Professional Documents
Culture Documents
Supplemental content
IMPORTANCE Visual assessment of amyloid positron emission tomographic (PET) images has
been approved by regulatory authorities for clinical use. Several immunoassays have been
developed to measure -amyloid (A) 42 in cerebrospinal fluid (CSF). The agreement
between CSF A42 measures from different immunoassays and visual PET readings may
influence the use of CSF biomarkers and/or amyloid PET assessment in clinical practice and
trials.
OBJECTIVE To determine the concordance between CSF A42 levels measured using 5
different immunoassays and visual amyloid PET analysis.
DESIGN, SETTING, AND PARTICIPANTS The study included 262 patients with mild cognitive
impairment or subjective cognitive decline from the Swedish BioFINDER (Biomarkers for
Identifying Neurodegenerative Disorders Early and Reliably) cohort (recruited from
September 1, 2010, through December 31, 2014) who had undergone flutemetamol F 18
([18F]flutemetamol)labeled PET. Levels of CSF A42 were analyzed using the classic
INNOTEST and the newer modified INNOTEST, fully automated Lumipulse (FL), EUROIMMUN
(EI), and Meso Scale Discovery (MSD) assays. Concentrations of CSF A were assessed using
an antibody-independent mass spectrometrybased reference measurement procedure.
MAIN OUTCOMES AND MEASURES The concordance of CSF A42 levels and A42:A40 and
A42:tau ratios with visual [18F]flutemetamol PET status.
RESULTS Of 262 participants (mean [SD] age, 70.9 [5.5] years), 108 were women (41.2%) and
154 were men (58.8%). The mass spectrometryderived A42 values showed higher
correlations with the modified A42-INNOTEST (r = 0.97), A42-FL (r = 0.93), A42-EI
(r = 0.93), and A42-MSD (r = 0.95) assays compared with the classic A42-INNOTEST assay
(r = 0.88; P .01). The signal in the classic A42-INNOTEST assay was partly quenched by
recombinant A1-40 peptide. However, the classic A42-INNOTEST assay showed better
concordance with visual [18F]flutemetamol PET status (area under the receiver operating
characteristic curve [AUC], 0.92) compared with the newer assays (AUCs, 0.87-0.89;
P .01). The accuracies of the newer assays improved significantly when A42:A40 (AUCs,
0.93-0.95; P .01), A42 to total tau (T-tau) (AUCs, 0.94; P .05), or A42 to
phosphorylated tau (P-tau) (AUCs, 0.94-0.95; P .001) ratios were used. A combination of
the A42:A40 ratio and T-tau or P-tau level did not improve the accuracy compared with the
ratio alone.
CONCLUSIONS AND RELEVANCE Concentrations of CSF A42 derived from the new
immunoassays (modified INNOTEST, FL, EI, and MSD) may correlate better with the
antibody-independent mass spectrometrybased reference measurement procedure and
may show improved agreement with visual [18F]flutemetamol PET assessment when using
the A42:A40 or A42:tau ratios. These findings suggest the benefit of implementing the Author Affiliations: Author
CSF A42:A40 or A42:tau ratios as a biomarker of amyloid deposition in clinical practice affiliations are listed at the end of this
article.
and trials.
Corresponding Author: Shorena
Janelidze, PhD, Clinical Memory
Research Unit, Department of Clinical
Sciences Malm, Lund University,
Slvegatan 19, BMC Building B11,
JAMA Neurol. doi:10.1001/jamaneurol.2017.2814 221 84 Lund, Sweden
Published online November 6, 2017. (shorena.janelidze@med.lu.se).
(Reprinted) E1
2017 American Medical Association. All rights reserved.
C
erebrospinal fluid (CSF) -amyloid (A) 42 and amy-
loid positron emission tomography (PET) have proved Key Points
to have high diagnostic accuracy for Alzheimer dis-
Question How well do the core cerebrospinal fluid biomarkers of
ease (AD) years before the onset of clinical symptoms and are Alzheimer disease measured using different immunoassays agree
becoming an important part of the diagnostic workup.1 Both with visual amyloid positron emission tomographic assessment?
biomarkers have shown correlations with postmortem plaque
Findings In this study of 262 individuals with mild cognitive
measurements,2-5 and previous studies have indicated 80%
complaints, the cerebrospinal fluid A42:A40 or A42:tau ratios
to 90% agreement between CSF A42 values and quantita- from several newer assays showed improved accuracy for
tive amyloid PET data.6-8 For most immunoassays, the con- detection of cortical -amyloid fibrils as measured by positron
cordance between the 2 biomarker modalities is further im- emission tomography.
proved with the CSF A42:A40 ratio (compared with CSF
Meaning These findings support implementation of the
A42 level alone), probably because the ratio corrects for cerebrospinal fluid A42:A40 and/or A42:tau ratios as
(1) interindividual variability in the overall A production; biomarkers of amyloid deposition in clinical practice and trials.
