cohorts of clinically diagnosed patients, analytical sources of errors, e.g., adsorption of peptides to comprising 651 CSF samples. For the Aβ 1-42/Aβ 1-40 ratio, CSF collection tubes and pipetting inaccuracy, it has also the cutpoint was determined by mixture model analysis of been suggested that it may serve to compensate for phys- 2,782 CSF samples. iological variation in the expression and processing of the Results: The LUMIPULSE G assays showed strong correla- amyloid precursor protein [8, 9]. tion to all other immunoassays (r>0.93 for all assays). The Enzyme-linked immunosorbent assays (ELISA), such as repeatability (intra-laboratory) CVs ranged between 2.0 and the INNOTEST β-amyloid 1-42, hTau Ag, and PHOSPHO- 5.6%, with the highest variation observed for β-amyloid TAU(181P) assays, have for a long time been used to measure 1-40. The reproducibility (inter-laboratory) CVs ranged the CSF biomarkers. A drawback of manually performed between 2.1 and 6.5%, with the highest variation observed ELISA assays is that they are prone to variability; minor for β-amyloid 1-42. The clinical cutpoints for AD were variations in the quality of the reagents, laboratory envi- determined to be 409 ng/L for total tau, 50.2 ng/L for pTau ronment, and execution of the assay protocol may signifi- 181, 526 ng/L for β-amyloid 1-42, and 0.072 for the Aβ 1-42/Aβ cantly affect the result, thereby making it difficult to obtain 1-40 ratio. consistent results over time and between labs [10]. Such Conclusions: Our results suggest that the LUMIPULSE G variation has been greatly decreased by the implementation assays for the CSF AD biomarkers are fit for purpose in of automated immunoassay platforms with pre-supplied clinical laboratory practice. Further, they corroborate ready-to-use reagents that have become available in recent earlier presented reference limits for the biomarkers. years from several assay manufacturers. The LUMIPULSE G (Fujirebio) is an automated system Keywords: Alzheimer’s disease; biomarkers; immuno- for bead-based immunoassays, which is able to measure assay; LUMIPULSE; validation. all four AD biomarkers; Aβ 1-42, tTau, pTau, and the Aβ 1-42/Aβ 1-40 ratio. Clinical validation studies of the LUMIPULSE G assays Introduction have shown good correlations in comparisons with other immunoassays, low intra- and inter-assay varitation, and Alzheimer’s disease (AD) is the most common cause of good concordance with clinical diagnosis [11–16], and dementia [1]. The neuropathological hallmarks of the dis- recent studies reported good concordance between LUMI- ease are amyloid plaques, composed of amyloid beta (Aβ) PULSE G amyloid β (1-42) or amyloid β (1-42)/total tau ratio, peptides [2], and intraneuronal neurofibrillary tangles, and amyloid PET status [17–20]. consisting of hyperphosphorylated tau protein (pTau) [3]. In this study, we evaluated the performance of LUMI- The most abundant form of Aβ in the extracellular plaques PULSE G in relation to established immunoassays for is a 42-amino acid peptide denoted Aβ 1-42 [4]. Aβ peptides measurement of the core AD biomarkers, and for LUMI- are formed through the cleavage of the transmembrane PULSE G amyloid β 1-42, performed a comparison with an Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF 209 LC-MS based certified reference method [21]. We also days. For some measurements, there was insufficient sample volume evaluated intra- and inter-laboratory and longitudinal avaliable; these measurements were then excluded from the com- parison. The standard deviations and coefficients of variation (CV) for variability, and by analysis of three cohorts of clinically intra-laboratory repeatability and inter-laboratory reproducibility diagnosed patients, determined clinical cutpoints for Aβ42, were calculated according to ISO 5725-2. Aβ42/Aβ40, pTau and tTau. Inter- and intra-lab variation were also assessed by analyzing samples within the Alzheimer’s Association Quality Control (AA-QC) program, administered at the Clinical Neurochemistry Laboratory at the University of Gothenburg, Sweden [10]. In the AA-QC program, CSF Materials and methods samples with known concentrations of the AD biomarkers are peri- odically dispatched to laboratories around the world for method LUMIPULSE validation to monitor inter-lab variability of AD biomarkers. For LUMIPULSE G, only results of β-amyloid 1-42 and total tau were For measurement of amyloid β 1-40, amyloid β 1-42, total tau and pTau available. Individual CSF samples were analyzed on seven different 181, a LUMIPULSE® G600II instrument was used (Fujirebio, Ghent, occasion. For β-amyloid 1-42 a longitudinal sample consisting of Belgium). It is a cartridge-based system that uses monoclonal pooled CSF was analysed at seven rounds with 4–19 participating antibody-coated beads for capture and monoclonal antibodies for laboratories. For tTau we used data from a pooled CSF sample ana- detection. For amyloid β 1-42 and total tau, streptavidin-conjugated lysed at four time points with 4–19 participating laboratories in each alkaline phosphatase (AP) is added after a washing step, which binds round. to the biotinylated monoclonal antibody. Amyloid