detailed comparison of amyloid pet and csf biomarkers for ... · csf biomarkers for identifying...
TRANSCRIPT
Sebastian Palmqvist MDPhD
Henrik Zetterberg MDPhD
Niklas Mattsson MDPhD
Per Johansson MD PhDFor the Alzheimerrsquos
Disease NeuroimagingInitiative
Lennart Minthon MDPhD
Kaj Blennow MD PhDMattias Olsson PhDFor the Swedish
BioFINDER studygroup
Oskar Hansson MDPhD
Correspondence toDr Palmqvistsebastianpalmqvistmedluseor Dr Hanssonoskarhanssonmedluse
Supplemental dataat Neurologyorg
Detailed comparison of amyloid PET andCSF biomarkers for identifying earlyAlzheimer disease
ABSTRACT
Objective To compare the diagnostic accuracy of CSF biomarkers and amyloid PET for diagnos-ing early-stage Alzheimer disease (AD)
Methods From the prospective longitudinal BioFINDER study we included 122 healthy elderly and34 patients with mild cognitive impairment who developed AD dementia within 3 years (MCI-AD)b-Amyloid (Ab) deposition in 9 brain regions was examined with [18F]-flutemetamol PET CSF wasanalyzed with INNOTEST and EUROIMMUN ELISAs The results were replicated in 146 controlsand 64 patients with MCI-AD from the Alzheimerrsquos Disease Neuroimaging Initiative study
Results The best CSF measures for identifying MCI-AD were Ab42total tau (t-tau) and Ab42hyperphosphorylated tau (p-tau) (area under the curve [AUC] 093ndash094) The best PET measuresperformed similarly (AUC 092ndash093 anterior cingulate posterior cingulateprecuneus andglobal neocortical uptake) CSF Ab42t-tau and Ab42p-tau performed better than CSF Ab42and Ab4240 (AUC difference 003ndash012 p 005) Using nonoptimized cutoffs CSF Ab42t-tau had the highest accuracy of all CSFPET biomarkers (sensitivity 97 specificity 83) Thecombination of CSF and PET was not better than using either biomarker separately
Conclusions Amyloid PET and CSF biomarkers can identify early AD with high accuracy Therewere no differences between the best CSF and PET measures and no improvement when combin-ing them Regional PET measures were not better than assessing the global Ab deposition Theresults were replicated in an independent cohort using another CSF assay and PET tracer Thechoice between CSF and amyloid PET biomarkers for identifying early AD can be based on avail-ability costs and doctorpatient preferences since both have equally high diagnostic accuracy
Classification of evidence This study provides Class III evidence that amyloid PET and CSF bio-markers identify early-stage AD equally accurately Neurologyreg 2015851240ndash1249
GLOSSARYAb 5 b-amyloid AD 5 Alzheimer disease ADNI 5 Alzheimerrsquos Disease Neuroimaging Initiative AUC 5 area under thereceiver operating characteristic curve CI 5 confidence interval MCI-AD 5 mild cognitive impairment later developing intoAD MCS 5 mild cognitive symptoms MSD 5 Meso Scale Discovery OR 5 odds ratio p-tau 5 hyperphosphorylated tauROC 5 receiver operating characteristic SUVR 5 standardized uptake value ratio t-tau 5 total tau VOI 5 volume ofinterest YI 5 Youden index
Biomarkers of cerebral b-amyloid (Ab) are used in the criteria for the early stages of Alzheimerdisease (AD)12 and are increasingly used in clinical trials3ndash5 This stresses the need for reliable andavailable biomarkers of brain Ab pathology Two Ab modalities have been establishedmdashCSFAb42 and amyloid PETmdashwhich both correlate highly with brain biopsy findings67 A potentialadvantage of amyloid PET over CSF Ab42 as an early diagnostic marker is the possibility to detectregional Ab depositions that might occur before the global neocortical signal becomes pathologicOn the other hand CSF analysis has the advantages that it may easily incorporate assessments
Authorsrsquo affiliations are provided at the end of the article
Coinvestigators are listed on the Neurologyreg Web site at Neurologyorg
Part of the data used in preparation of this article were obtained from the Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) database (adniloniuscedu) As such the investigators within the ADNI contributed to the design and implementation of ADNI or provided data but did notparticipate in analysis or writing of this report
Go to Neurologyorg for full disclosures Funding information and disclosures deemed relevant by the authors if any are provided at the end of the articleThe Article Processing Charge was paid by the Swedish BioFINDER study
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 downloading and sharing the work provided it is properly cited The work cannot be changed in any way or used commercially
1240 copy 2015 American Academy of Neurology
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
such as tau (a measure of neuronal degenera-tion8) and hyperphosphorylated tau (p-tau apotential marker of tau pathology)9
Several studies have examined the agree-ment between amyloid PET and CSFAb4210ndash22 but head-to-head studies compar-ing their diagnostic accuracy for incipient ADare scarce Very few studies have used clinicallyrelevant consecutively recruited patients Toour knowledge no previous study has com-pared the accuracy of regional amyloid PETand different CSF assays or ratios of CSF bio-markers such as Ab4240 Ab42total tau(t-tau) and Ab42p-tau when identifying caseswith incipient AD We therefore performed adetailed head-to-head comparison of regionaland global amyloid PET and CSF analysis with2 different assays in a clinical cohort of consec-utive patients with mild cognitive impairmentwho later developed AD dementia (MCI-AD)We also examined the diagnostic benefit ofcombining CSF and PET measures
METHODS This study conducts a head-to-head comparison
of the diagnostic accuracy of amyloid PET and CSF biomarkers
for identifying early-stage AD It provides Class III evidence
that amyloid PET and CSF biomarkers identify early-stage AD
equally accurately
Subjects The present study population was part of the prospectiveand longitudinal Swedish BioFINDER study which among other
cohorts consecutively enrolls patients without dementia with mild
cognitive symptoms (MCS) from 3 participating memory clinics
in Sweden More information about the design and populations is
available at biofinderse and in the online supplement (e-Methods
on the Neurologyreg Web site at Neurologyorg) We included
patients with MCS who had progressed to AD dementia during
the follow-up period (hereafter referred to as MCI-AD) This
resulted in a sample of 34 patients with MCI-AD The mean
follow-up time was 20 years (range 08ndash34) A consensus group
(Katarina Naumlgga PJ SP) determined the follow-up diagnosis
probable AD23 in September 2014 The group was blinded to all
biomarker data At baseline in the MCS cohort 3 patients (9)
had subjective cognitive decline and 31 (91) had MCI (48
amnestic single-domain 39 amnestic multidomain and 12
nonamnestic) A total of 122 cognitively healthy elderly from the
BioFINDER study were included as controls24
Standard protocol approvals registrations and patientconsents The Ethics Committee in Lund Sweden approved
the study All patients gave their written informed consent
Amyloid PET scanning and analysis Cerebral Ab deposition
was visualized with the PET tracer 18F-flutemetamol (approved
by the Food and Drug Administration and the European Medical
Agency)25 PETCT scanning of the brain was conducted at 2
sites using the same type of scanner (Gemini Philips Healthcare
Best the Netherlands) Baseline sum images from 90ndash110 mi-
nutes postinjection were analyzed using the software NeuroMarQ
(GE Healthcare Cleveland OH) A volume of interest (VOI)
template was applied for the following 9 bilateral regions pre-
frontal parietal lateral temporal medial temporal sensorimotor
occipital anterior cingulate posterior cingulateprecuneus and a
global neocortical composite region26 The standardized uptake
value ratio (SUVR) was defined as the uptake in a VOI normal-
ized for the cerebellar cortex uptake
CSF analysis The procedure and analysis of the CSF followed
the Alzheimerrsquos Association Flow Chart for CSF biomarkers8
Baseline lumbar CSF samples were collected at the 3 centers
and analyzed at one center on one occasion using single batch
analysis according to a standardized protocol820 CSF t-tau
Ab40 and Ab42 were analyzed by EUROIMMUN (EI) ELISAs
(EUROIMMUN AG Luumlbeck Germany) CSF Ab42 and
tau phosphorylated at Thr181 (p-tau) were analyzed with
INNOTEST (IT) ELISAs (Fujirebio Europe Ghent Belgium)
The following 8 variables were derived from the CSF analyses
Ab42IT Ab42EI Ab42ITAb40EI Ab42ITt-tauEI Ab42IT
p-tauIT Ab42EIAb40EI Ab42EIp-tauIT Ab42EIt-tauEI
Hippocampus volume and cognition All patients were
examined using a single 3T MRI scanner (Trio Siemens
Munich Germany) Hippocampal volume was analyzed with
FreeSurfer version 51 The smallest hippocampal volume (left
or right) was used Global cognition was assessed with the
Mini-Mental State Examination Memory was assessed with the
10-word delayed recall test from the Alzheimerrsquos Disease
Assessment Scalendashcognitive subscale27
Alzheimerrsquos Disease Neuroimaging Initiative cohort Tovalidate the results from BioFINDER in an independent cohort
we used data from the Alzheimerrsquos Disease Neuroimaging Initia-
tive (ADNI adniloniuscedu) The sample consisted of 64
patients with MCI-AD and 146 controls who had undergone
CSF sampling and Ab PET at baseline of ADNI-2 (table 1 and
reference 19) In sum the PET tracer 18F-florbetapir was used to
quantify Ab in different brain regions (and globally) normalized
for the cerebellar uptake CSF Ab42 t-tau and p-tau were
measured using xMAP Luminex (Luminex Corp Austin TX)
with the INNOBIA AlzBio3 kit (Innogenetics Ghent
Belgium)28 A consensus group blinded to the biomarker data
determined the follow-up diagnoses
Statistical analysis Group differences were calculated with the
Mann-Whitney U test (table 1) The area under the receiver
operating characteristic (ROC) curve (AUC) was used to
examine the diagnostic accuracy of the continuous CSF and
PET variables (table 2) The 95 confidence interval (CI) and
significance for differences between the AUCs were calculated
using bootstrap techniques29 The AUCs of the combined CSF
and PET variables were derived from logistic regressions
Nonoptimized and unbiased cutoffs were established using
mixture modeling30 A Youden index (YI sensitivity 1
specificity 2 1) was used for an easier comparison of
sensitivities and specificities Odds ratios (OR) were calculated
with multivariate logistic regression analysis (table 3) The
statistical analyses were performed with MedCalc version 14
(MedCalc Software MariaKerke Belgium) SPSS version 220
(SPSS Inc Chicago IL) MATLAB release 2014 Statistics
Toolbox (MathWorks Natick MA) and R version 302
RESULTS Baseline characteristics are shown in table1 There were no significant differences in ageAPOE4 or sex between the ADNI and BioFINDERcohorts but education differed between control andMCI-AD populations (higher in ADNI p 0001
Neurology 85 October 6 2015 1241
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
table 1) Biomarker data could not be directlycompared between the studies because of differentCSF assays and PET tracers (table 1 and table e-1)
CSF biomarkers for classification of MCI-AD and
controls The CSF biomarkers had diagnostic accura-cies for MCI-AD ranging from AUC 082 (CSFAb42EI) to AUC 093ndash094 (CSF Ab42t-tau andAb42p-tau ratios independent of assay table 2)CSF Ab42ITt-tauEI and Ab42ITp-tauIT hadsignificantly better accuracies than CSF Ab42IT
(AUC difference 004ndash005 p 5 002) andAb42EIAb40EI (AUC difference 008 p 0001)CSF Ab42EI had significantly lower AUC compared tomost other biomarkers but this could be partly overcomeby the ratio of Ab42EIAb40EI The diagnostic accuracyof CSF Ab42IT on the other hand was not improvedwhen used as a ratio with Ab40EI (table 2)
Regional and composite PET biomarkers for
classification of MCI-AD vs controls The AUCs ofthe amyloid PET biomarkers ranged from 075 to
Table 1 Baseline characteristics of MCI-AD patients and cognitively healthy elderly from the BioFINDER study and the ADNI study
BioFINDER MCI-AD(n 5 34)
BioFINDER controls(n 5 122)
p Value BioFINDERMCI-AD vs controls
ADNI MCI-AD(n 5 64)
ADNI controls(n 5 146)
p Value ADNIMCI-AD vs controls
Demographics
Age y (range) 727 (63ndash80) 735 (65ndash85) 029 721 (48ndash85) 732 (56ndash89) 079
Female 54 64 060 45 52 042
Education y 119 6 38 113 6 33 067 160 6 28 166 6 25 013
APOE e4 Dagger1 allele 61 24 0001 73 27 0001
MMSE points 267 6 15 290 6 09 0001 270 6 18 291 6 12 0001
10-word delayed recallerrors
72 6 23 20 6 20 0001
Hippocampus volume cm3 31 6 05 36 6 04 0001
PET regions
Globalcomposite 211 6 047 129 6 028 0001
Prefrontal 211 6 049 124 6 031 0001
Anterior cingulate 236 6 052 141 6 034 0001
Posterior cingulateprecuneus
226 6 048 139 6 033 0001
Parietal 198 6 044 123 6 026 0001
Lateral temporal 214 6 050 139 6 025 0001
Medial temporal 158 6 029 136 6 016 0001
Occipital 183 6 040 137 6 019 0001
Sensorimotor 181 6 042 131 6 018 0001
CSF analyses
Ab42EI 333 6 114 538 6 186 0001
Ab42IT 380 6 102 659 6 184 0001
Ab40EI 4881 6 1877 4516 6 1522 047
t-tauEI 581 6 213 318 6 113 0001
p-tauIT 924 6 339 535 6 180 0001
Ab42EIAb40EI 0072 6 0024 012 6 0038 0001
Ab42ITAb40EI 0086 6 0034 016 6 0052 0001
Ab42EIt-tauEI 063 6 027 187 6 076 0001
Ab42EIp-tauIT 396 6 173 110 6 442 0001
Ab42ITt-tauEI 073 6 033 231 6 089 0001
Ab42ITp-tauIT 459 6 20 137 6 539 0001
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative EI5 EUROIMMUN assay IT5 INNOTEST assay MCI-AD5 patients with mild cognitiveimpairment who developed Alzheimer disease within 3 years MMSE 5 Mini-Mental State Examination p-tau 5 hyperphosphorylated tau t-tau 5 total tauBiomarker data of the replication population (ADNI study) can be found in table e-1 As for comparisons between demographics in the BioFINDER and ADNIcohorts only education differed significantly (p 0001) Values are mean 6 SD unless otherwise specified CSF measures are given in pgmL and PETscore in mean standardized uptake value ratio
1242 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
092 (table 2) The AUCs of the composite PETSUVR and best regional PET SUVRs (anterior cin-gulate and posterior cingulateprecuneus) wereequally good (p 5 035ndash046) The prefrontaland parietal regional SUVRs had similar AUCs(p 5 049ndash099) The medial temporal SUVRperformed significantly worse than all otherPET measures (AUC difference 009ndash017 p 5
00001ndash002)
Comparison of CSF and PET biomarkers for
classification of MCI-AD vs controls The best CSF bio-markers (CSF Ab42ITt-tauEI and Ab42ITp-tauIT
ratios) had similar AUCs (093ndash094) as the bestPET measures (AUC 092ndash093 p 5 034ndash60)CSF Ab42IT also performed similar to the best PETmeasures (table 2) CSF Ab42EIAb40EI had anumerically poorer AUC compared to all PET varia-bles except for the sensorimotor occipital and medialtemporal regions but the differences were not signif-icant (p 5 009ndash040)
Combination of CSF and PET biomarkers To examinethe potential benefit of combining PET and CSFanalysis we tested models with CSF Ab42ITp-tauIT
and the composite PET SUVR entered separately andtogether as predictors of diagnosis in logistic regres-sion analyses When used together the AUC was096 (95 CI 092ndash097) and both variables wereindependent significant predictors (p 001) Thiswas numerically higher than for models using theindividual modalities but the differences were notsignificant (AUC difference 0021ndash0047 p 5
007ndash008) A combined model of the compositePET SUVR and CSF p-tauIT had equal AUC valueas CSF Ab42ITp-tauIT (both were 094 95 CI089ndash097)
Classification of incipient AD and controls at specific
cutoffs All Ab variables had a bimodal distributionsuitable for establishing nonoptimized unbiased cut-offs with mixture modeling except for CSF Ab42EIwhich was excluded from this analysis When usingthese cutoffs in the ROC analysis CSF Ab42ITt-tauEI
and CSF Ab42ITp-tauIT had the best sensitivities andspecificities of all CSF and PET measures (table 3)The 2 best PET measures were the prefrontal andthe posterior cingulateprecuneus regions Scatterplotsshow that the differences in specificities between CSFand PET are mostly caused by controls with normalPET and abnormal CSF values (figures e-1 and e-2)In logistic regressions CSF Ab42t-tauIT and EI hadthe highest OR when adjusting for age sex mem-ory function APOE e4 and hippocampal volume(table 3)
The diagnostic accuracy of biomarkers used in theclinic should preferably not be very sensitive tosmaller changes in cutoff values if they are to be gen-eralizable between different centers and settings Infigure 1 A and B continuous sensitivities and specif-icities of 4 CSF and PET measures are shown as afunction of the cutoff point The CSF and PETmeas-ures were not dependent on an optimized cutoff butprovide high accuracies from cutoff values spanningat least 1 SD in the current sample The exceptionwas CSF Ab42EI40EI which had a slightly narrowerinterval with near optimal YI
Comparison with the ADNI data Accuracies for CSFand PET biomarkers were also analyzed in the inde-pendent ADNI cohort (table 4) All CSF and PETvariables had similar AUCs ranging from 086 to 087and no significant differences were found (p5 017ndash093) As in BioFINDER CSF Ab42t-tau andAb42p-tau had higher AUCs than CSF Ab42 alone(both 087 vs 085) but in ADNI the differenceswere not significant (p 5 060ndash065) In ADNIthe AUCs of t-tau (081 95 CI 074ndash088) andp-tau (082 95 CI 075ndash088) were lower than in
Table 2 Classification of MCI-AD and healthy controls in BioFINDER based onROC analyses in BioFINDER
Variable (in order of AUCvalue) AUC (95 CI) AUC significantly better than
CSF Ab42ITp-tauIT 094 (089ndash097) PET medial temporal PET occipital CSFAb42EIAb40EI Ab42EI Ab42IT
CSF Ab42EIt-tauEI 093 (088ndash097) PET medial temporal CSF Ab42EIAb40EIAb42EI
CSF Ab42ITt-tauEI 093 (088ndash097) PET medial temporal PET occipital CSFAb42EIAb40EI Ab42EI Ab42IT
CSF Ab42EIp-tauIT 093 (088ndash096) PET medial temporal CSF Ab42EIAb40EIAb42EI
PET posterior cingulateprecuneus
093 (087ndash096) PET medial temporal PET occipital PETsensorimotor CSF Ab42EI
PET anterior cingulate 092 (087ndash096) PET medial temporal PET occipital CSFAb42EI
PET composite 092 (086ndash095) PET medial temporal PET occipital
PET prefrontal 091 (086ndash095) PET medial temporal CSF Ab42EI
PET parietal 091 (085ndash095) PET medial temporal PET occipital PETsensorimotor
CSF Ab42IT 090 (084ndash094) PET medial temporal CSF Ab42EI
PET lateral temporal 090 (084ndash094) PET medial temporal PET occipital
CSF Ab42ITAb40EI 088 (080ndash094) PET medial temporal
CSF Ab42EIAb40EI 086 (080ndash091) PET medial temporal
PET sensorimotor 085 (079ndash090) PET medial temporal
PET occipital 084 (077ndash089) PET medial temporal
CSF Ab42EI 082 (074ndash089)
PET medial temporal 075 (068ndash082)
Abbreviations AUC 5 area under the curve CI 5 confidence interval EI 5 EUROIMMUNassay IT 5 INNOTEST assay MCI-AD 5 patients with mild cognitive impairment who devel-oped Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauAUC was calculated with ROC analysis The 95 CI and significance for differencesbetween the AUCs were calculated using bootstrap techniques with 5000 bootstrapreplicas
Neurology 85 October 6 2015 1243
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
BioFINDER (t-tau 088 95 CI 082ndash093 andp-tau 087 95 CI 080ndash091 data not shown inthe tables) However no significant differences couldbe tested since the results were derived from 2 differ-ent cohorts
When combining CSF Ab42p-tau and compos-ite PET SUVR in ADNI the AUC was 087(95 CI 082ndash093) This did not differ significantlyfrom using the variables separately (AUC difference000ndash001 p 5 040ndash053)
DISCUSSION The main finding of this study wasthat the diagnostic accuracy of CSF and Ab PETbiomarkers to identify MCI-AD was similar whenusing 18F-flutemetamol amyloid PET and severaldifferent CSF biomarkers Specifically the best CSFmeasures (CSF Ab42t-tau and Ab42p-tau ratios)had similar diagnostic accuracies as the best PETmeasures (composite and cingulate SUVRs table2) We also found that no regional PET biomarkerwas better than the neocortical composite PET SUVR(table 2) For CSF biomarkers the CSF Ab42t-tauor Ab42p-tau ratios had significantly higherdiagnostic accuracy compared to using CSF Abbiomarkers alone When using unbiased cutoffs
CSF Ab42t-tau had the highest sensitivity andspecificity of all CSF and PET biomarkers (table 3)Finally the combination of the best CSF and PETbiomarkers did not provide any added diagnosticvalue compared to using either modality separatelyThe overall results were replicated in an independentcohort (ADNI)
Although we found that CSF Ab42IT and Ab42EIAb40EI performed similarly to the best SUVRs of18F-flutemetamol PET in terms of AUCs (table 2)these CSF biomarkers generally had lower specificitiesthan the PET biomarkers when using unbiased cut-offs (table 3) The addition of t-tau or p-tau to Ab42(as ratios) significantly increased the diagnostic accur-acy of CSF biomarkers (table 2) This supports thecommon usage of CSF Ab42 in combination with t-tau or p-tau in clinical practice and is in agreementwith previous studies31ndash34
The diagnostic accuracy of CSF Ab42 was lowerfor EUROIMMUN compared with INNOTEST(table 2 and figure e-3) This was partly overcomeby using the Ab4240 ratio which did not improvethe accuracy of Ab42 INNOTEST (table 2) Thisfinding has not been shown previously and needs tobe replicated in future studies since the causes are
Table 3 ROC analysis in BioFINDER of MCI-AD and healthy controls based on unbiased cutoffs
Variable (in order of Youden index) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b Adjusted OR (95 CI)d
CSF Ab42ITt-tauEI 125 080 (068ndash087) 97 83 62 (49ndash796)
CSF Ab42ITp-tauIT 724 079 (066ndash087) 94 85 34 (44ndash263)
CSF Ab42EIp-tauIT 567 074 (056ndash084) 88 86 20 (33ndash126)
PET prefrontal 148 074 (056ndash084) 88 86 29 (36ndash239)
CSF Ab42EIt-tauEI 096 073 (059ndash083) 91 82 44 (43ndash447)
PET posterior cingulateprecuneus 162 073 (056ndash083) 88 84 24 (33ndash182)
PET anterior cingulate 162 072 (056ndash081) 91 80 22 (29ndash170)
PET composite 151 072 (055ndash083) 85 87 29 (36ndash239)
PET lateral temporal 158 071 (054ndash083) 85 86 30 (36ndash250)