(2) interindividual variability in the CSF turnover; (3) changes
in global levels of all A isoforms owing to non-ADrelated
abnormal findings, such as white matter lesions; and (4) vari- immunoassays, the performance of A42 and the A42:
ability owing to preanalytical factors.9-11 A40, A42 to total tau (T-tau), and A42 to phosphorylated
Several enzyme-linked immunosorbent assays (ELISAs) are tau (P-tau) ratios were determined using visual assessment
commonly used for measurements of A42 levels in CSF. In of flutemetamol F 18labeled PET images as the criterion
general, the precision of an immunoassay may be compro- standard. Furthermore, an important advancement in
mised by matrix interference when endogenous biological fac- the CSF biomarker field was the recent development of a mass
tors in the sample interact with the analyte of interest or non- spectrometrybased reference measurement procedure
specifically bind to antibodies. Matrix interference affecting (RMP),20 which allows antibody-independent quantification
A42 quantification has been reported for the initially devel- of absolute CSF concentration of A42 with high accuracy and
oped INNOTEST A42 assay12 but seems to be minimized in the is now considered to be the criterion standard method
newer A immunoassays from EUROIMMUN, Fujirebio, Meso for CSF A42 measurement. 1 However, implementation
Scale Discovery, and Roche.13-15 A recent study suggested that of mass spectrometry in routine clinical laboratories is chal-
the diagnostic accuracy of CSF A biomarkers may vary de- lenging owing to the complexity of the technology, the need
pending on the assays used.16 More specifically, CSF A42 mea- for in-house method development and validation and per-
sured with the classic INNOTEST (Fujirebio) kit17 performed bet- sonal training, and high costs. Herein, we explored correla-
ter in distinguishing patients with early-stage AD from tion between CSF levels of amyloid biomarkers analyzed using
cognitively healthy control individuals than one of the newer immunoassays and antibody-independent mass spectrometry
immunoassays. At same time, using the CSF A42:A40 ratio based RMP.
improved the diagnostic accuracy of the newer EUROIMMUN
(EI; EUROIMMUN) A42 assay but not the classic A42
INNOTEST assay.16 Another report9 demonstrated higher con-
cordance between quantitative amyloid PET and the A42:
Methods
A40 ratio compared with CSF A42 level alone for the EI and Study Participants
Meso Scale Discovery (MSD; Meso Scale Discovery) assays. The study population included 262 patients with mild cogni-
In most of the studies that have compared amyloid PET tive impairment or subjective cognitive decline and no de-
with CSF A42 levels, the quantification of regional and global mentia from the prospective and longitudinal Swedish
amyloid PET ligand binding has used semiquantitative image BioFINDER (Biomarkers for Identifying Neurodegenerative Dis-
assessments of standardized uptake value ratio (SUVR). In clini- orders Early and Reliably) cohort (http://www.biofinder.se) who
cal practice, however, the standard approved by regulatory au- had undergone [ 18 F]flutemetamol PET evaluation from
thorities is visual classification of scans on a binary scale as September 1, 2010, through December 31, 2014. The diagnostic
positive or negative for amyloid, with a negative finding indi- criteria and cohort characteristics are described in the
cating that AD is unlikely. Visual readings of PET images agree eMethods and eTable 1 in the Supplement. The study was
with PET SUVR values and with brain amyloid burden at approved by the Regional Ethical Review Board in Lund,
autopsy.18,19 Nevertheless, concordance between the visual Sweden, and all study participants gave written informed
reading of PET and CSF A42 measures from different immu- consent to participate in the study.
noassays has not been established and may influence the use
of amyloid PET and/or CSF analysis in clinical practice and CSF Sampling and Analysis
trials. In the present study, we investigated the agreement be- The procedure and analysis of CSF samples were conducted
tween visual amyloid PET analysis and CSF A42 measured according to the Alzheimer Association Flowchart for CSF
using 5 different A42 immunoassays or protocols biomarkers.21 Lumbar CSF samples were collected at 3 memory
(INNOTEST A1-42 classical procedure, INNOTEST A1-42 clinics at the Skne University Hospital in Lund and Malm and
modified procedure, fully automated Lumipulse [FL; ngelholm Hospital in Sweden and analyzed according to a
Fujirebio], EI A1-42, and MSD AN-42). For each of the 5 A42 standardized protocol.8,21
Table 2. ROC Analysis of CSF As for Distinguishing Abnormal From Normal Visual Reading Assessments of Flutemetamol F 18Labeled PET
Cutoff (Cutoff for Mixture Youden J
Immunoassaya AUC (95% CI) Modeling Analysis)b Index Sensitivity Specificity
Classic A42-INNOTEST 0.92 (0.89-0.95) 548.00 (524.00) 0.78 0.96 0.82
Classic INNOTEST A42:A40 ratio 0.92 (0.88-0.95) 0.06 (0.05) 0.73 0.91 0.82
Classic INNOTEST A42:T-tau ratio 0.94 (0.91-0.97)c 1.79 (1.23) 0.84 0.96 0.87
Classic INNOTEST A42:P-tau ratio 0.95 (0.92-0.98)c,d 10.30 (8.11) 0.83 0.95 0.89
Modified A42-INNOTEST 0.87 (0.83-0.91)e 1091.00 (1129.00) 0.67 0.92 0.74
Modified INNOTEST A42:A40 ratio 0.93 (0.90-0.96)f 0.12 (0.10) 0.79 0.92 0.87
Modified INNOTEST A42:T-tau ratio 0.94 (0.91-0.97)g 3.30 (2.38) 0.83 0.96 0.88
Modified INNOTEST A42:P-tau ratio 0.95 (0.92-0.97)g 19.60 (16.50) 0.82 0.95 0.87
A42-EI 0.88 (0.84-0.92)e 449.00 (479.00) 0.63 0.82 0.80
A42-EI:A40 ratio 0.93 (0.90-0.96)h 0.10 (0.10) 0.81 0.93 0.88
A42-EI:T-tau ratio 0.94 (0.90-0.97)h 1.44 (1.11) 0.81 0.96 0.86
A42-EI:P-tau ratio 0.94 (0.91-0.96)i 9.59 (7.26) 0.81 0.95 0.87
A42-MSD 0.89 (0.85-0.93)j 506.00 (500.00) 0.70 0.94 0.76
A42-MSD:A40 ratio 0.95 (0.93-0.98)k 0.08 (0.09) 0.86 0.96 0.89
A42-MSD:T-tau ratio 0.94 (0.91-0.97)l 1.39 (1.12) 0.85 0.96 0.89
A42-MSD:P-tau ratio 0.95 (0.92-0.98)k 9.89 (7.84) 0.83 0.96 0.87
Abbreviations: A, amyloid ; AUC, area under the curve; CSF, cerebrospinal d
P .05 compared with classic A42-INNOTEST level.