β 1-40 and pTau Long-term consistency of measurments for all four assays, per- 181, the antibodies are directly conjugated to AP instead, and formed on a single LUMIPULSE G instrument was evaluated by 3-(2′-spiroadamantane)-4-methoxy-4-(3ʺ-phosphoryloxy) phenyl-1, analyzing aliquots of two CSF pools; one composed of patient samples 2-dioxetane disodium salt (AMPPD) is used as substrate and the with AD-like core biomarker profile, and one with normal biomarker resulting luminescence at 477 nm is measured. levels. Samples were analyzed 71 times approximately once per week during 18 months in Gothenburg. Method comparisons Clinical cutpoint determination Method comparisons were performed at the neurochemistry labora- tory, Sahlgrenska University Hospital/Mölndal, using de-identified For determination of clinical cutpoints for tTau, pTau, and Aβ42, CSF CSF-samples from the hosptal routine analysis. Samples were selected samples from three cohorts were used; from McGill University based on initial INNOTEST results and re-analyzed on the same (Translational Biomarkers for Aging and dementia (TRIAD), Canada), occasion by LUMIPULSE G and the respective assay. The use of surplus University of Perugia (Italy), and Brno/Praha (Czech Republic). CSF from the routine analysis was approved by the Ethics Committee Demographics and biomarker data of the samples are listed in Table 5. at the University of Gothenburg (EPN – Gothenburg, Aug 11, 2014). For the TRIAD cohort (n=101), lumbar puncture (LP) was per- INNOTEST β-amyloid (1-42), hTau Ag, and pTau 181 (Fujirebio) formed under local anesthesia, using an 18 ga. “introducer” to pene- assays were performed according to the manufacturer’s kit insert in- trate the interspinous ligaments, followed by dural puncture using the structions as described previously [22]. Samples with biomarker con- 24 ga. Sprotte atraumatic needle. Twenty nine milli-liter of fluid was centrations below the lower limit of quantification (LLoQ) as provided collected with polypropylene syringes into 10 mL polypropylene tubes by the manufacturer were excluded (β-amyloid (1-42)<225 ng/L, hTau (Sarstedt, part no. 62.9924.294). The first four ml was sent to the Ag<57 ng/L and pTau 181<20 ng/L). MSD electrochemiluminescence clinical laboratory for determination of albumin, total protein, glucose assay (MSD) was used according to the manufacturer’s instructions for and cells. The remaining 25 mL was transferred to polypropylene tubes measurement of Aβ 1-42/Aβ 1-40. and centrifuged at 20 °C for 10 min at 2,200×g. The CSF was then LC-MS analysis of Aβ 1-42 was performed using a certified refer- rapidly frozen for permanent storage at −80 °C until analysed on the ence method, as previously described [21, 23]. LUMIPULSE instrument. The Perugia cohort was sampled at the Center of Memory Dis- turbances of the University of Perugia. CSF samples were obtained Reproducibility from subjects that were consecutively recruited between January 2012 and June 2016 and followed up for at least two years. The cohort CSF samples having low, medium, and high concentrations of the four consisted of 58 patients with probable AD diagnosed according to the analytes were analyzed by LUMIPULSE G in three laboratories. The NIA-AA criteria [24], regardless of CSF biomarker profile, and 37 non- experiments were performed in two separate rounds: in the first round, demented controls. All patients underwent a baseline clinical exam- Total Tau was measured at the University of Gothenburg, Sweden (Lab ination by experienced neurologists, detailed neuropsychological “a” in Figure 2), at Radboud University, Netherlands (Lab “b”), and at assessment including Mini-Mental State Examination (MMSE), blood FujiRebio, Belgium (Lab “c”). In the second round, pTau 181, β-amy- chemistry, MRI and lumbar puncture (LP). Neurological controls loid 1-42, and β-amyloid 1-40 were measured at Radboud University, included cognitively normal subjects, with other neurological dis- FujiRebio, and Sant Pau- Biomedical Research Institute, Spain (Lab eases such as headache, epilepsy and polyneuropathies, who showed “d”). The homogeneity of the sample aliquots was verified by repeated no evidence of progression to dementia after at least two years of measurments by at the FujiRebio laboratory, Belgium. In each labo- follow-up. Patients with subjective memory complaints were not ratory, the samples were analyzed in triplicate, twice per day over five included in the control group. CSF samples were collected via LP from 210 Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF Figure 1: Regression analysis (Passing-Bablok) comparing LUMIPULSE G and: INNOTEST (A) β-AMYLOID (1-42), (B) hTAU Ag and (C) PHOSPHO TAU(181P); MSD (D) β-amyloid 1-42, (E) β-amyloid 1-40, and (F) 10 × β-amyloid 1-42/1-40; LC-MS (G) β-amyloid 1-42. The regression line is indicated in solid blue, and the identity line in dashed orange. The shaded blue area indicates a 95% confidence interval. Table : Method comparisons. Biomarker Reference system n Slope Intercept r β-AMYLOID (-) INNOTEST  . (., .) −. (−., −.) . hTAU Ag INNOTEST  . (., .) . (., .) . PHOSPHO-TAU (P) INNOTEST  . (., .) −. (−., −.) . Aβ - MSD  . (., .) −,. (−,, −,) . Aβ - MSD  . (., .) . (−., .) .  × Aβ -/Aβ - MSD  . (., .) . (., .) . Aβ - LC-MS  . (., .) − (−., −.) . Slope and Intercept denote Passing-Bablok regression parameters for comparisons between LUMIPULSE G and the respective reference system. R, Pearson’s regression coefficient. Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF 211 A low medium high 360 560 1240 Total Tau [ng/L] Total Tau [ng/L] Total Tau [ng/L] 540 340 1200 520 320 1160 500 300 480 1120 a b c a b c a b c Center Center Center B low medium high 25 115 pTau 181 [ng/L] pTau 181 [ng/L] pTau 181 [ng/L] 44 24 110 42 23 40 105 22 38 b c d b c d b c d Center Center Center C low medium high 380 650 950 β-Amyloid 1-42 [ng/L] β-Amyloid 1-42 [ng/L] β-Amyloid 1-42 [ng/L] 360 900 600 340 850 320 550 800 b c d b c d b c d Center Center Center D low medium high 4600 β-Amyloid 1-40 [ng/L] β-Amyloid 1-40 [ng/L] β-Amyloid 1-40 [ng/L] 8000 4400 11000 7500 4200 10500 7000 4000 10000 6500 3800 9500 6000 b c d b c d b c d Center Center Center E low medium high 10 x β-Amyloid 1-42 / 1-40 10 x β-Amyloid 1-42 / 1-40 10 x β-Amyloid 1-42 / 1-40 0.9 Figure 2: Inter- and intra-lab variation of 0.85 LUMIPULSE G (A) total tau, (B) pTau 181, 0.9 (C) β-amyloid 1-42, (D) β-amyloid 1-40, and 0.8 0.80 (E) β-amyloid 1-42/1-40, measured at three 0.8 (out of four) different laboratories desig- 0.7 nated a–d. 0.75 The black circle indicates value of the mean b c d b c d b c d and the bars represent ±1 standard Center Center Center deviation. 8:00 to 10:00 a.m. after overnight fasting, following a standardized samples were centrifuged at 2000×g for 10 min at room temperature, procedure [12] and according to international guidelines [25, 26]. CSF aliquoted (0.5 mL) in polypropylene tubes (Sarstedt cat. nr. (10–12 mL) was taken from the L3–L4 or L4–L5 interspace, immedi- 72.730.007) and stored at −80 °C pending analysis. ately collected in sterile polypropylene tubes (Sarstedt cat. nr. The Czech cohort was collected within the Czech Brain Aging 62.610.201), and gently mixed to avoid possible gradient effects. The Study (CBAS) [27]. CBAS is a prospective longitudinal memory clinic- 212 Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF Table : Inter- and intra-laboratory variation. Analyte concentration n Mean Repeatability Reproducibility (intra-laboratory) (inter-laboratory) SD CV, % SD CV, % Total tau, ng/L Low  . . . . . Medium  . . . . . High  ,. . . . . pTau , ng/L Low  . . . . . Medium  . . . . . High  . . . . . Amyloid β -, ng/L Low  . . . . . Medium  . . . . . High  . . . . . Amyloid β -, ng/L Low  ,. . . . . Medium  ,. . . ,. . High  ,. . . . .  × amyloid β -/- Low  . . . . . Medium  . . . . . High  . . . . . Analytical variation is expressed as standard deviation (SD) and coeficcient of variation (CV) of intra-laboratory repeatability and inter- laboratory variation, respectively. Figure 3: Inter-laboratory variation in the AA-QC program. (A and B) Variation of (A) β-amyloid 1-42 and (B) total tau for different samples measured in different rounds. The mean for each round is indicated by a circle and the whiskers indicate ±1 standard deviation. (C and D) Variation for (C) β-amyloid 1-42 and (D) total tau for a longitudinal sample analysed in seven and four rounds, respectively. The shaded gray area indicates a 95% confidence interval. Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF 213 based multicentre study recruiting non-demented adults 55+ years of , Eight individual CSF samples were measured on four occasions, with two samples measured per occasion. “n” denotes the number of laboratories participating in each round of measurments. .   .  OCT  OCT  age. Both CBAS centres in Prague and Brno work as a low-threshold facility; hence, the participants are mostly volunteers who enter by self-referral, with memory complaints expressed by themselves or the  .  OCT   .   .  family or who were referred by general practitioners, local specialists or the Czech Alzheimer Society to one of the memory clinics. All study participants underwent a standard set of procedures, including neurological and comprehensive neuropsychology examinations, as Aliquots of a single pooled CSF sample (“longitudinal sample”) were measured on seven (Aβ) and four (tTau) occasions, respectively to assess longitudinal stability. well as laboratory and vital function assessments. Sociodemographic,  .  MAY   .   .  personal, pharmacological and family history data were collected. Participants and their informants completed multiple questionnaires MAY  MAY  about cognitive complaints and lifestyle factors. MRI scans of 1.5 or 3 T were performed every 24 months or earlier when a participant con- verted to dementia or progressed towards cognitive impairment at an  .  FEB   .  unusual rate. Genotyping was carried out at baseline. In a subset, CSF    and/or amyloid PET was performed. The CBAS is complemented by a biological sample bank linked to data from the CBAS and CBAS Plus cohorts. The cerebrospinal fluid (CSF) collection and storage were carried out according to the widely recognised consensus protocol for  .  OCT   . OCT     . .   