ARTICLE OPEN ACCESS CLASS OF EVIDENCE

Total Aβ42Aβ40 ratio in plasma predictsamyloid-PET status independent of clinicalAD diagnosisJames D Doecke PhD Virginia Perez-Grijalba PhD Noelia Fandos PhD Christopher Fowler PhD

Victor L Villemagne MD PhD Colin L Masters MD PhD Pedro Pesini PhD and Manuel Sarasa PhD

for the AIBL Research Group

Neurologyreg 202094e1580-1591 doi101212WNL0000000000009240

Correspondence

Dr Pesini

ppesiniaracloncom

AbstractObjectiveTo explore whether the plasma total β-amyloid (Aβ) Aβ42Aβ40 ratio is a reliable predictor ofthe amyloid-PET status by exploring the association between these 2 variables in a subset of theAustralian Imaging Biomarkers and Lifestyle (AIBL) study of aging cohort

MethodsTaking plasma samples at 3 separate time points month 18 (n = 176) month 36 (n = 169) andmonth 54 (n = 135) we assessed the total Aβ42Aβ40 ratio in plasma (TP4240) with regard toneocortical Aβ burden via PET standardized uptake value ratio (SUVR) and investigated bothassociation with Aβ-PET status and correlation (and agreement) with SUVR

ResultsThe TP4240 plasma ratio was significantly reduced in amyloid-PETndashpositive participants at alltime points (p lt 00001) Adjusting for covariates age gender APOE e4 allele status andclinical classification clearly affects the significance with p values reduced and only comparisonsat 54 months retaining significance (p = 0006) Correlations with SUVR were similar acrosseach time point with Spearman ρ reaching minus064 (p lt 00001) Area under the curve valueswere highly reproducible over time points with values ranging from 0880 at 36 months to0913 at 54 months In assessments of the healthy control group only the same relationshipswere found

ConclusionsThe current study demonstrates reproducibility of the plasma assay to discriminate betweenamyloid-PET positive and negative over 3 time points which can help to substantially reducingthe screening rate of failure for clinical trials targeting preclinical or prodromal disease

Classification of evidenceThis study provides Class II evidence that plasma total Aβ42Aβ40 ratio is associated withneocortical amyloid burden as measured by PET SUVR

MORE ONLINE

Class of EvidenceCriteria for ratingtherapeutic and diagnosticstudies

NPuborgcoe

From the CSIRO Health and BiosecurityAustralian E-Health Research Centre (JDD) Royal Brisbane and Womenrsquos Hospital Herston Queensland Australia RampD Department (VP-G NF PP MS) Araclon Biotech Ltd Zaragoza Spain and The Florey Institute of Neuroscience and Mental Health (CF VLV CLM) University of Melbourne Parkville VictoriaAustralia

Go to NeurologyorgN for full disclosures Funding information and disclosures deemed relevant by the authors if any are provided at the end of the article

AIBL Research Group coinvestigators are listed in the Appendix 2 at the end of the article

The Article Processing Charge was funded by Araclon Biotech-Grifols

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 40 (CC BY-NC-ND) which permits downloadingand sharing the work provided it is properly cited The work cannot be changed in any way or used commercially without permission from the journal

e1580 Copyright copy 2020 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology

The current shift in the Alzheimer disease (AD) paradigm istransforming the therapeutic target population for clinicaltrials from people with dementia or mild cognitive impair-ment (MCI) to cognitively healthy people at risk Thesepeople are not easy to find in a community setting whenβ-amyloid (Aβ) positivity is a criterion for eligibility and thescreening rate of failure (SRF) rises to gt70

Under these conditions the use of Aβ-PET scans for screeningrepresents a huge burden on the budget of any clinical trialhampering its feasibility Furthermore the complex logistichandling of radiotracers coupled with the low availability ofPET scanners can seriously limit the follow-up of a populationto determine when and to whom to administer a preventivetreatment once one becomes available CSF analysis may besignificantly less expensive but a lumbar puncture reduces itssuitability for periodic population assessment

Thus the discovery and development of accessible and in-expensive biomarkers that may help to enrich a population-based sample reducing the SRF for prevention trials has beennoted as a top research priority to prevent and to effectivelytreat AD in the shortest possible time frame12

In the last decade accruing experimental results have shownthat lower Aβ42Aβ40 (Aβ4240) plasma ratio is associatedwith higher amyloid cortical burden and steeper accumulationtrajectories3ndash11 greater cognitive decline12 or increased riskof developing AD dementia at follow-up13ndash17 However somestudies were unable to replicate these findings introducingcontroversy and casting doubts on the reliability of blood-based biomarkers18ndash21 Nevertheless this issue is being cur-rently elucidated by a better understanding of the interactionsof Aβ peptides within the complex plasma matrix and the useof the cortical amyloid status instead of the clinical diagnosisas the gold standard to assess Aβ blood-based biomarkers22

In line with this Araclonrsquos team developed an ELISA assay(ABtest Araclon Biotech Ltd Zaragoza Spain) to assess thefree (FP) total (TP) and bound (BP) Aβ40 and Aβ42 levels inplasma23 Recently we used the ABtest to determine theTP4240 FP4240 and BP4240 (the difference betweenTP and FP) in a subcohort of healthy controls (HCs) fromthe Australian Imaging Biomarkers and Lifestyle (AIBL)study3 In that study we found that lower Aβ4240 plasmaratios (particularly TP4240) were associated with higher

cortical Aβ burden and faster Aβ accumulation rates In thepresent study we aim to assess the reproducibility of theseratios particularly TP4240 which had previously showna better performance in separating people with or withoutPET-confirmed amyloid cortical pathology over 3 time pointsin a larger population sample spanning the disease continuum

MethodsStandard protocol approvals registrationsand patient consentsThe AIBL study was approved by the institutional ethicscommittees of Austin Health St Vincentrsquos Health Holly-wood Private Hospital and Edith Cowan University and allvolunteers gave written informed consent before participatingin the study (further information is available in reference 24)

Study populationA subset of samples from AIBL with information pertaining toneocortical amyloid burden from PET imaging over 3 timepoints (18 36 and 54 months) were selected The AIBL studywas initiated in 2006 with the express aim to identify thosebiomarkers that were both associated with and predictive of ADpathology and clinical disease To this end the current studyused plasma from a subselection of participants who werefollowed up over at least 54 months Other information col-lected and used in this study includes results from cognitiveassessments to derive the AIBL Preclinical Alzheimerrsquos Cog-nitive Composite score the Mini-Mental State Examinationthe Clinical Dementia Rating score APOE e4 allele status agesex and relative information pertaining to the PET imaging