fluid; EI, EUROIMMUN; MSD, Meso Scale Discovery; PET, positron emission e
P .001 compared with classic A42-INNOTEST level.
tomography; P-tau, phosphorylated tau; ROC, receiver operating characteristic; f
P .01 compared with modified A42-INNOTEST level.
T-tau, total tau.
a
g
P .001 compared with modified A42-INNOTEST level.
Cerebrospinal fluid levels of T-tau and P-tau were quantified with
EUROIMMUN and INNOTEST enzyme-linked immunosorbent assay kits,
h
P .01 compared with A42-EI level.
respectively. These tau values and A42 values from the 5 different assays i
P .001 compared with A42-EI level.
(classic and modified INNOTESTs, fully automated Lumipulse, EI, and MSD) j
P .01 compared with classic A42-INNOTEST level.
were used for the calculations of the CSF A42:tau ratios.
b
k
P .001 compared with A42-MSD level.
Cutoffs for A42 are given in pg/mL. l
P .05 compared with A42-MSD level.
c
P .01 compared with classic INNOTEST A42:A40 ratio.
Concordance Between Visual and Quantitative not the A42:A40 or the A42:T-tau ratios exhibited a sig-
[18F]flutemetamol PET Assessments nificantly higher AUC than A42 alone in predicting visual clas-
With use of visual readings of [18F]flutemetamol PET, 113 pa- sification of [18F]flutemetamol PET (Table 2, eTable 2 in the
tients (43.1%) were classified as A positive and 149 patients Supplement, and Figure 1C). For all the newer assays (modi-
(56.9%) as A negative. When applying the SUVR cutoff, 132 fied A42-INNOTEST, A42-EI, and A42-MSD), all 3 ratios (ie,
patients (50.4%) showed abnormal amyloid deposition, A42:A40, A42:T-tau, and A42:P-tau) were comparable and
whereas composite SUVR values were within the reference significantly more accurate than A42 levels alone (Table 2,
range in 130 patients (49.6%). The Cohen was 0.81, suggest- eTable 2 in the Supplement, and Figure 1D-F). The results for
ing very good agreement between the visual and quantitative the A42-FL assay were comparable to those of the modified
[18F]flutemetamol PET assessment methods.27 Visual and A42-INNOTEST assay and are shown in eFigure 1 in the
quantitative [18F]flutemetamol PET data disagreed in 25 cases Supplement. Furthermore, using logistic regression analysis,
(9.5%), of which 22 (8.4%) were A positive based on the SUVR we found that a combination of the A42:A40 ratio and
cutoff but A negative based on visual analysis (Figure 1A). T-tau (AUCs, 0.92-0.95) or P-tau (AUCs, 0.92-0.93) for any of
the 4 A42 assays did not have better concordance with vi-
Visual A PET vs CSF AD Biomarker Assessments sual [18F]flutemetamol PET classification than the A42:
We investigated how accurately CSF A42 levels could pre- A40 ratio alone (AUCs, 0.92-0.95; P > .14 when comparing the
dict visual [18F]flutemetamol PET assessment when using the AUC of 2 ROC curves). The performance of T-tau and P-tau was
commercially available classic A42-INNOTEST assay or the similar in all the statistical tests and, therefore, only data for
newer modified A42-INNOTEST, A42-EI, and A42-MSD ass- P-tau are presented hereinafter.