the standardisation of CSF collection and biobanking. CSF was collected in polypropylene 10 mL tubes (Gama, part no. 400 942). FEB  FEB  Eighteen aliquots of 0.2 mL CSF were stored in polypropylene tubes for each participant. All samples were stored at −80 °C. The Czech and Italian cohorts were analyzed by LUMIPULSE G at    MAY   . MAY     . .   the respective sites where the samples were collected, while the Canadian cohort and the clinical routine samples used to establish clinical cutpoints for Aβ 1-42/Aβ 1-40 were analyzed at the University of Gothenburg in Sweden. Use of the samples for these studies was granted by the respective local ethics boards: the Czech cohort: No. EK-701/16, Date of EC Session: 25.5.2016; the Italian cohort: Prot. N.  .  FEB   . FEB     . .   1936908/AV del 09/10l2O08; the Canadian cohort: REB from Douglas OCT  Hospital Research Centre – Montreal – Canada, no. IUSMD-16-60. OCT  Determination of clinical cutpoint for Aβ 1-42/Aβ 1-40 was based on mixture model analysis of data from 2,782 de-identified patient  .  OCT   . OCT    .  .  samples analyzed at the Neurochemistry Laboratory at Sahlgrenska  University Hospital between December 2019 – October 2020. Statistical analysis Table : Inter-lab variation of LUMIPULSE G in the AA-QC program. Month of analysis Month of analysis Month of analysis Month of analysis Mean conc, ng/L Mean conc, ng/L Mean conc, ng/L Mean conc, ng/L Statistical data analysis was performed in R (version 3.6.1; http:// www.R-project.org/). Passing-Bablok regression analysis was per- formed using the R package mcr (https://CRAN.R-project.org/pack- CV, % CV, % CV, % CV, % age=mcr). Clinical cutpoints for tTau, pTau, and Aβ42 were n n n n determined by ROC curve analysis, using the ROC01 algorithm [28, 29], implemented in the R package “pROC” [30]. For Aβ42/Aβ40, the cut- point was determined by mixture model analysis, using the Longitudinal sample Longitudinal sample Individual samples Individual samples expectation-maximization (EM) algorithm for mixtures of normal distributions (normalmixEM), implemented in the R package “mix- tools” [31]. Results β-Amyloid - Method comparisons Total tau The LUMIPULSE G assays β-amyloid 1-42, β-amyloid 1-40, total tau, and pTau 181 were compared to the corresponding 214 Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF INNOTEST assays, β-AMYLOID (1-42), β-AMYLOID (1-40), of a single sample indicated no significant drift over time hTAU Ag, and PHOSPHO TAU (181P) (Figure 1A–C); to MSD for β-amyloid 1-42 (Figure 3C, Table 3), but again high CV (V-PLEX Plus Aβ Peptide Panel 1 (6E10)) for Aβ 1-42, Aβ40, variation between rounds (7.1–18.9%). For total tau and Aβ 1-42/Aβ40 (Figure 1D–F); an LC-MS reference (Figure 3D, Table 3), there was a slight downward trend method for Aβ 1-42 (Figure 1G), using regression analysis. over time, with the last measurement 20% lower than the The results are summarized in Table 1. The LUMIPULSE G first, while the CVs were more uniform between the mea- assays showed strong correlation to all other immunoassays surement rounds (4.3–8.9%) compared to β-amyloid 1-42. (r>0.93 for all assays). The results for LUMIPULSE G amyloid Long-term assay stability was also assessed by 1-42 and total tau were similar to the corresponding measuring aliquots of two CSF sample pools, analyzed on a INNOTEST ELISA assays, with small intercept and slope single LUMIPULSE G instrument in Gothenburg twice per close to one in the correlation plots. LUMIPULSE pTau 181 week over an 18-month period (Table 4). For the AD pool also had a small intercept but a positive bias (slope=1.6) (CSF Pool 2), the CV:s for all analytes were below 4%, while compared to INNOTEST PHOSPHO TAU(181P). Compared to for the normal pool (CSF Pool 1) they were slightly higher, the MSD V-PLEX assays for Aβ 1-42 and Aβ 1-40, LUMIPULSE but for all analytes within 8%. G also gave significantly higher values for both β-amyloid 1- 42 (slope=1.38, intercept=16.8) and β-amyloid 1-40 (slope=2.07, intercept=−1,724.6), and approximatley 20% lower Aβ 1-42/Aβ 1-40 ratio. LUMIPULSE G β-amyloid 1-42, Diagnostic performance and cut-off values which has been adjusted according to an LC-MS based To determine the diagnostic performance of the LUMI- reference method [21, 23], gave similar values to LC-MS in PULSE G β-amyloid 1-42, β-amyloid 1-42/1-40, total tau, the range of 400–800 ng/L but showed a slight negative and pTau 181 assays and establish cut-off values for use in bias at lower concentrations, and a positive bias at higher a clinical setting, we analyzed CSF from three cohorts of concentrations. AD patients and non-demented controls (Table 5). Cut-off values were determined by ROC curve analysis of data Reproducibility from all cohorts combined. Histograms depicting the distributions of the biomarker concentrations in the The intra-laboratory repeatability CVs (Figure 2, Table 2). groups and ROC curves are shown in Figure 4, and the were below 6% for all four analytes, with the largest varia- established cut-off values, sensitivities and specificities tion for a single analyte was observed for β-amyloid 1-40 are listed in Table 6, including also values for each cohort (6.2%), while the