PET parietal 143 070 (053ndash081) 85 84 25 (32ndash190)
PET sensorimotor 153 069 (051ndash081) 79 89 30 (38ndash231)
CSF Ab42EIAb40EI 0092 067 (050ndash078) 88 79 14 (25ndash82)
CSF Ab42IT 500 060 (042ndash072) 85 75 61 (12ndash30)
CSF Ab42ITAb40EI 010 059 (039ndash073) 76 83 82 (17ndash40)
PET occipital 168 049 (030ndash066) 56 93 27 (28ndash263)
PET medial temporal 169 020 (007ndash038) 23 97 Nonsignificant
Abbreviations CI5 confidence interval EI5 EUROIMMUN assay IT5 INNOTEST assay MCI-AD5 patients with mild cognitive impairment who developedAlzheimer disease within 3 years OR 5 odds ratio p-tau 5 hyperphosphorylated tau ROC 5 receiving operating characteristic t-tau 5 total tauArea under the curve was calculated with ROC analysis The 95 CI and significance for differences between the AUCs were calculated using bootstraptechniques with 5000 bootstrap replicas PET values are shown in standardized uptake value ratio and CSF levels in pgmL (except for the CSF ratios)a Established with mixture modeling analysis (described in Methods)b Based on the ROC analysis (not derived from the logistic regression analysis)c Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffsdA logistic regression analysis was performed with the diagnosis (MCI-AD or control) as the dependent variable and the dichotomized CSFPET variable asa covariate to yield an OR Age sex APOE e4 allele memory function and hippocampal volume were adjusted for in the model
1244 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
unknown Few studies have compared differentELISAs for CSF Ab42 to identify MCI-AD Hertzeet al33 found that CSF Ab42 analyzed with xMAP
AlzBio3 had higher diagnostic accuracy comparedwith the Meso Scale Discovery (MSD) assay but thiswas overcome by using the Ab42Ab40 MSD ratio
Figure 1 Sensitivities specificities and Youden indices of all possible cutoff points
(A B) The cutoffs have been transformed to z scores (SD) for easier comparison between variables Near optimal Youden indicesare found for a relatively wide range of cutoffs (1 SD) for all variables except for CSF Ab42Ab40EI which have a slightlynarrower span The different cutoffs within this near optimal range thus only change the relationship between the sensitivity andspecificity but not the overall classification accuracy Thiswide range of near optimal cutoffs suggests that the cutoffs are likely toproduce high diagnostic accuracies in other populations AUC 5 area under the receiver operating characteristic curve
Neurology 85 October 6 2015 1245
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
Similar to the present study they showed that Ab42t-tau was superior to Ab42 and Ab42Ab40
A possible advantage of Ab PET over CSF Ab42as an early marker for amyloid pathology is that AbPET may be able to identify early region-specificpathology However this was not supported by ourstudy since the global Ab uptake performed similarcompared to the best regional SUVRs (tables 2 and4) Only one previous study has examined this andfound similar results19 The similar AUCs for the bestPET regions support the notion that the Ab deposi-tion is uniformly distributed in the neocortical asso-ciation areas already at the MCI stage of AD35
We used classification cutoffs established withmixture modeling which is a robust way of determin-ing unbiased thresholds and used in several stud-ies203637 Even so the cutoffs (table 3) should notbe considered generalizable but study-specific forcomparative purposes However figure 1 shows thatalthough a cutoff is not optimized for the currentpopulation it can still provide good diagnostic accur-acy because of the broad range of high YI The sta-bility of cutoffs between populations is also supportedby a previous cross-validation study on CSF Ab42and amyloid PET cutoffs20 However even thoughthe classification accuracy stays the same a change incutoff will of course result in a higher sensitivitylowerspecificity or lower sensitivityhigher specificity andthis must be taken into consideration depending onthe clinical aim of the examination
The overall results were similar between the Bio-FINDER and ADNI cohorts In ADNI the samecomparable results between regional and compositeSUVRs as well as equal diagnostic accuracies ofCSF and PET measures were seen (table 4) This
similarity between studies is especially interestingconsidering the use of different PET tracers and dif-ferent CSF assays In both cohorts numerically high-er AUCs were seen for the CSF Ab42t-tau or p-tauratios compared with just Ab42 but in ADNI theincrease was not significant (tables 2 and 4) Thiscould be attributed to the poorer AUCs of t-tauand p-tau in ADNI (081 and 082 AlzBio3)compared with BioFINDER (088 and 087EUROIMMUN and INNOTEST) A similar differ-ence between INNOTEST and AlzBio3 regardingAb42tau ratios was also found in a previous study38
It was notable that the AUCs of all brain regions weresimilar in ADNI (AUC range 001 table 4) in con-trast to BioFINDER (AUC range 017 table 2) Thereason for this could be that in ADNI the regionswere coarser and not able to detect differencesbetween eg the medial and lateral temporal lobe
The diagnostic accuracy of Ab PET and CSF bio-markers to detect incipient AD has only been comparedhead-to-head in one previous study which partly usedthe same ADNI data used for replication in the presentstudy19 In that study the accuracy between stable MCI(2- to 3-year follow-up without progression) and MCI-AD was compared In the present study we insteadcompared healthy elderly and patients with MCI-ADwhich resulted in higher AUCs (on average about 005in the ADNI study compare reference 19 and table 4)The rationale behind comparing controls andMCI-ADis that5ndash10 years of follow-up is required before onecan say that a patient withMCI is truly stable36 Amongpatients with stable MCI with a short follow-up timethere are several cases with early-stage AD These pa-tients with stable MCI will in most cases be correctlyidentified as MCI-AD by the biomarkers but result in
Table 4 Classification of MCI-AD and healthy controls based on ROC analyses in ADNI
Variable (in order of AUC value) AUC (95 CI) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b
PET frontal 087 (082ndash091) 113 063 (051ndash073) 89 74
CSF Ab42t-tau 087 (080ndash092) 171 065 (053ndash076) 80 86
CSF Ab42p-tau 087 (080ndash092) 383 065 (052ndash075) 84 81
PET composite 086 (081ndash091) 113 066 (054ndash075) 91 75
PET parietal 086 (081ndash091) 111 059 (047ndash069) 91 68
PET temporal 086 (080ndash092) 108 066 (054ndash076) 89 77
PET cingulate 086 (080ndash091) 120 056 (044ndash066) 89 66
CSF Ab42 085 (079ndash090) 173 060 (050ndash070) 94 66
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative AUC5 area under the receiver operating characteristic curve CI5 confidence intervalMCI-AD 5 patients with mild cognitive impairment who developed Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauNo significant differences between the AUCs were found (p 5 017ndash093)a Established with mixture modeling analysis (described in Methods)bDerived from the ROC analysisc Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffs
1246 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
false low specificity (and a false low AUC) due to theincorrect clinical diagnosisshort follow-up We there-fore compared patients with MCI-AD and controlsgiven the relatively short follow-up data in the ADNIand BioFINDER populations for a more robust com-parison of Ab biomarkers
The novelties of the present study compared withthe previous study19 include a comparison betweenMCI-AD and controls a more detailed analysis ofregional Ab PET data analyses of ratios of CSFAb42Ab40 Ab42t-tau and Ab42p-tau a compar-ison of 2 different ELISAs for CSF Ab42 and evalu-ation of the combination of PET and CSF biomarkers
The similar results we found for CSF biomarkersand amyloid PET suggest that other factors than theirdiagnostic accuracy may be considered when decidingwhich biomarker to use CSF analysis has the advan-tages that it may easily incorporate other biomarkersto improve the differential diagnosis (eg leukocytesalbumin ratio neurofilament a-synuclein) requiresless advanced instruments than PET and is in somecountries more available in clinical practice AmyloidPET on the other hand is less invasive and has ahigher reliability in longitudinal examinations andbetween centers With appropriate standardizedprocedures203940 CSF analysis and amyloid PET per-form equally well and either method can be used in theclinical workup of AD for increased diagnosticaccuracy
AUTHOR AFFILIATIONSFrom the Clinical Memory Research Unit Department of Clinical Scien-
ces (SP LM OH) and the Department of Computational Biology
and Biological Physics (MO) Lund University the Department of
Neurology (SP) and the Memory Clinic (LM OH) Skaringne University
Hospital the Clinical Neurochemistry Laboratory (HZ NM KB)
Institute of Neuroscience and Physiology the Sahlgrenska Academy at
the University of Gothenburg Moumllndal Sweden the UCL Institute of
Neurology (HZ) London UK the Department of Radiology and Bio-
medical Imaging (NM) University of California San Francisco the
Department of Veterans Affairs Medical Center (NM) Center for Imag-
ing of Neurodegenerative Diseases San Francisco CA and the Depart-
ment of Internal Medicine (PJ) Sahlgrenska Academy University of
Gothenburg Sweden
AUTHOR CONTRIBUTIONSSebastian Palmqvist draftingrevising the manuscript study concept or
design analysis or interpretation of data accepts responsibility for con-
duct of research and final approval statistical analysis study supervision
Henrik Zetterberg draftingrevising the manuscript analysis or interpre-
tation of data accepts responsibility for conduct of research and
final approval acquisition of data obtaining funding Niklas Mattsson
draftingrevising the manuscript analysis or interpretation of data
accepts responsibility for conduct of research and final approval Per
Johansson draftingrevising the manuscript analysis or interpretation
of data accepts responsibility for conduct of research and final approval
acquisition of data study supervision Lennart Minthon draftingrevising
the manuscript accepts responsibility for conduct of research and final
approval obtaining funding Kaj Blennow draftingrevising the manu-
script study concept or design accepts responsibility for conduct of
research and final approval study supervision obtaining funding Mattias
Ohlsson draftingrevising the manuscript analysis or interpretation of
data accepts responsibility for conduct of research and final approval sta-
tistical analysis Oskar Hansson draftingrevising the manuscript study
concept or design analysis or interpretation of data accepts responsibility
for conduct of research and final approval contribution of vital reagents
toolspatients acquisition of data study supervision obtaining funding
ACKNOWLEDGMENTThe authors thank the collaborators of this study and the entire BioFINDER
study group (wwwbiofinderse) including Ulf Andreasson Susanna Vestberg
for classifying the patients with MCI-AD into MCI subgroups Erik Stomrud
and Katarina Naumlgga for clinical evaluations of cognitively healthy individuals
Christer Nilsson for clinical evaluations of patients with mild cognitive symptoms
Per Wollmer and Douglas Haumlgerstroumlm for help with 18F-flutemetamol PET
imaging Olof Lindberg for analyzing the hippocampal volumes and Karin
Nilsson Rosita Nordkvist Ida Friberg Malin Ottheacuten and Johanna Fredlund
for organizing inclusions and assessments
STUDY FUNDINGWork in the authorsrsquo laboratory was supported by the European Research
Council the Swedish Research Council the Strategic Research Area Multi-
Park (Multidisciplinary Research in Parkinsonrsquos disease) at Lund University
the Crafoord Foundation the Swedish Brain Foundation the Skaringne Uni-
versity Hospital Foundation the Swedish Alzheimer Association Stiftelsen
foumlr Gamla Tjaumlnarinnor and the Swedish federal government under the ALF
agreement The Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) data
collection and sharing for this project was funded by ADNI (NIH grant
U01 AG024904) and DOD ADNI (Department of Defense award num-
ber W81XWH-12-2-0012) ADNI is funded by the National Institute on
Aging the National Institute of Biomedical Imaging and Bioengineering
and through contributions from the following Alzheimerrsquos Association
Alzheimerrsquos Drug Discovery Foundation Araclon Biotech BioClinica
Inc Biogen Idec Inc Bristol-Myers Squibb Company Eisai Inc Elan
Pharmaceuticals Inc Eli Lilly and Company EuroImmun F Hoffmann-
La Roche Ltd and its affiliated company Genentech Inc Fujirebio GE
Healthcare IXICO Ltd Janssen Alzheimer Immunotherapy Research amp
Development LLC Johnson amp Johnson Pharmaceutical Research amp
Development LLC Medpace Inc Merck amp Co Inc Meso Scale Diag-
nostics LLC NeuroRx Research Neurotrack Technologies Novartis Phar-
maceuticals Corporation Pfizer Inc Piramal Imaging Servier Synarc Inc
and Takeda Pharmaceutical Company The Canadian Institutes of Health
Research is providing funds to support ADNI clinical sites in Canada
Private sector contributions are facilitated by the Foundation for the
NIH (wwwfnihorg) The grantee organization is the Northern California
Institute for Research and Education and the study is coordinated by the
Alzheimerrsquos Disease Cooperative Study at the University of California San
Diego ADNI data are disseminated by the Laboratory for NeuroImaging at
the University of Southern California
DISCLOSURES Palmqvist H Zetterberg N Mattsson P Johansson and L Minthon
report no disclosures relevant to the manuscript K Blennow has served
at advisory boards for Koyowa Kirin Pharma Eli Lilly Pfizer and Roche
Doses of 18F-flutemetamol were sponsored by GE Healthcare M Ohlsson
and O Hansson report no disclosures relevant to the manuscript Go to
Neurologyorg for full disclosures
Received February 4 2015 Accepted in final form June 3 2015
REFERENCES1 Albert MS DeKosky ST Dickson D et al The diagnosis of
mild cognitive impairment due to Alzheimerrsquos disease recom-
mendations from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for
Alzheimerrsquos disease Alzheimers Dement 20117270ndash279
2 Sperling RA Aisen PS Beckett LA et al Toward defining
the preclinical stages of Alzheimerrsquos disease recommenda-
tions from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for Alz-
heimerrsquos disease Alzheimers Dement 20117280ndash292
Neurology 85 October 6 2015 1247
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
such as tau (a measure of neuronal degenera-tion8) and hyperphosphorylated tau (p-tau apotential marker of tau pathology)9
Several studies have examined the agree-ment between amyloid PET and CSFAb4210ndash22 but head-to-head studies compar-ing their diagnostic accuracy for incipient ADare scarce Very few studies have used clinicallyrelevant consecutively recruited patients Toour knowledge no previous study has com-pared the accuracy of regional amyloid PETand different CSF assays or ratios of CSF bio-markers such as Ab4240 Ab42total tau(t-tau) and Ab42p-tau when identifying caseswith incipient AD We therefore performed adetailed head-to-head comparison of regionaland global amyloid PET and CSF analysis with2 different assays in a clinical cohort of consec-utive patients with mild cognitive impairmentwho later developed AD dementia (MCI-AD)We also examined the diagnostic benefit ofcombining CSF and PET measures
METHODS This study conducts a head-to-head comparison
of the diagnostic accuracy of amyloid PET and CSF biomarkers
for identifying early-stage AD It provides Class III evidence
that amyloid PET and CSF biomarkers identify early-stage AD
equally accurately
Subjects The present study population was part of the prospectiveand longitudinal Swedish BioFINDER study which among other
cohorts consecutively enrolls patients without dementia with mild
cognitive symptoms (MCS) from 3 participating memory clinics
in Sweden More information about the design and populations is
available at biofinderse and in the online supplement (e-Methods
on the Neurologyreg Web site at Neurologyorg) We included
patients with MCS who had progressed to AD dementia during
the follow-up period (hereafter referred to as MCI-AD) This
resulted in a sample of 34 patients with MCI-AD The mean
follow-up time was 20 years (range 08ndash34) A consensus group
(Katarina Naumlgga PJ SP) determined the follow-up diagnosis
probable AD23 in September 2014 The group was blinded to all
biomarker data At baseline in the MCS cohort 3 patients (9)
had subjective cognitive decline and 31 (91) had MCI (48
amnestic single-domain 39 amnestic multidomain and 12
nonamnestic) A total of 122 cognitively healthy elderly from the
BioFINDER study were included as controls24
Standard protocol approvals registrations and patientconsents The Ethics Committee in Lund Sweden approved
the study All patients gave their written informed consent
Amyloid PET scanning and analysis Cerebral Ab deposition
was visualized with the PET tracer 18F-flutemetamol (approved
by the Food and Drug Administration and the European Medical
Agency)25 PETCT scanning of the brain was conducted at 2
sites using the same type of scanner (Gemini Philips Healthcare
Best the Netherlands) Baseline sum images from 90ndash110 mi-
nutes postinjection were analyzed using the software NeuroMarQ
(GE Healthcare Cleveland OH) A volume of interest (VOI)
template was applied for the following 9 bilateral regions pre-
frontal parietal lateral temporal medial temporal sensorimotor
occipital anterior cingulate posterior cingulateprecuneus and a
global neocortical composite region26 The standardized uptake
value ratio (SUVR) was defined as the uptake in a VOI normal-
ized for the cerebellar cortex uptake
CSF analysis The procedure and analysis of the CSF followed
the Alzheimerrsquos Association Flow Chart for CSF biomarkers8
Baseline lumbar CSF samples were collected at the 3 centers
and analyzed at one center on one occasion using single batch
analysis according to a standardized protocol820 CSF t-tau
Ab40 and Ab42 were analyzed by EUROIMMUN (EI) ELISAs
(EUROIMMUN AG Luumlbeck Germany) CSF Ab42 and
tau phosphorylated at Thr181 (p-tau) were analyzed with
INNOTEST (IT) ELISAs (Fujirebio Europe Ghent Belgium)
The following 8 variables were derived from the CSF analyses
Ab42IT Ab42EI Ab42ITAb40EI Ab42ITt-tauEI Ab42IT
p-tauIT Ab42EIAb40EI Ab42EIp-tauIT Ab42EIt-tauEI
Hippocampus volume and cognition All patients were
examined using a single 3T MRI scanner (Trio Siemens
Munich Germany) Hippocampal volume was analyzed with
FreeSurfer version 51 The smallest hippocampal volume (left
or right) was used Global cognition was assessed with the
Mini-Mental State Examination Memory was assessed with the
10-word delayed recall test from the Alzheimerrsquos Disease
Assessment Scalendashcognitive subscale27
Alzheimerrsquos Disease Neuroimaging Initiative cohort Tovalidate the results from BioFINDER in an independent cohort
we used data from the Alzheimerrsquos Disease Neuroimaging Initia-
tive (ADNI adniloniuscedu) The sample consisted of 64
patients with MCI-AD and 146 controls who had undergone
CSF sampling and Ab PET at baseline of ADNI-2 (table 1 and
reference 19) In sum the PET tracer 18F-florbetapir was used to
quantify Ab in different brain regions (and globally) normalized
for the cerebellar uptake CSF Ab42 t-tau and p-tau were
measured using xMAP Luminex (Luminex Corp Austin TX)
with the INNOBIA AlzBio3 kit (Innogenetics Ghent
Belgium)28 A consensus group blinded to the biomarker data
determined the follow-up diagnoses
Statistical analysis Group differences were calculated with the
Mann-Whitney U test (table 1) The area under the receiver
operating characteristic (ROC) curve (AUC) was used to
examine the diagnostic accuracy of the continuous CSF and
PET variables (table 2) The 95 confidence interval (CI) and
significance for differences between the AUCs were calculated
using bootstrap techniques29 The AUCs of the combined CSF
and PET variables were derived from logistic regressions
Nonoptimized and unbiased cutoffs were established using
mixture modeling30 A Youden index (YI sensitivity 1
specificity 2 1) was used for an easier comparison of
sensitivities and specificities Odds ratios (OR) were calculated
with multivariate logistic regression analysis (table 3) The
statistical analyses were performed with MedCalc version 14
(MedCalc Software MariaKerke Belgium) SPSS version 220
(SPSS Inc Chicago IL) MATLAB release 2014 Statistics
Toolbox (MathWorks Natick MA) and R version 302
RESULTS Baseline characteristics are shown in table1 There were no significant differences in ageAPOE4 or sex between the ADNI and BioFINDERcohorts but education differed between control andMCI-AD populations (higher in ADNI p 0001
Neurology 85 October 6 2015 1241
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
table 1) Biomarker data could not be directlycompared between the studies because of differentCSF assays and PET tracers (table 1 and table e-1)
CSF biomarkers for classification of MCI-AD and
controls The CSF biomarkers had diagnostic accura-cies for MCI-AD ranging from AUC 082 (CSFAb42EI) to AUC 093ndash094 (CSF Ab42t-tau andAb42p-tau ratios independent of assay table 2)CSF Ab42ITt-tauEI and Ab42ITp-tauIT hadsignificantly better accuracies than CSF Ab42IT
(AUC difference 004ndash005 p 5 002) andAb42EIAb40EI (AUC difference 008 p 0001)CSF Ab42EI had significantly lower AUC compared tomost other biomarkers but this could be partly overcomeby the ratio of Ab42EIAb40EI The diagnostic accuracyof CSF Ab42IT on the other hand was not improvedwhen used as a ratio with Ab40EI (table 2)
Regional and composite PET biomarkers for
classification of MCI-AD vs controls The AUCs ofthe amyloid PET biomarkers ranged from 075 to
Table 1 Baseline characteristics of MCI-AD patients and cognitively healthy elderly from the BioFINDER study and the ADNI study
BioFINDER MCI-AD(n 5 34)
BioFINDER controls(n 5 122)
p Value BioFINDERMCI-AD vs controls
ADNI MCI-AD(n 5 64)
ADNI controls(n 5 146)
p Value ADNIMCI-AD vs controls
Demographics
Age y (range) 727 (63ndash80) 735 (65ndash85) 029 721 (48ndash85) 732 (56ndash89) 079