says. The results of ROC curve analysis are shown in Table 2
and eTable 2 in the Supplement. We found that classic A42- Cutoff for the Different Immunoassays
INNOTEST was more accurate than the newer assays (modi- The A42 levels in CSF are most often bimodally distributed,28
fied A42-INNOTEST, A42-EI, and A42-MSD) in distinguish- and more reliable cutoffs likely can be derived when bio-
ing individuals with normal and abnormal visual readings marker levels show 2 clearly distinct populations.29 Fre-
(Figure 1B). Next, we studied whether the A42:A40, A42: quency plots revealed improved bimodal separation of
T-tau, or A42:P-tau ratios improved the accuracy of the 4 as- [18F]flutemetamol PETpositive and negative populations
says. For the classic A42-INNOTEST, only the A42:P-tau but based on the A42:A40 and A42:P-tau ratios compared with
Figure 1. Cerebrospinal Fluid (CSF) Alzheimer Disease (AD) Biomarkers and Flutemetamol F 18Labeled Positron Emission Tomography (PET)
3.5 1.0
3.0
0.8
[18F]flutemetamol SUVR
2.5
0.6
Sensitivity
2.0
0.4
1.5 A42-INC, AUC = 0.92
A42-INM, AUC = 0.87
0.2 A42-EI, AUC = 0.88
1.0
A42-MSD, AUC = 0.89
0.5 0
Negative Positive 0 0.2 0.4 0.6 0.8 1.0
Visual PET Analysis 1Specificity
1.0 1.0
0.8 0.8
0.6 0.6
Sensitivity
Sensitivity
0 0
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
1Specificity 1Specificity
E ROC curve for A42:T-tau ratio F ROC curve for A42:P-tau ratio
1.0 1.0
0.8 0.8
0.6 0.6
Sensitivity
Sensitivity
0 0
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
1Specificity 1Specificity
A, Concordance between visual and quantitative amyloid PET analysis. used A40 kits from the assay vendors Fujirebio (INC and INM), EUROIMMUN
Composite standardized uptake value ratio (SUVR) of [18F]flutemetamol data in (EI), and Meso Scale Discovery (MSD). Levels of CSF total tau (T-tau) and
patients who were classified as negative (n = 149) or positive (n = 113) for phosphorylated tau (P-tau) were measured with the EI and INNOTEST assays,
amyloid findings with visual PET assessment are shown. The dotted line respectively. Receiver operating characteristic (ROC) curves were generated for
represents a cutoff of greater than 1.42 SUVR. Data points indicate individuals. A42 level (B-F), the A42:A40 ratio (C [INC] and D [INM, EI, and MSD]), the
B-F, Correlation of CSF AD biomarkers with [18F]flutemetamol PET status A42:T-tau ratio (C [INC] and E [INM, EI, and MSD]), and the A42:P-tau ratio
according to visual analysis. Levels of CSF -amyloid (A) 42 were analyzed with (C [INC] and F [INM, EI, and MSD]) to determine their accuracy in differentiating
the classic INNOTEST (INC), modified INNOTEST (INM), EUROIMMUN (EI) and A-negative (n = 149) and A-positive (n = 113) visual readings. AUC indicates
Meso Scale Discovery (MSD) assays. To calculate the CSF A42:A40 ratio, we area under the ROC curve.
Figure 2. Frequency Plots of Cerebrospinal Fluid (CSF) -Amyloid (A) 42 Levels and A42:A40 and A42 to Phosphorylated Tau (P-tau) Ratios
No. of Values
No. of Values
No. of Values
20
15
10
10
10
5
5
0 0 0
0 500 1000 1500 0 0.05 0.10 0.15 0.20 0 10 20 30 40
A42-INC, pg/mL A42-INC:A40 A42-INC:P-tau
30 20 20
15 15
No. of Values
No. of Values
No. of Values
20
10 10
10
5 5
0 0 0
0 1000 2000 3000 4000 0 0.1 0.2 0.3 0 20 40 60
A42-INM, pg/mL A42-INM:A40 A42-INM:P tau
15
No. of Values
No. of Values
No. of Values
20 20
10
10 10
5
0 0 0
0 500 1000 1500 2000 0 0.05 0.10 0.15 0.20 0.25 0 5 10 15 20 25 30 35
A42-EI, pg/mL A42-EI:A40 A42-EI:P-tau
20
No. of Values
No. of Values
No. of Values
20 20
15
10
10 10
5
0 0 0
0 500 1000 1500 0 0.05 0.10 0.15 0.20 0 10 20 30 40
A42-MSD, pg/mL A42-MSD:A40 A42-MSD:P tau
Histograms of frequency distribution for CSF A42 levels and the A42:A40 with the Youden J index. EI indicates EUROIMMUN; INC, classic INNOTEST;
and A42:P-tau ratios across groups with A-positive (n = 113) and A-negative INM, modified INNOTEST; and MSD, Meso Scale Discovery.
(n = 149) visual ratings. Veritical dashed lines indicate cutoff points associated
the A42 measurement alone when using the modified A42- come. Mixture modeling, which is independent of the
INNOTEST, A42-EI, and A42-MSD assays (Figure 2D-F, G-I, standard of truth (ie, PET imaging results), produced similar
and J-L). This effect was less pronounced for the classic A42- cutoff values (Table 2). Analytical imprecision and within-
INNOTEST assay (Figure 2A-C). The cutoffs for each bio- participant variability may increase the risk for misclassifica-
marker were determined based on the Youden J index in which tion, especially when sensitivity and/or specificity markedly
visual read of [18F]flutemetamol PET was used as the out- decrease around cutoff points. In our study, the trends for sen-
sitivity and/or specificity near the cutoff points appeared to the newer modified A42-INNOTEST, A42-FL, A42-EI, and
be less steep for the A42:A40 and A42:P-tau ratios than for A42-MSD immunoassays. However, the CSF A42:A40
A42 levels alone when A42 was measured using the newer ratios from these assays performed better than the correspond-
assays (modified A42-INNOTEST, A42-EI, or A42-MSD) ing A42 levels alone and had comparable performance to the
(eFigure 2B-D, F-H, and J-L in the Supplement), but this was classic A42-INNOTEST level. The accuracy of the classic A42-
not the case for the classic A42-INNOTEST assay (eFigure 2A, INNOTEST assay did not improve by using the A42:A40 ra-
E, and I in the Supplement). tio. For all the A42 assays, the addition of T-tau and P-tau to
To further assess the concordance between CSF biomark- A42:A40 did not improve the ability to predict cortical A
ers and visual PET assessment in terms of classification per- accumulation compared with the A42:A40 ratios alone.
formance, we dichotomized CSF A variables into abnormal The INNOTEST ELISA is one of the most extensively used
(CSF A-positive) and normal (CSF A-negative) values based assays for assessment of A42 levels in CSF. However, this
on the optimal cutoff points corresponding to the highest assay is limited by the matrix interference that may lead to in-
Youden J indices. The number of cases with discordant CSF accurate estimates of A42 concentrations.12 The matrix ef-
A status compared with visual PET assessments was higher fects are reduced in the newer assays (modified INNOTEST, FL,
for the newer assays (48 [18.3%] for modified A42- EI, and MSD), which are based on the analysis of diluted CSF
INNITEST; 49 [18.7%] for A42-EI; and 42 [16.0%] for A42- samples.13-15 Accordingly, we found that CSF concentrations
MSD) than for the classic A42-INNOTEST (31 [11.8%]) and con- of A42 derived from antibody-independent mass spectro-
sisted of mainly CSF A42-positive and visual PET-negative metry procedure matched better with the data from the newer
cases (eTable 3 in the Supplement). However, the number of assays than the classic INNOTEST assay.
discordant cases for the newer assays (modified A42- Visual rating is the only approved procedure for assess-
INNOTEST, A42-EI, and A42-MSD) was reduced signifi- ment of amyloid PET scans in the clinic and when selecting AD
cantly when using the A42:A40 or A42:P-tau ratios (eTable cases for clinical trials.1 Although quantitative approaches pro-
3 in the Supplement). As shown in Figure 3, using cutoffs ob- vide more informative data that might be critical for investi-
tained from the 2 ratios (A42:A40 or A42:P-tau) resulted gating longitudinal changes in amyloid burden and treat-
in better separation of visual PET-positive and -negative cases ment effects, the concordance between visual and quantitative
than CSF A42 levels alone, and the ratios mainly reduced the methods for establishing amyloid status is very high. Similar
number of cases in the discordant group that were positive for to previous studies,19,30 we found that only 25 cases (9.5%)
CSF A42 and negative for visual PET findings (eTable 3 in the were discordant according to visual and quantitative PET analy-
Supplement). The results were similar for CSF biomarkers sis with very good intermethod agreement (Cohen = 0.81).
measured using antibody-independent mass spectrometry Most discordant cases were found in individuals who had been
based RMP and quantitative PET assessment (eResults and classified as amyloid negative by the visual analysis but as amy-
eFigure 3 in the Supplement). loid positive by SUVR analysis, demonstrating that visual analy-
sis is more conservative than SUVR analysis; this finding is con-
Spiking of CSF Samples With A40 sistent with those of previous studies.19
Addition of recombinant A1-40 peptide to CSF samples caused Cerebrospinal fluid A42 is a biomarker of amyloid dep-
a spike leveldependent decrease in measured CSF A42 con- osition that is often used interchangeably or in combination
centrations by as much as 62% when measured using the clas- with PET imaging in the diagnostic workup in the clinic or in
sic INNOTEST assay (eFigure 4A and D in the Supplement). In clinical trials.1 The agreement between CSF A42 and quan-
contrast, no differences in A42 concentrations were ob- titative amyloid PET data has ranged from 80% to 90%,6-8 and
served when the samples were analyzed using the newer modi- this agreement might be improved by using the A42:A40 or
fied INNOTEST procedure (eFigure 4B and E in the Supple- A42:tau ratios.9,10,31 However, for implementation of CSF AD
ment) and MSD assay (eFigure 4C and F in the Supplement). biomarkers (A42 level, A42:A40 ratio, and/or A42:tau ra-
tio) in routine clinical practice, establishing the concordance
with the clinically approved visual amyloid PET ratings, and
studying whether the concordance rate is affected by the choice
Discussion of the immunoassay for CSF A42 measurements are impor-
In this study, we showed that the new modified A42- tant. Herein, we showed that newer A42-EI and A42-MSD
INNOTEST, A42-FL, A42-EI, and A42-MSD assays corre- assays have similar performance with respect to association
lated better with A42 values obtained with antibody- of visual read outcome and exhibit acceptable accuracy with
independent mass spectrometrybased RMP compared with visual [18F]flutemetamol PET status when using the A42:
the classic A42-INNOTEST assay. The classic A42- A40 ratio. With the classic INNOTEST assay, the AUC for A42
INNOTEST instead correlated better with the A42:A40 ra- alone was close to the A42:A40 ratios from the EI and MSD
tio obtained with mass spectrometrybased RMP compared assays and did not increase any further when the classic
with the other A42 assays, a phenomenon that might be INNOTEST A42:A40 ratio was used. The modification of the
explained by the fact that the signal in the classic A42- classic INNOTEST assay, which improved the correlation with
INNOTEST assay is somewhat quenched by A40 levels. Of mass spectrometrybased RMP A42 values, resulted in de-
interest, the classic A42-INNOTEST assay had a higher accu- creased accuracy of A42 levels alone but improved perfor-
racy to predict visual PET assessment outcome compared with mance for the A42:A40 ratio, as was the case for other A42
Ratio
15 000 125
cutoff
100
10 000
75
50
5000
25
0 0
0 250 500 750 1000 1250 0 250 500 750 1000 1250
A42-INC Level, pg/mL A42-INC Level, pg/mL
C A42-INM vs A40 D A42-INM vs P-tau
30 000 200
A42 level cutoff Ratio A42 level cutoff
cutoff 175 Ratio
25 000 cutoff
150
P-tau Level, pg/mL
A40 Level, pg/mL
20 000
125
15 000 100
75
10 000
50
5000
25
0 0
0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 3000 3500
A42-INM Level, pg/mL A42-INM Level, pg/mL
E A42-EI vs A40 F A42-EI vs P-tau
15 000 200
Ratio A42 level cutoff
A42 level cutoff cutoff 175 Ratio
12 500 cutoff
150
P-tau Level, pg/mL
A40 Level, pg/mL
10 000
125
7500 100
75
5000
50
2500
25
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
A42-EI Level, pg/mL A42-EI Level, pg/mL
G A42-MSD vs A40 H A42-MSD vs P-tau
10 000
125 cutoff phosphorylated tau (P-tau) ratios.