β-amyloid 1-42/1-40 ratio had a CV of 6.2%. separately. The optimal cut-off values were <409 ng/L för Inter-laboratory reproducibility CVs were similar in magni- total tau, <50.2 ng/L for pTau 181, and >526 ng/L for tude, with the largest CV observed for β-amyloid 1-42 (6.5%). β-amyloid 1-42. Because of the high sensitivity and spec- Reproducibility was also studied within the AA-QC ificity of the Aβ42/Aβ40 ratio, this biomarker displays a program. β-amyloid 1-42 and total tau were measured in clear bimodal distribution in patient populations. Therefore, (different) CSF samples, analysed on seven occasions in it was possible to use an alternative approach to determining different laboratories (Figure 3A, B, Table 3). The inter- the cut-off value, by using mixture model analysis of data laboratory variation varied significantly between mea- from a large number of patient CSF samples (n=2,782) surement rounds, from 7.6–21% for β-amyloid 1-42, and analyzed in a clinical routine laboratory setting (Figure 5, from 3.5–10.6% for Total Tau. Longitudinal measurement Table 6), resulting in a cut-off of >0.72. Table : Longitudinal stability of the LUMIPULSE G assays on a single instrument. Amyloid β -, ng/L Amyloid β -, ng/L  × amyloid β -/- Tau, ng/L pTau, ng/L CSF pool  Mean: ,. . . . . CV, % . . . . . CSF pool  Mean: ,. . . . . CV, % . . . . . Aliquots of two CSF pools; one with normal (Pool ) and one with AD-type (Pool ) core biomarker profile, were measured once a week over  months (n=) at the Neurochmeistry Laboratory in Gothenburg. Table : Patient demographics. All Czech McGills Perugia Overall AD, n= CN, n= AD, n= CN, n= AD, n= CN, n= AD, n= CN, n= AD, n= CN, n= Sex Female  (.%)  (.%)  (.%)  (.%)  (.%)  (.%)  (.%)  (.%)  (.%)  (.%) Male  (.%)  (.%)  (.%)  (.%)  (.%)  (.%)  (.%)  (.%)  (.%)  (.%) Age, years Mean, SD . (.) . (.) . (.) . (.) . (.) . (.) . (.) . (.) . (.) . (.) Median [min, . [., . [., . [., . [., . [., . [., . [., . [., . [., . [., max] .] .] .] .] .] .] .] .] .] .] Missing  (.%)  (.%)  (.%)  (.%)  (%)  (%)  (.%)  (%)  (.%)  (.%) Amyloid β - Mean, SD  ()  ()  ()  ()  ()  ()  ()  ()  ()  () Median [min,  [,  [,  [,  [,  [, ]  [, ,]  [, ]  [,  [,  [, max] ,] ,] ,] ,] ,] ,] ,] Amyloid β - Mean, SD , (,) , (,) , (,) , (,) , (,) , (,) , (,) , (,) , (,) , (,) Median [min, , [,, , [,, , [,, , [,, , [,, , [,, , [,, , [,, , [,, , [,, max] ,] ,] ,] ,] ,] ,] ,] ,] ,] ,] Amyloid β -/- Mean, SD . (.) . (.) . (.) . (.) . (.) . (.) . (.) . (.) . (.) . (.) Median [min, . [., . [., . [., . [., . [., . [., . [., . [., . [., . [., max] .] .] .] .] .] .] .] .] .] .] Total tau Mean, SD  ()  ()  ()  ()  ()  ()  ()  ()  ()  () Median [min,  [.,  [.,  [,  [.,  [.,  [, ]  [, ,]  [,  [.,  [., max] ,] ,] ,] ,] ,] ,] ,] ,] pTau  Mean, SD  (.) . (.) . (.) . (.)  (.) . (.)  (.) . (.)  (.) . (.) Median [min, . [., . [., . [., . [., . [., ] . [., ]  [., ] . [., . [., . [., max] ] ] ] ] .] ] ] Demographic information of the patient cohorts used for cut-off determination of tTau, pTau and Aβ. AD, Alzheimer’s disease patients; CN, controls. Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF 215 216 Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF A 0.72 409 AD Control (0.201, 0.723) 0.8 40 Sensi vity 2 count 0.4 20 density AUC: 0.795 (0.76, 0.83) 0.0 0 1 0 1000 2000 0.0 0.4 0.8 Total Tau [ng/L] 1-Specificity B 50.2 AD (0.192, 0.818) 75 Control 0.8 0 Sensi vity 0.5 1.0 count 50 10 x β-Amyloid 1-42 / 1-40 0.4 25 AUC: 0.854 (0.824, 0.885) Figure 5: Mixture model analysis of β-amyloid 1-42/1-40 of data from patient CSF samples (n=2,782) analyzed in a routine laboratory 0.0 0 0 100 200 300 400 0.0 0.4 0.8 setting. pTau 181 [ng/L] 1-Specificity C 40 526 AD Control (0.303, 0.739) Discussion 0.8 30 Sensi vity In the present study, we show good diagnostic perfor- count 20 mance for LUMIPULSE G β-amyloid 1-42, β-amyloid 0.4 AUC: 0.784 (0.749, 0.82) 1-42/40, total tau and pTau 181. All LUMIPULSE G assays 10 show strong linear correlation with INNOTEST amyloid 0.0 0 β 1-42, hTAU Ag and pTau 181, and with MSD for Aβ42/Aβ40, 500 1000 1500 2000 0.0 0.4 0.8 β-Amyloid 1-42 [ng/L] 1-Specificity as well as good correlation with the LC-MS reference method D for Aβ 1-42. A slight deviation from linearity can be seen for 0.0 0.2 0.4 0.6 0.8 1.0 0.582 AD (0.204, 0.852) pTau 181, for which data points in the lower range, below 30 Control 50 ng/L, appear to follow a regression line with a more shallow slope. Notably, this feature is also visible in the Sensi vity count 20 control group data in a recent study by Leitao et al. [14]. 