Female 54 64 060 45 52 042
Education y 119 6 38 113 6 33 067 160 6 28 166 6 25 013
APOE e4 Dagger1 allele 61 24 0001 73 27 0001
MMSE points 267 6 15 290 6 09 0001 270 6 18 291 6 12 0001
10-word delayed recallerrors
72 6 23 20 6 20 0001
Hippocampus volume cm3 31 6 05 36 6 04 0001
PET regions
Globalcomposite 211 6 047 129 6 028 0001
Prefrontal 211 6 049 124 6 031 0001
Anterior cingulate 236 6 052 141 6 034 0001
Posterior cingulateprecuneus
226 6 048 139 6 033 0001
Parietal 198 6 044 123 6 026 0001
Lateral temporal 214 6 050 139 6 025 0001
Medial temporal 158 6 029 136 6 016 0001
Occipital 183 6 040 137 6 019 0001
Sensorimotor 181 6 042 131 6 018 0001
CSF analyses
Ab42EI 333 6 114 538 6 186 0001
Ab42IT 380 6 102 659 6 184 0001
Ab40EI 4881 6 1877 4516 6 1522 047
t-tauEI 581 6 213 318 6 113 0001
p-tauIT 924 6 339 535 6 180 0001
Ab42EIAb40EI 0072 6 0024 012 6 0038 0001
Ab42ITAb40EI 0086 6 0034 016 6 0052 0001
Ab42EIt-tauEI 063 6 027 187 6 076 0001
Ab42EIp-tauIT 396 6 173 110 6 442 0001
Ab42ITt-tauEI 073 6 033 231 6 089 0001
Ab42ITp-tauIT 459 6 20 137 6 539 0001
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative EI5 EUROIMMUN assay IT5 INNOTEST assay MCI-AD5 patients with mild cognitiveimpairment who developed Alzheimer disease within 3 years MMSE 5 Mini-Mental State Examination p-tau 5 hyperphosphorylated tau t-tau 5 total tauBiomarker data of the replication population (ADNI study) can be found in table e-1 As for comparisons between demographics in the BioFINDER and ADNIcohorts only education differed significantly (p 0001) Values are mean 6 SD unless otherwise specified CSF measures are given in pgmL and PETscore in mean standardized uptake value ratio
1242 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
092 (table 2) The AUCs of the composite PETSUVR and best regional PET SUVRs (anterior cin-gulate and posterior cingulateprecuneus) wereequally good (p 5 035ndash046) The prefrontaland parietal regional SUVRs had similar AUCs(p 5 049ndash099) The medial temporal SUVRperformed significantly worse than all otherPET measures (AUC difference 009ndash017 p 5
00001ndash002)
Comparison of CSF and PET biomarkers for
classification of MCI-AD vs controls The best CSF bio-markers (CSF Ab42ITt-tauEI and Ab42ITp-tauIT
ratios) had similar AUCs (093ndash094) as the bestPET measures (AUC 092ndash093 p 5 034ndash60)CSF Ab42IT also performed similar to the best PETmeasures (table 2) CSF Ab42EIAb40EI had anumerically poorer AUC compared to all PET varia-bles except for the sensorimotor occipital and medialtemporal regions but the differences were not signif-icant (p 5 009ndash040)
Combination of CSF and PET biomarkers To examinethe potential benefit of combining PET and CSFanalysis we tested models with CSF Ab42ITp-tauIT
and the composite PET SUVR entered separately andtogether as predictors of diagnosis in logistic regres-sion analyses When used together the AUC was096 (95 CI 092ndash097) and both variables wereindependent significant predictors (p 001) Thiswas numerically higher than for models using theindividual modalities but the differences were notsignificant (AUC difference 0021ndash0047 p 5
007ndash008) A combined model of the compositePET SUVR and CSF p-tauIT had equal AUC valueas CSF Ab42ITp-tauIT (both were 094 95 CI089ndash097)
Classification of incipient AD and controls at specific
cutoffs All Ab variables had a bimodal distributionsuitable for establishing nonoptimized unbiased cut-offs with mixture modeling except for CSF Ab42EIwhich was excluded from this analysis When usingthese cutoffs in the ROC analysis CSF Ab42ITt-tauEI
and CSF Ab42ITp-tauIT had the best sensitivities andspecificities of all CSF and PET measures (table 3)The 2 best PET measures were the prefrontal andthe posterior cingulateprecuneus regions Scatterplotsshow that the differences in specificities between CSFand PET are mostly caused by controls with normalPET and abnormal CSF values (figures e-1 and e-2)In logistic regressions CSF Ab42t-tauIT and EI hadthe highest OR when adjusting for age sex mem-ory function APOE e4 and hippocampal volume(table 3)
The diagnostic accuracy of biomarkers used in theclinic should preferably not be very sensitive tosmaller changes in cutoff values if they are to be gen-eralizable between different centers and settings Infigure 1 A and B continuous sensitivities and specif-icities of 4 CSF and PET measures are shown as afunction of the cutoff point The CSF and PETmeas-ures were not dependent on an optimized cutoff butprovide high accuracies from cutoff values spanningat least 1 SD in the current sample The exceptionwas CSF Ab42EI40EI which had a slightly narrowerinterval with near optimal YI
Comparison with the ADNI data Accuracies for CSFand PET biomarkers were also analyzed in the inde-pendent ADNI cohort (table 4) All CSF and PETvariables had similar AUCs ranging from 086 to 087and no significant differences were found (p5 017ndash093) As in BioFINDER CSF Ab42t-tau andAb42p-tau had higher AUCs than CSF Ab42 alone(both 087 vs 085) but in ADNI the differenceswere not significant (p 5 060ndash065) In ADNIthe AUCs of t-tau (081 95 CI 074ndash088) andp-tau (082 95 CI 075ndash088) were lower than in
Table 2 Classification of MCI-AD and healthy controls in BioFINDER based onROC analyses in BioFINDER
Variable (in order of AUCvalue) AUC (95 CI) AUC significantly better than
CSF Ab42ITp-tauIT 094 (089ndash097) PET medial temporal PET occipital CSFAb42EIAb40EI Ab42EI Ab42IT
CSF Ab42EIt-tauEI 093 (088ndash097) PET medial temporal CSF Ab42EIAb40EIAb42EI
CSF Ab42ITt-tauEI 093 (088ndash097) PET medial temporal PET occipital CSFAb42EIAb40EI Ab42EI Ab42IT
CSF Ab42EIp-tauIT 093 (088ndash096) PET medial temporal CSF Ab42EIAb40EIAb42EI
PET posterior cingulateprecuneus
093 (087ndash096) PET medial temporal PET occipital PETsensorimotor CSF Ab42EI
PET anterior cingulate 092 (087ndash096) PET medial temporal PET occipital CSFAb42EI
PET composite 092 (086ndash095) PET medial temporal PET occipital
PET prefrontal 091 (086ndash095) PET medial temporal CSF Ab42EI
PET parietal 091 (085ndash095) PET medial temporal PET occipital PETsensorimotor
CSF Ab42IT 090 (084ndash094) PET medial temporal CSF Ab42EI
PET lateral temporal 090 (084ndash094) PET medial temporal PET occipital
CSF Ab42ITAb40EI 088 (080ndash094) PET medial temporal
CSF Ab42EIAb40EI 086 (080ndash091) PET medial temporal
PET sensorimotor 085 (079ndash090) PET medial temporal
PET occipital 084 (077ndash089) PET medial temporal
CSF Ab42EI 082 (074ndash089)
PET medial temporal 075 (068ndash082)
Abbreviations AUC 5 area under the curve CI 5 confidence interval EI 5 EUROIMMUNassay IT 5 INNOTEST assay MCI-AD 5 patients with mild cognitive impairment who devel-oped Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauAUC was calculated with ROC analysis The 95 CI and significance for differencesbetween the AUCs were calculated using bootstrap techniques with 5000 bootstrapreplicas
Neurology 85 October 6 2015 1243
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
BioFINDER (t-tau 088 95 CI 082ndash093 andp-tau 087 95 CI 080ndash091 data not shown inthe tables) However no significant differences couldbe tested since the results were derived from 2 differ-ent cohorts
When combining CSF Ab42p-tau and compos-ite PET SUVR in ADNI the AUC was 087(95 CI 082ndash093) This did not differ significantlyfrom using the variables separately (AUC difference000ndash001 p 5 040ndash053)
DISCUSSION The main finding of this study wasthat the diagnostic accuracy of CSF and Ab PETbiomarkers to identify MCI-AD was similar whenusing 18F-flutemetamol amyloid PET and severaldifferent CSF biomarkers Specifically the best CSFmeasures (CSF Ab42t-tau and Ab42p-tau ratios)had similar diagnostic accuracies as the best PETmeasures (composite and cingulate SUVRs table2) We also found that no regional PET biomarkerwas better than the neocortical composite PET SUVR(table 2) For CSF biomarkers the CSF Ab42t-tauor Ab42p-tau ratios had significantly higherdiagnostic accuracy compared to using CSF Abbiomarkers alone When using unbiased cutoffs
CSF Ab42t-tau had the highest sensitivity andspecificity of all CSF and PET biomarkers (table 3)Finally the combination of the best CSF and PETbiomarkers did not provide any added diagnosticvalue compared to using either modality separatelyThe overall results were replicated in an independentcohort (ADNI)
Although we found that CSF Ab42IT and Ab42EIAb40EI performed similarly to the best SUVRs of18F-flutemetamol PET in terms of AUCs (table 2)these CSF biomarkers generally had lower specificitiesthan the PET biomarkers when using unbiased cut-offs (table 3) The addition of t-tau or p-tau to Ab42(as ratios) significantly increased the diagnostic accur-acy of CSF biomarkers (table 2) This supports thecommon usage of CSF Ab42 in combination with t-tau or p-tau in clinical practice and is in agreementwith previous studies31ndash34
The diagnostic accuracy of CSF Ab42 was lowerfor EUROIMMUN compared with INNOTEST(table 2 and figure e-3) This was partly overcomeby using the Ab4240 ratio which did not improvethe accuracy of Ab42 INNOTEST (table 2) Thisfinding has not been shown previously and needs tobe replicated in future studies since the causes are
Table 3 ROC analysis in BioFINDER of MCI-AD and healthy controls based on unbiased cutoffs
Variable (in order of Youden index) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b Adjusted OR (95 CI)d
CSF Ab42ITt-tauEI 125 080 (068ndash087) 97 83 62 (49ndash796)
CSF Ab42ITp-tauIT 724 079 (066ndash087) 94 85 34 (44ndash263)
CSF Ab42EIp-tauIT 567 074 (056ndash084) 88 86 20 (33ndash126)
PET prefrontal 148 074 (056ndash084) 88 86 29 (36ndash239)
CSF Ab42EIt-tauEI 096 073 (059ndash083) 91 82 44 (43ndash447)
PET posterior cingulateprecuneus 162 073 (056ndash083) 88 84 24 (33ndash182)
PET anterior cingulate 162 072 (056ndash081) 91 80 22 (29ndash170)
PET composite 151 072 (055ndash083) 85 87 29 (36ndash239)
PET lateral temporal 158 071 (054ndash083) 85 86 30 (36ndash250)
PET parietal 143 070 (053ndash081) 85 84 25 (32ndash190)
PET sensorimotor 153 069 (051ndash081) 79 89 30 (38ndash231)
CSF Ab42EIAb40EI 0092 067 (050ndash078) 88 79 14 (25ndash82)
CSF Ab42IT 500 060 (042ndash072) 85 75 61 (12ndash30)
CSF Ab42ITAb40EI 010 059 (039ndash073) 76 83 82 (17ndash40)
PET occipital 168 049 (030ndash066) 56 93 27 (28ndash263)
PET medial temporal 169 020 (007ndash038) 23 97 Nonsignificant
Abbreviations CI5 confidence interval EI5 EUROIMMUN assay IT5 INNOTEST assay MCI-AD5 patients with mild cognitive impairment who developedAlzheimer disease within 3 years OR 5 odds ratio p-tau 5 hyperphosphorylated tau ROC 5 receiving operating characteristic t-tau 5 total tauArea under the curve was calculated with ROC analysis The 95 CI and significance for differences between the AUCs were calculated using bootstraptechniques with 5000 bootstrap replicas PET values are shown in standardized uptake value ratio and CSF levels in pgmL (except for the CSF ratios)a Established with mixture modeling analysis (described in Methods)b Based on the ROC analysis (not derived from the logistic regression analysis)c Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffsdA logistic regression analysis was performed with the diagnosis (MCI-AD or control) as the dependent variable and the dichotomized CSFPET variable asa covariate to yield an OR Age sex APOE e4 allele memory function and hippocampal volume were adjusted for in the model
1244 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
unknown Few studies have compared differentELISAs for CSF Ab42 to identify MCI-AD Hertzeet al33 found that CSF Ab42 analyzed with xMAP
AlzBio3 had higher diagnostic accuracy comparedwith the Meso Scale Discovery (MSD) assay but thiswas overcome by using the Ab42Ab40 MSD ratio
Figure 1 Sensitivities specificities and Youden indices of all possible cutoff points
(A B) The cutoffs have been transformed to z scores (SD) for easier comparison between variables Near optimal Youden indicesare found for a relatively wide range of cutoffs (1 SD) for all variables except for CSF Ab42Ab40EI which have a slightlynarrower span The different cutoffs within this near optimal range thus only change the relationship between the sensitivity andspecificity but not the overall classification accuracy Thiswide range of near optimal cutoffs suggests that the cutoffs are likely toproduce high diagnostic accuracies in other populations AUC 5 area under the receiver operating characteristic curve
Neurology 85 October 6 2015 1245
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
Similar to the present study they showed that Ab42t-tau was superior to Ab42 and Ab42Ab40
A possible advantage of Ab PET over CSF Ab42as an early marker for amyloid pathology is that AbPET may be able to identify early region-specificpathology However this was not supported by ourstudy since the global Ab uptake performed similarcompared to the best regional SUVRs (tables 2 and4) Only one previous study has examined this andfound similar results19 The similar AUCs for the bestPET regions support the notion that the Ab deposi-tion is uniformly distributed in the neocortical asso-ciation areas already at the MCI stage of AD35
We used classification cutoffs established withmixture modeling which is a robust way of determin-ing unbiased thresholds and used in several stud-ies203637 Even so the cutoffs (table 3) should notbe considered generalizable but study-specific forcomparative purposes However figure 1 shows thatalthough a cutoff is not optimized for the currentpopulation it can still provide good diagnostic accur-acy because of the broad range of high YI The sta-bility of cutoffs between populations is also supportedby a previous cross-validation study on CSF Ab42and amyloid PET cutoffs20 However even thoughthe classification accuracy stays the same a change incutoff will of course result in a higher sensitivitylowerspecificity or lower sensitivityhigher specificity andthis must be taken into consideration depending onthe clinical aim of the examination
The overall results were similar between the Bio-FINDER and ADNI cohorts In ADNI the samecomparable results between regional and compositeSUVRs as well as equal diagnostic accuracies ofCSF and PET measures were seen (table 4) This
similarity between studies is especially interestingconsidering the use of different PET tracers and dif-ferent CSF assays In both cohorts numerically high-er AUCs were seen for the CSF Ab42t-tau or p-tauratios compared with just Ab42 but in ADNI theincrease was not significant (tables 2 and 4) Thiscould be attributed to the poorer AUCs of t-tauand p-tau in ADNI (081 and 082 AlzBio3)compared with BioFINDER (088 and 087EUROIMMUN and INNOTEST) A similar differ-ence between INNOTEST and AlzBio3 regardingAb42tau ratios was also found in a previous study38
It was notable that the AUCs of all brain regions weresimilar in ADNI (AUC range 001 table 4) in con-trast to BioFINDER (AUC range 017 table 2) Thereason for this could be that in ADNI the regionswere coarser and not able to detect differencesbetween eg the medial and lateral temporal lobe
The diagnostic accuracy of Ab PET and CSF bio-markers to detect incipient AD has only been comparedhead-to-head in one previous study which partly usedthe same ADNI data used for replication in the presentstudy19 In that study the accuracy between stable MCI(2- to 3-year follow-up without progression) and MCI-AD was compared In the present study we insteadcompared healthy elderly and patients with MCI-ADwhich resulted in higher AUCs (on average about 005in the ADNI study compare reference 19 and table 4)The rationale behind comparing controls andMCI-ADis that5ndash10 years of follow-up is required before onecan say that a patient withMCI is truly stable36 Amongpatients with stable MCI with a short follow-up timethere are several cases with early-stage AD These pa-tients with stable MCI will in most cases be correctlyidentified as MCI-AD by the biomarkers but result in
Table 4 Classification of MCI-AD and healthy controls based on ROC analyses in ADNI
Variable (in order of AUC value) AUC (95 CI) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b
PET frontal 087 (082ndash091) 113 063 (051ndash073) 89 74
CSF Ab42t-tau 087 (080ndash092) 171 065 (053ndash076) 80 86
CSF Ab42p-tau 087 (080ndash092) 383 065 (052ndash075) 84 81
PET composite 086 (081ndash091) 113 066 (054ndash075) 91 75
PET parietal 086 (081ndash091) 111 059 (047ndash069) 91 68
PET temporal 086 (080ndash092) 108 066 (054ndash076) 89 77
PET cingulate 086 (080ndash091) 120 056 (044ndash066) 89 66
CSF Ab42 085 (079ndash090) 173 060 (050ndash070) 94 66
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative AUC5 area under the receiver operating characteristic curve CI5 confidence intervalMCI-AD 5 patients with mild cognitive impairment who developed Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauNo significant differences between the AUCs were found (p 5 017ndash093)a Established with mixture modeling analysis (described in Methods)bDerived from the ROC analysisc Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffs
1246 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
false low specificity (and a false low AUC) due to theincorrect clinical diagnosisshort follow-up We there-fore compared patients with MCI-AD and controlsgiven the relatively short follow-up data in the ADNIand BioFINDER populations for a more robust com-parison of Ab biomarkers
The novelties of the present study compared withthe previous study19 include a comparison betweenMCI-AD and controls a more detailed analysis ofregional Ab PET data analyses of ratios of CSFAb42Ab40 Ab42t-tau and Ab42p-tau a compar-ison of 2 different ELISAs for CSF Ab42 and evalu-ation of the combination of PET and CSF biomarkers
The similar results we found for CSF biomarkersand amyloid PET suggest that other factors than theirdiagnostic accuracy may be considered when decidingwhich biomarker to use CSF analysis has the advan-tages that it may easily incorporate other biomarkersto improve the differential diagnosis (eg leukocytesalbumin ratio neurofilament a-synuclein) requiresless advanced instruments than PET and is in somecountries more available in clinical practice AmyloidPET on the other hand is less invasive and has ahigher reliability in longitudinal examinations andbetween centers With appropriate standardizedprocedures203940 CSF analysis and amyloid PET per-form equally well and either method can be used in theclinical workup of AD for increased diagnosticaccuracy
AUTHOR AFFILIATIONSFrom the Clinical Memory Research Unit Department of Clinical Scien-
ces (SP LM OH) and the Department of Computational Biology
and Biological Physics (MO) Lund University the Department of
Neurology (SP) and the Memory Clinic (LM OH) Skaringne University
Hospital the Clinical Neurochemistry Laboratory (HZ NM KB)
Institute of Neuroscience and Physiology the Sahlgrenska Academy at
the University of Gothenburg Moumllndal Sweden the UCL Institute of
Neurology (HZ) London UK the Department of Radiology and Bio-
medical Imaging (NM) University of California San Francisco the
Department of Veterans Affairs Medical Center (NM) Center for Imag-
ing of Neurodegenerative Diseases San Francisco CA and the Depart-
ment of Internal Medicine (PJ) Sahlgrenska Academy University of
Gothenburg Sweden
AUTHOR CONTRIBUTIONSSebastian Palmqvist draftingrevising the manuscript study concept or
design analysis or interpretation of data accepts responsibility for con-
duct of research and final approval statistical analysis study supervision
Henrik Zetterberg draftingrevising the manuscript analysis or interpre-
tation of data accepts responsibility for conduct of research and
final approval acquisition of data obtaining funding Niklas Mattsson
draftingrevising the manuscript analysis or interpretation of data
accepts responsibility for conduct of research and final approval Per
Johansson draftingrevising the manuscript analysis or interpretation
of data accepts responsibility for conduct of research and final approval
acquisition of data study supervision Lennart Minthon draftingrevising
the manuscript accepts responsibility for conduct of research and final
approval obtaining funding Kaj Blennow draftingrevising the manu-
script study concept or design accepts responsibility for conduct of
research and final approval study supervision obtaining funding Mattias
Ohlsson draftingrevising the manuscript analysis or interpretation of
data accepts responsibility for conduct of research and final approval sta-
tistical analysis Oskar Hansson draftingrevising the manuscript study
concept or design analysis or interpretation of data accepts responsibility
for conduct of research and final approval contribution of vital reagents
toolspatients acquisition of data study supervision obtaining funding
ACKNOWLEDGMENTThe authors thank the collaborators of this study and the entire BioFINDER
study group (wwwbiofinderse) including Ulf Andreasson Susanna Vestberg
for classifying the patients with MCI-AD into MCI subgroups Erik Stomrud
and Katarina Naumlgga for clinical evaluations of cognitively healthy individuals
Christer Nilsson for clinical evaluations of patients with mild cognitive symptoms
Per Wollmer and Douglas Haumlgerstroumlm for help with 18F-flutemetamol PET
imaging Olof Lindberg for analyzing the hippocampal volumes and Karin
Nilsson Rosita Nordkvist Ida Friberg Malin Ottheacuten and Johanna Fredlund
for organizing inclusions and assessments
STUDY FUNDINGWork in the authorsrsquo laboratory was supported by the European Research
Council the Swedish Research Council the Strategic Research Area Multi-
Park (Multidisciplinary Research in Parkinsonrsquos disease) at Lund University
the Crafoord Foundation the Swedish Brain Foundation the Skaringne Uni-
versity Hospital Foundation the Swedish Alzheimer Association Stiftelsen
foumlr Gamla Tjaumlnarinnor and the Swedish federal government under the ALF
agreement The Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) data
collection and sharing for this project was funded by ADNI (NIH grant
U01 AG024904) and DOD ADNI (Department of Defense award num-
ber W81XWH-12-2-0012) ADNI is funded by the National Institute on
