Plasma p-tau181 and amyloid markers in Alzheimer ’ s disease: A comparison between Lumipulse and SIMOA

and


Introduction
Cerebrospinal fluid (CSF) biomarkers are informative, sensitive, and specific for detection of Alzheimer's diseases (AD) already in early stages (Hansson, 2021).The recent development of plasma biomarkers using Single Molecule Array (SIMOA) technique is changing the era of the diagnosis of AD potentially shifting from CSF to blood assessments (Blennow, 2021;Leuzy et al., 2021a;Zetterberg and Blennow, 2021).
Blood biomarkers may provide significant advantages for clinical practice, as they are less invasive and have the potential for greater suitability in widespread applications (Alawode et al., 2021).
A number of assays have been developed to measure p-tau isoforms in blood, which were recently reviewed in Karikari et al. (Karikari et al., 2021).Currently, the most promising blood biomarkers developed by SIMOA technique for detecting AD pathology are phosphorylated tau species (including p-tau181, p-tau217 and p-tau231) (Bayoumy et al., 2021) and amyloid Aβ42/40 ratio (Ashton et al., 2021b;Gonzalez-Ortiz et al., 2023b;Karikari et al., 2020;Mielke et al., 2022;Palmqvist et al., 2019;Schindler et al., 2019;Simrén et al., 2021;Suárez-Calvet et al., 2020).Most published studies focused on p-tau species have employed immunoassays on either the Meso Scale Discovery (MSD) platform (Janelidze et al., 2020;Mielke et al., 2018;Thijssen et al., 2020) or the SIMOA system (Karikari et al., 2020) both with analytical sensitivity below the picomolar range.The recent development of assays by using the chemiluminescent enzyme immunoassay (CLEIA) technology (including the fully-automated Lumipulse platform) represent an attractive further step for their application and wider use in clinical practice (Martínez-Dubarbie et al., 2023).Preliminary data showed a high concordance between Lumipulse plasma p-tau181 levels and amyloid positron emission tomography (PET) and AD diagnosis (Mielke et al., 2021).Moreover, the species identified by the specific antibodies used for Lumipulse automatic detection had proven to be highly effective for the clinical diagnosis of AD compared to SIMOA and MSD assays (Ashton et al., 2023).Despite the growing evidence in the field, there is still a need for validation of Lumipulse plasma biomarkers in real-life scenarios to challenge their technical and biological consistency and validity, in comparison with the more established SIMOA testing.
Specifically, the low fold change (around a 10 % reduction) in the Aβ42/40 ratio in brain amyloidosis positive cases (Janelidze et al., 2021) results in a low robustness of clinical classifications (Rabe et al., 2023) and the performance of this biomarker is also affected by pre-analytical handling of samples (Benedet et al., 2022) and commonly used drugs may result in false positives (Brum et al., 2023b), further supporting the need of real-life studies.
The aim of our study was thus to evaluate the newly developed plasma p-tau181, Aβ42, and Aβ40 biomarkers on LUMIPULSE G600II automated platform and assess their technical validity and correlations with CSF biomarkers in consecutively recruited participants with neurodegenerative disorders.A secondary aim of the research was to evaluate the correlations between biomarkers tested with CLEIA and SIMOA to challenge their biological consistency and explore their diagnostic value in exploratory analyses.