10 Two previous studies have reported large bias between AUC: 0.873 (0.845, 0.901) the INNOTEST and LUMIPULSE G β-amyloid 1-42 assays 0 [15, 16] but since December 2018, the LUMIPULSE G assay 0.3 0.6 0.9 1.2 0.0 0.2 0.4 0.6 0.8 1.0 has been adjusted to harmonize the measurements to the 10 x β-Amyloid 1-42 /1-40 1-Specificity certified reference materials (CRMs) for Aβ 1-42 from the Figure 4: Diagnostic performance of LUMIPULSE and cutpoints. International Federation of Clinical Chemistry and Labo- Histograms and ROC curves for (A) tTau, (B) pTau 181, (C) β-amyloid ratory Medicine (IFCC). The previous values, as measured 1-42, (D) 10 × β-Amyloid 1-42/1-40 for separating Alzheimer patients before the update, are divided by 1.46 in order to be and controls, in three combined cohorts. harmonized with the CRM. This translates to a decrease of about 31.5%, which to a great extent explains the bias observed in previous studies. Total tau and pTau 181 had AUC:s of 80 and 86%, The CSF Aβ 1-42/1-40 ratio correlates well between respectively, with sensitivity and specificity of 72%/80% LUMIPULSE G and MSD. However, if the individual con- for total tau of, and 82%/81% for pTau 181. β-amyloid 1-42 centrations of Aβ 1-40 and Aβ 1-42 are compared for had AUC of 79% with sensitivity and specificity figures of correlation, there is an approximate two-fold higher con- 85%/84%. For amyloid 1-42/1-40, the sensitivity and centration of Aβ 1-40 in LUMIPULSE. This discrepancy specificity, as determined by mixture modeling [31], was highlights the need for a CRM for Aβ40 to harmonize results 92%/71%. between assay platforms. Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF 217 Table : Diagnostic performance and cut-points of LUMIPULSE G. Analyte Cohort ROC AUC Sensitivity Specificity Cut-off value Total tau Czech . . (., .) . (., .) < ng/L Perugia . . (., .) . (., .) < ng/L McGills . . (., .) . (., .) < ng/L Combined . . (., .) . (., .) < ng/L pTau  Czech . . (., .) . (., .) <. ng/L Perugia . . (., )  (., −) <. ng/L McGills . . (., .) . (., .) < ng/L Combined . . (., .) . (., .) <. ng/L β-Amyloid - Czech . . (., .) . (., .) > ng/L Perugia . . (., .) . (., .) > ng/L McGills . . (., .) . (., .) > ng/L Combined . . (., .) . (., .) > ng/L  × β-Amyloid -/ Czech . . (., .) . (., .) >. Perugia . . (., .) . (., .) >. McGills . . (., .) . (., .) >. Combined . . (., .) . (., .) >.  × β-Amyloid -/ Gothenburg Mixture model analysis . . >. ROC curve analysis and cut-point determination for tTau, pTau, Aβ, and Aβ/Aβ based on data recorded from three cohorts of AD patients and controls, for each conhort separately as well as combined. For Aβ/Aβ, an alternative method for cut-point determination was used, based on mixture model analysis of clinical patient data from Gothenburg. An important aspect of immunoassay platforms for are almost identical to those of Leiato et al. (50.2 vs. diagnostic use is the ability to generate reproducible re- 50.6 ng/L). The cut-points for Aβ42 (526 vs. 543 ng/L) and sults over time and between labs. In the past, the limited Aβ42/Aβ40 (0.72 vs. 0.68) are also similar, whereas that of ability of many immunoassay methods for the AD bio- tTau is 22% higher (409 vs. 335 ng/L). markers has made it difficult to compare results reported in The histogram in Figure 4D shows that there are several different studies, with reproducibility CVs of over 30% subjects in the control group that have low Aβ42/Aβ40 ratio. described [32]. Our results show remarkably similar varia- These are possibly individuals with incipient amyloid tion for repeatability and reproducibility comparisons, pathology who do not yet manifest AD symptoms. In the with reproducibility CVs ranging between 5.4 and 6.5% for ROC curve analysis, they may lead to underestimation of Aβ42 and between 2.1 and 2.8% for tTau. These results the optimal cutpoint. For Aβ42/Aβ40, the a clear bimodal indicate that the use of an automated system with pre- distribution made it possible to use mixture model analysis supplied reagents in a closed cartridge format can greatly to calculate the cutpoint, thus resulting in a slightly higher improve reproducibility of results between labs. value compared to that obtained by ROC curve analysis. Data from the AA-QC program show a wide spread of In conclusion, the results presented here suggest that reproducibility CVs when comparisons were performed, the fully automated LUMIPULSE assays for the CSF AD ranging between 7.6 and 21% for Aβ42 and 3.5–10.6% for biomarkers are fit for purpose in clinical laboratory prac- tTau. A possible reason for this difference is that the tice. Further, they corroborate earlier presented reference reproducibility study was performed under optimal con- limits for the biomarkers. ditions, using the same calibrator and kit lots in all labs and analyzing the samples as part of a study. In the AA-QC Research funding: Brno team is supported by the project no. program, in contrast, the samples may be measured with LQ1605 from the National Program of Sustainability II (MEYS different reagent lots and are handled as in clinical routine, CR) and Praha team is supported by Ministry of Health of the where analytical errors are more likely to occur. The lon- Czech Republic, grant no. 19-04-00560. MMV is supported by gitudinal sample of pooled CSF showed no deviation over the BIONIC project (nr. 733050822), which has been made time for Aβ42 or tTau. At this point, longitudinal data is not possible by ZonMW (part of the Dutch national ‘Deltaplan for available for LUMIPULSE G pTau 181 or β-Amyloid 1-40, as Dementia’; zonmw.nl/dementiaresearch), and by grants they were only recently included in the program. from the Selfridges Group Foundation, and National Cut-off values were determined in a previous study by Institutes of Health, USA [grant number 5R01NS104147- Leitao et al. For pTau, the values established in our study 02]. JG is supported by Alzheimerfonden (AF-930934), 218 Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF Åhléns-stiftelsen, and Stiftelsen för Gamla tjänarinnor. HZ advisory boards for Fujirebio-Europe, Roche, Biogen, and is a Wallenberg Scholar supported by grants from the Nutricia and has a patent WO2019175379 A1 Markers of Swedish Research Council (#2018-02532), the European synaptopathy in neurodegenerative disease issued. DA has Research Council (#681712), Swedish State Support for served as a consultant or at advisory boards for Krka Clinical Research (#ALFGBG-720931), the Alzheimer Drug Farmacéutica S.L., Fujirebio-Europe, Roche Diagnostics, Discovery Foundation (ADDF), USA (#201809-2016862), the Zambon S.A.U., Esteve Pharmaceuticals S.A. and Nutricia European Union’s Horizon 2020 research and innovation and has a patent WO2019175379 A1 Markers of synaptopathy programme under the Marie Skłodowska-Curie grant in neurodegenerative disease issued. agreement No 860197 (MIRIADE), and the UK Dementia Informed consent: Informed consent was obtained from all Research Institute at UCL. KB is supported by the Swedish individuals included in this study. Research Council (#2017-00915), the Alzheimer Drug Ethical approval: Use of the samples for these studies was Discovery Foundation (ADDF), USA (#RDAPB-201809- granted by the respective local ethics boards: Czech cohort: 2016615), the Swedish Alzheimer Foundation (#AF-742881), No. EK-701/16, Date of EC Session: 25.5.2016 Italian cohort : Hjärnfonden, Sweden (#FO2017-0243), the Swedish State Prot. N. 1936908/AV del 09/10l2O08. Under the Agreement Between the Swedish Government and the County Councils, the ALF-Agreement (#ALFGBG-715986), and European Union Joint Program for Neurodegenerative References Disorders (JPND2019-466-236). PRN is supported by the Canadian Institutes of Health Research (CIHR) [MOP-11-51-31; 1. Blennow K, de Leon MJ, Zetterberg H. Alzheimer’s disease. Lancet 2006;368:387–403. RFN 152985, 159815, 162303], Canadian Consortium of 2. Masters CL, Simms G, Weinman NA, Multhaup G, McDonald BL, Neurodegeneration and Aging (CCNA; MOP-11-51-31 -team 1), Beyreuther K. Amyloid plaque core protein in Alzheimer disease and Weston Brain Institute, the Alzheimer’s Association down syndrome. Proc Natl Acad Sci USA 1985;82:4245–9. [NIRG-12-92090, NIRP-12-259245], Brain Canada Foundation 3. Grundke-Iqbal I, Iqbal K, Tung YC, Quinlan M, Wisniewski HM, (CFI Project 34874; 33397), the Fonds de Recherche du Binder LI. Abnormal phosphorylation of the microtubule- associated protein tau (tau) in Alzheimer cytoskeletal pathology. Québec – Santé (FRQS; Chercheur Boursier, 2020-VICO- Proc Natl Acad Sci USA 1986;83:4913–7. 279314). 4. Blennow K, Hampel H, Weiner M, Zetterberg H. Cerebrospinal Author contributions: All authors have accepted responsibility fluid and plasma biomarkers in Alzheimer disease. Nat Rev for the entire content of this manuscript and approved its Neurol 2010;6:131–44. submission. 5. Barthet G, Jorda-Siquier T, Rumi-Masante J, Bernadou F, Muller U, Competing interests: JH has served at scientific advisory Mulle C. Presenilin-mediated cleavage of APP regulates synaptotagmin-7 and presynaptic plasticity. Nat Commun 2018; board for Alzheon, Agora, Alzheimerchain, Axon, Biogen 9:4780. and has given lectures in symposia sponsored by Schwabe, 6. Grundke-Iqbal I, Iqbal K, Quinlan M, Tung YC, Zaidi MS, Lundbeck, Egis, and GE. KS has served at scientific advisory Wisniewski HM. Microtubule-associated protein tau. A board for Biogen and Alzheimerchain and has given lectures component of Alzheimer paired helical filaments. J Biol Chem in symposia sponsored by Schwabe and Lundbeck. HZ 1986;261:6084–9. 7. Cognat E, Mouton Liger F, Troussiere AC, Wallon D, Dumurgier J, has served at scientific advisory boards for Denali, Roche Magnin E, et al. What is the clinical impact of cerebrospinal fluid Diagnostics, Wave, Samumed, Siemens Healthineers, biomarkers on final diagnosis and management in patients with Pinteon Therapeutics and CogRx, has given lectures in mild cognitive impairment in clinical practice? Results from a symposia sponsored by Fujirebio, Alzecure and Biogen, nation-wide prospective survey in France. BMJ Open 2019;9: and is a co-founder of Brain Biomarker Solutions in e026380. Gothenburg AB (BBS), which is a part of the GU Ventures 8. Lewczuk P, Lelental N, Spitzer P, Maler JM, Kornhuber J. Amyloid- beta 42/40 cerebrospinal fluid concentration ratio in the Incubator Program (outside submitted work). KB has diagnostics of Alzheimer’s disease: validation of two novel served as a consultant, at advisory boards, or at data assays. J Alzheimer’s Dis 2015;43:183–91. monitoring committees for Abcam, Axon, Biogen, JOMDD/ 9. Janelidze S, Zetterberg H, Mattsson N, Palmqvist S, Shimadzu. Julius Clinical, Lilly, MagQu, Novartis, Roche Vanderstichele H, Lindberg O, et al. CSF Abeta42/Abeta40 and Diagnostics, and Siemens Healthineers, and is a co-founder Abeta42/Abeta38 ratios: better diagnostic markers of Alzheimer disease. Ann Clin Transl Neurol 2016;3:154–65. of Brain