Scatterplots of CSF A42 levels
7500 100 against A40 and P-tau levels; A42
75 was measured using the classic
5000 INNOTEST (INC), modified INNOTEST
50 (INM), EUROIMMUN (EI), and Meso
2500 Scale Discovery (MSD) assays. Dotted
25
lines indicate Youden J index cutoffs
0 0 for CSF A42 levels. Solid lines
0 250 500 750 1000 1250 1500 0 250 500 750 1000 1250 1500
indicated Youden J index cutoffs
A42-MSD Level, pg/mL A42-MSD Level, pg/mL for the CSF A42:A40 and
A42:P-tau ratios.
assays evaluated herein. These results could in part be ex- factors than A peptides.33-35 Consequently, variations in pre-
plained by our spiking data, which suggested that A42 con- analytical handling should in theory be better compensated for
centrations obtained using the classic INNOTEST were influ- by using A42:A40 ratios11 than A42:tau ratios, suggesting
enced by A1-40 and that in individuals with higher CSF levels that the A42:A40 ratios might be more useful in clinical prac-
of A1-40, these effects might be more pronounced, resulting tice as well as for tracking the emergence of amyloid deposi-
in a lower signal. Further analysis revealed that the CSF A42: tion in longitudinal studies.
A40 ratios from the newer immunoassays (modified
INNOTEST, EI, and MSD) had a more pronounced bimodal dis- Limitations
tribution allowing improved separation of [18F]flutemetamol One limitation that could have possibly influenced the
PET-positive and -negative populations compared with CSF result of the present study is that PET images were assessed
A42 levels alone. Moreover, compared with A42 levels alone, by a single rater. However, this is unlikely considering that
the rate of decrease in sensitivity and specificity values around the findings were the same for quantitative measures
the cutoff points was less steep when using the A42:A40 ra- (SUVR) of amyloid PET.
tios. However, we did not observe differences in the distribu-
tions and behavior of sensitivity and specificity values near cut-
off between A42 level and A42:A40 ratio for the classic
INNOTEST assay. These results indicate that the A42:A40
Conclusions
ratio cutoffs for the new assays (modified INNOTEST, EI, and We showed that the newer A42 assays (modified INNOTEST,
MSD) may be more reliable in terms of performance that is more EI, and MSD) correlated better with the antibody-free RMP
robust with respect to small changes in cutoff values. In agree- than did the classic INNOTEST assay, possibly because the
ment with our findings, the CSF A42:A40 ratio has also been signal in the classic INNOTEST assay is partly quenched by
shown to provide a better prediction of prodromal AD com- A1-40. However, the accuracy to correlate with visual
pared with A42 level alone when using the A42-EI and A42- [18F]flutemetamol PET status was decreased in the newer A42
MSD assays but not the classic A42-INNOTEST or xMAP assays, a limitation that is overcome by using a A42:A40 or
AlzBio3 (Fujirebio) assays.16,32 A42:tau ratios. The CSF A42:A40 or A42:tau ratios from
In the present study, the A42:tau ratios were better pre- the newer assays showed improved accuracy for detection of
dictors of visual [18F]flutemetamol PET classification than were cortical A fibrils as measured by PET. Moreover, the sensi-
A42 levels alone. However, we did not observe improved ac- tivities and specificities of these newer assays were less influ-
curacy when the A42:A40 ratios were combined with T-tau enced by moderate changes in the cutoffs when A42:A40
or P-tau compared with the A42:A40 ratios alone. Biochemi- or A42:tau ratios were used, a finding that is important
cal mechanisms underlying the better performance of the A42: when samples will be analyzed consecutively over time. The
A40 and A42:tau ratios are not known. Interindividual vari- precision of the A42:A40 ratios in differentiating
ability in the production of CSF, the production and secretions [18F]flutemetamol PETpositive and negative visual reads did
of proteins by neurons such as A and tau, and preanalytical fac- not improve further when combined with T-tau or P-tau
tors may affect the accuracy of CSF A42 levels in predicting values. Thus, our study provides a comprehensive overview
amyloid status. Although all these effects could be partly com- of the correlation of the performance of CSF biomarkers across
pensated by using the ratios to A40 or tau, at present, which different immunoassays with amyloid PET status, which may
of the mechanisms are mainly responsible for the better per- influence the selection of immunoassays and biomarkers
formance of the ratios and whether these mechanisms differ for in future studies. Furthermore, our findings support imple-
ratios to A40 or tau remain unclear. For example, some data mentation of the CSF A42:A40 and/or the A42:tau ratios
suggest that CSF levels of tau are less affected by preanalytical as biomarkers of amyloid deposition in clinical practice.
ARTICLE INFORMATION Blennow); Department of Molecular Neuroscience, Obtained funding: Zetterberg, Blennow, Hansson.
Accepted for Publication: July 24, 2017. University College London Institute of Neurology, Administrative, technical, or material support:
Queen Square, London, England (Zetterberg); Mikulskis, Chiao, Zetterberg, Hansson.
Published Online: November 6, 2017. Memory Clinic, Skne University Hospital, Malm, Study supervision: Hansson.
doi:10.1001/jamaneurol.2017.2814 Sweden (Hansson). Conflict of Interest Disclosures: Drs Mikulskis and
Open Access: This article is published under the Author Contributions: Drs Janelidze and Hansson Chiao report employment by Biogen. Dr Zetterberg
JN-OA license and is free to read on the day of had full access to all the data in the study and take reports serving on advisory boards for Eli Lilly,
publication. responsibility for the integrity of the data and the Roche Diagnostics, and Pharmasum Therapeutics.
Author Affiliations: Clinical Memory Research accuracy of the data analysis. Drs Zetterberg and Blennow report serving as
Unit, Department of Clinical Sciences Malm, Lund Study concept and design: Mikulskis, Chiao, cofounders of Brain Biomarker Solutions in
University, Lund, Sweden (Janelidze, Hansson); Blennow, Hansson. Gothenburg AB, a GU Venturesbased platform
Institute of Neuroscience and Physiology, Acquisition, analysis, or interpretation of data: All company at the University of Gothenburg. Dr
Department of Psychiatry and Neurochemistry, the authors. Blennow reports serving on advisory boards or as a
Sahlgrenska Academy at the University of Drafting of the manuscript: Janelidze, Chiao, consultant (unrelated to the work presented in the
Gothenburg, Gothenburg, Sweden (Pannee, Hansson. present study) for Eli Lilly and Company, Fujirebio
Zetterberg, Blennow); Biogen, Cambridge, Critical revision of the manuscript for important Europe, IBL International, Novartis, and Roche
Massachusetts (Mikulskis, Chiao); Clinical intellectual content: Pannee, Mikulskis, Zetterberg, Diagnostics. Dr Hansson reports serving on
Neurochemistry Laboratory, Sahlgrenska University Blennow, Hansson. advisory boards or as a consultant for Eli Lilly and
Hospital, Gothenburg, Sweden (Zetterberg, Statistical analysis: Janelidze. Company and receiving research support from
Roche, GE Healthcare, and AVID and A42:A38 ratios: better diagnostic markers of 22. Pannee J, Portelius E, Minthon L, et al.
Radiopharmaceuticals. No other disclosures were Alzheimer disease. Ann Clin Transl Neurol. 2016;3 Reference measurement procedure for CSF amyloid
reported. (3):154-165. (A)1-42 and the CSF A1-42:A1-40 ratio:
Funding/Support: This study was supported by the 10. Leuzy A, Chiotis K, Hasselbalch SG, et al. a cross-validation study against amyloid PET.
European Research Council, the Swedish Research Pittsburgh compound B imaging and cerebrospinal J Neurochem. 2016;139(4):651-658.
Council, the Swedish Alzheimer Foundation, the fluid amyloid- in a multicentre European memory 23. A Language and Environment for Statistical
Swedish Brain Foundation, the Marianne and clinic study. Brain. 2016;139(Pt 9):2540-2553. Computing. Vienna, Austria: R Foundation for
Marcus Wallenberg Foundation, and the Swedish 11. Vanderstichele HM, Janelidze S, Demeyer L, Statistical Computing; 2014. http://www
Federal Government under the ALF Agreement. et al. Optimized standard operating procedures for .R-project.org. Accessed September 17, 2016.
Doses of [18]F-flutemetamol injection were the analysis of cerebrospinal fluid A42 and the 24. Meng XL, Rosenthal R, Rubin DB. Comparing
sponsored by GE Healthcare. EUROIMMUN A42, ratios of A isoforms using low protein binding correlated correlation-coefficients. Psychol Bull.
A40, and T-tau kits were provided by tubes. J Alzheimers Dis. 2016;53(3):1121-1132. 1992;111(1):172-175.
EUROIMMUN. Analysis of cerebrospinal fluid
samples using INNOTEST and Lumipulse kits were 12. Cullen VC, Fredenburg RA, Evans C, Conliffe PR, 25. Robin X, Turck N, Hainard A, et al. pROC: an
performed at Fujirebio Europe. Solomon ME. Development and advanced open-source package for R and S+ to analyze and
validation of an optimized method for the compare ROC curves. BMC Bioinformatics. 2011;12:
Role of the Funder/Sponsor: The sponsors had no quantitation of A42 in human cerebrospinal fluid. 77.
role in the design and conduct of the study; AAPS J. 2012;14(3):510-518.
collection, management, analysis, and 26. Benaglia T, Chauveau D, Hunter DR, Young DS.
interpretation of the data; and preparation, review, 13. Bittner T, Zetterberg H, Teunissen CE, et al. mixtools: an R package for analyzing finite mixture
or approval of the manuscript; and decision to Technical performance of a novel, fully automated models. J Stat Softw. 2009;32(6):1-29.
submit the manuscript for publication. electrochemiluminescence immunoassay for the 27. Altman DG. Practical Statistics for Medical
quantitation of -amyloid (1-42) in human Research. London, England: Chapman & Hall:CRC;
Additional Contributions: The authors thank the cerebrospinal fluid. Alzheimers Dement. 2016;12(5):
study participants and the research nurses involved 1999.
517-526.
in the study for their invaluable contributions. 28. Buchhave P, Minthon L, Zetterberg H, Wallin
14. Pan C, Korff A, Galasko D, et al. Diagnostic AK, Blennow K, Hansson O. Cerebrospinal fluid
REFERENCES values of cerebrospinal fluid T-tau and A42 using levels of -amyloid 1-42, but not of tau, are fully
Meso Scale Discovery assays for Alzheimers changed already 5 to 10 years before the onset of
1. Blennow K, Mattsson N, Schll M, Hansson O, disease. J Alzheimers Dis. 2015;45(3):709-719.
Zetterberg H. Amyloid biomarkers in Alzheimers Alzheimer dementia. Arch Gen Psychiatry. 2012;69
disease. Trends Pharmacol Sci. 2015;36(5):297-309. 15. Vandijck M, Moonen R, Andreasson U, et al. (1):98-106.
Correlation of the modified innotest -amyloid 1-42 29. De Meyer G, Shapiro F, Vanderstichele H, et al;
2. Ikonomovic MD, Klunk WE, Abrahamson EE, with a lc-ms:ms candidate reference method.
et al. Post-mortem correlates of in vivo PiB-PET Alzheimers Disease Neuroimaging Initiative.
Alzheimers Dement. 2015;11(7)(suppl):385. Diagnosis-independent Alzheimer disease
amyloid imaging in a typical case of Alzheimers
disease. Brain. 2008;131(pt 6):1630-1645. 16. Palmqvist S, Zetterberg H, Mattsson N, et al; biomarker signature in cognitively normal elderly
Alzheimers Disease Neuroimaging Initiative; people. Arch Neurol. 2010;67(8):949-956.
3. Strozyk D, Blennow K, White LR, Launer LJ. CSF Swedish BioFINDER Study Group. Detailed
A 42 levels correlate with amyloid- 30. Zwan MD, Ossenkoppele R, Tolboom N, et al.
comparison of amyloid PET and CSF biomarkers for Comparison of simplified parametric methods for
neuropathology in a population-based autopsy identifying early Alzheimer disease. Neurology.
study. Neurology. 2003;60(4):652-656. visual interpretation of 11C-Pittsburgh compound-B
2015;85(14):1240-1249. PET images. J Nucl Med. 2014;55(8):1305-1307.
4. Tapiola T, Alafuzoff I, Herukka SK, et al. 17. Andreasen N, Hesse C, Davidsson P, et al.
Cerebrospinal fluid -amyloid 42 and tau proteins 31. Wang MJ, Yi S, Han JY, et al. Analysis of
Cerebrospinal fluid -amyloid(1-42) in Alzheimer cerebrospinal fluid and [11C]PIB PET biomarkers for
as biomarkers of Alzheimer-type pathologic disease: differences between early- and late-onset
changes in the brain. Arch Neurol. 2009;66(3):382- Alzheimers disease with updated protocols.
Alzheimer disease and stability during the course of J Alzheimers Dis. 2016;52(4):1403-1413.
389. disease. Arch Neurol. 1999;56(6):673-680.
5. Wolk DA, Grachev ID, Buckley C, et al. 32. Hertze J, Minthon L, Zetterberg H,
18. Clark CM, Pontecorvo MJ, Beach TG, et al; Vanmechelen E, Blennow K, Hansson O. Evaluation
Association between in vivo fluorine 18-labeled AV-45-A16 Study Group. Cerebral PET with
flutemetamol amyloid positron emission of CSF biomarkers as predictors of Alzheimers
florbetapir compared with neuropathology at disease: a clinical follow-up study of 4.7 years.
tomography imaging and in vivo cerebral cortical autopsy for detection of neuritic amyloid- plaques:
histopathology. Arch Neurol. 2011;68(11):1398-1403. J Alzheimers Dis. 2010;21(4):1119-1128.
a prospective cohort study. Lancet Neurol. 2012;11
6. Landau SM, Lu M, Joshi AD, et al; Alzheimers (8):669-678. 33. Le Bastard N, De Deyn PP, Engelborghs S.
Disease Neuroimaging Initiative. Comparing Importance and impact of preanalytical variables on
19. Schreiber S, Landau SM, Fero A, Schreiber F, Alzheimer disease biomarker concentrations in
positron emission tomography imaging and Jagust WJ; Alzheimers Disease Neuroimaging
cerebrospinal fluid measurements of -amyloid. cerebrospinal fluid. Clin Chem. 2015;61(5):734-743.
Initiative. Comparison of visual and quantitative
Ann Neurol. 2013;74(6):826-836. florbetapir F 18 positron emission tomography 34. Toombs J, Paterson RW, Lunn MP, et al.
7. Mattsson N, Insel PS, Donohue M, et al; analysis in predicting mild cognitive impairment Identification of an important potential confound in
Alzheimers Disease Neuroimaging Initiative. outcomes. JAMA Neurol. 2015;72(10):1183-1190. CSF AD studies: aliquot volume. Clin Chem Lab Med.
Independent information from cerebrospinal fluid 2013;51(12):2311-2317.
20. Leinenbach A, Pannee J, Dlffer T, et al; IFCC
amyloid- and florbetapir imaging in Alzheimers Scientific Division Working Group on CSF proteins. 35. Toombs J, Paterson RW, Nicholas JM, Petzold
disease. Brain. 2015;138(pt 3):772-783. Mass spectrometrybased candidate reference A, Schott JM, Zetterberg H. The impact of Tween
8. Palmqvist S, Zetterberg H, Blennow K, et al. measurement procedure for quantification of 20 on repeatability of amyloid and tau
Accuracy of brain amyloid detection in clinical amyloid- in cerebrospinal fluid. Clin Chem. 2014; measurements in cerebrospinal fluid. Clin Chem Lab
practice using cerebrospinal fluid -amyloid 42: 60(7):987-994. Med. 2015;53(12):e329-e332.
a cross-validation study against amyloid positron 21. Blennow K, Hampel H, Weiner M, Zetterberg H.
emission tomography. JAMA Neurol. 2014;71(10): Cerebrospinal fluid and plasma biomarkers in
1282-1289. Alzheimer disease. Nat Rev Neurol. 2010;6(3):131-
9. Janelidze S, Zetterberg H, Mattsson N, et al; 144.
Swedish BioFINDER study group. CSF A42:A40