Study population
The study included participants with mild cognitive impairment (MCI) or mild dementia who underwent CSF assessment at the outpatient Neurodegenerative clinic of the Brescia University Hospital, Italy.Participants who fulfilled a clinical diagnosis of AD according to NIA-AA criteria (Jack et al., 2018), dementia with Lewy bodies (DLB) according to DLB consortium criteria (McKeith et al., 2020(McKeith et al., , 2017) ) or frontotemporal dementia (FTD) according to FTD consortium criteria (Gorno-Tempini et al., 2011;Rascovsky et al., 2011) were consecutively enrolled in the study.A standardized full cognitive and behavioural assessment, including Mini-Mental State Examination (MMSE), Neuropsychiatric Inventory (NPI) and Clinical Dementia Rating Scale (CDR) was performed in each participant.Only participants with A+T+N+ biological confirmation of AD were included.The internal cut-off values used in the laboratory were Aβ42 < 650 pg/mL, p-tau > 60 pg/mL, t-tau > 400 pg/mL and p-tau181/Aβ42 ratio > 0.09 (Pilotto et al., 2022).Patients with a clinical diagnosis of FTD and DLB were included in the study only when the p-tau181/Aβ42 ratio resulted negative to (<0.09); participants with MCI with negative CSF amyloid markers were defined as MCI-NDD.A group of neurologically and cognitively normal (CN) individuals were recruited from participants' caregivers and were included as reference group for plasma analyses.

CSF collection and analysis
Each patient underwent lumbar puncture (LP) in fasting condition according to the standardized protocol of the outpatient Neurodegenerative clinic.The CSF specimens were collected in 15 mL polypropylene sterile tubes, gently mixed to avoid gradient effects, and sent directly to the hospital laboratory for routine assessments.The remaining CSF volume was stored in 2 mL polypropylene tubes at − 20 • C, and afterwards placed at − 80 • C. CSF Aβ42, Aβ40, p-tau181, and total tau were measured using the Lumipulse G assays (Fujirebio) on the LUMIPULSE G600II for diagnostic standard analyses.As by the guidelines delivered by the Consensus of the Task Force on Biological Markers in Psychiatry of the World Federation of Societies of Biological Psychiatry, CSF samples were subjected to a maximum of two freeze-thaw cycles (Lewczuk et al., 2018).

Plasma collection and analysis
Blood samples were collected in 7.5 mL tubes with K2ethylenediaminetetraacetic acid (K2-EDTA) per each participant.The blood was mixed by gently inverting the tubes 5-10 times.The tubes were thereafter centrifuged at 2500×g for 10 min at room temperature (RT).Plasma aliquots of 0.5 mL were pipetted into polypropylene cryotubes and frozen at ultra-low temperature freezing (ULTF) − 80 • C for storage.
The Lumipulse technical reliability and the accuracy of plasma Immunoreaction Cartridges RUO were investigated using an intra-and inter-day testing schemes according to Vidali et al. and based on CLSI EP15 (Neill Carey et al., 2014;Vidali et al., 2016).Briefly, three different plasma aliquots from the same participant were tested as independent samples for 5 different runs (intra-day testing), to verify the testing repeatability.Furthermore, to assess the inter-day repeatability of the test, other three plasma aliquots of the same participant were tested in 5 days (inter-day testing), for 15 total runs of the same sample.

Statistical analyses
Normality distribution was evaluated using the Shapiro-Wilk test and Q-Q plots.Clinical and demographic characteristics as well as cognitive assessments and CSF and plasma markers comparisons between diagnostic groups (AD, NDD and CN) were performed using the Kruskal-Wallis test and the ANOVA test, as appropriate.The comparison between AD and NDD/CN was performed by the Mann-Whitney U test.For biomarkers comparison, the significance levels were set to p=0.01 as multiple comparison correction.The comparability between the two analytical platforms was assessed using Passing-Bablok regression, while their imprecision was assessed by calculating the laboratory's coefficient of variation (CV).The relationship between plasma and CSF biomarkers was calculated with Spearman's test in a correlation matrix.The accuracy of plasma biomarkers in terms of specificity and sensibility to discriminate between AD and non-AD participants was analysed by receiver-operating characteristic (ROC) approach.We computed areas under the curve (AUCs) and their 95 % confidence intervals (CI).Comparison between AUC ROC performance was evaluated using DeLong non-parametric Method (DeLong et al., 1988) All analyses were performed with R Statistical Software (https://www.r-project.org/).
Statistical significance was set as α=0.05, and all tests were two-tailed.

Precision and repeatability of Lumipulse G600II testing
A set of K2EDTA plasma samples collected from one AD CSFconfirmed patient (CSF t-tau=809 pg/mL, p-tau181=132.6pg/mL, Aβ42=446 pg/mL and Aβ40=9128 pg/mL) were aliquoted and tested as fifteen independent samples.The CVs of p-tau181, Aβ42 and Aβ40 were 2.4 %, 3.1 % and 2.8 %, respectively, for 15 repeated measures (see Suppl.Table 1).The inter-day reproducibility of the Lumipulse platform was verified using the ready-to-use controls Level 1 (L1) and Level 2 (L2) given by the manufacturer tested in 5 different days.The mean interassay CV was less than 5 % for all three analytes.Specifically, for L1, p-tau181 had a CV of 3.58 %, Aβ42 had a CV of 2.64 %, and Aβ40 had a CV of 0.89 %.For L2, p-tau181 had a CV of 1.89 %, Aβ42 had a CV of 3.19 %, and Aβ40 had a CV of 2.15 % (see Suppl.Table 2).In addition, 15 independent plasma samples, aliquoted from the same patient, were first collected from ULTF − 80 • C and then stored at − 20 • C for testing along different consecutive days.P-tau181 levels exhibited good stability, with a CV around 20 % for all three aliquots along the 5 days (aliquot-1 CV=20.75 %; aliquot-2 CV=19.99 %; aliquot-3 CV=18.82%; Suppl.Table 2).Aβ42 and Aβ40 samples showed lower stability starting with the third day of assessment.Specifically, all three aliquots exhibited a fair stability during the first two days (Aβ42 day1 =15.65 pg/mL at day1, Aβ42 day2 =11.43 pg/mL; Aβ40 day1 =237.68 pg/mL; Aβ40 day2 =183.10 pg/mL), with a similar trend of decrease in total levels between 75 % and 85 % from day 1 to day 3 for Aβ42 and 78-86 % for Aβ40 (see supplementary materials for detailed day per day levels, Suppl.Table 2).

Cerebrospinal fluid and plasma biomarkers analyses
The clinical study validation comprised 133 participants, namely 83 patients and 50 controls.The clinical assessment and CSF AD markers allowed the classification of patients in 55 AD (40 with MCI and 15 having mild dementia) and 28 other neurodegenerative disorders (including 8 DLB and 5 FTD and 15 MCI with negative AD markers, MCI-NDD).Clinical and demographics data and CSF core biomarkers are indicated in Table 1; age and sex were used as covariates given the slight but significant distribution between-groups.Table 2 and Fig. 1 showed the between-groups distribution of plasma biomarkers assessed by Lumipulse and SIMOA.Plasma levels of p-tau181 measured by Lumipulse were around two-fold higher in patients with AD compared to NDD and CN (p<0.001,Table 2 and Fig. 1, Panel A).Aβ42 levels were lower in patients with AD compared to both NDD patients (p=0.006) and CN (p<0.001),whereas no differences were detected for Aβ40 plasma between groups.The ratio of p-tau181/Aβ42 was significantly higher in AD compared to both NDD and CN groups (p<0.001,Table 2 and Fig. 1, panel B); Aβ42/40 ratio was lower in AD compared to NDD and CN (p<0.001,Table 2 and Fig. 1, panel C).Plasma biomarkers measured by SIMOA showed very similar between-group differences for each biomarker compared to Lumipulse.Specifically, p-tau181 levels and p-tau181/Aβ42 ratio were higher in AD compared to NDD and CN (Fig. 2).Plasma Aβ42 and Aβ40 levels were lower in AD compared to CN but similar compared to NDD.Plasma Aβ42/40 ratio was lower in AD compared to CN (Fig. 2).
The correlation matrix evaluating the correlations between Lumipulse and SIMOA and CSF core biomarkers is presented in Fig. 3.The Passing-Bablok regression was applied to compare the two methods for plasma levels of Aβ40, Aβ42 and p-tau181.

Discussion
This study demonstrated the excellent stability and clinical validity of p-tau181, and amyloid species detected in plasma with Lumipulse assays in comparison with the SIMOA technique.The clinical application indicated a high diagnostic performance of p-tau181 and its ratio in diagnosing AD, with very similar discriminative values of both techniques, extending previous findings with similar techniques in different settings (Bellomo et al., 2024;Gonzalez-Ortiz et al., 2023a).
The study analyzed plasma and CSF biomarkers at the same timepoint in a clinical setting scenario, thus including participants with different diseases and ages and not defined by a priori selection; therefore, confirming the wide applicability of such techniques in real-life settings.The technical validity of Lumipulse platform was assessed by using repeated runs to challenge the intraday and inter-day variability and stability of analytes.The technical comparison between Lumipulse G600II and SIMOA SR-X platforms demonstrated a good correlation between analyses using the Passing-Bablok regression, in line with previous data evaluating several techniques for the diagnosis of AD in clinical settings (Karikari et al., 2022;Suárez-Calvet et al., 2020).The intraday laboratory validation for Lumipulse indicated a CV under 4 % for all the analytes, evidencing the robustness and stability of such technique in standard conditions even in unselected old participants.P-tau species also appeared to be stable through inter-day testing, at least for the first two days of testing.Of note, while p-tau181 levels remained stable for multiple days testing, an important decrease of both amyloid species concentrations was observed after the third day of testing.A possible explanation might be related to the high sensitivity of amyloid species to storage conditions and temperature changes, as the shift from − 80 • C to − 20 • C storage could have affected protein concentration.It has been demonstrated that for samples stored at − 20 • C, the observed variations in concentration (bias%) are greater than those observed for other storage temperatures (Musso et al., 2023).These findings highlighted once again the higher instability of amyloid species compared to p-tau species also in plasma, in line with few validations on plasma (Rózga et al., 2019;Chang et al., 2020;Walter et al., 2020;Verberk et al., 2022;Mansilla et al., 2023;Sunde et al., 2023) and large data of CSF (Lewczuk et al., 2018).Further technical validations for the specific techniques are thus warranted to challenge the stability of biomarkers in vivo in different settings, as the testing performed directly after − 80 • C storage is not always possible in different settings.
When applied in a clinical setting, plasma assessment confirmed a high biological validity, with a high discrimination accuracy for the diagnosis of AD.Both techniques showed a significantly increased levels of p-tau181 in plasma compared to both controls and participants with other neurodegenerative conditions.Plasma Aβ42/40 ratio appeared significantly reduced in AD compared to CN and NDD only using the Lumipulse technique.The effect size of tau and amyloid markers observed is similar to those described for large MESO Scale and SIMOA based validations.The findings indeed showed an increase of 2-4-fold of p-tau181 already in the early stage of the disease, with a very strong correlation with CSF p-tau values.Of note, the p-tau CSF-plasma Fig. 2. Plasma p-tau181, Aβ42, Aβ40, Aβ42/40 and p-tau181/Aβ42 levels AD, CN, and NDD groups measured using Simoa.p-tau181 is significantly higher in AD compared to both CN and NDD.Aβ42 and Aβ40 is significantly lower in AD in respect to CN; Aβ420 did not show any significant difference among the three tested groups.Noteworthy, both ratios Aβ42/40 and p-tau181/Aβ42 can effectively discriminate AD towards CN and NDD.AD, Alzheimer's disease; CN, cognitively normal; NDD, non-Alzheimer neurodegenerative disorders.;p-tau181_S, phosphorylated tau 181 tested on SIMOA (S); Aβ42_S, β Amyloid Protein Fragment 1-42 tested on SIMOA (S); Aβ40_S, Amyloid β Protein Fragment 1-40 tested on SIMOA (S).correlations were opening the discussion about the source of plasma ptau once AD pathology is deposited.In contrast, Amyloid Aβ42/40 exhibited only a 10 %-12 % average difference in plasma of AD patients, with no correlation with CSF amyloid species levels, a finding in line with the literature, which limit the use of Aβ42/40 ratio alone for AD diagnosis.Of note, amyloid species evaluated in plasma by Lumipulse and SIMOA techniques appeared to correlate with each other thus confirming that the detection of Aβ42 and Aβ40 is biologically valid, despite it definitively does not linearly reflect the CSF amyloid-related pathology (Benedet et al., 2022;Brum et al., 2023b).This is likely due to the fact that a major portion of Aβ in plasma does not come from the brain but from peripheral tissues (Bustamante and Zaidi, 2023;Colvin et al., 2015), which question the real value of this biomarker in plasma.Also, the low fold change of Aβ42/40, compared with CSF (and other plasma markers) limit the clinical application of amyloid ratio in clinical practice.The findings also showed that the combination of amyloid and p-tau species using the p-tau181/Aβ42 ratio had the highest diagnostic accuracy using both techniques, with slightly higher value compared to p-tau181 only and very similar values in term of sensitivity and specificity for the different techniques.This highlights, again, the general validity of Lumipulse as a promising alternative technique to SIMOA and Mesoscale, especially when considering the shift from research to clinical use (Ashton et al., 2021a;Karikari et al., 2020;O'Connor et al., 2021;Pilotto et al., 2022).Despite the small sample size, the study confirmed the diagnostic value of plasma markers using both Lumipulse and SIMOA, with similar accuracy compared to previous studies also in a non-selected population (Martínez-Dubarbie et al., 2023;Verde et al., 2023).Further on-going validation studies will explore the diagnostic accuracy, clinical impact, and cost-effectiveness impact of a wider application of plasma biomarkers in the daily routine practice for diagnosing neurodegenerative conditions (Leuzy et al., 2022;O'Connor et al., 2021;Pereira et al., 2021).
One limitation of our study is that a more comprehensive methodological assessment of storage temperature could have not been performed.The validation of new detection methods for p-tau217, which has been demonstrated to have a higher diagnostic accuracy in research settings compared to p-tau181 (Leuzy et al., 2021b), will be also an important step forward for a wider use of such biomarkers in clinical practice.The use of plasma biomarkers within clinical assessment has been now supported by large amount of literature and will definitively change the management of patients, especially when considering the development of disease-modifying treatments (Brum et al., 2023a;Mattsson-Carlgren et al., 2023).

Fig. 3 .
Fig. 3. Correlation matrix based on Spearman's test showing the most robust relations.Biomarkers are differentiated with "L" as for Lumipulse, instead of "S" which stands for SIMOA.Different biological samples are indicated as plasma and CSF, Cerebral Spinal Fluid.

Fig. 4 .
Fig.4.Scatter plots with density curves for associated variables and 95 % CI (dashed lines).The association of CSF and plasma for p-tau181, is showed in panels A and B, which reflect CSF samples in contrast with plasma tested on Lumipulse G600II and on Simoa SR-X, respectively.Panels E and F shows the robust association of plasma p-tau181 and Aβ42/40 between Lumipulse G600II and Simoa SR-X; while the correlation of plasma Aβ42 and Aβ40 between Lumipulse and Simoa is plotted in panels C and D. L, Lumipulse; S, SIMOA; CSF, Cerebral spinal fluid; p-tau181_S, phosphorylated tau 181; Aβ42_S, β Amyloid Protein Fragment 1-42; Aβ40_S, Amyloid β Protein Fragment 1-40.

Table 1
Participants' characteristics.Data are expressed as mean (M) and standard deviation (SD) p-values showed the difference between AD CSF core biomarkers profile groups and were computed with a Mann-Whitney U test (age, MMSE, NPI, AD CSF core biomarkers), or a chi-square (sex, APOE ε4 carrier status).Plasma biomarkers assessed by Lumipulse and SIMOA platforms.