Biomarker Solutions in Gothenburg AB (BBS), 10. Mattsson N, Andreasson U, Persson S, Arai H, Batish SD, which is a part of the GU Ventures Incubator Program Bernardini S, et al. The Alzheimer’s Association external quality (outside submitted work). PRN is consultant for CERVEAU control program for cerebrospinal fluid biomarkers. Alzheimer’s radiopharmaeuticals. AL has served as a consultant or at Dementia 2011;7:386–95.e6. Gobom et al.: Validation of the LUMIPULSE immunoassay for AD biomarkers in CSF 219 11. Bayart JL, Hanseeuw B, Ivanoiu A, van Pesch V. Analytical and 21. Leinenbach A, Pannee J, Dulffer T, Huber A, Bittner T, Andreasson clinical performances of the automated Lumipulse cerebrospinal U, et al. Mass spectrometry-based candidate reference fluid Abeta42 and T-tau assays for Alzheimer’s disease diagnosis. measurement procedure for quantification of amyloid-beta in J Neurol 2019;266:2304–11. cerebrospinal fluid. Clin Chem 2014;60:987–94. 12. Chiasserini D, Biscetti L, Farotti L, Eusebi P, Salvadori N, Lisetti V, 22. Le Bastard N, Aerts L, Sleegers K, Martin JJ, Van Broeckhoven C, et al. Performance evaluation of an automated ELISA system for De Deyn PP, et al. Longitudinal stability of cerebrospinal fluid Alzheimer’s disease detection in clinical routine. J Alzheimer’s biomarker levels: fulfilled requirement for pharmacodynamic Dis 2016;54:55–67. markers in Alzheimer’s disease. J Alzheimer’s Dis 2013;33:807–22. 13. Dakterzada F, Lopez-Ortega R, Arias A, Riba-Llena I, Ruiz-Julian 23. Blennow K, Mattsson N, Scholl M, Hansson O, Zetterberg H. M, Huerto R, et al. Assessment of the concordance and Amyloid biomarkers in Alzheimer’s disease. Trends Pharmacol diagnostic accuracy between elecsys and Lumipulse fully Sci 2015;36:297–309. automated platforms and innotest. Front Aging Neurosci 2021; 24. McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR Jr., 13:604119. Kawas CH, et al. The diagnosis of dementia due to Alzheimer’s 14. Leitao MJ, Silva-Spinola A, Santana I, Olmedo V, Nadal A, disease: recommendations from the National Institute on Le Bastard N, et al. Clinical validation of the Lumipulse G Aging-Alzheimer’s Association Workgroups on diagnostic cerebrospinal fluid assays for routine diagnosis of Alzheimer’s guidelines for Alzheimer’s disease. Alzheimer’s Dementia 2011;7: disease. Alzheimer’s Res Ther 2019;11:91. 263–9. 15. Paciotti S, Sepe FN, Eusebi P, Farotti L, Cataldi S, Gatticchi L, et al. 25. del Campo M, Mollenhauer B, Bertolotto A, Engelborghs S, Diagnostic performance of a fully automated chemiluminescent Hampel H, Simonsen AH, et al. Recommendations to standardize enzyme immunoassay for Alzheimer’s disease diagnosis. Clin preanalytical confounding factors in Alzheimer’s and Parkinson’s Chim Acta 2019;494:74–8. disease cerebrospinal fluid biomarkers: an update. Biomarkers 16. Zecca C, Brescia V, Piccininni M, Capozzo R, Barone R, Barulli MR, Med 2012;6:419–30. et al. Comparative evaluation of two immunoassays for 26. Teunissen CE, Petzold A, Bennett JL, Berven FS, Brundin L, cerebrospinal fluid beta-amyloid1-42 measurement. Clin Chim Comabella M, et al. A consensus protocol for the standardization Acta 2019;493:107–11. of cerebrospinal fluid collection and biobanking. Neurology 17. Alcolea D, Pegueroles J, Munoz L, Camacho V, Lopez-Mora D, 2009;73:1914–22. Fernandez-Leon A, et al. Agreement of amyloid PET and CSF 27. Sheardova K, Vyhnalek M, Nedelska Z, Laczo J, Andel R, Marciniak biomarkers for Alzheimer’s disease on Lumipulse. Ann Clin Transl R, et al. Czech Brain Aging Study (CBAS): prospective multicentre Neurol 2019;6:1815–24. cohort study on risk and protective factors for dementia in the 18. Kaplow J, Vandijck M, Gray J, Kanekiyo M, Huyck E, Traynham CJ, Czech Republic. BMJ Open 2019;9:e030379. et al. Concordance of Lumipulse cerebrospinal fluid t-tau/ 28. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978; Abeta42 ratio with amyloid PET status. Alzheimer’s Dementia 8:283–98. 2020;16:144–52. 29. Vermont J, Bosson JL, Francois P, Robert C, Rueff A, Demongeot J. 19. Keshavan A, Wellington H, Chen Z, Khatun A, Chapman M, Hart M, Strategies for graphical threshold determination. Comput et al. Concordance of CSF measures of Alzheimer’s pathology Methods Progr Biomed 1991;35:141–50. with amyloid PET status in a preclinical cohort: a comparison of 30. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. Lumipulse and established immunoassays. Alzheimer’s pROC: an open-source package for R and S+ to analyze and Dementia 2020;12:e12097. compare ROC curves. BMC Bioinf 2011;12:77. 20. Moon S, Kim S, Mankhong S, Choi SH, Vandijck M, Kostanjevecki 31. Benaglia T, Chauveau D, Hunter DR, Young DS. mixtools: an R V, et al. Alzheimer’s cerebrospinal biomarkers from Lumipulse package for analyzing finite mixture models. J Stat Software fully automated immunoassay: concordance with amyloid-beta 2009;32:1–29. PET and manual immunoassay in Koreans: CSF AD biomarkers 32. Mattsson N, Zetterberg H, Blennow K. Lessons from multicenter measured by Lumipulse in Koreans. Alzheimer’s Res Ther 2021; studies on CSF biomarkers for Alzheimer’s disease. Int J 13:22. Alzheimer’s Dis 2010;2010:610613.