The primary research question of the present work is to ex-plore the association between plasma TP4240 ratio andneocortical amyloid burden as measured by PET standardizeduptake value ratio (SUVR) which has received a Class IIclassification of evidence

Amyloid PET imagingPET information using the Pittsburgh compound B radiotracerwas collected for the cohort for at least 1 of the 3 time pointsWhen PET measurements were not collected at a corre-sponding time point data from the last known PET mea-surement were used In brief the quantitative representation ofneocortical amyloid plaques was determined by taking the sumof the spatially normalized PET images to create a standardized

GlossaryAβ = β-amyloid Aβ4240 = Aβ42Aβ40 ratio AD = Alzheimer disease AIBL = Australian Imaging Biomarkers and LifestyleAUC = area under the curve BP4220 = bound plasma Aβ42Aβ40 ratioCV = coefficient of variation FP4220 = free plasmaAβ42Aβ40 ratio GLMM = generalized linear mixed models HC = healthy control ICC = intraclass correlation coefficientMCI =mild cognitive impairmentNPV = negative predictive value PPV = positive predictive valueROC = receiver-operatingcharacteristic SRF = screening rate of failure SUVR = standardized uptake value ratio TP4220 = total plasma Aβ42Aβ40ratio

NeurologyorgN Neurology | Volume 94 Number 15 | April 14 2020 e1581

update value Standardized uptake values were normalized tothe cerebellar cortex25 to form the SUVR SUVR values werethen transformed to a binary scale (Aβ-PET+veAβ-PETminusve)via the precalculated threshold (14)

Plasma Aβ40 and Aβ42 quantificationPlasma samples were collected with ethylene-diamine-tetra-acetic acid used as the anticoagulant and conserved at minus70degCuntil analysis following AIBL procedures26 Levels of Aβ40 andAβ42 were quantified with the ABtest40 and ABtest42 re-spectively (Araclon Biotech Ltd) 2 validated colorimetricassays based on the sandwich ELISA technique as describedelsewhere23 Each plasma sample was analyzed both undilutedand diluted one-third in a proprietary samplestandard diluentspecifically formulated to disrupt the interactions between Aβand other plasma components As a result FP and TP levels ofAβ40 and Aβ42 were determined The difference between theconcentrations of TP and FP corresponds to the amyloidpeptide bound to plasma components (BP) The Aβ4240ratios in each of these plasma fractions (FP4240 TP4240and BP4240) were calculated with the TP4240 ratio beingthe target plasma biomarker assessed in this study Free andtotal Aβ plasma levels were always analyzed in duplicates andthe 4 determinations (FP40 TP40 FP42 and TP42) from 1sample were measured intra-assay to reduce variability and toavoid extra freezethaw cycles The analyses were always per-formed in a coded manner to ensure blindness of the operator

Plasma Aβ data used in this study come from 2 set of assayscarried out in June to July 20143 and January to February 2016respectively 98 samples from 33 individuals were included inboth sets of ELISAs and were used to assess test-retest re-producibility Different batches of these kits were used in eachELISA set The average coefficient of variation (CV) the av-erage relative difference (percent) and the intraclass correla-tion coefficient (ICC) between both sets of analyses werecalculated An ICC gt075 indicates excellent reproducibility04 le ICC le075 indicates fair to good reproducibility In ad-dition the percentage difference of the results in each pair ofrepeated samples was calculated as 100 times (repeatedminus original)original Following regulatory guidelines for validation of bio-assays27 the acceptability criteria for incurred sample reanalysis(4-6-30) is that at least 67 of the repeated samples resultsshould be within 30 of the original results

Statistical analysisSample demographic and clinical characteristics were in-vestigated with a range of statistical methods includingindependent-samples t test Kruskal-Wallis ranks test and theχ2 test with the Fisher exact approximation when necessaryPlasma TP4240 means were compared between Aβ-PETgroups with generalized linear models Random effects due tothe 2 assay sets were assessing with generalized linear mixedmodels (GLMMs) Assessment of potential confounders agesex and APOE e4 allele status was performed in sepa-rate analyses (within the GLMM) with clinical classificationadded as a surrogate for cognitive staging Receiver operating

characteristic (ROC) analyses was used to calculate thresholdsfrom fitted GLMM via the Youden28 method (Youden maxi-mum) Ensuing sensitivity specificity positive predictive value(PPV) and negative predictive value (NPV) were calculatedusing the chosen thresholds without adjustment for populationprevalence To ensure that the calculated thresholds had utilitywe let the ROC analyses find the optimum threshold given theYouden index at each time point and then used the averagethreshold for plotting correlation (Spearman ρ) and agreementresults (for both unadjusted and adjusted analyses) GLMMcomparisons were performed with analysis of variance analysesof deviance with p values determined with χ2 distribution with1 df29 Predictive ROC models were compared with theDeLong method30 Values of p from comparisons of biomarkermeans were compared with a Bonferroni-adjusted α value (α =005k number of tests the main biomarker TP4240 wasassessed at 3 time points k = 3 00167) All statistical analyseswere performed with the R Statistical Environment (R Foun-dation for Statistical Computing Vienna Austria)31

Data availabilityDeidentified participant data used for this article together with thestudy protocol and statisticalmethods used will bemade availableafter the article publication date and for 5 years for any scientist byrequest to the authors for the only purpose of assessing replica-bility of the results published in the present article

ResultsSample demographicsClinical characteristics including APOE e4 allele status Pre-clinical Alzheimerrsquos Cognitive Composite score Mini-MentalState Examination and Clinical Dementia Rating score alongwith sample demographics age and sex were assessed betweenAβ-PET+ve and Aβ-PETminusve groups at the 18- 36- and 54-month time points stratified for clinical classification Becausethere were no Aβ-PETminusve participants in the AD group nocomparisons were made At all time points there were moreAPOE e4ndashpositive participants in the Aβ-PET+ve groupscompared with the Aβ-PETminusve groups (table 1) Compared toAβ-PETminusve participants those in the Aβ-PET+ve group wereolder and had lower cognitive performance at all time pointshowever this varied between clinical classifications There wereno sex differences between Aβ-PET groups