Aging the National Institute of Biomedical Imaging and Bioengineering
and through contributions from the following Alzheimerrsquos Association
Alzheimerrsquos Drug Discovery Foundation Araclon Biotech BioClinica
Inc Biogen Idec Inc Bristol-Myers Squibb Company Eisai Inc Elan
Pharmaceuticals Inc Eli Lilly and Company EuroImmun F Hoffmann-
La Roche Ltd and its affiliated company Genentech Inc Fujirebio GE
Healthcare IXICO Ltd Janssen Alzheimer Immunotherapy Research amp
Development LLC Johnson amp Johnson Pharmaceutical Research amp
Development LLC Medpace Inc Merck amp Co Inc Meso Scale Diag-
nostics LLC NeuroRx Research Neurotrack Technologies Novartis Phar-
maceuticals Corporation Pfizer Inc Piramal Imaging Servier Synarc Inc
and Takeda Pharmaceutical Company The Canadian Institutes of Health
Research is providing funds to support ADNI clinical sites in Canada
Private sector contributions are facilitated by the Foundation for the
NIH (wwwfnihorg) The grantee organization is the Northern California
Institute for Research and Education and the study is coordinated by the
Alzheimerrsquos Disease Cooperative Study at the University of California San
Diego ADNI data are disseminated by the Laboratory for NeuroImaging at
the University of Southern California
DISCLOSURES Palmqvist H Zetterberg N Mattsson P Johansson and L Minthon
report no disclosures relevant to the manuscript K Blennow has served
at advisory boards for Koyowa Kirin Pharma Eli Lilly Pfizer and Roche
Doses of 18F-flutemetamol were sponsored by GE Healthcare M Ohlsson
and O Hansson report no disclosures relevant to the manuscript Go to
Neurologyorg for full disclosures
Received February 4 2015 Accepted in final form June 3 2015
REFERENCES1 Albert MS DeKosky ST Dickson D et al The diagnosis of
mild cognitive impairment due to Alzheimerrsquos disease recom-
mendations from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for
Alzheimerrsquos disease Alzheimers Dement 20117270ndash279
2 Sperling RA Aisen PS Beckett LA et al Toward defining
the preclinical stages of Alzheimerrsquos disease recommenda-
tions from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for Alz-
heimerrsquos disease Alzheimers Dement 20117280ndash292
Neurology 85 October 6 2015 1247
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
table 1) Biomarker data could not be directlycompared between the studies because of differentCSF assays and PET tracers (table 1 and table e-1)
CSF biomarkers for classification of MCI-AD and
controls The CSF biomarkers had diagnostic accura-cies for MCI-AD ranging from AUC 082 (CSFAb42EI) to AUC 093ndash094 (CSF Ab42t-tau andAb42p-tau ratios independent of assay table 2)CSF Ab42ITt-tauEI and Ab42ITp-tauIT hadsignificantly better accuracies than CSF Ab42IT
(AUC difference 004ndash005 p 5 002) andAb42EIAb40EI (AUC difference 008 p 0001)CSF Ab42EI had significantly lower AUC compared tomost other biomarkers but this could be partly overcomeby the ratio of Ab42EIAb40EI The diagnostic accuracyof CSF Ab42IT on the other hand was not improvedwhen used as a ratio with Ab40EI (table 2)
Regional and composite PET biomarkers for
classification of MCI-AD vs controls The AUCs ofthe amyloid PET biomarkers ranged from 075 to
Table 1 Baseline characteristics of MCI-AD patients and cognitively healthy elderly from the BioFINDER study and the ADNI study
BioFINDER MCI-AD(n 5 34)
BioFINDER controls(n 5 122)
p Value BioFINDERMCI-AD vs controls
ADNI MCI-AD(n 5 64)
ADNI controls(n 5 146)
p Value ADNIMCI-AD vs controls
Demographics
Age y (range) 727 (63ndash80) 735 (65ndash85) 029 721 (48ndash85) 732 (56ndash89) 079
Female 54 64 060 45 52 042
Education y 119 6 38 113 6 33 067 160 6 28 166 6 25 013
APOE e4 Dagger1 allele 61 24 0001 73 27 0001
MMSE points 267 6 15 290 6 09 0001 270 6 18 291 6 12 0001
10-word delayed recallerrors
72 6 23 20 6 20 0001
Hippocampus volume cm3 31 6 05 36 6 04 0001
PET regions
Globalcomposite 211 6 047 129 6 028 0001
Prefrontal 211 6 049 124 6 031 0001
Anterior cingulate 236 6 052 141 6 034 0001
Posterior cingulateprecuneus
226 6 048 139 6 033 0001
Parietal 198 6 044 123 6 026 0001
Lateral temporal 214 6 050 139 6 025 0001
Medial temporal 158 6 029 136 6 016 0001
Occipital 183 6 040 137 6 019 0001
Sensorimotor 181 6 042 131 6 018 0001
CSF analyses
Ab42EI 333 6 114 538 6 186 0001
Ab42IT 380 6 102 659 6 184 0001
Ab40EI 4881 6 1877 4516 6 1522 047
t-tauEI 581 6 213 318 6 113 0001
p-tauIT 924 6 339 535 6 180 0001
Ab42EIAb40EI 0072 6 0024 012 6 0038 0001
Ab42ITAb40EI 0086 6 0034 016 6 0052 0001
Ab42EIt-tauEI 063 6 027 187 6 076 0001
Ab42EIp-tauIT 396 6 173 110 6 442 0001
Ab42ITt-tauEI 073 6 033 231 6 089 0001
Ab42ITp-tauIT 459 6 20 137 6 539 0001
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative EI5 EUROIMMUN assay IT5 INNOTEST assay MCI-AD5 patients with mild cognitiveimpairment who developed Alzheimer disease within 3 years MMSE 5 Mini-Mental State Examination p-tau 5 hyperphosphorylated tau t-tau 5 total tauBiomarker data of the replication population (ADNI study) can be found in table e-1 As for comparisons between demographics in the BioFINDER and ADNIcohorts only education differed significantly (p 0001) Values are mean 6 SD unless otherwise specified CSF measures are given in pgmL and PETscore in mean standardized uptake value ratio
1242 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
092 (table 2) The AUCs of the composite PETSUVR and best regional PET SUVRs (anterior cin-gulate and posterior cingulateprecuneus) wereequally good (p 5 035ndash046) The prefrontaland parietal regional SUVRs had similar AUCs(p 5 049ndash099) The medial temporal SUVRperformed significantly worse than all otherPET measures (AUC difference 009ndash017 p 5
00001ndash002)
Comparison of CSF and PET biomarkers for
classification of MCI-AD vs controls The best CSF bio-markers (CSF Ab42ITt-tauEI and Ab42ITp-tauIT
ratios) had similar AUCs (093ndash094) as the bestPET measures (AUC 092ndash093 p 5 034ndash60)CSF Ab42IT also performed similar to the best PETmeasures (table 2) CSF Ab42EIAb40EI had anumerically poorer AUC compared to all PET varia-bles except for the sensorimotor occipital and medialtemporal regions but the differences were not signif-icant (p 5 009ndash040)
Combination of CSF and PET biomarkers To examinethe potential benefit of combining PET and CSFanalysis we tested models with CSF Ab42ITp-tauIT
and the composite PET SUVR entered separately andtogether as predictors of diagnosis in logistic regres-sion analyses When used together the AUC was096 (95 CI 092ndash097) and both variables wereindependent significant predictors (p 001) Thiswas numerically higher than for models using theindividual modalities but the differences were notsignificant (AUC difference 0021ndash0047 p 5
007ndash008) A combined model of the compositePET SUVR and CSF p-tauIT had equal AUC valueas CSF Ab42ITp-tauIT (both were 094 95 CI089ndash097)
Classification of incipient AD and controls at specific
cutoffs All Ab variables had a bimodal distributionsuitable for establishing nonoptimized unbiased cut-offs with mixture modeling except for CSF Ab42EIwhich was excluded from this analysis When usingthese cutoffs in the ROC analysis CSF Ab42ITt-tauEI
and CSF Ab42ITp-tauIT had the best sensitivities andspecificities of all CSF and PET measures (table 3)The 2 best PET measures were the prefrontal andthe posterior cingulateprecuneus regions Scatterplotsshow that the differences in specificities between CSFand PET are mostly caused by controls with normalPET and abnormal CSF values (figures e-1 and e-2)In logistic regressions CSF Ab42t-tauIT and EI hadthe highest OR when adjusting for age sex mem-ory function APOE e4 and hippocampal volume(table 3)
The diagnostic accuracy of biomarkers used in theclinic should preferably not be very sensitive tosmaller changes in cutoff values if they are to be gen-eralizable between different centers and settings Infigure 1 A and B continuous sensitivities and specif-icities of 4 CSF and PET measures are shown as afunction of the cutoff point The CSF and PETmeas-ures were not dependent on an optimized cutoff butprovide high accuracies from cutoff values spanningat least 1 SD in the current sample The exceptionwas CSF Ab42EI40EI which had a slightly narrowerinterval with near optimal YI
Comparison with the ADNI data Accuracies for CSFand PET biomarkers were also analyzed in the inde-pendent ADNI cohort (table 4) All CSF and PETvariables had similar AUCs ranging from 086 to 087and no significant differences were found (p5 017ndash093) As in BioFINDER CSF Ab42t-tau andAb42p-tau had higher AUCs than CSF Ab42 alone(both 087 vs 085) but in ADNI the differenceswere not significant (p 5 060ndash065) In ADNIthe AUCs of t-tau (081 95 CI 074ndash088) andp-tau (082 95 CI 075ndash088) were lower than in
Table 2 Classification of MCI-AD and healthy controls in BioFINDER based onROC analyses in BioFINDER
Variable (in order of AUCvalue) AUC (95 CI) AUC significantly better than
CSF Ab42ITp-tauIT 094 (089ndash097) PET medial temporal PET occipital CSFAb42EIAb40EI Ab42EI Ab42IT
CSF Ab42EIt-tauEI 093 (088ndash097) PET medial temporal CSF Ab42EIAb40EIAb42EI
CSF Ab42ITt-tauEI 093 (088ndash097) PET medial temporal PET occipital CSFAb42EIAb40EI Ab42EI Ab42IT
CSF Ab42EIp-tauIT 093 (088ndash096) PET medial temporal CSF Ab42EIAb40EIAb42EI
PET posterior cingulateprecuneus
093 (087ndash096) PET medial temporal PET occipital PETsensorimotor CSF Ab42EI
PET anterior cingulate 092 (087ndash096) PET medial temporal PET occipital CSFAb42EI
PET composite 092 (086ndash095) PET medial temporal PET occipital
PET prefrontal 091 (086ndash095) PET medial temporal CSF Ab42EI
PET parietal 091 (085ndash095) PET medial temporal PET occipital PETsensorimotor
CSF Ab42IT 090 (084ndash094) PET medial temporal CSF Ab42EI
PET lateral temporal 090 (084ndash094) PET medial temporal PET occipital
CSF Ab42ITAb40EI 088 (080ndash094) PET medial temporal
CSF Ab42EIAb40EI 086 (080ndash091) PET medial temporal
PET sensorimotor 085 (079ndash090) PET medial temporal
PET occipital 084 (077ndash089) PET medial temporal
CSF Ab42EI 082 (074ndash089)
PET medial temporal 075 (068ndash082)
Abbreviations AUC 5 area under the curve CI 5 confidence interval EI 5 EUROIMMUNassay IT 5 INNOTEST assay MCI-AD 5 patients with mild cognitive impairment who devel-oped Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauAUC was calculated with ROC analysis The 95 CI and significance for differencesbetween the AUCs were calculated using bootstrap techniques with 5000 bootstrapreplicas
Neurology 85 October 6 2015 1243
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
BioFINDER (t-tau 088 95 CI 082ndash093 andp-tau 087 95 CI 080ndash091 data not shown inthe tables) However no significant differences couldbe tested since the results were derived from 2 differ-ent cohorts
When combining CSF Ab42p-tau and compos-ite PET SUVR in ADNI the AUC was 087(95 CI 082ndash093) This did not differ significantlyfrom using the variables separately (AUC difference000ndash001 p 5 040ndash053)
DISCUSSION The main finding of this study wasthat the diagnostic accuracy of CSF and Ab PETbiomarkers to identify MCI-AD was similar whenusing 18F-flutemetamol amyloid PET and severaldifferent CSF biomarkers Specifically the best CSFmeasures (CSF Ab42t-tau and Ab42p-tau ratios)had similar diagnostic accuracies as the best PETmeasures (composite and cingulate SUVRs table2) We also found that no regional PET biomarkerwas better than the neocortical composite PET SUVR(table 2) For CSF biomarkers the CSF Ab42t-tauor Ab42p-tau ratios had significantly higherdiagnostic accuracy compared to using CSF Abbiomarkers alone When using unbiased cutoffs
CSF Ab42t-tau had the highest sensitivity andspecificity of all CSF and PET biomarkers (table 3)Finally the combination of the best CSF and PETbiomarkers did not provide any added diagnosticvalue compared to using either modality separatelyThe overall results were replicated in an independentcohort (ADNI)
Although we found that CSF Ab42IT and Ab42EIAb40EI performed similarly to the best SUVRs of18F-flutemetamol PET in terms of AUCs (table 2)these CSF biomarkers generally had lower specificitiesthan the PET biomarkers when using unbiased cut-offs (table 3) The addition of t-tau or p-tau to Ab42(as ratios) significantly increased the diagnostic accur-acy of CSF biomarkers (table 2) This supports thecommon usage of CSF Ab42 in combination with t-tau or p-tau in clinical practice and is in agreementwith previous studies31ndash34
The diagnostic accuracy of CSF Ab42 was lowerfor EUROIMMUN compared with INNOTEST(table 2 and figure e-3) This was partly overcomeby using the Ab4240 ratio which did not improvethe accuracy of Ab42 INNOTEST (table 2) Thisfinding has not been shown previously and needs tobe replicated in future studies since the causes are
Table 3 ROC analysis in BioFINDER of MCI-AD and healthy controls based on unbiased cutoffs
Variable (in order of Youden index) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b Adjusted OR (95 CI)d
CSF Ab42ITt-tauEI 125 080 (068ndash087) 97 83 62 (49ndash796)
CSF Ab42ITp-tauIT 724 079 (066ndash087) 94 85 34 (44ndash263)
CSF Ab42EIp-tauIT 567 074 (056ndash084) 88 86 20 (33ndash126)
PET prefrontal 148 074 (056ndash084) 88 86 29 (36ndash239)
CSF Ab42EIt-tauEI 096 073 (059ndash083) 91 82 44 (43ndash447)
PET posterior cingulateprecuneus 162 073 (056ndash083) 88 84 24 (33ndash182)
PET anterior cingulate 162 072 (056ndash081) 91 80 22 (29ndash170)
PET composite 151 072 (055ndash083) 85 87 29 (36ndash239)
PET lateral temporal 158 071 (054ndash083) 85 86 30 (36ndash250)
PET parietal 143 070 (053ndash081) 85 84 25 (32ndash190)
PET sensorimotor 153 069 (051ndash081) 79 89 30 (38ndash231)
CSF Ab42EIAb40EI 0092 067 (050ndash078) 88 79 14 (25ndash82)
CSF Ab42IT 500 060 (042ndash072) 85 75 61 (12ndash30)
CSF Ab42ITAb40EI 010 059 (039ndash073) 76 83 82 (17ndash40)
PET occipital 168 049 (030ndash066) 56 93 27 (28ndash263)
PET medial temporal 169 020 (007ndash038) 23 97 Nonsignificant
Abbreviations CI5 confidence interval EI5 EUROIMMUN assay IT5 INNOTEST assay MCI-AD5 patients with mild cognitive impairment who developedAlzheimer disease within 3 years OR 5 odds ratio p-tau 5 hyperphosphorylated tau ROC 5 receiving operating characteristic t-tau 5 total tauArea under the curve was calculated with ROC analysis The 95 CI and significance for differences between the AUCs were calculated using bootstraptechniques with 5000 bootstrap replicas PET values are shown in standardized uptake value ratio and CSF levels in pgmL (except for the CSF ratios)a Established with mixture modeling analysis (described in Methods)b Based on the ROC analysis (not derived from the logistic regression analysis)c Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffsdA logistic regression analysis was performed with the diagnosis (MCI-AD or control) as the dependent variable and the dichotomized CSFPET variable asa covariate to yield an OR Age sex APOE e4 allele memory function and hippocampal volume were adjusted for in the model
1244 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
unknown Few studies have compared differentELISAs for CSF Ab42 to identify MCI-AD Hertzeet al33 found that CSF Ab42 analyzed with xMAP
AlzBio3 had higher diagnostic accuracy comparedwith the Meso Scale Discovery (MSD) assay but thiswas overcome by using the Ab42Ab40 MSD ratio
Figure 1 Sensitivities specificities and Youden indices of all possible cutoff points
(A B) The cutoffs have been transformed to z scores (SD) for easier comparison between variables Near optimal Youden indicesare found for a relatively wide range of cutoffs (1 SD) for all variables except for CSF Ab42Ab40EI which have a slightlynarrower span The different cutoffs within this near optimal range thus only change the relationship between the sensitivity andspecificity but not the overall classification accuracy Thiswide range of near optimal cutoffs suggests that the cutoffs are likely toproduce high diagnostic accuracies in other populations AUC 5 area under the receiver operating characteristic curve
Neurology 85 October 6 2015 1245
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
Similar to the present study they showed that Ab42t-tau was superior to Ab42 and Ab42Ab40
A possible advantage of Ab PET over CSF Ab42as an early marker for amyloid pathology is that AbPET may be able to identify early region-specificpathology However this was not supported by ourstudy since the global Ab uptake performed similarcompared to the best regional SUVRs (tables 2 and4) Only one previous study has examined this andfound similar results19 The similar AUCs for the bestPET regions support the notion that the Ab deposi-tion is uniformly distributed in the neocortical asso-ciation areas already at the MCI stage of AD35
We used classification cutoffs established withmixture modeling which is a robust way of determin-ing unbiased thresholds and used in several stud-ies203637 Even so the cutoffs (table 3) should notbe considered generalizable but study-specific forcomparative purposes However figure 1 shows thatalthough a cutoff is not optimized for the currentpopulation it can still provide good diagnostic accur-acy because of the broad range of high YI The sta-bility of cutoffs between populations is also supportedby a previous cross-validation study on CSF Ab42and amyloid PET cutoffs20 However even thoughthe classification accuracy stays the same a change incutoff will of course result in a higher sensitivitylowerspecificity or lower sensitivityhigher specificity andthis must be taken into consideration depending onthe clinical aim of the examination
The overall results were similar between the Bio-FINDER and ADNI cohorts In ADNI the samecomparable results between regional and compositeSUVRs as well as equal diagnostic accuracies ofCSF and PET measures were seen (table 4) This
similarity between studies is especially interestingconsidering the use of different PET tracers and dif-ferent CSF assays In both cohorts numerically high-er AUCs were seen for the CSF Ab42t-tau or p-tauratios compared with just Ab42 but in ADNI theincrease was not significant (tables 2 and 4) Thiscould be attributed to the poorer AUCs of t-tauand p-tau in ADNI (081 and 082 AlzBio3)compared with BioFINDER (088 and 087EUROIMMUN and INNOTEST) A similar differ-ence between INNOTEST and AlzBio3 regardingAb42tau ratios was also found in a previous study38
It was notable that the AUCs of all brain regions weresimilar in ADNI (AUC range 001 table 4) in con-trast to BioFINDER (AUC range 017 table 2) Thereason for this could be that in ADNI the regionswere coarser and not able to detect differencesbetween eg the medial and lateral temporal lobe
The diagnostic accuracy of Ab PET and CSF bio-markers to detect incipient AD has only been comparedhead-to-head in one previous study which partly usedthe same ADNI data used for replication in the presentstudy19 In that study the accuracy between stable MCI(2- to 3-year follow-up without progression) and MCI-AD was compared In the present study we insteadcompared healthy elderly and patients with MCI-ADwhich resulted in higher AUCs (on average about 005in the ADNI study compare reference 19 and table 4)The rationale behind comparing controls andMCI-ADis that5ndash10 years of follow-up is required before onecan say that a patient withMCI is truly stable36 Amongpatients with stable MCI with a short follow-up timethere are several cases with early-stage AD These pa-tients with stable MCI will in most cases be correctlyidentified as MCI-AD by the biomarkers but result in
Table 4 Classification of MCI-AD and healthy controls based on ROC analyses in ADNI
Variable (in order of AUC value) AUC (95 CI) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b
PET frontal 087 (082ndash091) 113 063 (051ndash073) 89 74
CSF Ab42t-tau 087 (080ndash092) 171 065 (053ndash076) 80 86
CSF Ab42p-tau 087 (080ndash092) 383 065 (052ndash075) 84 81
PET composite 086 (081ndash091) 113 066 (054ndash075) 91 75
PET parietal 086 (081ndash091) 111 059 (047ndash069) 91 68
PET temporal 086 (080ndash092) 108 066 (054ndash076) 89 77
PET cingulate 086 (080ndash091) 120 056 (044ndash066) 89 66
CSF Ab42 085 (079ndash090) 173 060 (050ndash070) 94 66
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative AUC5 area under the receiver operating characteristic curve CI5 confidence intervalMCI-AD 5 patients with mild cognitive impairment who developed Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauNo significant differences between the AUCs were found (p 5 017ndash093)a Established with mixture modeling analysis (described in Methods)bDerived from the ROC analysisc Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffs
1246 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
false low specificity (and a false low AUC) due to theincorrect clinical diagnosisshort follow-up We there-fore compared patients with MCI-AD and controlsgiven the relatively short follow-up data in the ADNIand BioFINDER populations for a more robust com-parison of Ab biomarkers
The novelties of the present study compared withthe previous study19 include a comparison betweenMCI-AD and controls a more detailed analysis ofregional Ab PET data analyses of ratios of CSFAb42Ab40 Ab42t-tau and Ab42p-tau a compar-ison of 2 different ELISAs for CSF Ab42 and evalu-ation of the combination of PET and CSF biomarkers
The similar results we found for CSF biomarkersand amyloid PET suggest that other factors than theirdiagnostic accuracy may be considered when decidingwhich biomarker to use CSF analysis has the advan-tages that it may easily incorporate other biomarkersto improve the differential diagnosis (eg leukocytesalbumin ratio neurofilament a-synuclein) requiresless advanced instruments than PET and is in somecountries more available in clinical practice AmyloidPET on the other hand is less invasive and has ahigher reliability in longitudinal examinations andbetween centers With appropriate standardizedprocedures203940 CSF analysis and amyloid PET per-form equally well and either method can be used in theclinical workup of AD for increased diagnosticaccuracy
AUTHOR AFFILIATIONSFrom the Clinical Memory Research Unit Department of Clinical Scien-
ces (SP LM OH) and the Department of Computational Biology
and Biological Physics (MO) Lund University the Department of
Neurology (SP) and the Memory Clinic (LM OH) Skaringne University
Hospital the Clinical Neurochemistry Laboratory (HZ NM KB)
Institute of Neuroscience and Physiology the Sahlgrenska Academy at
the University of Gothenburg Moumllndal Sweden the UCL Institute of
Neurology (HZ) London UK the Department of Radiology and Bio-
medical Imaging (NM) University of California San Francisco the
Department of Veterans Affairs Medical Center (NM) Center for Imag-
ing of Neurodegenerative Diseases San Francisco CA and the Depart-
ment of Internal Medicine (PJ) Sahlgrenska Academy University of
Gothenburg Sweden
AUTHOR CONTRIBUTIONSSebastian Palmqvist draftingrevising the manuscript study concept or
design analysis or interpretation of data accepts responsibility for con-
duct of research and final approval statistical analysis study supervision
Henrik Zetterberg draftingrevising the manuscript analysis or interpre-
tation of data accepts responsibility for conduct of research and
final approval acquisition of data obtaining funding Niklas Mattsson
draftingrevising the manuscript analysis or interpretation of data
accepts responsibility for conduct of research and final approval Per
Johansson draftingrevising the manuscript analysis or interpretation
of data accepts responsibility for conduct of research and final approval
acquisition of data study supervision Lennart Minthon draftingrevising
the manuscript accepts responsibility for conduct of research and final
approval obtaining funding Kaj Blennow draftingrevising the manu-
script study concept or design accepts responsibility for conduct of
research and final approval study supervision obtaining funding Mattias
Ohlsson draftingrevising the manuscript analysis or interpretation of
data accepts responsibility for conduct of research and final approval sta-
tistical analysis Oskar Hansson draftingrevising the manuscript study
concept or design analysis or interpretation of data accepts responsibility
for conduct of research and final approval contribution of vital reagents
toolspatients acquisition of data study supervision obtaining funding
ACKNOWLEDGMENTThe authors thank the collaborators of this study and the entire BioFINDER
study group (wwwbiofinderse) including Ulf Andreasson Susanna Vestberg
for classifying the patients with MCI-AD into MCI subgroups Erik Stomrud
and Katarina Naumlgga for clinical evaluations of cognitively healthy individuals
Christer Nilsson for clinical evaluations of patients with mild cognitive symptoms
Per Wollmer and Douglas Haumlgerstroumlm for help with 18F-flutemetamol PET
imaging Olof Lindberg for analyzing the hippocampal volumes and Karin
Nilsson Rosita Nordkvist Ida Friberg Malin Ottheacuten and Johanna Fredlund
for organizing inclusions and assessments
STUDY FUNDINGWork in the authorsrsquo laboratory was supported by the European Research
Council the Swedish Research Council the Strategic Research Area Multi-
Park (Multidisciplinary Research in Parkinsonrsquos disease) at Lund University
the Crafoord Foundation the Swedish Brain Foundation the Skaringne Uni-
versity Hospital Foundation the Swedish Alzheimer Association Stiftelsen
foumlr Gamla Tjaumlnarinnor and the Swedish federal government under the ALF
agreement The Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) data
collection and sharing for this project was funded by ADNI (NIH grant
U01 AG024904) and DOD ADNI (Department of Defense award num-
ber W81XWH-12-2-0012) ADNI is funded by the National Institute on
Aging the National Institute of Biomedical Imaging and Bioengineering
and through contributions from the following Alzheimerrsquos Association
Alzheimerrsquos Drug Discovery Foundation Araclon Biotech BioClinica
Inc Biogen Idec Inc Bristol-Myers Squibb Company Eisai Inc Elan
Pharmaceuticals Inc Eli Lilly and Company EuroImmun F Hoffmann-
La Roche Ltd and its affiliated company Genentech Inc Fujirebio GE
Healthcare IXICO Ltd Janssen Alzheimer Immunotherapy Research amp
Development LLC Johnson amp Johnson Pharmaceutical Research amp
Development LLC Medpace Inc Merck amp Co Inc Meso Scale Diag-
nostics LLC NeuroRx Research Neurotrack Technologies Novartis Phar-
maceuticals Corporation Pfizer Inc Piramal Imaging Servier Synarc Inc
and Takeda Pharmaceutical Company The Canadian Institutes of Health
Research is providing funds to support ADNI clinical sites in Canada
Private sector contributions are facilitated by the Foundation for the
NIH (wwwfnihorg) The grantee organization is the Northern California
Institute for Research and Education and the study is coordinated by the
Alzheimerrsquos Disease Cooperative Study at the University of California San
Diego ADNI data are disseminated by the Laboratory for NeuroImaging at
the University of Southern California
DISCLOSURES Palmqvist H Zetterberg N Mattsson P Johansson and L Minthon
report no disclosures relevant to the manuscript K Blennow has served
at advisory boards for Koyowa Kirin Pharma Eli Lilly Pfizer and Roche
Doses of 18F-flutemetamol were sponsored by GE Healthcare M Ohlsson
and O Hansson report no disclosures relevant to the manuscript Go to
Neurologyorg for full disclosures
Received February 4 2015 Accepted in final form June 3 2015
REFERENCES1 Albert MS DeKosky ST Dickson D et al The diagnosis of
mild cognitive impairment due to Alzheimerrsquos disease recom-
mendations from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for
Alzheimerrsquos disease Alzheimers Dement 20117270ndash279
2 Sperling RA Aisen PS Beckett LA et al Toward defining
the preclinical stages of Alzheimerrsquos disease recommenda-
tions from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for Alz-
heimerrsquos disease Alzheimers Dement 20117280ndash292
Neurology 85 October 6 2015 1247
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
092 (table 2) The AUCs of the composite PETSUVR and best regional PET SUVRs (anterior cin-gulate and posterior cingulateprecuneus) wereequally good (p 5 035ndash046) The prefrontaland parietal regional SUVRs had similar AUCs(p 5 049ndash099) The medial temporal SUVRperformed significantly worse than all otherPET measures (AUC difference 009ndash017 p 5
00001ndash002)
Comparison of CSF and PET biomarkers for
classification of MCI-AD vs controls The best CSF bio-markers (CSF Ab42ITt-tauEI and Ab42ITp-tauIT
ratios) had similar AUCs (093ndash094) as the bestPET measures (AUC 092ndash093 p 5 034ndash60)CSF Ab42IT also performed similar to the best PETmeasures (table 2) CSF Ab42EIAb40EI had anumerically poorer AUC compared to all PET varia-bles except for the sensorimotor occipital and medialtemporal regions but the differences were not signif-icant (p 5 009ndash040)
Combination of CSF and PET biomarkers To examinethe potential benefit of combining PET and CSFanalysis we tested models with CSF Ab42ITp-tauIT
and the composite PET SUVR entered separately andtogether as predictors of diagnosis in logistic regres-sion analyses When used together the AUC was096 (95 CI 092ndash097) and both variables wereindependent significant predictors (p 001) Thiswas numerically higher than for models using theindividual modalities but the differences were notsignificant (AUC difference 0021ndash0047 p 5
007ndash008) A combined model of the compositePET SUVR and CSF p-tauIT had equal AUC valueas CSF Ab42ITp-tauIT (both were 094 95 CI089ndash097)
Classification of incipient AD and controls at specific
cutoffs All Ab variables had a bimodal distributionsuitable for establishing nonoptimized unbiased cut-offs with mixture modeling except for CSF Ab42EIwhich was excluded from this analysis When usingthese cutoffs in the ROC analysis CSF Ab42ITt-tauEI
and CSF Ab42ITp-tauIT had the best sensitivities andspecificities of all CSF and PET measures (table 3)The 2 best PET measures were the prefrontal andthe posterior cingulateprecuneus regions Scatterplotsshow that the differences in specificities between CSFand PET are mostly caused by controls with normalPET and abnormal CSF values (figures e-1 and e-2)In logistic regressions CSF Ab42t-tauIT and EI hadthe highest OR when adjusting for age sex mem-ory function APOE e4 and hippocampal volume(table 3)
The diagnostic accuracy of biomarkers used in theclinic should preferably not be very sensitive tosmaller changes in cutoff values if they are to be gen-eralizable between different centers and settings Infigure 1 A and B continuous sensitivities and specif-icities of 4 CSF and PET measures are shown as afunction of the cutoff point The CSF and PETmeas-ures were not dependent on an optimized cutoff butprovide high accuracies from cutoff values spanningat least 1 SD in the current sample The exceptionwas CSF Ab42EI40EI which had a slightly narrowerinterval with near optimal YI
Comparison with the ADNI data Accuracies for CSFand PET biomarkers were also analyzed in the inde-pendent ADNI cohort (table 4) All CSF and PETvariables had similar AUCs ranging from 086 to 087and no significant differences were found (p5 017ndash093) As in BioFINDER CSF Ab42t-tau andAb42p-tau had higher AUCs than CSF Ab42 alone(both 087 vs 085) but in ADNI the differenceswere not significant (p 5 060ndash065) In ADNIthe AUCs of t-tau (081 95 CI 074ndash088) andp-tau (082 95 CI 075ndash088) were lower than in
Table 2 Classification of MCI-AD and healthy controls in BioFINDER based onROC analyses in BioFINDER
Variable (in order of AUCvalue) AUC (95 CI) AUC significantly better than
CSF Ab42ITp-tauIT 094 (089ndash097) PET medial temporal PET occipital CSFAb42EIAb40EI Ab42EI Ab42IT
CSF Ab42EIt-tauEI 093 (088ndash097) PET medial temporal CSF Ab42EIAb40EIAb42EI
CSF Ab42ITt-tauEI 093 (088ndash097) PET medial temporal PET occipital CSFAb42EIAb40EI Ab42EI Ab42IT
CSF Ab42EIp-tauIT 093 (088ndash096) PET medial temporal CSF Ab42EIAb40EIAb42EI
PET posterior cingulateprecuneus
093 (087ndash096) PET medial temporal PET occipital PETsensorimotor CSF Ab42EI
PET anterior cingulate 092 (087ndash096) PET medial temporal PET occipital CSFAb42EI
PET composite 092 (086ndash095) PET medial temporal PET occipital
PET prefrontal 091 (086ndash095) PET medial temporal CSF Ab42EI
PET parietal 091 (085ndash095) PET medial temporal PET occipital PETsensorimotor
CSF Ab42IT 090 (084ndash094) PET medial temporal CSF Ab42EI
PET lateral temporal 090 (084ndash094) PET medial temporal PET occipital
CSF Ab42ITAb40EI 088 (080ndash094) PET medial temporal
CSF Ab42EIAb40EI 086 (080ndash091) PET medial temporal
PET sensorimotor 085 (079ndash090) PET medial temporal
PET occipital 084 (077ndash089) PET medial temporal
CSF Ab42EI 082 (074ndash089)
PET medial temporal 075 (068ndash082)
Abbreviations AUC 5 area under the curve CI 5 confidence interval EI 5 EUROIMMUNassay IT 5 INNOTEST assay MCI-AD 5 patients with mild cognitive impairment who devel-oped Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauAUC was calculated with ROC analysis The 95 CI and significance for differencesbetween the AUCs were calculated using bootstrap techniques with 5000 bootstrapreplicas
Neurology 85 October 6 2015 1243
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
BioFINDER (t-tau 088 95 CI 082ndash093 andp-tau 087 95 CI 080ndash091 data not shown inthe tables) However no significant differences couldbe tested since the results were derived from 2 differ-ent cohorts
When combining CSF Ab42p-tau and compos-ite PET SUVR in ADNI the AUC was 087(95 CI 082ndash093) This did not differ significantlyfrom using the variables separately (AUC difference000ndash001 p 5 040ndash053)
DISCUSSION The main finding of this study wasthat the diagnostic accuracy of CSF and Ab PETbiomarkers to identify MCI-AD was similar whenusing 18F-flutemetamol amyloid PET and severaldifferent CSF biomarkers Specifically the best CSFmeasures (CSF Ab42t-tau and Ab42p-tau ratios)had similar diagnostic accuracies as the best PETmeasures (composite and cingulate SUVRs table2) We also found that no regional PET biomarkerwas better than the neocortical composite PET SUVR(table 2) For CSF biomarkers the CSF Ab42t-tauor Ab42p-tau ratios had significantly higherdiagnostic accuracy compared to using CSF Abbiomarkers alone When using unbiased cutoffs
CSF Ab42t-tau had the highest sensitivity andspecificity of all CSF and PET biomarkers (table 3)Finally the combination of the best CSF and PETbiomarkers did not provide any added diagnosticvalue compared to using either modality separatelyThe overall results were replicated in an independentcohort (ADNI)
Although we found that CSF Ab42IT and Ab42EIAb40EI performed similarly to the best SUVRs of18F-flutemetamol PET in terms of AUCs (table 2)these CSF biomarkers generally had lower specificitiesthan the PET biomarkers when using unbiased cut-offs (table 3) The addition of t-tau or p-tau to Ab42(as ratios) significantly increased the diagnostic accur-acy of CSF biomarkers (table 2) This supports thecommon usage of CSF Ab42 in combination with t-tau or p-tau in clinical practice and is in agreementwith previous studies31ndash34
The diagnostic accuracy of CSF Ab42 was lowerfor EUROIMMUN compared with INNOTEST(table 2 and figure e-3) This was partly overcomeby using the Ab4240 ratio which did not improvethe accuracy of Ab42 INNOTEST (table 2) Thisfinding has not been shown previously and needs tobe replicated in future studies since the causes are
Table 3 ROC analysis in BioFINDER of MCI-AD and healthy controls based on unbiased cutoffs
Variable (in order of Youden index) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b Adjusted OR (95 CI)d
CSF Ab42ITt-tauEI 125 080 (068ndash087) 97 83 62 (49ndash796)
CSF Ab42ITp-tauIT 724 079 (066ndash087) 94 85 34 (44ndash263)
CSF Ab42EIp-tauIT 567 074 (056ndash084) 88 86 20 (33ndash126)
PET prefrontal 148 074 (056ndash084) 88 86 29 (36ndash239)
CSF Ab42EIt-tauEI 096 073 (059ndash083) 91 82 44 (43ndash447)
PET posterior cingulateprecuneus 162 073 (056ndash083) 88 84 24 (33ndash182)
PET anterior cingulate 162 072 (056ndash081) 91 80 22 (29ndash170)
PET composite 151 072 (055ndash083) 85 87 29 (36ndash239)
PET lateral temporal 158 071 (054ndash083) 85 86 30 (36ndash250)
PET parietal 143 070 (053ndash081) 85 84 25 (32ndash190)
PET sensorimotor 153 069 (051ndash081) 79 89 30 (38ndash231)
CSF Ab42EIAb40EI 0092 067 (050ndash078) 88 79 14 (25ndash82)
CSF Ab42IT 500 060 (042ndash072) 85 75 61 (12ndash30)
CSF Ab42ITAb40EI 010 059 (039ndash073) 76 83 82 (17ndash40)
PET occipital 168 049 (030ndash066) 56 93 27 (28ndash263)
PET medial temporal 169 020 (007ndash038) 23 97 Nonsignificant
Abbreviations CI5 confidence interval EI5 EUROIMMUN assay IT5 INNOTEST assay MCI-AD5 patients with mild cognitive impairment who developedAlzheimer disease within 3 years OR 5 odds ratio p-tau 5 hyperphosphorylated tau ROC 5 receiving operating characteristic t-tau 5 total tauArea under the curve was calculated with ROC analysis The 95 CI and significance for differences between the AUCs were calculated using bootstraptechniques with 5000 bootstrap replicas PET values are shown in standardized uptake value ratio and CSF levels in pgmL (except for the CSF ratios)a Established with mixture modeling analysis (described in Methods)b Based on the ROC analysis (not derived from the logistic regression analysis)c Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffsdA logistic regression analysis was performed with the diagnosis (MCI-AD or control) as the dependent variable and the dichotomized CSFPET variable asa covariate to yield an OR Age sex APOE e4 allele memory function and hippocampal volume were adjusted for in the model
1244 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
unknown Few studies have compared differentELISAs for CSF Ab42 to identify MCI-AD Hertzeet al33 found that CSF Ab42 analyzed with xMAP
AlzBio3 had higher diagnostic accuracy comparedwith the Meso Scale Discovery (MSD) assay but thiswas overcome by using the Ab42Ab40 MSD ratio
Figure 1 Sensitivities specificities and Youden indices of all possible cutoff points
(A B) The cutoffs have been transformed to z scores (SD) for easier comparison between variables Near optimal Youden indicesare found for a relatively wide range of cutoffs (1 SD) for all variables except for CSF Ab42Ab40EI which have a slightlynarrower span The different cutoffs within this near optimal range thus only change the relationship between the sensitivity andspecificity but not the overall classification accuracy Thiswide range of near optimal cutoffs suggests that the cutoffs are likely toproduce high diagnostic accuracies in other populations AUC 5 area under the receiver operating characteristic curve
Neurology 85 October 6 2015 1245
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
Similar to the present study they showed that Ab42t-tau was superior to Ab42 and Ab42Ab40
A possible advantage of Ab PET over CSF Ab42as an early marker for amyloid pathology is that AbPET may be able to identify early region-specificpathology However this was not supported by ourstudy since the global Ab uptake performed similarcompared to the best regional SUVRs (tables 2 and4) Only one previous study has examined this andfound similar results19 The similar AUCs for the bestPET regions support the notion that the Ab deposi-tion is uniformly distributed in the neocortical asso-ciation areas already at the MCI stage of AD35
We used classification cutoffs established withmixture modeling which is a robust way of determin-ing unbiased thresholds and used in several stud-ies203637 Even so the cutoffs (table 3) should notbe considered generalizable but study-specific forcomparative purposes However figure 1 shows thatalthough a cutoff is not optimized for the currentpopulation it can still provide good diagnostic accur-acy because of the broad range of high YI The sta-bility of cutoffs between populations is also supportedby a previous cross-validation study on CSF Ab42and amyloid PET cutoffs20 However even thoughthe classification accuracy stays the same a change incutoff will of course result in a higher sensitivitylowerspecificity or lower sensitivityhigher specificity andthis must be taken into consideration depending onthe clinical aim of the examination
The overall results were similar between the Bio-FINDER and ADNI cohorts In ADNI the samecomparable results between regional and compositeSUVRs as well as equal diagnostic accuracies ofCSF and PET measures were seen (table 4) This
similarity between studies is especially interestingconsidering the use of different PET tracers and dif-ferent CSF assays In both cohorts numerically high-er AUCs were seen for the CSF Ab42t-tau or p-tauratios compared with just Ab42 but in ADNI theincrease was not significant (tables 2 and 4) Thiscould be attributed to the poorer AUCs of t-tauand p-tau in ADNI (081 and 082 AlzBio3)compared with BioFINDER (088 and 087EUROIMMUN and INNOTEST) A similar differ-ence between INNOTEST and AlzBio3 regardingAb42tau ratios was also found in a previous study38
It was notable that the AUCs of all brain regions weresimilar in ADNI (AUC range 001 table 4) in con-trast to BioFINDER (AUC range 017 table 2) Thereason for this could be that in ADNI the regionswere coarser and not able to detect differencesbetween eg the medial and lateral temporal lobe
The diagnostic accuracy of Ab PET and CSF bio-markers to detect incipient AD has only been comparedhead-to-head in one previous study which partly usedthe same ADNI data used for replication in the presentstudy19 In that study the accuracy between stable MCI(2- to 3-year follow-up without progression) and MCI-AD was compared In the present study we insteadcompared healthy elderly and patients with MCI-ADwhich resulted in higher AUCs (on average about 005in the ADNI study compare reference 19 and table 4)The rationale behind comparing controls andMCI-ADis that5ndash10 years of follow-up is required before onecan say that a patient withMCI is truly stable36 Amongpatients with stable MCI with a short follow-up timethere are several cases with early-stage AD These pa-tients with stable MCI will in most cases be correctlyidentified as MCI-AD by the biomarkers but result in
Table 4 Classification of MCI-AD and healthy controls based on ROC analyses in ADNI
Variable (in order of AUC value) AUC (95 CI) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b
PET frontal 087 (082ndash091) 113 063 (051ndash073) 89 74
CSF Ab42t-tau 087 (080ndash092) 171 065 (053ndash076) 80 86
CSF Ab42p-tau 087 (080ndash092) 383 065 (052ndash075) 84 81
PET composite 086 (081ndash091) 113 066 (054ndash075) 91 75
PET parietal 086 (081ndash091) 111 059 (047ndash069) 91 68
PET temporal 086 (080ndash092) 108 066 (054ndash076) 89 77
PET cingulate 086 (080ndash091) 120 056 (044ndash066) 89 66
CSF Ab42 085 (079ndash090) 173 060 (050ndash070) 94 66
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative AUC5 area under the receiver operating characteristic curve CI5 confidence intervalMCI-AD 5 patients with mild cognitive impairment who developed Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauNo significant differences between the AUCs were found (p 5 017ndash093)a Established with mixture modeling analysis (described in Methods)bDerived from the ROC analysisc Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffs
1246 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
false low specificity (and a false low AUC) due to theincorrect clinical diagnosisshort follow-up We there-fore compared patients with MCI-AD and controlsgiven the relatively short follow-up data in the ADNIand BioFINDER populations for a more robust com-parison of Ab biomarkers
The novelties of the present study compared withthe previous study19 include a comparison betweenMCI-AD and controls a more detailed analysis ofregional Ab PET data analyses of ratios of CSFAb42Ab40 Ab42t-tau and Ab42p-tau a compar-ison of 2 different ELISAs for CSF Ab42 and evalu-ation of the combination of PET and CSF biomarkers
The similar results we found for CSF biomarkersand amyloid PET suggest that other factors than theirdiagnostic accuracy may be considered when decidingwhich biomarker to use CSF analysis has the advan-tages that it may easily incorporate other biomarkersto improve the differential diagnosis (eg leukocytesalbumin ratio neurofilament a-synuclein) requiresless advanced instruments than PET and is in somecountries more available in clinical practice AmyloidPET on the other hand is less invasive and has ahigher reliability in longitudinal examinations andbetween centers With appropriate standardizedprocedures203940 CSF analysis and amyloid PET per-form equally well and either method can be used in theclinical workup of AD for increased diagnosticaccuracy
AUTHOR AFFILIATIONSFrom the Clinical Memory Research Unit Department of Clinical Scien-
ces (SP LM OH) and the Department of Computational Biology
and Biological Physics (MO) Lund University the Department of
Neurology (SP) and the Memory Clinic (LM OH) Skaringne University
Hospital the Clinical Neurochemistry Laboratory (HZ NM KB)
Institute of Neuroscience and Physiology the Sahlgrenska Academy at
the University of Gothenburg Moumllndal Sweden the UCL Institute of
Neurology (HZ) London UK the Department of Radiology and Bio-
medical Imaging (NM) University of California San Francisco the
Department of Veterans Affairs Medical Center (NM) Center for Imag-
ing of Neurodegenerative Diseases San Francisco CA and the Depart-
ment of Internal Medicine (PJ) Sahlgrenska Academy University of
Gothenburg Sweden
AUTHOR CONTRIBUTIONSSebastian Palmqvist draftingrevising the manuscript study concept or
design analysis or interpretation of data accepts responsibility for con-
duct of research and final approval statistical analysis study supervision
Henrik Zetterberg draftingrevising the manuscript analysis or interpre-
tation of data accepts responsibility for conduct of research and
final approval acquisition of data obtaining funding Niklas Mattsson
draftingrevising the manuscript analysis or interpretation of data
accepts responsibility for conduct of research and final approval Per
Johansson draftingrevising the manuscript analysis or interpretation
of data accepts responsibility for conduct of research and final approval
acquisition of data study supervision Lennart Minthon draftingrevising
the manuscript accepts responsibility for conduct of research and final
approval obtaining funding Kaj Blennow draftingrevising the manu-
script study concept or design accepts responsibility for conduct of
research and final approval study supervision obtaining funding Mattias
Ohlsson draftingrevising the manuscript analysis or interpretation of
data accepts responsibility for conduct of research and final approval sta-
tistical analysis Oskar Hansson draftingrevising the manuscript study
concept or design analysis or interpretation of data accepts responsibility
for conduct of research and final approval contribution of vital reagents
toolspatients acquisition of data study supervision obtaining funding
ACKNOWLEDGMENTThe authors thank the collaborators of this study and the entire BioFINDER
study group (wwwbiofinderse) including Ulf Andreasson Susanna Vestberg
for classifying the patients with MCI-AD into MCI subgroups Erik Stomrud
and Katarina Naumlgga for clinical evaluations of cognitively healthy individuals
Christer Nilsson for clinical evaluations of patients with mild cognitive symptoms
Per Wollmer and Douglas Haumlgerstroumlm for help with 18F-flutemetamol PET
imaging Olof Lindberg for analyzing the hippocampal volumes and Karin
Nilsson Rosita Nordkvist Ida Friberg Malin Ottheacuten and Johanna Fredlund
for organizing inclusions and assessments
STUDY FUNDINGWork in the authorsrsquo laboratory was supported by the European Research
Council the Swedish Research Council the Strategic Research Area Multi-
Park (Multidisciplinary Research in Parkinsonrsquos disease) at Lund University
the Crafoord Foundation the Swedish Brain Foundation the Skaringne Uni-
versity Hospital Foundation the Swedish Alzheimer Association Stiftelsen
foumlr Gamla Tjaumlnarinnor and the Swedish federal government under the ALF
agreement The Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) data
collection and sharing for this project was funded by ADNI (NIH grant
U01 AG024904) and DOD ADNI (Department of Defense award num-
ber W81XWH-12-2-0012) ADNI is funded by the National Institute on
Aging the National Institute of Biomedical Imaging and Bioengineering
and through contributions from the following Alzheimerrsquos Association
Alzheimerrsquos Drug Discovery Foundation Araclon Biotech BioClinica
Inc Biogen Idec Inc Bristol-Myers Squibb Company Eisai Inc Elan
Pharmaceuticals Inc Eli Lilly and Company EuroImmun F Hoffmann-
La Roche Ltd and its affiliated company Genentech Inc Fujirebio GE
Healthcare IXICO Ltd Janssen Alzheimer Immunotherapy Research amp
Development LLC Johnson amp Johnson Pharmaceutical Research amp
Development LLC Medpace Inc Merck amp Co Inc Meso Scale Diag-
nostics LLC NeuroRx Research Neurotrack Technologies Novartis Phar-
maceuticals Corporation Pfizer Inc Piramal Imaging Servier Synarc Inc
and Takeda Pharmaceutical Company The Canadian Institutes of Health
Research is providing funds to support ADNI clinical sites in Canada
Private sector contributions are facilitated by the Foundation for the
NIH (wwwfnihorg) The grantee organization is the Northern California
Institute for Research and Education and the study is coordinated by the
Alzheimerrsquos Disease Cooperative Study at the University of California San
Diego ADNI data are disseminated by the Laboratory for NeuroImaging at
the University of Southern California
DISCLOSURES Palmqvist H Zetterberg N Mattsson P Johansson and L Minthon
report no disclosures relevant to the manuscript K Blennow has served
at advisory boards for Koyowa Kirin Pharma Eli Lilly Pfizer and Roche
Doses of 18F-flutemetamol were sponsored by GE Healthcare M Ohlsson
and O Hansson report no disclosures relevant to the manuscript Go to
Neurologyorg for full disclosures
Received February 4 2015 Accepted in final form June 3 2015
REFERENCES1 Albert MS DeKosky ST Dickson D et al The diagnosis of
mild cognitive impairment due to Alzheimerrsquos disease recom-
mendations from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for
Alzheimerrsquos disease Alzheimers Dement 20117270ndash279
2 Sperling RA Aisen PS Beckett LA et al Toward defining
the preclinical stages of Alzheimerrsquos disease recommenda-
tions from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for Alz-
heimerrsquos disease Alzheimers Dement 20117280ndash292
Neurology 85 October 6 2015 1247
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
BioFINDER (t-tau 088 95 CI 082ndash093 andp-tau 087 95 CI 080ndash091 data not shown inthe tables) However no significant differences couldbe tested since the results were derived from 2 differ-ent cohorts
When combining CSF Ab42p-tau and compos-ite PET SUVR in ADNI the AUC was 087(95 CI 082ndash093) This did not differ significantlyfrom using the variables separately (AUC difference000ndash001 p 5 040ndash053)
DISCUSSION The main finding of this study wasthat the diagnostic accuracy of CSF and Ab PETbiomarkers to identify MCI-AD was similar whenusing 18F-flutemetamol amyloid PET and severaldifferent CSF biomarkers Specifically the best CSFmeasures (CSF Ab42t-tau and Ab42p-tau ratios)had similar diagnostic accuracies as the best PETmeasures (composite and cingulate SUVRs table2) We also found that no regional PET biomarkerwas better than the neocortical composite PET SUVR(table 2) For CSF biomarkers the CSF Ab42t-tauor Ab42p-tau ratios had significantly higherdiagnostic accuracy compared to using CSF Abbiomarkers alone When using unbiased cutoffs
CSF Ab42t-tau had the highest sensitivity andspecificity of all CSF and PET biomarkers (table 3)Finally the combination of the best CSF and PETbiomarkers did not provide any added diagnosticvalue compared to using either modality separatelyThe overall results were replicated in an independentcohort (ADNI)
Although we found that CSF Ab42IT and Ab42EIAb40EI performed similarly to the best SUVRs of18F-flutemetamol PET in terms of AUCs (table 2)these CSF biomarkers generally had lower specificitiesthan the PET biomarkers when using unbiased cut-offs (table 3) The addition of t-tau or p-tau to Ab42(as ratios) significantly increased the diagnostic accur-acy of CSF biomarkers (table 2) This supports thecommon usage of CSF Ab42 in combination with t-tau or p-tau in clinical practice and is in agreementwith previous studies31ndash34
The diagnostic accuracy of CSF Ab42 was lowerfor EUROIMMUN compared with INNOTEST(table 2 and figure e-3) This was partly overcomeby using the Ab4240 ratio which did not improvethe accuracy of Ab42 INNOTEST (table 2) Thisfinding has not been shown previously and needs tobe replicated in future studies since the causes are
Table 3 ROC analysis in BioFINDER of MCI-AD and healthy controls based on unbiased cutoffs
Variable (in order of Youden index) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b Adjusted OR (95 CI)d
CSF Ab42ITt-tauEI 125 080 (068ndash087) 97 83 62 (49ndash796)
CSF Ab42ITp-tauIT 724 079 (066ndash087) 94 85 34 (44ndash263)
CSF Ab42EIp-tauIT 567 074 (056ndash084) 88 86 20 (33ndash126)
PET prefrontal 148 074 (056ndash084) 88 86 29 (36ndash239)
CSF Ab42EIt-tauEI 096 073 (059ndash083) 91 82 44 (43ndash447)
PET posterior cingulateprecuneus 162 073 (056ndash083) 88 84 24 (33ndash182)
PET anterior cingulate 162 072 (056ndash081) 91 80 22 (29ndash170)
PET composite 151 072 (055ndash083) 85 87 29 (36ndash239)
PET lateral temporal 158 071 (054ndash083) 85 86 30 (36ndash250)
PET parietal 143 070 (053ndash081) 85 84 25 (32ndash190)
PET sensorimotor 153 069 (051ndash081) 79 89 30 (38ndash231)
CSF Ab42EIAb40EI 0092 067 (050ndash078) 88 79 14 (25ndash82)
CSF Ab42IT 500 060 (042ndash072) 85 75 61 (12ndash30)
CSF Ab42ITAb40EI 010 059 (039ndash073) 76 83 82 (17ndash40)
PET occipital 168 049 (030ndash066) 56 93 27 (28ndash263)
PET medial temporal 169 020 (007ndash038) 23 97 Nonsignificant
Abbreviations CI5 confidence interval EI5 EUROIMMUN assay IT5 INNOTEST assay MCI-AD5 patients with mild cognitive impairment who developedAlzheimer disease within 3 years OR 5 odds ratio p-tau 5 hyperphosphorylated tau ROC 5 receiving operating characteristic t-tau 5 total tauArea under the curve was calculated with ROC analysis The 95 CI and significance for differences between the AUCs were calculated using bootstraptechniques with 5000 bootstrap replicas PET values are shown in standardized uptake value ratio and CSF levels in pgmL (except for the CSF ratios)a Established with mixture modeling analysis (described in Methods)b Based on the ROC analysis (not derived from the logistic regression analysis)c Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffsdA logistic regression analysis was performed with the diagnosis (MCI-AD or control) as the dependent variable and the dichotomized CSFPET variable asa covariate to yield an OR Age sex APOE e4 allele memory function and hippocampal volume were adjusted for in the model
1244 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
unknown Few studies have compared differentELISAs for CSF Ab42 to identify MCI-AD Hertzeet al33 found that CSF Ab42 analyzed with xMAP
AlzBio3 had higher diagnostic accuracy comparedwith the Meso Scale Discovery (MSD) assay but thiswas overcome by using the Ab42Ab40 MSD ratio
Figure 1 Sensitivities specificities and Youden indices of all possible cutoff points
(A B) The cutoffs have been transformed to z scores (SD) for easier comparison between variables Near optimal Youden indicesare found for a relatively wide range of cutoffs (1 SD) for all variables except for CSF Ab42Ab40EI which have a slightlynarrower span The different cutoffs within this near optimal range thus only change the relationship between the sensitivity andspecificity but not the overall classification accuracy Thiswide range of near optimal cutoffs suggests that the cutoffs are likely toproduce high diagnostic accuracies in other populations AUC 5 area under the receiver operating characteristic curve
Neurology 85 October 6 2015 1245
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
Similar to the present study they showed that Ab42t-tau was superior to Ab42 and Ab42Ab40
A possible advantage of Ab PET over CSF Ab42as an early marker for amyloid pathology is that AbPET may be able to identify early region-specificpathology However this was not supported by ourstudy since the global Ab uptake performed similarcompared to the best regional SUVRs (tables 2 and4) Only one previous study has examined this andfound similar results19 The similar AUCs for the bestPET regions support the notion that the Ab deposi-tion is uniformly distributed in the neocortical asso-ciation areas already at the MCI stage of AD35
We used classification cutoffs established withmixture modeling which is a robust way of determin-ing unbiased thresholds and used in several stud-ies203637 Even so the cutoffs (table 3) should notbe considered generalizable but study-specific forcomparative purposes However figure 1 shows thatalthough a cutoff is not optimized for the currentpopulation it can still provide good diagnostic accur-acy because of the broad range of high YI The sta-bility of cutoffs between populations is also supportedby a previous cross-validation study on CSF Ab42and amyloid PET cutoffs20 However even thoughthe classification accuracy stays the same a change incutoff will of course result in a higher sensitivitylowerspecificity or lower sensitivityhigher specificity andthis must be taken into consideration depending onthe clinical aim of the examination
The overall results were similar between the Bio-FINDER and ADNI cohorts In ADNI the samecomparable results between regional and compositeSUVRs as well as equal diagnostic accuracies ofCSF and PET measures were seen (table 4) This
similarity between studies is especially interestingconsidering the use of different PET tracers and dif-ferent CSF assays In both cohorts numerically high-er AUCs were seen for the CSF Ab42t-tau or p-tauratios compared with just Ab42 but in ADNI theincrease was not significant (tables 2 and 4) Thiscould be attributed to the poorer AUCs of t-tauand p-tau in ADNI (081 and 082 AlzBio3)compared with BioFINDER (088 and 087EUROIMMUN and INNOTEST) A similar differ-ence between INNOTEST and AlzBio3 regardingAb42tau ratios was also found in a previous study38
It was notable that the AUCs of all brain regions weresimilar in ADNI (AUC range 001 table 4) in con-trast to BioFINDER (AUC range 017 table 2) Thereason for this could be that in ADNI the regionswere coarser and not able to detect differencesbetween eg the medial and lateral temporal lobe
The diagnostic accuracy of Ab PET and CSF bio-markers to detect incipient AD has only been comparedhead-to-head in one previous study which partly usedthe same ADNI data used for replication in the presentstudy19 In that study the accuracy between stable MCI(2- to 3-year follow-up without progression) and MCI-AD was compared In the present study we insteadcompared healthy elderly and patients with MCI-ADwhich resulted in higher AUCs (on average about 005in the ADNI study compare reference 19 and table 4)The rationale behind comparing controls andMCI-ADis that5ndash10 years of follow-up is required before onecan say that a patient withMCI is truly stable36 Amongpatients with stable MCI with a short follow-up timethere are several cases with early-stage AD These pa-tients with stable MCI will in most cases be correctlyidentified as MCI-AD by the biomarkers but result in
Table 4 Classification of MCI-AD and healthy controls based on ROC analyses in ADNI
Variable (in order of AUC value) AUC (95 CI) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b
PET frontal 087 (082ndash091) 113 063 (051ndash073) 89 74
CSF Ab42t-tau 087 (080ndash092) 171 065 (053ndash076) 80 86
CSF Ab42p-tau 087 (080ndash092) 383 065 (052ndash075) 84 81
PET composite 086 (081ndash091) 113 066 (054ndash075) 91 75
PET parietal 086 (081ndash091) 111 059 (047ndash069) 91 68
PET temporal 086 (080ndash092) 108 066 (054ndash076) 89 77
PET cingulate 086 (080ndash091) 120 056 (044ndash066) 89 66
CSF Ab42 085 (079ndash090) 173 060 (050ndash070) 94 66
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative AUC5 area under the receiver operating characteristic curve CI5 confidence intervalMCI-AD 5 patients with mild cognitive impairment who developed Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauNo significant differences between the AUCs were found (p 5 017ndash093)a Established with mixture modeling analysis (described in Methods)bDerived from the ROC analysisc Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffs
1246 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
false low specificity (and a false low AUC) due to theincorrect clinical diagnosisshort follow-up We there-fore compared patients with MCI-AD and controlsgiven the relatively short follow-up data in the ADNIand BioFINDER populations for a more robust com-parison of Ab biomarkers
The novelties of the present study compared withthe previous study19 include a comparison betweenMCI-AD and controls a more detailed analysis ofregional Ab PET data analyses of ratios of CSFAb42Ab40 Ab42t-tau and Ab42p-tau a compar-ison of 2 different ELISAs for CSF Ab42 and evalu-ation of the combination of PET and CSF biomarkers
The similar results we found for CSF biomarkersand amyloid PET suggest that other factors than theirdiagnostic accuracy may be considered when decidingwhich biomarker to use CSF analysis has the advan-tages that it may easily incorporate other biomarkersto improve the differential diagnosis (eg leukocytesalbumin ratio neurofilament a-synuclein) requiresless advanced instruments than PET and is in somecountries more available in clinical practice AmyloidPET on the other hand is less invasive and has ahigher reliability in longitudinal examinations andbetween centers With appropriate standardizedprocedures203940 CSF analysis and amyloid PET per-form equally well and either method can be used in theclinical workup of AD for increased diagnosticaccuracy
AUTHOR AFFILIATIONSFrom the Clinical Memory Research Unit Department of Clinical Scien-
ces (SP LM OH) and the Department of Computational Biology
and Biological Physics (MO) Lund University the Department of
Neurology (SP) and the Memory Clinic (LM OH) Skaringne University
Hospital the Clinical Neurochemistry Laboratory (HZ NM KB)
Institute of Neuroscience and Physiology the Sahlgrenska Academy at
the University of Gothenburg Moumllndal Sweden the UCL Institute of
Neurology (HZ) London UK the Department of Radiology and Bio-
medical Imaging (NM) University of California San Francisco the
Department of Veterans Affairs Medical Center (NM) Center for Imag-
ing of Neurodegenerative Diseases San Francisco CA and the Depart-
ment of Internal Medicine (PJ) Sahlgrenska Academy University of
Gothenburg Sweden
AUTHOR CONTRIBUTIONSSebastian Palmqvist draftingrevising the manuscript study concept or
design analysis or interpretation of data accepts responsibility for con-
duct of research and final approval statistical analysis study supervision
Henrik Zetterberg draftingrevising the manuscript analysis or interpre-
tation of data accepts responsibility for conduct of research and
final approval acquisition of data obtaining funding Niklas Mattsson
draftingrevising the manuscript analysis or interpretation of data
accepts responsibility for conduct of research and final approval Per
Johansson draftingrevising the manuscript analysis or interpretation
of data accepts responsibility for conduct of research and final approval
acquisition of data study supervision Lennart Minthon draftingrevising
the manuscript accepts responsibility for conduct of research and final
approval obtaining funding Kaj Blennow draftingrevising the manu-
script study concept or design accepts responsibility for conduct of
research and final approval study supervision obtaining funding Mattias
Ohlsson draftingrevising the manuscript analysis or interpretation of
data accepts responsibility for conduct of research and final approval sta-
tistical analysis Oskar Hansson draftingrevising the manuscript study
concept or design analysis or interpretation of data accepts responsibility
for conduct of research and final approval contribution of vital reagents
toolspatients acquisition of data study supervision obtaining funding
ACKNOWLEDGMENTThe authors thank the collaborators of this study and the entire BioFINDER
study group (wwwbiofinderse) including Ulf Andreasson Susanna Vestberg
for classifying the patients with MCI-AD into MCI subgroups Erik Stomrud
and Katarina Naumlgga for clinical evaluations of cognitively healthy individuals
Christer Nilsson for clinical evaluations of patients with mild cognitive symptoms
Per Wollmer and Douglas Haumlgerstroumlm for help with 18F-flutemetamol PET
imaging Olof Lindberg for analyzing the hippocampal volumes and Karin
Nilsson Rosita Nordkvist Ida Friberg Malin Ottheacuten and Johanna Fredlund
for organizing inclusions and assessments
STUDY FUNDINGWork in the authorsrsquo laboratory was supported by the European Research
Council the Swedish Research Council the Strategic Research Area Multi-
Park (Multidisciplinary Research in Parkinsonrsquos disease) at Lund University
the Crafoord Foundation the Swedish Brain Foundation the Skaringne Uni-
versity Hospital Foundation the Swedish Alzheimer Association Stiftelsen
foumlr Gamla Tjaumlnarinnor and the Swedish federal government under the ALF
agreement The Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) data
collection and sharing for this project was funded by ADNI (NIH grant
U01 AG024904) and DOD ADNI (Department of Defense award num-
ber W81XWH-12-2-0012) ADNI is funded by the National Institute on
Aging the National Institute of Biomedical Imaging and Bioengineering
and through contributions from the following Alzheimerrsquos Association
Alzheimerrsquos Drug Discovery Foundation Araclon Biotech BioClinica
Inc Biogen Idec Inc Bristol-Myers Squibb Company Eisai Inc Elan
Pharmaceuticals Inc Eli Lilly and Company EuroImmun F Hoffmann-
La Roche Ltd and its affiliated company Genentech Inc Fujirebio GE
Healthcare IXICO Ltd Janssen Alzheimer Immunotherapy Research amp
Development LLC Johnson amp Johnson Pharmaceutical Research amp
Development LLC Medpace Inc Merck amp Co Inc Meso Scale Diag-
nostics LLC NeuroRx Research Neurotrack Technologies Novartis Phar-
maceuticals Corporation Pfizer Inc Piramal Imaging Servier Synarc Inc
and Takeda Pharmaceutical Company The Canadian Institutes of Health
Research is providing funds to support ADNI clinical sites in Canada
Private sector contributions are facilitated by the Foundation for the
NIH (wwwfnihorg) The grantee organization is the Northern California
Institute for Research and Education and the study is coordinated by the
Alzheimerrsquos Disease Cooperative Study at the University of California San
Diego ADNI data are disseminated by the Laboratory for NeuroImaging at
the University of Southern California
DISCLOSURES Palmqvist H Zetterberg N Mattsson P Johansson and L Minthon
report no disclosures relevant to the manuscript K Blennow has served
at advisory boards for Koyowa Kirin Pharma Eli Lilly Pfizer and Roche
Doses of 18F-flutemetamol were sponsored by GE Healthcare M Ohlsson
and O Hansson report no disclosures relevant to the manuscript Go to
Neurologyorg for full disclosures
Received February 4 2015 Accepted in final form June 3 2015
REFERENCES1 Albert MS DeKosky ST Dickson D et al The diagnosis of
mild cognitive impairment due to Alzheimerrsquos disease recom-
mendations from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for
Alzheimerrsquos disease Alzheimers Dement 20117270ndash279
2 Sperling RA Aisen PS Beckett LA et al Toward defining
the preclinical stages of Alzheimerrsquos disease recommenda-
tions from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for Alz-
heimerrsquos disease Alzheimers Dement 20117280ndash292
Neurology 85 October 6 2015 1247
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
unknown Few studies have compared differentELISAs for CSF Ab42 to identify MCI-AD Hertzeet al33 found that CSF Ab42 analyzed with xMAP
AlzBio3 had higher diagnostic accuracy comparedwith the Meso Scale Discovery (MSD) assay but thiswas overcome by using the Ab42Ab40 MSD ratio
Figure 1 Sensitivities specificities and Youden indices of all possible cutoff points
(A B) The cutoffs have been transformed to z scores (SD) for easier comparison between variables Near optimal Youden indicesare found for a relatively wide range of cutoffs (1 SD) for all variables except for CSF Ab42Ab40EI which have a slightlynarrower span The different cutoffs within this near optimal range thus only change the relationship between the sensitivity andspecificity but not the overall classification accuracy Thiswide range of near optimal cutoffs suggests that the cutoffs are likely toproduce high diagnostic accuracies in other populations AUC 5 area under the receiver operating characteristic curve
Neurology 85 October 6 2015 1245
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
Similar to the present study they showed that Ab42t-tau was superior to Ab42 and Ab42Ab40
A possible advantage of Ab PET over CSF Ab42as an early marker for amyloid pathology is that AbPET may be able to identify early region-specificpathology However this was not supported by ourstudy since the global Ab uptake performed similarcompared to the best regional SUVRs (tables 2 and4) Only one previous study has examined this andfound similar results19 The similar AUCs for the bestPET regions support the notion that the Ab deposi-tion is uniformly distributed in the neocortical asso-ciation areas already at the MCI stage of AD35
We used classification cutoffs established withmixture modeling which is a robust way of determin-ing unbiased thresholds and used in several stud-ies203637 Even so the cutoffs (table 3) should notbe considered generalizable but study-specific forcomparative purposes However figure 1 shows thatalthough a cutoff is not optimized for the currentpopulation it can still provide good diagnostic accur-acy because of the broad range of high YI The sta-bility of cutoffs between populations is also supportedby a previous cross-validation study on CSF Ab42and amyloid PET cutoffs20 However even thoughthe classification accuracy stays the same a change incutoff will of course result in a higher sensitivitylowerspecificity or lower sensitivityhigher specificity andthis must be taken into consideration depending onthe clinical aim of the examination
The overall results were similar between the Bio-FINDER and ADNI cohorts In ADNI the samecomparable results between regional and compositeSUVRs as well as equal diagnostic accuracies ofCSF and PET measures were seen (table 4) This
similarity between studies is especially interestingconsidering the use of different PET tracers and dif-ferent CSF assays In both cohorts numerically high-er AUCs were seen for the CSF Ab42t-tau or p-tauratios compared with just Ab42 but in ADNI theincrease was not significant (tables 2 and 4) Thiscould be attributed to the poorer AUCs of t-tauand p-tau in ADNI (081 and 082 AlzBio3)compared with BioFINDER (088 and 087EUROIMMUN and INNOTEST) A similar differ-ence between INNOTEST and AlzBio3 regardingAb42tau ratios was also found in a previous study38
It was notable that the AUCs of all brain regions weresimilar in ADNI (AUC range 001 table 4) in con-trast to BioFINDER (AUC range 017 table 2) Thereason for this could be that in ADNI the regionswere coarser and not able to detect differencesbetween eg the medial and lateral temporal lobe
The diagnostic accuracy of Ab PET and CSF bio-markers to detect incipient AD has only been comparedhead-to-head in one previous study which partly usedthe same ADNI data used for replication in the presentstudy19 In that study the accuracy between stable MCI(2- to 3-year follow-up without progression) and MCI-AD was compared In the present study we insteadcompared healthy elderly and patients with MCI-ADwhich resulted in higher AUCs (on average about 005in the ADNI study compare reference 19 and table 4)The rationale behind comparing controls andMCI-ADis that5ndash10 years of follow-up is required before onecan say that a patient withMCI is truly stable36 Amongpatients with stable MCI with a short follow-up timethere are several cases with early-stage AD These pa-tients with stable MCI will in most cases be correctlyidentified as MCI-AD by the biomarkers but result in
Table 4 Classification of MCI-AD and healthy controls based on ROC analyses in ADNI
Variable (in order of AUC value) AUC (95 CI) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b
PET frontal 087 (082ndash091) 113 063 (051ndash073) 89 74
CSF Ab42t-tau 087 (080ndash092) 171 065 (053ndash076) 80 86
CSF Ab42p-tau 087 (080ndash092) 383 065 (052ndash075) 84 81
PET composite 086 (081ndash091) 113 066 (054ndash075) 91 75
PET parietal 086 (081ndash091) 111 059 (047ndash069) 91 68
PET temporal 086 (080ndash092) 108 066 (054ndash076) 89 77
PET cingulate 086 (080ndash091) 120 056 (044ndash066) 89 66
CSF Ab42 085 (079ndash090) 173 060 (050ndash070) 94 66
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative AUC5 area under the receiver operating characteristic curve CI5 confidence intervalMCI-AD 5 patients with mild cognitive impairment who developed Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauNo significant differences between the AUCs were found (p 5 017ndash093)a Established with mixture modeling analysis (described in Methods)bDerived from the ROC analysisc Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffs
1246 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
false low specificity (and a false low AUC) due to theincorrect clinical diagnosisshort follow-up We there-fore compared patients with MCI-AD and controlsgiven the relatively short follow-up data in the ADNIand BioFINDER populations for a more robust com-parison of Ab biomarkers
The novelties of the present study compared withthe previous study19 include a comparison betweenMCI-AD and controls a more detailed analysis ofregional Ab PET data analyses of ratios of CSFAb42Ab40 Ab42t-tau and Ab42p-tau a compar-ison of 2 different ELISAs for CSF Ab42 and evalu-ation of the combination of PET and CSF biomarkers
The similar results we found for CSF biomarkersand amyloid PET suggest that other factors than theirdiagnostic accuracy may be considered when decidingwhich biomarker to use CSF analysis has the advan-tages that it may easily incorporate other biomarkersto improve the differential diagnosis (eg leukocytesalbumin ratio neurofilament a-synuclein) requiresless advanced instruments than PET and is in somecountries more available in clinical practice AmyloidPET on the other hand is less invasive and has ahigher reliability in longitudinal examinations andbetween centers With appropriate standardizedprocedures203940 CSF analysis and amyloid PET per-form equally well and either method can be used in theclinical workup of AD for increased diagnosticaccuracy
AUTHOR AFFILIATIONSFrom the Clinical Memory Research Unit Department of Clinical Scien-
ces (SP LM OH) and the Department of Computational Biology
and Biological Physics (MO) Lund University the Department of
Neurology (SP) and the Memory Clinic (LM OH) Skaringne University
Hospital the Clinical Neurochemistry Laboratory (HZ NM KB)
Institute of Neuroscience and Physiology the Sahlgrenska Academy at
the University of Gothenburg Moumllndal Sweden the UCL Institute of
Neurology (HZ) London UK the Department of Radiology and Bio-
medical Imaging (NM) University of California San Francisco the
Department of Veterans Affairs Medical Center (NM) Center for Imag-
ing of Neurodegenerative Diseases San Francisco CA and the Depart-
ment of Internal Medicine (PJ) Sahlgrenska Academy University of
Gothenburg Sweden
AUTHOR CONTRIBUTIONSSebastian Palmqvist draftingrevising the manuscript study concept or
design analysis or interpretation of data accepts responsibility for con-
duct of research and final approval statistical analysis study supervision
Henrik Zetterberg draftingrevising the manuscript analysis or interpre-
tation of data accepts responsibility for conduct of research and
final approval acquisition of data obtaining funding Niklas Mattsson
draftingrevising the manuscript analysis or interpretation of data
accepts responsibility for conduct of research and final approval Per
Johansson draftingrevising the manuscript analysis or interpretation
of data accepts responsibility for conduct of research and final approval
acquisition of data study supervision Lennart Minthon draftingrevising
the manuscript accepts responsibility for conduct of research and final
approval obtaining funding Kaj Blennow draftingrevising the manu-
script study concept or design accepts responsibility for conduct of
research and final approval study supervision obtaining funding Mattias
Ohlsson draftingrevising the manuscript analysis or interpretation of
data accepts responsibility for conduct of research and final approval sta-
tistical analysis Oskar Hansson draftingrevising the manuscript study
concept or design analysis or interpretation of data accepts responsibility
for conduct of research and final approval contribution of vital reagents
toolspatients acquisition of data study supervision obtaining funding
ACKNOWLEDGMENTThe authors thank the collaborators of this study and the entire BioFINDER
study group (wwwbiofinderse) including Ulf Andreasson Susanna Vestberg
for classifying the patients with MCI-AD into MCI subgroups Erik Stomrud
and Katarina Naumlgga for clinical evaluations of cognitively healthy individuals
Christer Nilsson for clinical evaluations of patients with mild cognitive symptoms
Per Wollmer and Douglas Haumlgerstroumlm for help with 18F-flutemetamol PET
imaging Olof Lindberg for analyzing the hippocampal volumes and Karin
Nilsson Rosita Nordkvist Ida Friberg Malin Ottheacuten and Johanna Fredlund
for organizing inclusions and assessments
STUDY FUNDINGWork in the authorsrsquo laboratory was supported by the European Research
Council the Swedish Research Council the Strategic Research Area Multi-
Park (Multidisciplinary Research in Parkinsonrsquos disease) at Lund University
the Crafoord Foundation the Swedish Brain Foundation the Skaringne Uni-
versity Hospital Foundation the Swedish Alzheimer Association Stiftelsen
foumlr Gamla Tjaumlnarinnor and the Swedish federal government under the ALF
agreement The Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) data
collection and sharing for this project was funded by ADNI (NIH grant
U01 AG024904) and DOD ADNI (Department of Defense award num-
ber W81XWH-12-2-0012) ADNI is funded by the National Institute on
Aging the National Institute of Biomedical Imaging and Bioengineering
and through contributions from the following Alzheimerrsquos Association
Alzheimerrsquos Drug Discovery Foundation Araclon Biotech BioClinica
Inc Biogen Idec Inc Bristol-Myers Squibb Company Eisai Inc Elan
Pharmaceuticals Inc Eli Lilly and Company EuroImmun F Hoffmann-
La Roche Ltd and its affiliated company Genentech Inc Fujirebio GE
Healthcare IXICO Ltd Janssen Alzheimer Immunotherapy Research amp
Development LLC Johnson amp Johnson Pharmaceutical Research amp
Development LLC Medpace Inc Merck amp Co Inc Meso Scale Diag-
nostics LLC NeuroRx Research Neurotrack Technologies Novartis Phar-
maceuticals Corporation Pfizer Inc Piramal Imaging Servier Synarc Inc
and Takeda Pharmaceutical Company The Canadian Institutes of Health
Research is providing funds to support ADNI clinical sites in Canada
Private sector contributions are facilitated by the Foundation for the
NIH (wwwfnihorg) The grantee organization is the Northern California
Institute for Research and Education and the study is coordinated by the
Alzheimerrsquos Disease Cooperative Study at the University of California San
Diego ADNI data are disseminated by the Laboratory for NeuroImaging at
the University of Southern California
DISCLOSURES Palmqvist H Zetterberg N Mattsson P Johansson and L Minthon
report no disclosures relevant to the manuscript K Blennow has served
at advisory boards for Koyowa Kirin Pharma Eli Lilly Pfizer and Roche
Doses of 18F-flutemetamol were sponsored by GE Healthcare M Ohlsson
and O Hansson report no disclosures relevant to the manuscript Go to
Neurologyorg for full disclosures
Received February 4 2015 Accepted in final form June 3 2015
REFERENCES1 Albert MS DeKosky ST Dickson D et al The diagnosis of
mild cognitive impairment due to Alzheimerrsquos disease recom-
mendations from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for
Alzheimerrsquos disease Alzheimers Dement 20117270ndash279
2 Sperling RA Aisen PS Beckett LA et al Toward defining
the preclinical stages of Alzheimerrsquos disease recommenda-
tions from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for Alz-
heimerrsquos disease Alzheimers Dement 20117280ndash292
Neurology 85 October 6 2015 1247
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
Similar to the present study they showed that Ab42t-tau was superior to Ab42 and Ab42Ab40
A possible advantage of Ab PET over CSF Ab42as an early marker for amyloid pathology is that AbPET may be able to identify early region-specificpathology However this was not supported by ourstudy since the global Ab uptake performed similarcompared to the best regional SUVRs (tables 2 and4) Only one previous study has examined this andfound similar results19 The similar AUCs for the bestPET regions support the notion that the Ab deposi-tion is uniformly distributed in the neocortical asso-ciation areas already at the MCI stage of AD35
We used classification cutoffs established withmixture modeling which is a robust way of determin-ing unbiased thresholds and used in several stud-ies203637 Even so the cutoffs (table 3) should notbe considered generalizable but study-specific forcomparative purposes However figure 1 shows thatalthough a cutoff is not optimized for the currentpopulation it can still provide good diagnostic accur-acy because of the broad range of high YI The sta-bility of cutoffs between populations is also supportedby a previous cross-validation study on CSF Ab42and amyloid PET cutoffs20 However even thoughthe classification accuracy stays the same a change incutoff will of course result in a higher sensitivitylowerspecificity or lower sensitivityhigher specificity andthis must be taken into consideration depending onthe clinical aim of the examination
The overall results were similar between the Bio-FINDER and ADNI cohorts In ADNI the samecomparable results between regional and compositeSUVRs as well as equal diagnostic accuracies ofCSF and PET measures were seen (table 4) This
similarity between studies is especially interestingconsidering the use of different PET tracers and dif-ferent CSF assays In both cohorts numerically high-er AUCs were seen for the CSF Ab42t-tau or p-tauratios compared with just Ab42 but in ADNI theincrease was not significant (tables 2 and 4) Thiscould be attributed to the poorer AUCs of t-tauand p-tau in ADNI (081 and 082 AlzBio3)compared with BioFINDER (088 and 087EUROIMMUN and INNOTEST) A similar differ-ence between INNOTEST and AlzBio3 regardingAb42tau ratios was also found in a previous study38
It was notable that the AUCs of all brain regions weresimilar in ADNI (AUC range 001 table 4) in con-trast to BioFINDER (AUC range 017 table 2) Thereason for this could be that in ADNI the regionswere coarser and not able to detect differencesbetween eg the medial and lateral temporal lobe
The diagnostic accuracy of Ab PET and CSF bio-markers to detect incipient AD has only been comparedhead-to-head in one previous study which partly usedthe same ADNI data used for replication in the presentstudy19 In that study the accuracy between stable MCI(2- to 3-year follow-up without progression) and MCI-AD was compared In the present study we insteadcompared healthy elderly and patients with MCI-ADwhich resulted in higher AUCs (on average about 005in the ADNI study compare reference 19 and table 4)The rationale behind comparing controls andMCI-ADis that5ndash10 years of follow-up is required before onecan say that a patient withMCI is truly stable36 Amongpatients with stable MCI with a short follow-up timethere are several cases with early-stage AD These pa-tients with stable MCI will in most cases be correctlyidentified as MCI-AD by the biomarkers but result in
Table 4 Classification of MCI-AD and healthy controls based on ROC analyses in ADNI
Variable (in order of AUC value) AUC (95 CI) Unbiased cutoffa Youden index (95 CI)bc Sensitivity ()b Specificity ()b
PET frontal 087 (082ndash091) 113 063 (051ndash073) 89 74
CSF Ab42t-tau 087 (080ndash092) 171 065 (053ndash076) 80 86
CSF Ab42p-tau 087 (080ndash092) 383 065 (052ndash075) 84 81
PET composite 086 (081ndash091) 113 066 (054ndash075) 91 75
PET parietal 086 (081ndash091) 111 059 (047ndash069) 91 68
PET temporal 086 (080ndash092) 108 066 (054ndash076) 89 77
PET cingulate 086 (080ndash091) 120 056 (044ndash066) 89 66
CSF Ab42 085 (079ndash090) 173 060 (050ndash070) 94 66
Abbreviations ADNI5 Alzheimerrsquos Disease Neuroimaging Initiative AUC5 area under the receiver operating characteristic curve CI5 confidence intervalMCI-AD 5 patients with mild cognitive impairment who developed Alzheimer disease within 3 years p-tau 5 hyperphosphorylated tau ROC 5 receivingoperating characteristic t-tau 5 total tauNo significant differences between the AUCs were found (p 5 017ndash093)a Established with mixture modeling analysis (described in Methods)bDerived from the ROC analysisc Youden index (sensitivity 1 specificity 2 1) is provided for an easier comparison of the combined value of the sensitivity and specificity eg thediagnostic accuracy of CSF and PET measures as dichotomized variables The value is based on the unbiased cutoffs
1246 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
false low specificity (and a false low AUC) due to theincorrect clinical diagnosisshort follow-up We there-fore compared patients with MCI-AD and controlsgiven the relatively short follow-up data in the ADNIand BioFINDER populations for a more robust com-parison of Ab biomarkers
The novelties of the present study compared withthe previous study19 include a comparison betweenMCI-AD and controls a more detailed analysis ofregional Ab PET data analyses of ratios of CSFAb42Ab40 Ab42t-tau and Ab42p-tau a compar-ison of 2 different ELISAs for CSF Ab42 and evalu-ation of the combination of PET and CSF biomarkers
The similar results we found for CSF biomarkersand amyloid PET suggest that other factors than theirdiagnostic accuracy may be considered when decidingwhich biomarker to use CSF analysis has the advan-tages that it may easily incorporate other biomarkersto improve the differential diagnosis (eg leukocytesalbumin ratio neurofilament a-synuclein) requiresless advanced instruments than PET and is in somecountries more available in clinical practice AmyloidPET on the other hand is less invasive and has ahigher reliability in longitudinal examinations andbetween centers With appropriate standardizedprocedures203940 CSF analysis and amyloid PET per-form equally well and either method can be used in theclinical workup of AD for increased diagnosticaccuracy
AUTHOR AFFILIATIONSFrom the Clinical Memory Research Unit Department of Clinical Scien-
ces (SP LM OH) and the Department of Computational Biology
and Biological Physics (MO) Lund University the Department of
Neurology (SP) and the Memory Clinic (LM OH) Skaringne University
Hospital the Clinical Neurochemistry Laboratory (HZ NM KB)
Institute of Neuroscience and Physiology the Sahlgrenska Academy at
the University of Gothenburg Moumllndal Sweden the UCL Institute of
Neurology (HZ) London UK the Department of Radiology and Bio-
medical Imaging (NM) University of California San Francisco the
Department of Veterans Affairs Medical Center (NM) Center for Imag-
ing of Neurodegenerative Diseases San Francisco CA and the Depart-
ment of Internal Medicine (PJ) Sahlgrenska Academy University of
Gothenburg Sweden
AUTHOR CONTRIBUTIONSSebastian Palmqvist draftingrevising the manuscript study concept or
design analysis or interpretation of data accepts responsibility for con-
duct of research and final approval statistical analysis study supervision
Henrik Zetterberg draftingrevising the manuscript analysis or interpre-
tation of data accepts responsibility for conduct of research and
final approval acquisition of data obtaining funding Niklas Mattsson
draftingrevising the manuscript analysis or interpretation of data
accepts responsibility for conduct of research and final approval Per
Johansson draftingrevising the manuscript analysis or interpretation
of data accepts responsibility for conduct of research and final approval
acquisition of data study supervision Lennart Minthon draftingrevising
the manuscript accepts responsibility for conduct of research and final
approval obtaining funding Kaj Blennow draftingrevising the manu-
script study concept or design accepts responsibility for conduct of
research and final approval study supervision obtaining funding Mattias
Ohlsson draftingrevising the manuscript analysis or interpretation of
data accepts responsibility for conduct of research and final approval sta-
tistical analysis Oskar Hansson draftingrevising the manuscript study
concept or design analysis or interpretation of data accepts responsibility
for conduct of research and final approval contribution of vital reagents
toolspatients acquisition of data study supervision obtaining funding
ACKNOWLEDGMENTThe authors thank the collaborators of this study and the entire BioFINDER
study group (wwwbiofinderse) including Ulf Andreasson Susanna Vestberg
for classifying the patients with MCI-AD into MCI subgroups Erik Stomrud
and Katarina Naumlgga for clinical evaluations of cognitively healthy individuals
Christer Nilsson for clinical evaluations of patients with mild cognitive symptoms
Per Wollmer and Douglas Haumlgerstroumlm for help with 18F-flutemetamol PET
imaging Olof Lindberg for analyzing the hippocampal volumes and Karin
Nilsson Rosita Nordkvist Ida Friberg Malin Ottheacuten and Johanna Fredlund
for organizing inclusions and assessments
STUDY FUNDINGWork in the authorsrsquo laboratory was supported by the European Research
Council the Swedish Research Council the Strategic Research Area Multi-
Park (Multidisciplinary Research in Parkinsonrsquos disease) at Lund University
the Crafoord Foundation the Swedish Brain Foundation the Skaringne Uni-
versity Hospital Foundation the Swedish Alzheimer Association Stiftelsen
foumlr Gamla Tjaumlnarinnor and the Swedish federal government under the ALF
agreement The Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) data
collection and sharing for this project was funded by ADNI (NIH grant
U01 AG024904) and DOD ADNI (Department of Defense award num-
ber W81XWH-12-2-0012) ADNI is funded by the National Institute on
Aging the National Institute of Biomedical Imaging and Bioengineering
and through contributions from the following Alzheimerrsquos Association
Alzheimerrsquos Drug Discovery Foundation Araclon Biotech BioClinica
Inc Biogen Idec Inc Bristol-Myers Squibb Company Eisai Inc Elan
Pharmaceuticals Inc Eli Lilly and Company EuroImmun F Hoffmann-
La Roche Ltd and its affiliated company Genentech Inc Fujirebio GE
Healthcare IXICO Ltd Janssen Alzheimer Immunotherapy Research amp
Development LLC Johnson amp Johnson Pharmaceutical Research amp
Development LLC Medpace Inc Merck amp Co Inc Meso Scale Diag-
nostics LLC NeuroRx Research Neurotrack Technologies Novartis Phar-
maceuticals Corporation Pfizer Inc Piramal Imaging Servier Synarc Inc
and Takeda Pharmaceutical Company The Canadian Institutes of Health
Research is providing funds to support ADNI clinical sites in Canada
Private sector contributions are facilitated by the Foundation for the
NIH (wwwfnihorg) The grantee organization is the Northern California
Institute for Research and Education and the study is coordinated by the
Alzheimerrsquos Disease Cooperative Study at the University of California San
Diego ADNI data are disseminated by the Laboratory for NeuroImaging at
the University of Southern California
DISCLOSURES Palmqvist H Zetterberg N Mattsson P Johansson and L Minthon
report no disclosures relevant to the manuscript K Blennow has served
at advisory boards for Koyowa Kirin Pharma Eli Lilly Pfizer and Roche
Doses of 18F-flutemetamol were sponsored by GE Healthcare M Ohlsson
and O Hansson report no disclosures relevant to the manuscript Go to
Neurologyorg for full disclosures
Received February 4 2015 Accepted in final form June 3 2015
REFERENCES1 Albert MS DeKosky ST Dickson D et al The diagnosis of
mild cognitive impairment due to Alzheimerrsquos disease recom-
mendations from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for
Alzheimerrsquos disease Alzheimers Dement 20117270ndash279
2 Sperling RA Aisen PS Beckett LA et al Toward defining
the preclinical stages of Alzheimerrsquos disease recommenda-
tions from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for Alz-
heimerrsquos disease Alzheimers Dement 20117280ndash292
Neurology 85 October 6 2015 1247
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
false low specificity (and a false low AUC) due to theincorrect clinical diagnosisshort follow-up We there-fore compared patients with MCI-AD and controlsgiven the relatively short follow-up data in the ADNIand BioFINDER populations for a more robust com-parison of Ab biomarkers
The novelties of the present study compared withthe previous study19 include a comparison betweenMCI-AD and controls a more detailed analysis ofregional Ab PET data analyses of ratios of CSFAb42Ab40 Ab42t-tau and Ab42p-tau a compar-ison of 2 different ELISAs for CSF Ab42 and evalu-ation of the combination of PET and CSF biomarkers
The similar results we found for CSF biomarkersand amyloid PET suggest that other factors than theirdiagnostic accuracy may be considered when decidingwhich biomarker to use CSF analysis has the advan-tages that it may easily incorporate other biomarkersto improve the differential diagnosis (eg leukocytesalbumin ratio neurofilament a-synuclein) requiresless advanced instruments than PET and is in somecountries more available in clinical practice AmyloidPET on the other hand is less invasive and has ahigher reliability in longitudinal examinations andbetween centers With appropriate standardizedprocedures203940 CSF analysis and amyloid PET per-form equally well and either method can be used in theclinical workup of AD for increased diagnosticaccuracy
AUTHOR AFFILIATIONSFrom the Clinical Memory Research Unit Department of Clinical Scien-
ces (SP LM OH) and the Department of Computational Biology
and Biological Physics (MO) Lund University the Department of
Neurology (SP) and the Memory Clinic (LM OH) Skaringne University
Hospital the Clinical Neurochemistry Laboratory (HZ NM KB)
Institute of Neuroscience and Physiology the Sahlgrenska Academy at
the University of Gothenburg Moumllndal Sweden the UCL Institute of
Neurology (HZ) London UK the Department of Radiology and Bio-
medical Imaging (NM) University of California San Francisco the
Department of Veterans Affairs Medical Center (NM) Center for Imag-
ing of Neurodegenerative Diseases San Francisco CA and the Depart-
ment of Internal Medicine (PJ) Sahlgrenska Academy University of
Gothenburg Sweden
AUTHOR CONTRIBUTIONSSebastian Palmqvist draftingrevising the manuscript study concept or
design analysis or interpretation of data accepts responsibility for con-
duct of research and final approval statistical analysis study supervision
Henrik Zetterberg draftingrevising the manuscript analysis or interpre-
tation of data accepts responsibility for conduct of research and
final approval acquisition of data obtaining funding Niklas Mattsson
draftingrevising the manuscript analysis or interpretation of data
accepts responsibility for conduct of research and final approval Per
Johansson draftingrevising the manuscript analysis or interpretation
of data accepts responsibility for conduct of research and final approval
acquisition of data study supervision Lennart Minthon draftingrevising
the manuscript accepts responsibility for conduct of research and final
approval obtaining funding Kaj Blennow draftingrevising the manu-
script study concept or design accepts responsibility for conduct of
research and final approval study supervision obtaining funding Mattias
Ohlsson draftingrevising the manuscript analysis or interpretation of
data accepts responsibility for conduct of research and final approval sta-
tistical analysis Oskar Hansson draftingrevising the manuscript study
concept or design analysis or interpretation of data accepts responsibility
for conduct of research and final approval contribution of vital reagents
toolspatients acquisition of data study supervision obtaining funding
ACKNOWLEDGMENTThe authors thank the collaborators of this study and the entire BioFINDER
study group (wwwbiofinderse) including Ulf Andreasson Susanna Vestberg
for classifying the patients with MCI-AD into MCI subgroups Erik Stomrud
and Katarina Naumlgga for clinical evaluations of cognitively healthy individuals
Christer Nilsson for clinical evaluations of patients with mild cognitive symptoms
Per Wollmer and Douglas Haumlgerstroumlm for help with 18F-flutemetamol PET
imaging Olof Lindberg for analyzing the hippocampal volumes and Karin
Nilsson Rosita Nordkvist Ida Friberg Malin Ottheacuten and Johanna Fredlund
for organizing inclusions and assessments
STUDY FUNDINGWork in the authorsrsquo laboratory was supported by the European Research
Council the Swedish Research Council the Strategic Research Area Multi-
Park (Multidisciplinary Research in Parkinsonrsquos disease) at Lund University
the Crafoord Foundation the Swedish Brain Foundation the Skaringne Uni-
versity Hospital Foundation the Swedish Alzheimer Association Stiftelsen
foumlr Gamla Tjaumlnarinnor and the Swedish federal government under the ALF
agreement The Alzheimerrsquos Disease Neuroimaging Initiative (ADNI) data
collection and sharing for this project was funded by ADNI (NIH grant
U01 AG024904) and DOD ADNI (Department of Defense award num-
ber W81XWH-12-2-0012) ADNI is funded by the National Institute on
Aging the National Institute of Biomedical Imaging and Bioengineering
and through contributions from the following Alzheimerrsquos Association
Alzheimerrsquos Drug Discovery Foundation Araclon Biotech BioClinica
Inc Biogen Idec Inc Bristol-Myers Squibb Company Eisai Inc Elan
Pharmaceuticals Inc Eli Lilly and Company EuroImmun F Hoffmann-
La Roche Ltd and its affiliated company Genentech Inc Fujirebio GE
Healthcare IXICO Ltd Janssen Alzheimer Immunotherapy Research amp
Development LLC Johnson amp Johnson Pharmaceutical Research amp
Development LLC Medpace Inc Merck amp Co Inc Meso Scale Diag-
nostics LLC NeuroRx Research Neurotrack Technologies Novartis Phar-
maceuticals Corporation Pfizer Inc Piramal Imaging Servier Synarc Inc
and Takeda Pharmaceutical Company The Canadian Institutes of Health
Research is providing funds to support ADNI clinical sites in Canada
Private sector contributions are facilitated by the Foundation for the
NIH (wwwfnihorg) The grantee organization is the Northern California
Institute for Research and Education and the study is coordinated by the
Alzheimerrsquos Disease Cooperative Study at the University of California San
Diego ADNI data are disseminated by the Laboratory for NeuroImaging at
the University of Southern California
DISCLOSURES Palmqvist H Zetterberg N Mattsson P Johansson and L Minthon
report no disclosures relevant to the manuscript K Blennow has served
at advisory boards for Koyowa Kirin Pharma Eli Lilly Pfizer and Roche
Doses of 18F-flutemetamol were sponsored by GE Healthcare M Ohlsson
and O Hansson report no disclosures relevant to the manuscript Go to
Neurologyorg for full disclosures
Received February 4 2015 Accepted in final form June 3 2015
REFERENCES1 Albert MS DeKosky ST Dickson D et al The diagnosis of
mild cognitive impairment due to Alzheimerrsquos disease recom-
mendations from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for
Alzheimerrsquos disease Alzheimers Dement 20117270ndash279
2 Sperling RA Aisen PS Beckett LA et al Toward defining
the preclinical stages of Alzheimerrsquos disease recommenda-
tions from the National Institute on Aging-Alzheimerrsquos
Association workgroups on diagnostic guidelines for Alz-
heimerrsquos disease Alzheimers Dement 20117280ndash292
Neurology 85 October 6 2015 1247
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
3 Blennow K Zetterberg H Haass C Finucane T Sema-
gacestatrsquos fall where next for AD therapies Nat Med
2013191214ndash1215
4 Eli Lilly and Company Progress of Mild Alzheimerrsquos
Disease in Participants on Solanezumab Versus Placebo
(EXPEDITION 3) In ClinicalTrialsgov [Internet]
Bethesda MD National Library of Medicine Cited
January 20 2015 Available at httpsclinicaltrialsgovct2
showNCT01900665 NLM Identifier NCT01900665
5 Karran E Hardy J Antiamyloid therapy for Alzheimerrsquos
disease are we on the right road N Engl J Med 2014370
377ndash378
6 Seppala TT Nerg O Koivisto AM et al CSF biomarkers
for Alzheimer disease correlate with cortical brain biopsy
findings Neurology 2012781568ndash1575
7 Wolk DA Grachev ID Buckley C et al Association
between in vivo fluorine 18-labeled flutemetamol amyloid
positron emission tomography imaging and in vivo cerebral
cortical histopathology Arch Neurol 2011681398ndash1403
8 Blennow K Hampel H Weiner M Zetterberg H Cere-
brospinal fluid and plasma biomarkers in Alzheimer dis-
ease Nat Rev Neurol 20106131ndash144
9 Buerger K Ewers M Pirttila T et al CSF phosphorylated
tau protein correlates with neocortical neurofibrillary pathol-
ogy in Alzheimerrsquos disease Brain 20061293035ndash3041
10 Degerman Gunnarsson M Lindau M Wall A et al Pitts-
burgh compound-B and Alzheimerrsquos disease biomarkers in
CSF plasma and urine an exploratory study Dement
Geriatr Cogn Disord 201029204ndash212
11 Fagan AM Mintun MA Mach RH et al Inverse relation
between in vivo amyloid imaging load and cerebrospinal
fluid Abeta42 in humans Ann Neurol 200659512ndash519
12 Fagan AM Mintun MA Shah AR et al Cerebrospinal
fluid tau and ptau(181) increase with cortical amyloid
deposition in cognitively normal individuals implications
for future clinical trials of Alzheimerrsquos disease EMBOMol
Med 20091371ndash380
13 Forsberg A Almkvist O Engler H Wall A Langstrom B
Nordberg A High PiB retention in Alzheimerrsquos disease is
an early event with complex relationship with CSF bio-
markers and functional parameters Curr Alzheimer Res
2010756ndash66
14 Grimmer T Riemenschneider M Forstl H et al Beta
amyloid in Alzheimerrsquos disease increased deposition in
brain is reflected in reduced concentration in cerebrospinal
fluid Biol Psychiatry 200965927ndash934
15 Jagust WJ Landau SM Shaw LM et al Relationships
between biomarkers in aging and dementia Neurology
2009731193ndash1199
16 Koivunen J Pirttila T Kemppainen N et al PET amyloid
ligand [11C]PIB uptake and cerebrospinal fluid beta-
amyloid in mild cognitive impairment Dement Geriatr
Cogn Disord 200826378ndash383
17 Landau SM Lu M Joshi AD et al Comparing PET
imaging and CSF measurements of Ab Ann Neurol
201374826ndash836
18 Mattsson N Insel P Donohue M et al Independent infor-
mation from cerebrospinal fluid b-amyloid and florbetapir
imaging in Alzheimerrsquos disease Brain 2015138772ndash783
19 Mattsson N Insel P Landau S et al Diagnostic accuracy
of CSF Ab42 and florbetapir PET for Alzheimerrsquos disease
Ann Clin Transl Neurol 20141534ndash543
20 Palmqvist S Zetterberg H Blennow K et al Accuracy
of brain amyloid detection in clinical practice using
cerebrospinal fluid beta-amyloid 42 a cross-validation
study against amyloid positron emission tomography JA-
MA Neurol 2014711282ndash1289
21 Tolboom N van der Flier WM Yaqub M et al
Relationship of cerebrospinal fluid markers to 11C-
PiB and 18F-FDDNP binding J Nucl Med 200950
1464ndash1470
22 Weigand SD Vemuri P Wiste HJ et al Transforming
cerebrospinal fluid Abeta42 measures into calculated
Pittsburgh Compound B units of brain Abeta amyloid
Alzheimers Dement 20117133ndash141
23 McKhann GM Knopman DS Chertkow H et al The
diagnosis of dementia due to Alzheimerrsquos disease rec-
ommendations from the National Institute on Aging-
Alzheimerrsquos Association workgroups on diagnostic
guidelines for Alzheimerrsquos disease Alzheimers Dement
20117263ndash269
24 Haftenberger M Schuit AJ Tormo MJ et al Physical
activity of subjects aged 50ndash64 years involved in the Euro-
pean Prospective Investigation into Cancer and Nutrition
(EPIC) Public Health Nutr 200251163ndash1176
25 Nelissen N Van Laere K Thurfjell L et al Phase 1 study of
the Pittsburgh compound B derivative 18F-flutemetamol in
healthy volunteers and patients with probable Alzheimer
disease J Nucl Med 2009501251ndash1259
26 Lundqvist R Lilja J Thomas BA et al Implementation
and validation of an adaptive template registration method
for 18F-flutemetamol imaging data J Nucl Med 201354
1472ndash1478
27 Rosen WG Mohs RC Davis KL A new rating scale for
Alzheimerrsquos disease Am J Psychiatry 19841411356ndash1364
28 Shaw LM Vanderstichele H Knapik-Czajka M et al
Cerebrospinal fluid biomarker signature in Alzheimerrsquos
disease neuroimaging initiative subjects Ann Neurol
200965403ndash413
29 Robin X Turck N Hainard A et al pROC an open-
source package for R and S1 to analyze and compare
ROC curves BMC Bioinformatics 20111277
30 Benaglia T Chauveau D Hunter DR Young D Mix-
tools an R package for analyzing finite mixture models
J Stat Softw 2009321ndash29
31 Fagan AM Roe CM Xiong C Mintun MA Morris JC
Holtzman DM Cerebrospinal fluid taubeta-amyloid(42)
ratio as a prediction of cognitive decline in nondemented
older adults Arch Neurol 200764343ndash349
32 Hansson O Zetterberg H Buchhave P Londos E
Blennow K Minthon L Association between CSF bio-
markers and incipient Alzheimerrsquos disease in patients with
mild cognitive impairment a follow-up study Lancet
Neurol 20065228ndash234
33 Hertze J Minthon L Zetterberg H Vanmechelen E
Blennow K Hansson O Evaluation of CSF biomarkers
as predictors of Alzheimerrsquos disease a clinical follow-up
study of 47 years J Alzheimers Dis 2010211119ndash1128
34 Tapiola T Alafuzoff I Herukka SK et al Cerebrospinal
fluid beta-amyloid 42 and tau proteins as biomarkers of
Alzheimer-type pathologic changes in the brain Arch
Neurol 200966382ndash389
35 Thal DR Attems J Ewers M Spreading of amyloid tau
and microvascular pathology in Alzheimerrsquos disease find-
ings from neuropathological and neuroimaging studies
J Alzheimers Dis 201442S421ndashS429
36 Buchhave P Minthon L Zetterberg H Wallin AK
Blennow K Hansson O Cerebrospinal fluid levels of
1248 Neurology 85 October 6 2015
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
beta-amyloid 1-42 but not of tau are fully changed
already 5 to 10 years before the onset of Alzheimer demen-
tia Arch Gen Psychiatry 20126998ndash106
37 De Meyer G Shapiro F Vanderstichele H et al
Diagnosis-independent Alzheimer disease biomarker signa-
ture in cognitively normal elderly people Arch Neurol
201067949ndash956
38 Le Bastard N Coart E Vanderstichele H Vanmechelen E
Martin JJ Engelborghs S Comparison of two analytical
platforms for the clinical qualification of Alzheimerrsquos dis-
ease biomarkers in pathologically-confirmed dementia
J Alzheimers Dis 201333117ndash131
39 Klunk WE Koeppe RA Price JC et al The Centiloid
project standardizing quantitative amyloid plaque estima-
tion by PET Alzheimers Dement 2015111ndash15e1ndash4
40 Vanderstichele H Bibl M Engelborghs S et al Standard-
ization of preanalytical aspects of cerebrospinal fluid bio-
marker testing for Alzheimerrsquos disease diagnosis a
consensus paper from the Alzheimerrsquos Biomarkers Stan-
dardization Initiative Alzheimers Dement 2012865ndash73
Upcoming Dates amp DeadlinesDonrsquot miss these important dates for the 2016 AAN Annual Meeting taking place April 15-212016 in Vancouver BC Canada Learn more at AANcomviewAM16
bull Abstract Submission Deadline October 26 2015
bull Awards Application Deadline October 28 2015
bull Registration Opens November 2015
Enjoy Big Savings on NEW 2015 AAN PracticeManagement Webinar Subscriptions
The American Academy of Neurology offers 14 cost-effective Practice Management Webinars forwhich you can attend live or listen to recordings posted online AAN members can purchase onewebinar for $149 or subscribe to the entire series for only $199mdasha big savings from the 2015nonmember price of $199 per webinar or $649 for the subscription Register today for upcomingwebinars access recorded webinars and see the rest of the 2015 schedule at AANcomviewpmw15
bull October 13 Fundamentals of EM Coding
bull November 10 Getting More Bang for Your Buck with Your EHR
Neurology 85 October 6 2015 1249
ordf 2015 American Academy of Neurology Unauthorized reproduction of this article is prohibited
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
DOI 101212WNL00000000000019912015851240-1249 Published Online before print September 9 2015Neurology
Sebastian Palmqvist Henrik Zetterberg Niklas Mattsson et al Alzheimer disease
Detailed comparison of amyloid PET and CSF biomarkers for identifying early
This information is current as of September 9 2015
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology
ServicesUpdated Information amp
httpnneurologyorgcontent85141240fullincluding high resolution figures can be found at
Supplementary Material
991DC3httpnneurologyorgcontentsuppl20151215WNL0000000000001
991DC2httpnneurologyorgcontentsuppl20150930WNL0000000000001
991DC1httpnneurologyorgcontentsuppl20150909WNL0000000000001Supplementary material can be found at
References httpnneurologyorgcontent85141240fullref-list-1
This article cites 39 articles 6 of which you can access for free at
Citations httpnneurologyorgcontent85141240fullotherarticles
This article has been cited by 9 HighWire-hosted articles
Subspecialty Collections
httpnneurologyorgcgicollectionpetPET
httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)
httpnneurologyorgcgicollectiondiagnostic_test_assessment_Diagnostic test assessment
httpnneurologyorgcgicollectioncerebrospinal_fluidCerebrospinal Fluid
httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers diseasefollowing 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
rights reserved Print ISSN 0028-3878 Online ISSN 1526-632X1951 it is now a weekly with 48 issues per year Copyright copy 2015 American Academy of Neurology All
reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology