ADNI Biomarker Core

Report
ADNI Biomarker Core
Leslie M Shaw
John Q Trojanowski
ICAD, Honolulu
July 10, 2010
•
•
•
•
•
•
•
•
•
•
•
ADNI Biomarker Core
Established and maintained ADNI Biofluid Bank at UPenn, 24/7
Established reproducible CSF Ab42, t-tau, ptau181 immunoassays using Luminex xMAP/Innogenetics system; interlab
study involved 7 ISAB and academic laboratories
Developed AD – like biomarker signature/cutpoints using CSFs from ADNI-independent autopsy-based AD cases & agematched controls and determined incidence in ADNI cohorts (Ann Neurol, 2009)
Detected abnormal Ab42 CSF concentration in 34% of cognitively normal subjects
Assessments of longitudinal changes underway, an ADNI extension study
Determined predictive performance of individual and combined CSF biomarkers for MCI → AD progression
Further validation of the AD biomarker signature:
– DeMeyer et al diagnosis-independent “mixture modeling” approach (Arch Neurol, 2010)
– Significantly > 1 yr rate of hippocampal atrophy and LV volume  in subjects with baseline Ab42 below cutpoint
– Okonkwo et al characterized the relationship between CSF biomarker abnormalities and rate of decline in everyday
function across the AD spectrum (Arch Neurol, 2010)
Continue collaborative studies of data integration including imaging and biochemical biomarkers and neuropsychological
tests (eg, Vemuri et al, Neurol, 2009; Landau et al, Neurol, 2010; Davitskos et al, NBoA, 2010; Nettiksimmons et al, NBoA,
2010)
Developed and qualified an HPLC/MS-MS method for isoprostanes in CSF, BASELINE CSF data uploaded on ADNI website;
completed plasma homocysteine data through year 2; CSF HCy BASELINE measured at Merck, uploaded
Working closely with Kai Blennow on the International CSF QC program (sponsored by Alz Assn) to further promote the
standardization of pre-analytical and analytical steps involved in CSF biomarker measurements.
Plasma biomarker initiatives:
– Plasma Ab42/40 interlab study, using Luminex/Innogenetics reagents, data analyses: involving 12 ISAB & academic
labs, completed and posted on ADNI web site; ADNI longitudinal sample analyses initiated during 2010
– ISAB-sponsored studies of qualified RBM panels underway for 1066 ADNI BL & yr1 plasma in 2010
CSF sample collection for ADNI:
•After overnight fast
•Collect into polypropylene tube
•Transfer to polypropylene transfer tube
•No centrifugation
•Freeze at site, thaw & aliquot at UPenn,
storage at -80 0C
Number of biofluids collected as of 6/30/2010
13,122
Number of aliquots in biofluid bank
126,681
CSF through year 4
Plasma through year 4
N
948
2621
Average time
(mins)
35.11
67.94
95% CI (mins)
32.46-37.78
66.32-69.57
ADNI CSF Biochemical Biomarker Interlaboratory
Study (All Data Are on ADNI Website)
Participating Centers & Investigators
University of Pennsylvania: Leslie M Shaw, John Trojanowski, Virginia M-Y
Lee, Margaret Knapik-Czajka
Innogenetics: Hugo Vanderstichele
Sahlgrenska University Hospital: Kaj Blennow
Friedrich-Alexander-Universitat Erlangen-Nurnberg: Jens Wiltfang, Piotr
Lewczuk
Pfizer Global Research & Development: Holly Soares, Nancy Raha
Eli Lilly & Company: Robert A Dean, Eric Siemers, Richard Lachno, Brent
Salfen, (Linco)
Merck Research Laboratories: Adam Simon, William Potter
Reproducibility of xMAP Immunoassay (Innogenetics reagents/Luminex) in
the ADNI biomarker core lab
Avg test/re-test %CV:
Ab42,
5.7%
t-tau,
5.6%
p-tau181, 11.5%
CSF Biomarker Cutpoints Established Using CSFs Collected from ADNIIndependent Autopsy-Based AD and Age-Matched Cognitively Normal
Subjects
Tau
Ab1-42
p-Tau181p
Tau/Ab1-42
0.913
0.753
0.917
0.856
0.938
192 ng/mL
23 ng/mL
0.39
0.10
0.22
p-tau181p/Ab1-42
LR TAA
ROC AUC
0.831
Threshold values
93 ng/mL
Sensitivity (%)
69.6
6.4
67.9
85.7
91.1
100
Specificity (%)
92.3
76.9
73.1
84.6
71.2
76.9
Test accuracy (%)
80.6
87.0
73.1
85.2
81.5
88.9
Positive predictive
value (%)
90.7
81.8
67.9
85.7
77.3
82.4
Negative
predictive value
(%)
73.8
95.2
70.4
84.6
88.1
100
UPenn Autopsy-Based CSF Biomarker
AD Signature in the ADNI Study Cohorts
% of ADNI patients in whom biomarker
signature was detected using ROC
cutpoints
AD
MCI
NC
Ab1-42
91
78
34
Tau/Ab1-42 ratio
89
69
34
LRTAA model
89
70
31
Ab1-42 & p-tau181 abnormal
82
64
21
Ab1-42 abn but p-tau181 norm
9
10
17
Ab1-42 norm but p-tau181 abn
5
6
15
Ab1-42 & p-tau181 normal
4
20
47
LRTAA = a logistic regression model that includes Tau, Ab42, and ApoE e4
allele number; ROC = receiver operating characteristic curve
Survival analyses for ADNI MCI subjects:
progression to AD for BASELINE CSF biomarkers > or < cutpoints
Ab42<192 pg/mL
t-tau/Ab42>0.39
As of June 28, 2010
riskTAA2>0.34
Autopsy-based CSF biomarker AD signature in
ADNI MCI→AD progressors
N
MCI→AD
MCI→MCI
original MCI
82
109
191
Ab1-42
(<192 pg/mL)
90.2%
64.2%
75.4%
t-tau
(>93 pg/mL)
58.5%
35.8%
45.5%
p-tau181 (>23 pg/mL)
85.4%
59.6%
70.7%
t-tau/Ab1-42 (>0.39)
92.7%
56.9%
70.7%
p-tau181/Ab1-42 (>0.1)
92.7%
67.0%
78.0%
riskTAA2
89.0%
58.7%
71.7%
(>0.22)
Frequency distribution plots for Ab1-42
ADNI data as of 6/30/2010
Further validation of the AD biomarker signature
• DeMeyer, et al, diagnosis-independent analysis of ADNI CSF biomarker data: correctly
•
•
classified as AD 94% of autopsy-proven cases of AD in pre-mortem CSF (Arch Neurol,
67:1-8, 2010):
– 91% of ADNI AD subjects had the AD CSF signature
– Confirmed the AD signature more than 1/3rd of the elderly cognitively normal
control
We showed that there was a significantly greater rate of hippocampal atrophy and
lateral ventricular volume expansion in subjects with baseline Ab42 below the cutpoint
Okonkwo, et al, showed that CSF abnormalities are associated with functional decline
(FAQ) in NC and MCI subjects but do not predict functional decline in AD. Persons with
t-tau and Ab42 abnormalities are at greatest risk of functional impairment (Arch Neurol
67:688-696, 2010).
ADNI Data Integration is
off and Running
MRI and CSF biomarkers in normal,
MCI, and AD subjects
Diagnostic discrimination and cognitive correlations
Temporal Ordering of Biomarkers of AD
Reflects Disease Progression
Predictive performance of
biomarkers is time dependent
with Ab42 the earliest index of
disease pathology and tau a later
index. This time course of
biomarker pathology provides a
useful framework for
interpretation of biochemical
and imaging biomarker data.
Shaw, et al., 2007; Jack, et al., 2010; Trojanowski, et al, 2010
Plasma Ab42/40 interlaboratory qualification study
GROUP
CENTER
INVESTIGATOR
Pharmaceutical
Merck
Mary Savage
Pfizer
Holly Soares, now with BMS
Eli Lilly
Robert Dean
University Clinic, Erlangen
Piotr Lewczuk
Radboud University, Nijmegen
Mariel Verbeek
University Hospital, Amsterdam
Rien Blankenstein
Washington University, St Louis
David Holtzman
Molndal University, Goteborg
Kaj Blennow
Lille, France
Luc Buee
Mayo Clinic, Jacksonville
Steve Younkin
University of Penn, Philadelphia
Leslie Shaw
Academic
Diagnostic company Innogenetics, Ghent
Poster presentation at ICAD 2010
Hugo Vanderstichele
ADNI 2 Biomarker Core Update on the Recently Completed
Interlab Study of Plasma Aß Measures in Non-ADNI Pooled
Plasma Samples
•
•
•
•
•
Statistical analyses of the 12 center interlaboratory "round-robin" study report for
the INNO-BIA plasma Aß forms immunoassay(Innogenetics, Belgium) showed:
Total reproducibility across these 12 centers (combined within-center and between-center %CV)
was less than 16.7% for Aß42 and Aß40 and within individual centers the overall reproducibility is
even better.
The ADNI Biomarker Core laboratory which is funded by ADNI 1 and GO to measure plasma Aß42
and Aß40, achieved total analytical error (combined within- and between-day %CVs) ranging from
0.9 to 4.9% for plasma Aß42, 1.2 to 6.6% for Aß40 and up to 7.6% for the ratio of Aß42/Aß40.
Thus the ADNI Biomarker core has initiated measurement of plasma Aß42 and Aß40
concentrations in longitudinal samples of ADNI subjects, using this analytically qualified
methodology, to determine the changes in these biomarkers over time and determine their value as
indices of progression from MCI to AD.
110 INNO-BIA plasma Aß forms immunoassay kits have already been provided as an in-kind
contribution to ADNI study by INNOGENETICS.
This will help to significantly resolve controversies surrounding the significance of measuring
plasma Aß as an MCI/AD biomarker.
ADNI GO and ADNI 2
•
•
•
•
•
•
Receive, bank and monitor 24/7 CSF, plasma and serum samples
from ADNI GO and ADNI 2 subjects
Apply qualified Ab42, t-tau and p-tau181 to all ADNI GO and ADNI
2 subjects at baseline
Extend to earlier timepoint in the MCI to AD progression
biomarker assessments
Plasma Ab42 data on serial plasma samples from ADNI subjects
will be generated, analyzed
Apply Rules Based Medicine panel results from ADNI 1 for
plasma and CSF
Continue collaboration on the integration of imaging and
biochemical biomarker data in a variety of statistical models
Efforts underway to improve standardization
• ADNI
• Alz Assn International qc program
• UPenn/Wash U collaboration underway on
•
•
ELISA/xMAP relationships
UPenn/BIOCARD collaboration with Marilyn Alberts
Innogenetics-sponsored but investigator-led
workgroup on standardization guidelines
It takes a great team effort!
John Q Trojanowski
Uwe Christians
Kaj Blennow
Virginia M-Y Lee
Supported by the NIH/NIA and families
Holly
Soares
Chris Clark
of our patients
Adam Simon
Steve Arnold
Robert Dean
Hugo Vanderstichele
Eric Siemers
Magdalena Korecka
Piotr Lewczuk
Margaret Knapik-Czajka William Potter
Magdalena Brylska
ADNI investigators include
(complete listing available at
Teresa Waligorska
www.loni.ucla.edu\ADNI\Coll
Michal Figurski
aboration\ADNI_Manuscript_
Ravi Patel
Citations.pdf).
Leona Fields

similar documents