Mean biomarker differences between Aβ-PET groupsFigure 1 shows that the TP4240 ratio is consistently lower inboth Aβ-PET+ve participants and those with AD comparedwithAβ-PETminusve participants and HCs across all time points Com-parisons of mean TP4240 biomarker values (mean plusmn SD) be-tween Aβ-PET groups with unadjusted p values are shown intable 2 (p values from adjustedmodels are presented in table e-1available from Dryad doiorg105061dryadf300d56) Forboth the complete study population and HC group only it isclear that at all time points the TP4240 ratio is significantly

e1582 Neurology | Volume 94 Number 15 | April 14 2020 NeurologyorgN

Table 1 Sample demographics per collection

Characteristic

All groups 18 mo HC only MCI only AD only

Totalsample Aβ2 Aβ+

Totalsample Aβ2 Aβ+

Totalsample Aβ2 Aβ+

Total sample(Aβ+)

No () 176 99 (56) 77 (44) 130 92 (71) 38 (29) 24 7 (29) 17(71)

22

Male n () 90 (51) 46 (46) 44 (57) 66 (51) 43 (47) 23 (61) 13 (54) 3 (43) 10(59)

11 (50)

Mean age (SD) y 737 (72) 727(69)

75 (74)a 729 (7) 723(68)

743(73)

78 (63) 773(76)

783(6)

737 (8)

APOE e4 carriagen ()

77 (44) 24 (24) 53 (69)b 50 (38) 24 (26) 26 (68)b 11 (46) 0 (0) 11(65)b

16 (73)

Mean PACC score(SD)

minus16 (42) 03(24)

minus42(47)b

03 (23) 06(21)

minus05(26)a

minus54 (23) minus4 (26) minus6 (2) minus99 (23)

MedianMMSE (IQR) 29 (28) 29 (2) 27 (4)b 29 (14) 29 (2) 29 (2) 27 (19) 28 (25) 27 (1) 24 (34)

Median CDR score(IQR)

0 (03) 0 (0) 05 (05)b

0 (01) 0 (0) 0 (0) 05 (01) 05 (0) 05 (0) 05 (04)

All groups 36 mo HC only MCI only AD only

Totalsample Aβ2 Aβ+

Totalsample Aβ2 Aβ+

Totalsample Aβ2 Aβ+

Total sample(Aβ+)

No () 169 94 (56) 75 (44) 123 85 (69) 38 (31) 22 9 (41) 13 (59) 24

Male n () 86 (51) 44 (47) 42 (56) 60 (49) 39 (46) 21 (55) 15 (68) 5 (56) 10 (77) 11 (46)

Mean age (SD) y 75 (71) 739(69)

763(71)a

743 (7) 736(69)

76 (69) 783 (68) 769(67)

792 (7) 751 (75)

APOE e4 carriagen ()

69 (41) 19 (20) 50 (67)b 40 (33) 17 (20) 23 (61)b 12 (55) 2 (22) 10 (77)a 17 (71)

Mean PACC score(SD)

minus17 (47) 03(25)

minus45(55)b

03 (25) 07(22)

minus08(27)b

minus52 (3) minus32(25)

minus71(22)b

minus115 (31)

Median MMSE(IQR)

285 (45) 29 (2) 27 (5)b 29 (12) 29 (2) 29 (25)a 27 (21) 28 (1) 25 (2)a 20 (69)

Median CDR score(IQR)

0 (05) 0 (0) 05 (1)b 0 (01) 0 (0) 0 (0) 05 (01) 05 (0) 05 (0) 1 (06)

All groups 54 mo HC only MCI only AD only

Totalsample Aβ2 Aβ+

Totalsample Aβ2 Aβ+

Totalsample Aβ2 Aβ+

Total sample(Aβ+)

No () 135 81 (60) 54 (40) 104 72(69)

32 (31) 17 9 (53) 8 (47) 14

Male n () 70 (52) 41 (51) 29 (54) 52 (50) 35(49)

17 (53) 12 (71) 6 (67) 6 (75) 6 (43)

Mean age (SD) y 769 (71) 759(7)

785(71)a

763 (71) 758(7)

775(73)

788 (72) 772(72)

805(72)

796 (66)

APOE e4 carriagen ()

53 (39) 18 (22) 35 (65)b 37 (36) 17(24)

20 (62)b 6 (35) 1 (11) 5 (62)a 10 (71)

Mean PACC score(SD)

minus09 (4) 05(26)

minus34(47)b

05 (24) 1 (2) minus08(28)b

minus47 (3) minus39(29)

minus56 (3) minus107 (14)

Median MMSE (IQR) 29 (37) 29 (2) 28 (4)b 29 (12) 30 (2) 29 (2) 28 (29) 29 (1) 265 (3) 21 (58)

Median CDR score(IQR)

0 (04) 0 (0) 0 (05)b 0 (0) 0 (0) 0 (0) 05 (0) 05 (0) 05 (0) 1 (05)

Abbreviations Aβ = β-amyloid AD = Alzheimer disease CDR = Clinical Dementia Rating HC = healthy control IQR = interquartile range MCI = mild cognitiveimpairment MMSE = Mini-Mental State Examination PACC = Preclinical Alzheimerrsquos Cognitive Compositea p lt 005 p values from testing between Aβminus and Aβ+ groupsb p lt 001 p values from testing between Aβminus and Aβ+ groups

NeurologyorgN Neurology | Volume 94 Number 15 | April 14 2020 e1583

lower in the Aβ-PET+ve group compared with the Aβ-PETminusvegroup (unadjusted marginal means) The degree of these differ-ences however is clearly affected by the confounding factors andsignificance is diminished in adjusted models (18-month un-adjusted p = 000001 adjusted p = 0057 table e-1) The con-founders APOE e4 allele status and clinical classification wereassociated with PET-Aβ status in all models (p lt 002) Assessingthe HC group specifically we saw similar relationships indicatingthat the TP4240 ratio in plasma is consistent with cerebral am-yloid pathology not with the clinical stage although a tendency tolower levels in patients with AD is appreciated for the plasma ratio(figure 1) A comparison of the base model including only thecovariates age sexAPOE e4 allele status and clinical classificationwith the base model plus the TP4240 biomarker showed thatadding the biomarker performed significantly better than the basemodel alone at all time points (18months p= 000068 36monthsp = 0016 54 months p lt 00001)

Correlation and agreement between plasmaTP4240 ratio and SUVRAβ-PETWe approached the correlation and agreement between theplasma biomarker and PET biomarker in 2 separate ways (1)given the p value attenuation due to confounders (mainly

APOE e4 allele status and clinical classification) we used thefitted values from each model (fitted biomarker) to assessboth the correlation with SUVR and the agreement with Aβ-PET after threshold derivation (figure 2) and (2) we assessedthe raw plasma biomarker data against SUVR (figure e-1available from Dryad doiorg105061dryadf300d56)

The correlation between fitted model values including TP4240 and SUVR for the whole study population (figure 2 AndashC)was similarly strong at the 3 time points (ρ = asympminus063 minus064minus064 p lt 00001 at 18 36 and 54 months respectively)Similar correlations albeit slightly weaker (ρ = asympminus047 minus040minus050 p lt 00001 at 18 36 and 54 months respectively)were seen for the HC group only (figure 2 DndashF) Derivedthresholds from the adjusted model were 0433 0488 and0361 Taking the mean of these we plotted the overallthreshold (0428) on both the complete and HC data foragreement statistics calculations Correlation between the rawplasma biomarker and SUVR was weaker but still highlysignificant across all time points for all participants and theHC-only sample (figure e-1 available fromDryad doiorg105061dryadf300d56)

Figure 1 TP4240 plots for Aβ-PET and clinical classification groups

Box and whisker plots of total plasma β-amyloid (Aβ)42Aβ40 (TP4240) ratio between the 3 time points and (A) PET Aβ groups or (B) clinical classification Rawdata are presented on a box-and-whisker plot background Middle line of the box represents the median lower and upper lines represent first and thirdquartiles respectively In panel A blue represents those participants who are PET-Aβminusve and red represents those participants who are PET-Aβ+ve In panelB blue represents those participants who are in the healthy control (HC) group orange represents those participants who are in the mild cognitiveimpairment (MCI) group and red represents those participants who are in the Alzheimer disease (AD) group

e1584 Neurology | Volume 94 Number 15 | April 14 2020 NeurologyorgN

Table 3 shows the agreement statistics that align with thequartiles seen in figure 2 Overall agreement for TP4240 wasquite consistent throughout the 3 time points ranging be-tween 83 and 85 for the complete group and between 82and 86 in the HC group Regarding discriminating capa-bility each test had a better capability to predict those withsubthreshold cortical amyloid burden particularly in the HCgroup with NPVs ranging from 85 to 93 compared withthe PPVs which were lower (64ndash78) Agreement resultsfor the plasma biomarker alone were stronger in the HC-onlygroup than in the all-participants group with overall percentagreement higher by at least 2 at each time point (figure e-1and table e-2 available from Dryad doiorg105061dryadf300d56)

ROC analysesROC analyses for the TP4240 biomarker adjusted forconfounders (age sex APOE e4 allele status and clinicalclassification) are shown in figure 3 Results are consistent ateach time point with the adjusted area under the curve(AUC) for the TP4240 plasma ratio ranging from 088 to0913 for all groups and from 0808 to 0898 for the HCgroup The adjusted ROC models (DeLong method) con-taining the TP4240 biomarker were significantly strongerthan the base model (age sex APOE e4 allele status andclinical classification) at both 18 and 54 months (p = 0043and p = 0002 respectively) but not at 36 months (p =0497) (table e-3 available from Dryad doiorg105061dryadf300d56) Results for the HC group analyses weresimilar (18 months p = 002 36 months p = 0686 54 monthsp = 00007)

Test-retest studyBecause 98 samples of this study had already been analyzed ina previous study we could evaluate the test-retest re-producibility of our assay between both studies The results ofthe test-retest reproducibility study are summarized in table 4On average the CV between the 2 ELISA sets which were

separated by 18 months was lt20 for the 4 determinationsassayed (FP40 TP40 FP42 and TP42) The relative differ-ence between determinations in the 2 set of assays was alsolt20 for all of these determinations The ICC was gt08 forFP40 TP40 and FP42 exhibiting an excellent re-producibility TP42 presented an ICC = 0653 which is alsoconsidered fair to good In addition following regulatoryguidelines for the validation of ligand binding assays 3 of theAβ determinations (table 4) met acceptability criteria (4-6-30criteria) for incurred samples reanalysis while TP42 resultswere borderline

DiscussionIn this study we have found that the TP4240 plasma ratiois consistently associated with Aβ-PET status over the 3time points assayed Results from all time points showhighly significant lower plasma TP4240 ratio in the Aβ-PET+ve group than in the Aβ-PETminusve group before ad-justment for covariates (table e-1 available from Dryad doiorg105061dryadf300d56) We observed that signifi-cance was markedly reduced due to the strong associationbetween both APOE e4 allele status and clinical classifica-tion with Aβ-PET The role of the confounders in the as-sociation between the plasma biomarker and Aβ-PET statuscould be a concern if changes in both variables were due tocommon causes However assuming that TP4240 dependsexclusively on levels of brain Aβ burden (not from othersources) and that confounders such as clinical classificationare upstream to brain amyloid accumulation and conse-quently to both biomarkers we considered that pre-sentation of results without adjustment for confoundingfactors was necessary

On the other hand concerning the correlation and agree-ment of TP4240 with Aβ-PET and its performance fordiscriminating Aβ-PET status results are presented both

Table 2 Results from TP4240 plasma ratio comparisons between Aβ-PET groups

Fraction Collection mo Aβ2 n Aβ+ n Aβ2 mean (SD) Aβ+ mean (SD) p Value

All groups

TP4240 18 99 77 0092 (0027) 0075 (002) 000001

36 94 75 0094 (0028) 0078 (0022) 000001

54 81 54 0087 (0025) 0073 (0015) 00001

HC only

TP4240 18 92 38 0114 (0051) 0088 (0042) 00042

36 85 38 0111 (0043) 0092 (0059) 00132

54 72 32 010 (0043) 0073 (0025) 00029

Abbreviations Aβ = β-amyloid HC = healthy control TP4240 = total plasma Aβ42Aβ40 ratioUnadjusted comparison of marginal means without adjustment for confounding variables

NeurologyorgN Neurology | Volume 94 Number 15 | April 14 2020 e1585

Figure 2 Correlation and agreement between model fitted values including TP4240 and amyloid-PET SUVR

Correlation and threshold plot for the fitted values from the amyloid Aβ42Aβ40 (TP4240) model (generalized linear mixed model [GLMM]) includingcovariates as represented in table e-1 (available fromDryad doiorg105061dryadf300d56) (y-axis inverse with values close to 1 indicating high probabilitythat they are plasmaAβ-ve and values close to 0 indicating high probability that they are plasmaAβ+ ve) vs standardized uptake value ratio (SUVR) (x-axis) (AndashC) Relationships for all participants and (DndashF) relationships for healthy control (HC) participants only (A and D) Relationships at 18 months (B and E)relationships at 36 months and (C and F) relationships at 54 months Shown on each are the quadratic fit lines representing the relationship between fittedvalues from the TP4240model (from a full GLMM including adjustment for confounders) and SUVR Spearman ρ valuewith associated p value from the samedata is shown in the top right of each plot Circles represent those participants from the HC group squares represent those participants from the mildcognitive impairment group triangles represent those participants from the Alzheimer disease group Blue points represent those participants who are PET-Aβ-ve and red points represent those participants who are PET-Aβ+ve

e1586 Neurology | Volume 94 Number 15 | April 14 2020 NeurologyorgN

adjusted and unadjusted for the relevant covariates In thiscase we included the clinical classification such that anyperson with normal cognition MCI or AD can be tested forcortical amyloid positivity with our test in plasma Never-theless given the interest in blood-based biomarkers forsecondary prevention clinical trials and management ofpreclinical individuals we explored those variables in the HCgroup alone in which obviously clinical classification wasnot used

Along these lines the result of the adjusted model containingTP4240 showed a consistent inverse correlation with theSUVR at the 3 time points (ρ =asympminus063 p lt 00001) and anoverall agreement with Aβ-PET status ranging from 83 to85 with PPVs of 80 to 81 and NPVs of 86 to 88 Theresults for the HC group were similar at each time point(overall agreement with Aβ-PET status ranging from82ndash86) even outperforming the NPV (85ndash93) withregard to all participants (despite the reduction in the samplesize) which is very relevant for any screening test Theagreement of the unadjusted plasma ratio (table e-2 and figuree-1 available from Dryad doiorg105061dryadf300d56)with the Aβ-PET status was poorer ranging from 63 to 65in the all-participants group Nevertheless sensitivity(81ndash89) and NPV (64ndash84) in the HC group couldstill lead to a substantial reduction in the number of amyloid-PET scans in a screening scenario (see below)

These results are consistent with our previous findingsshowing an inverse association between the Aβ4240 plasma

ratios particularly TP4240 and cortical Aβ burden in anAIBL subcohort of cognitively normal controls3 and otherindependent cohorts1032 In agreement with this it has pre-viously been reported that the plasma Aβ4240 ratio corre-lated directly with Aβ42 CSF levels and inversely withAβndashPET SUVR4ndash9 Moreover our present results supportprevious community-based studies in healthy people report-ing an association between lower plasma Aβ4240 ratio andgreater cognitive decline12 or increased risk of developing ADdementia at follow-up13ndash17 More recently an associationbetween cortical Aβ burden and the Aβ4240 plasma ratio asdetermined by liquid chromatography tandem mass spec-trometry has also been found in HCs and patients with MCIand AD3334 Several studies have provided mechanisticdescriptions supporting this association by demonstrating theexistence of an Aβ-specific molecular transporter at the blood-brain barrier35ndash37 and positive clearance of Aβ peptides fromthe brain to the peripheral vasculature in humans3839 Thusmounting evidence coming from varied experimental designsand different analytical methods supports the existence of anassociation between the Aβ4240 plasma ratio and the cor-tical amyloid burden Furthermore the ROC curve analysis inthe present work demonstrated the reproducibility of ourplasma assay to separate Aβ-PET groups over 3 time pointswith an AUC for the TP4240ndashadjusted model ranging from0880 to 0913 in the all-participants group and from 0808 to0898 in the HC-only group Thus the consistency of ourassay demonstrated through follow-up confers reliability toTP4240 as a biomarker for cortical amyloidosis which isadditionally supported by the test-retest results obtained with

Table 3 Percentage agreement between TP4240 fitted model and Aβ-PET groups

Collection Plasma ratio Overall agreement Sensitivity Specificity PPV NPV

All groups mo

18 Base 8125 7013 8990 8438 7946

36 8284 6800 9468 9107 7876

54 7926 7037 8519 7600 8118

18 TP4240 8409 8312 8485 8101 8660

36 8343 8267 8404 8052 8587

54 8519 8333 8642 8036 8861

HC only mo

18 Base 7769 7105 8043 6000 8706

36 7642 6842 8000 6047 8500

54 7596 6562 8056 6000 8406

18 TP4240 8231 8684 8043 6471 9367

36 8211 6579 8941 7353 8539

54 8654 7812 9028 7813 9028

Abbreviations Aβ = β-amyloid HC = healthy control NPV = negative predictive value PPV = positive predictive value TP = total plasma Aβ42Aβ40 ratioldquoBaserdquo refers to the model including the demographic covariables but not the plasma TP4240 ratio

NeurologyorgN Neurology | Volume 94 Number 15 | April 14 2020 e1587

Figure 3 ROC curves for the TP4240 plasma ratio vs the base model

Plots show the receiver operating characteristic (ROC) curves from 2 models to predict PET β-amyloid (Aβ) status the base model with covariates only and the basemodel plus the total plasmaAβ42Aβ40 (TP4240) biomarker (A C and E) ROC curves fromall participants at 18 36 and53months respectively (B D and F) ROCcurvesfromhealthycontrol (HC) participantsonlyat 18 36 and53months Shownoneachpanel are theROCcurveswith the calculatedareaunder thecurve (AUC) value in thebottom right corner at 18 months (A and B) 36 months (C and D) and 54 months (E and F)

e1588 Neurology | Volume 94 Number 15 | April 14 2020 NeurologyorgN

the 98 samples assayed in the 2 sets of ELISAs carried out 15years apart The CV (lt20) for these test-retests was withinthe criteria recommended for interassay reproducibility andconfirm previous ABtest validation results following regula-tory agencies guidelines27

Various studies from other groups have failed to replicate thisassociation between plasma Aβ levels and clinical or patho-logic aspects of AD18ndash21 This disparity in results can beexplained at least in part by the use of the clinical diagnosis(instead of brain Aβ burden) as the gold standard to assessperformance of Aβ blood-based tests Our results show thatthe TP4240 plasma ratio is consistent with cortical amyloidpathology as visualized by PET and less so with the clinicaldiagnosis which itself has shown sensitivities ranging from709 to 873 and specificities from 443 to 7081140

This relatively poor performance for a gold standard can se-riously skew the results of any testing and is almost certainlya relevant source of variability between studies

Concordance of Aβ blood tests among different studies mayalso be hindered by the relatively small difference in theAβ4240 plasma ratio among Aβ-PET groups In the presentstudy the TP4240 was on average asymp17 lower in the Aβ-PET+ve participants than in the Aβ-PETminusve (asymp22 loweramong the HCs table 2) whereas in the CSF Aβ4240 ratiodifferences between those 2 groups are asymp5033 Thusstringent adherence to the protocols including preassayhandling of the samples is of maximum relevance to mini-mize the variability of determinations in the highly complexplasma matrix that may blur relatively small but meaningfuldifferences In this regard it deserves to be underlined thatcorrelations between TP4240 plasma levels and Aβ-PETSUVR found in the present study (ρ = asympminus063 p lt 00001)were stronger than those found in a recent study using anultrasensitive single-molecule assay (Aβ4240 vs [18F]flu-temetamol SUVR ρ = minus0167 p = 0002)4 On the otherhand the performance of the TP4240ndashadjusted model to

discriminate Aβ-PET status in the present work (AUCranging from 088ndash091) was very similar to that obtained byliquid chromatography tandem mass spectrometry (AUC088)3334 The concordance of our results with these studiesusing 2 technically different modes of assessments againstrongly supports the reliability of the plasma Aβ4240 ratiomeasurement as a biomarker for predicting amyloid-PETresults

Furthermore we have reported that in a recruitment scenariotargeting cognitively normal Aβ-PET+ve participants theTP4240 ratio could be used as a prescreening tool able toreduce the number of individuals undergoing Aβ-PET scansby asymp503 Assuming a 30 prevalence of Aβ-PET in the HCgroup of this particular study a recruitment based exclusivelyon amyloid-PET scans would need to test 500 individuals torecruit 150 Aβ-PET+ve with an SRF of asymp70 (150 chosen forthe sake of simplicity in the calculation any pivotal secondaryprevention clinical trial would most probably have a samplesize closer to an order of magnitude greater)

In the present study the average values for sensitivity and PPVfrom the 3 time points analyzed within the HC group were77 and 72 respectively Thus to find those 150 individualsusing our plasma prescreening tool (sensitivity 77) wewould need a population containing 195 Aβ-PET+ve (150 times10077) which at a 30 prevalence would mean 650 HCs(195 times 10030) Because our plasmamarker model had a 72PPV we would have 58 plasma false positives (150 times 10072)together with the pursued 150 Thus the total number ofamyloid-PET required to recruit those 150 individuals wouldbe reduced from 500 to 208 In addition this 2-step screeningstrategy would reduce the SRF at the amyloid-PET scan visitfrom 70 to asymp28 reducing substantially the patientsrsquo bur-den and overall budget for secondary prevention trials foramyloid-targeting therapies Moreover the high NPV (aver-age for the 3 time points asymp90 in the HC group) indicatesthat only a small fraction of the suitable Aβ-PET+ve

Table 4 Test-retest reliability between 2 ELISA sets of analysis

FP40 TP40 FP42 TP42

Repeated samples n 94 97 82 83

Interstudy variability CV 1492 1105 1821 1939

Relative difference 1609 591 482 1928

ICC 0804a 0893a 0850a 0653a

95 CI 0705ndash0870 0839ndash0928 0768ndash0904 0463ndash0775

Repeated samples results within the 30 of their originals 78 82 70 65

Abbreviations CI = confidence interval CV = coefficient of variation FP40 = free plasma β-amyloid40 FP42 = free plasma β-amyloid42 ICC = intraclasscorrelation coefficient TP40 = total plasma β-amyloid40 TP42 = total plasma β-amyloid42Data represent the mean value of n samples obtained from 33 different individuals at different time points Relative difference was calculated consideringthe results of 2014 as the reference a positive value implies that on average the concentration obtained in 2016 was higher Excellent reproducibility wasachieved for all markers regarding both ICC and 4-6-30 criteria for incurred samples except TP42 which was fair to gooda p lt 0001 in the correlation study

NeurologyorgN Neurology | Volume 94 Number 15 | April 14 2020 e1589

individuals will be missed during prescreening as plasma testfalse negative contributing to shortening of the recruitmentperiod

A strength of the present study is the large proportion of HCparticipants across all time points (asymp75) The Aβ-PETpositivity rate for this group across each of the 3 time points isasymp30 which is higher than previously published for othercohorts such as BioFinder and the Alzheimerrsquos Disease Neu-roimaging Initiative41 Given this higher prevalence of Aβ-PETpositivity it is possible that the PPV could have been over-estimated and the NPV underestimated However we usedthe PPV and NPV calculations without the addition of thesample or population prevalence such that it would be anunbiased calculation Furthermore the strong NPV values forthe plasma ratio in the HC-only group provide strong evi-dence for real-life inference across healthy and clinically im-paired people gt65 years of age

In addition we demonstrate the strength of the differences inTP4240 between Aβ-PET groups both with and withoutpotential confounders The presence of an APOE e4 allelewithin the modeling appeared to play quite a strong role inexplaining variance in Aβ-PET groups with higher prevalencein the Aβ-PET+ groups however its importance was de-creased in the later time point for both complete andHC-onlysamples Regulating Aβ aggregation and clearance in thebrain42 variants in the APOE gene are important to accountfor in biomarker analyses especially so when considering theage of participants Given these underlying associations withage APOE and amyloid it is important then to account forthese factors when looking at blood-based biomarkers help-ing to better understand a pathologic picture of the diseaseAdding a representative of the clinical form of the disease(ie clinical classification) improves our estimates of wherea person lies on the disease trajectory which is very useful inboth clinical trial design and the clinic

These results show that Aβ peptides can be measured inplasma with enough reproducibility and consistency to im-plement TP4240 as a reliable biomarker discriminating Aβ-PET status The use of TP4240 could facilitate the screeningprocess for preventive clinical trials in AD avoiding invasivetesting in a significant number of clinically healthy volunteersand saving substantial amounts of money and time

Nevertheless we acknowledge the potential weakness of thisstudy due to the relatively small sample size and the variabilityof determinations of Aβ in plasma largely related to thecharacteristics of the plasma matrix in each individual Thisvariability together with the differences and overlappinglevels of TP4240 between the Aβ-PET+ve and Aβ-PETminusvegroups still hampers the use of markers in plasma Largerpopulation studies should be addressed to overcome suchweakness and to demonstrate sufficient precision to be pro-spectively used in secondary prevention clinical trials

Study fundingThis work has been financed by Araclon Biotech Ltd

DisclosureJ Doecke reports no disclosures relevant to the manu-script V Perez-Grijalba and N Fandos are full-timeemployees at Araclon Biotech Ltd C Fowler reports nodisclosures relevant to the manuscript V Villemagne re-ceived funding for a trip to attend the CTAD 2018 con-gress C Masters reports no disclosures relevant to themanuscript P Pesini is a full-time employee at AraclonBiotech Ltd M Sarasa is a full-time employee at anda shareholder of Araclon Biotech Ltd Go to NeurologyorgN for full disclosures

Publication historyReceived by Neurology January 9 2019 Accepted in final formOctober 24 2019

Appendix 1 Authors

Name Location Contribution

James DDoecke PhD

CSIRO Health andBiosecurityAustralian E-Health Research CentreRoyal Brisbane andWomenrsquos HospitalHerston QueenslandAustralia

Contributed to designingthe study performed thestatistical analysis draftedthe manuscript forintellectual content wrotethe final version

VirginiaPerez-GrijalbaPhD

RampD Department AraclonBiotech Ltd ZaragozaSpain

Contributed to designingthe study major role inthe acquisition of data

NoeliaFandos PhD

RampD Department AraclonBiotech Ltd ZaragozaSpain

Major role in theacquisition of datarevised the manuscriptfor intellectualcontent

ChristopherFowler PhD

The Florey Institute ofNeuroscience and MentalHealth University ofMelbourne ParkvilleVictoria Australia

Major role in theacquisition of data

Victor LVillemagneMD PhD

The Florey Institute ofNeuroscience and MentalHealth University ofMelbourne ParkvilleVictoria Australia

Interpreted the datarevised the manuscript forintellectual content

Colin LMastersMD PhD

The Florey Institute ofNeuroscience and MentalHealth University ofMelbourne ParkvilleVictoria Australia

Revised the manuscript forintellectual content

PedroPesini PhD

RampD Department AraclonBiotech Ltd ZaragozaSpain

Contributed to designingthe study drafted themanuscript for intellectualcontent wrote the finalversion

ManuelSarasa PhD

RampD Department AraclonBiotech Ltd ZaragozaSpain

Revised the manuscript forintellectual content

e1590 Neurology | Volume 94 Number 15 | April 14 2020 NeurologyorgN

References1 Blennow K A review of fluid biomarkers for Alzheimerrsquos disease moving fromCSF to

blood Neurol Ther 2017615ndash242 Alber J Lee AKW Menard W Monast D Salloway SP Recruitment of at-risk par-

ticipants for clinical trials a major paradigm shift for Alzheimerrsquos disease preventionJ Prev Alzheimers Dis 20174213ndash214

3 Fandos N Perez-Grijalba V Pesini P et al Plasma amyloid beta 4240 ratios asbiomarkers for amyloid beta cerebral deposition in cognitively normal individualsAlzheimers Dement (Amst) 20178179ndash187

4 Janelidze S Stomrud E Palmqvist S et al Plasma beta-amyloid in Alzheimerrsquos diseaseand vascular disease Sci Rep 2016626801

5 Rembach A Faux NG Watt AD et al Changes in plasma amyloid beta in a longitu-dinal study of aging and Alzheimerrsquos disease Alzheimers Dement 20141053ndash61

6 Toledo JB Vanderstichele H Figurski M et al Factors affecting Abeta plasma levelsand their utility as biomarkers in ADNI Acta Neuropathol 2011122401ndash413

7 Devanand DP Schupf N Stern Y et al Plasma Abeta and PET PiB binding areinversely related in mild cognitive impairment Neurology 201177125ndash131

8 Lui JK Laws SM Li QX et al Plasma amyloid-beta as a biomarker in Alzheimerrsquosdisease the AIBL study of aging J Alzheimers Dis 2010201233ndash1242

9 Swaminathan S Risacher SL Yoder KK et al Association of plasma and corticalamyloid beta is modulated by APOE epsilon4 status Alzheimers Dement 201410e9ndashe18

10 Perez-Grijalba V Romero J Pesini P et al Plasma AB4240 ratio detects early stagesof Alzheimerrsquos disease and correlates with CSF and neuroimaging biomarkers in theAB255 study JPAD 2019634ndash41

11 Palmqvist S Janelidze S Stomrud E et al Performance of fully automated plasmaassays as screening tests for Alzheimer disease-related beta-amyloid status JAMANeurol Epub 2019 June 24

12 Yaffe K Weston A Graff-Radford NR et al Association of plasma beta-amyloid leveland cognitive reserve with subsequent cognitive decline JAMA 2011305261ndash266

13 Abdullah L Luis C Paris D et al Serum Abeta levels as predictors of conversion tomild cognitive impairmentAlzheimer disease in an ADAPT subcohort Mol Med200915432ndash437

14 Chouraki V Beiser A Younkin L et al Plasma amyloid-beta and risk of Alzheimerrsquosdisease in the Framingham Heart Study Alzheimers Dement 201511249ndash257

15 Graff-Radford NR Crook JE Lucas J et al Association of low plasma Abeta42Abeta40 ratios with increased imminent risk for mild cognitive impairment andAlzheimer disease Arch Neurol 200764354ndash362

16 Lambert JC Schraen-Maschke S Richard F et al Association of plasma amyloid betawith risk of dementia the prospective Three-City Study Neurology 200973847ndash853

17 van OijenM Hofman A Soares HD Koudstaal PJ Breteler MM Plasma Abeta(1-40)and Abeta(1-42) and the risk of dementia a prospective case-cohort study LancetNeurol 20065655ndash660

18 Hansson O Zetterberg H Vanmechelen E et al Evaluation of plasma Abeta(40) andAbeta(42) as predictors of conversion to Alzheimerrsquos disease in patients with mildcognitive impairment Neurobiol Aging 201031357ndash367

19 Le Bastard N Aerts L Leurs J BlommeW De Deyn PP Engelborghs S No correlationbetween time-linked plasma and CSF Abeta levels Neurochem Int 200955820ndash825

20 Lopez OL Kuller LH Mehta PD et al Plasma amyloid levels and the risk of AD innormal subjects in the Cardiovascular Health Study Neurology 2008701664ndash1671

21 LovheimH Elgh F Johansson A et al Plasma concentrations of free amyloid-beta cannotpredict the development of Alzheimerrsquos disease Alzheimers Dement 201713778ndash782

22 Pesini P Sarasa M Beyond the controversy on Aszlig blood-based biomarkers J Prev AlzDis 2015251ndash55

23 Perez-Grijalba V Fandos N Canudas J et al Validation of immunoassay-based toolsfor the comprehensive quantification of Abeta40 and Abeta42 peptides in plasmaJ Alzheimers Dis 201654751ndash762

24 Ellis KA Bush AI Darby D et al The Australian Imaging Biomarkers and Lifestyle (AIBL)study of aging methodology and baseline characteristics of 1112 individuals recruited fora longitudinal study of Alzheimerrsquos disease Int Psychogeriatr 200921672ndash687

25 Lopresti BJ Klunk WE Mathis CA et al Simplified quantification of Pittsburghcompound B amyloid imaging PET studies a comparative analysis J Nucl Med 2005461959ndash1972

26 Rembach A Watt AD Wilson WJ et al Plasma amyloid-beta levels are significantlyassociated with a transition toward Alzheimerrsquos disease as measured by cognitivedecline and change in neocortical amyloid burden J Alzheimers Dis 20144095ndash104

27 Hoffman D Statistical considerations for assessment of bioanalytical incurred samplereproducibility AAPS J 200911570ndash580

28 Youden WJ Index for rating diagnostic tests Cancer 1950332ndash3529 Hastie TJ Pregibon D Generalized linear models In Chambers JM Hastie TJ

editors Statistical Models in S Wadsworth amp BrooksCole 1992chapter 630 DeLong ER DeLong DM Clarke-Pearson DL Comparing the areas under two or

more correlated receiver operating characteristic curves a nonparametric approachBiometrics 198844837ndash845

31 R Computing Team A Language and Environment for Statistical Computing ViennaR Foundation for Statistical Computing 2016

32 De Rojas I Romero J Rodriguez-Gomez O et al Correlations between plasma andPET beta-amyloid levels in individuals with subjective cognitive decline the FundacioACE Healthy Brain Initiative (FACEHBI) Alzheimerrsquos Res Ther 201810119

33 Ovod V Ramsey KN Mawuenyega KG et al Amyloid beta concentrations and stableisotope labeling kinetics of human plasma specific to central nervous system amy-loidosis Alzheimerrsquos Demen 201713841ndash849

34 Nakamura A Kaneko N Villemagne VL et al High performance plasma amyloid-betabiomarkers for Alzheimerrsquos disease Nature 2018554249ndash254

35 Cirrito JR Deane R Fagan AM et al P-glycoprotein deficiency at the blood-brainbarrier increases amyloid-beta deposition in an Alzheimer disease mouse model J ClinInvest 20051153285ndash3290

36 Sagare A Deane R Bell RD et al Clearance of amyloid-beta by circulating lipoproteinreceptors Nat Med 2007131029ndash1031

37 Shibata M Yamada S Kumar SR et al Clearance of Alzheimerrsquos amyloid-Ab(1-40)peptide from brain by LDL receptor-related protein-1 at the blood-brain barrier J ClinInvest 20001061489ndash1499

38 Mawuenyega KG Sigurdson W Ovod V et al Decreased clearance of CNS beta-amyloid in Alzheimerrsquos disease Science 20103301774

39 Roberts KF Elbert DL Kasten TP et al Amyloid-beta efflux from the CNS into theplasma Ann Neurol 201476837ndash844

40 Beach TG Monsell SE Phillips LE Kukull W Accuracy of the clinical diagnosis ofAlzheimer disease at National Institute on Aging Alzheimer Disease centersJ Neuropathol Exp Neurol 201271266ndash273

41 Hansson O Seibyl J Stomrud E et al CSF biomarkers of Alzheimerrsquos disease concordwith amyloid-beta PET and predict clinical progression a study of fully automatedimmunoassays in BioFINDER and ADNI cohorts Alzheimers Dement 2018141470ndash1481

42 Zhong N Weisgraber KH Understanding the association of apolipoprotein E4 withAlzheimer disease clues from its structure J Biol Chem 20092846027ndash6031

Appendix 2 Coinvestigators

Name Location Role Contribution

David Ames University ofMelbourneVictoria Australia

AIBL study leader Overalldesign andmanagementof AIBL study

BelindaBrown

MurdochUniversityWestern Australia

Part of thelifestyle streamwithin AIBL

Overalldesign andmanagementof AIBL study

RalphMartins

Edith CowanUniversityWestern Australia

Leads thediagnostics andbiomarkersprogram aimedat early-stageidentification

Overalldesign andmanagementof AIBL study

Paul Maruff Professor at FloreyInstitute ofNeuroscience andMental HealthVictoria Australia

Cochair of theAIBL clinicalworking group

Overalldesign andmanagementof AIBL study

StephanieRainey-Smith

Edith CowanUniversityWestern Australia

Coordinated theWesternAustralian arm ofAIBL

Overalldesign andmanagementof AIBL study

ChristopherRowe

Department ofMolecular Imagingand TherapyAustin Health andUniversity ofMelbourneVictoria Australia

Imaging leaderfor AIBL

Overalldesign andmanagementof AIBL study

OlivierSalvado

Australian eHealthResearch CentreRoyal Brisbane andWomenrsquos Hospital

Leading theCSIRO BiomedicalInformaticsGroup

Overalldesign andmanagementof AIBL study

KevinTaddei

Edith CowanUniversityWestern Australia

ResearchManager withinthe NeuroscienceResearch Group

Overalldesign andmanagementof AIBL study

Larry Ward MedicineDevelopment forGlobal HealthSouthbankVictoria Australia

Head of businessdevelopment forAIBL

Overalldesign andmanagementof AIBL study

NeurologyorgN Neurology | Volume 94 Number 15 | April 14 2020 e1591

DOI 101212WNL0000000000009240202094e1580-e1591 Published Online before print March 16 2020Neurology

James D Doecke Virginia Peacuterez-Grijalba Noelia Fandos et al AD diagnosis

ratio in plasma predicts amyloid-PET status independent of clinical40βA42βTotal A

This information is current as of March 16 2020

ServicesUpdated Information amp

httpnneurologyorgcontent9415e1580fullincluding high resolution figures can be found at

References httpnneurologyorgcontent9415e1580fullref-list-1

This article cites 39 articles 6 of which you can access for free at

Citations httpnneurologyorgcontent9415e1580fullotherarticles

This article has been cited by 1 HighWire-hosted articles

Subspecialty Collections

y_cohort_case_controlhttpnneurologyorgcgicollectionclinical_trials_observational_studClinical trials Observational study (Cohort Case control)

httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers disease

httpnneurologyorgcgicollectionall_clinical_neurologyAll Clinical Neurologyfollowing collection(s) This article along with others on similar topics appears in the

Permissions amp Licensing

httpwwwneurologyorgaboutabout_the_journalpermissionsits entirety can be found online atInformation about reproducing this article in parts (figurestables) or in

Reprints

httpnneurologyorgsubscribersadvertiseInformation about ordering reprints can be found online

ISSN 0028-3878 Online ISSN 1526-632XWolters Kluwer Health Inc on behalf of the American Academy of Neurology All rights reserved Print1951 it is now a weekly with 48 issues per year Copyright Copyright copy 2020 The Author(s) Published by

reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology