PET Core

PIB vs CSF Biomarkers: A
Total N = 55 (11 Control
Control, 34 MCI
MCI, 10 AD)
300
MCI
AD
Control
CSF A 1-42
250
200
Penn Autopsy
Sample (56 AD, 52
Cog normal)
192 pg/ml
150
100
50.0
1
1.2
1.4
1.6
1.8
2
Mean Cortical SUVR
2.2
2.4
PIB vs CSF Biomarkers: p-Tau181
90.0
80.0
MCI
AD
Control
CS
SF P-tau
70.0
60.0
50.0
40.0
30.0
Penn Autopsy
Sample (56 AD, 52
Cog normal)
23 pg/ml
20.0
10 0
10.0
1
1.2
1.4
1.6
1.8
2
Mean Cortical SUVR
2.2
2.4
PIB vs FDG
1.60
1.50
Composiite FDG R
ROI
1.40
FDG Cutoff:
from ROC
analysis of
separate
ADNI
subjects (AD
vs Controls)
1.30
1.20
1.10
1.00
0.900
0.800
MCI
AD
Control
1
1.2
1.4
1.6
1.8
2
Mean Cortical SUVR
2.2
2.4
Biomarker Agreement
(kappa)
CSF A1-42
CSF t-tau
CSF p-tau
PIB
0.71
0.21
0.50
FDG
0.23
0.28
0.25
FDG
0.12
Biomarkers vs MMSE
FDG
CSF A1-42
35.0
35.0
30.0
30.0
30.0
25.0
25.0
25.0
20.0
20.0
15.0
10.0
MMSE
35.0
MMSE
MMSE
PIB
15.0
R 02
R=0.2
1
1.2
1.4
20.0
15.0
R=0.2
1.6
1.8
2
Mean Cortical SUVR
2.2
2.4
10.0
50
100
150
200
CSF A1-42
250
R=0.6
300
10.0
0.8
0.9
1
1.1
1.2
1.3
FDG Composite ROI
Linear regressions combining all 3 subject
s bject groups:
gro ps
MCI, AD, Controls
1.4
1.5
1.6
Simple Rules for Classification
98% specificity
p
y for FTD in autopsy-confirmed
p y
cases;; Foster et al. Brain 2007;; 130: 2616-35.
FTD May Contaminate AD Clinical
Ti l
Trials
10 of 93 subjects (10.8%) had FTD-like
metabolic pattern at baseline
Topographic Extent of
Cerebral Hypometabolism Is
Intermediate in MCI
# of Significant Pixels
Diagnosis at Baseline
MCI (n = 161)
217 ± 307
AD (n = 75)
579 ± 865 (p < 0.001)
CN (n = 80)
86 ± 152 (p < 0.01)
Median and quartile plots
Lower CMRgl in AD & MCI
92 AD<104 NC
184 MCI<104 NC
Reiman et al
Banner Alzheimer Institute
from Langbaum et al, NeuroImage (in press)
Correlations Between Lower CMRgl & Clinical Severity
CDR
298 subjects
MMSE
298 subjects
MMSE
74 AD patients
Reiman et al
Banner Alzheimer Institute
Langbaum et al, NeuroImage (in press)
12-month CMRgl Declines in AD & MCI
69 AD patients
154 MCI patients
Reiman et al
Banner Alzheimer Institute
Empirically pre-defined statistical ROI for the
assessment of 12-Month CMRgl declines in AD patients
Defined using data from 27 training-set patients using bootstrap with replacement
Number of AD patients per group needed in a 12-month multi-center RCT to detect a
25% treatment effect with power=80%, p=0.05 & no need to correct for multiple
comparisons
FDG PET
61
ADAS-COG11
612
Characterized in 29 test-set patients (excluding HiRez & HRRT scanners)
MMSE
493
Reiman et al
Banner Alzheimer Institute
Empirically pre-defined statistical ROI for the
assessment of 12-Month CMRgl declines in MCI patients
Defined using data from 50 training-set patients using bootstrap with replacement
Number of MCI patients per group needed in a 12-month multi-center RCT to detect a
25% treatment effect with power=80%, p=0.05 & no need to correct for multiple comparisons
FDG PET
217
ADAS-COG11
4402
MMSE
5749
Characterized in 74 test-set patients (excluding HiRez & HRRT scanners)
Reiman et al
Banner Alzheimer Institute
FDG-PET and Clinical Outcomes
Composite
p
ROI - Based on p
published coordinates (smoothed)
(
)
Multivariate Random Effects Models:
Baseline FDG vs ADAS-Cog Change
Baseline and Longitudinal FDG vs ADAS-Cog Change
Longitudinal FDG Sample
(1319 scans!)
!)
N
AD
MCI
N
Normal
l
Baseline
95
206
102
6 Months
86
188
94
12 Months
74
176
85
18 Months
NA
76
NA
24 Months
26
66
45
Composite FDG ROI and ADAS-Cog Change
Dependent: ADAS-cog change
AD
Model 1
MCI
AD, MCI, N
β
p
β
p
β
p
-1.96
0.003
-0.54
0.012
-0.63
<0.001
FDG-ROIs Baseline
-1.95
0.002
-0.66
0.003
-0.80
<0.001
FDG ROI Change
FDG-ROIs
Ch
-3.25
<0.001
-1.08
<0.001
-1.73
<0.001
FDG-ROIs Baseline
Model 2
Mixed effects models
Random slope & intercept to account for individual variability in starting
point and change
Controlled for age, education, # ApoE 4 alleles, group membership
Standardized FDG measure - change in ADAS-Cog per 1 SD difference
(either at baseline or change) in FDG ROI
MCI Subjects
Converters
N
Gender (M/F)
Cognitive
tests
28
57
19/9
37/20
p-value
ns
mean
SD
mean
SD
Age (yrs)
78.3
7.5
78.0
7.4
ns
Education (yrs)
16.4
2.6
16.3
2.8
ns
Total followup time (yrs)
1.9
0.4
1.9
0.4
ns
ADAS-cog (baseline)
13.2
4.6
10.3
3.9
0.003
MMSE (baseline)
26.4
1.7
27.3
1.6
0.03
APOE ε4 frequency
CSF
biomarkers
(pg/mL)
Non-converters
0.41
0.25
Cont
Dichot
0.07
ns
FDG-PET
1.13
0.10
1.22
0.14
0.002
0.05
3)
Hi
Hipp
V
Volume
l
(mm
(
2883
558
3187
527
0 009
0.009
0 07
0.07
Aß
149.7
45.3
165.7
57.9
ns
ns
p-tau
37.1
10.7
34.3
17.8
ns
0.02
t-tau
94 0
94.0
28 1
28.1
100 6
100.6
55 3
55.3
ns
ns
p-tau/Aß
0.27
0.12
0.25
0.18
ns
0.01
t-tau/Aß
0.68
0.27
0.75
0.62
ns
ns
AVLT Recall
26.4
6.6
32.2
8.1
0.001
0.01
AVLT Del
1.4
2.2
2.8
3.1
0.03
0.10
LogMem Recall
6.3
3.4
7.7
2.9
0.04
ns
LogMem Del
2.9
2.6
4.6
2.7
0.005
ns
univariate
B +/-SE
HR
p
B +/-SE
FDG-PET HR
p
B +/-SE
Hc
HR
p
B +/-SE
+/ SE
Aβ
HR
p
B +/-SE
Ptau
HR
p
B +/-SE
+/ SE
Tau
HR
p
B +/-SE
Ptau/Aβ HR
p
B +/-SE
/ SE
Tau/Aβ HR
p
B +/-SE
AVLT-Rc HR
p
B +/-SE
AVLT-Del HR
p
B +/-SE
LogMem-Rc HR
p
B +/-SE
LogMemHR
Del
p
ApoE4
C
S
F
b
i
o
m
a
r
k
e
r
s
M
e
m
o
r
y
F
x
n
0.39 +- 0.19
1.47
0.04
-0 73 +- 0.22
-0.73
0 22
0.48
0.001
-0.81 +- 0.24
0.45
0.001
multivariate
ns
-0.91
-0
91 +- 0.26
0 26
0.40
<0.001
-0.74 +- 0.26
0.48
0.005
Predicting
g conversion
to AD
ns
ns
ns
ns
ns
ns
-0.75 +- 0.23
0.47
0.001
-0.59 +- 0.26
0.56
0.021
-0.48 +- 0.21
0.62
0.020
-0.69 +- 0.21
0.50
0.001
-0.60 +- 0.20
0.55
0.002
ns
Predictor variables in continuous form
Cox proportional hazards model
Controlled for age, education, sex
univariate
B +/-SE
HR
p
B +/-SE
FDG-PET HR
p
B +/-SE
Hc
HR
p
B +/-SE
+/ SE
Aβ
HR
p
B +/-SE
Ptau
HR
p
B +/-SE
+/ SE
Tau
HR
p
B +/-SE
Ptau/Aβ HR
p
B +/-SE
/ SE
Tau/Aβ HR
p
B +/-SE
AVLT-Rc HR
p
B +/-SE
AVLT-Del HR
p
B +/-SE
LogMem-Rc HR
p
B +/-SE
LogMemHR
Del
p
ApoE4
C
S
F
b
i
o
m
a
r
k
e
r
s
M
e
m
o
r
y
F
x
n
0.66 +- .40
1.94
0.10
1 08 +- .045
1.08
045
2.94
0.016
0.91 +- 0.45
2.49
0.040
multivariate
ns
1.08
1
08 +- 0.46
0 46
2.95
0.018
Predicting
g conversion
to AD
ns
ns
1.06 +- 0.50
2.88
0.033
Joint hazard ratio: 15.03
15 03
ns
1.38 +- 0.62
3.99
0.025
ns
ns
1.54 +- 0.63
4.68
0.014
1.01 +- 0.51
2.75
0.047
1.63 +- 0.63
5.08
0.010
ns
ns
ns
Predictor variables in dichotomous form
Cox proportional hazards model
Controlled for age
age, education
education, sex
Hypometabolism is More
Extensive in MCI Subjects
Converting to AD by 1 Year
# off Si
Significant
ifi
t Pixels
Pi l
Remain MCI at 1 year
N=138 (92 men
men, 46 women)
Age: 75.4 ± 6.5
Pixels: 217 ± 307
Convert to AD by 1 year
N=23 (14 men, 9 women)
Age: 75.3 ± 6.5
Pixels 511 ± 627 (p < 0.02)
Distribution of Significant Pixels at Baseline and
Clinical Outcome in MCI Subjects
Baseline PIB Subjects
PiB(+) Definition
D fi iti
ADNI ROI and 4 Region Avg
3.500
3.5
Ctrl
MCI
AD
2.500
2.5
SUVR 50-70
SUV
VR 50-7
70
3.000
3.0
2.000
2.0
Cut-off
9
1 500
1.500
1.5
1
5
10
47 17
18 2
C ON
MC I
AD
1.0
1.000
ACG
0.500
FC
PAR
PRC 4 Reg Avg
Aizenstein et al., Arch Neurol 2008; 65:1509-17
Mathis et al
1 Year Changes in PIB SUVR Values by
Subject
j
Group
p
Longitudinal
Mean of 4 Region Average
2 50
2.50
Baseline
2.0
2.00
1.5
PiB SUVR 50-70
SUVR 5
S
50-70
1 year
1.50
B/L
Year1
1.0
1.00
05
0.5
0.50
0
0.00
Ctrl
C ON
MCI
PiB-
MC I
PiB(-)
AD
AD
Ctrl
C ON
MCI
MC I
PiB+
PiB(+)
AD
AD
Mathis et al
ADNI Subjects with Reliable Change
Reliable change defined as >1 SD of test-retest
variability or SUVR change >0.215
Ct l
Ctrl
MCI
# >0.215
#
>0.215
PiB(-)
6
0
15
PiB(+)
8
4
25
AD
#
>0.215
0
1
0
4
10
3
Mathis et al
Follo Up of PIB-Positive
Follow-Up
PIB Positi e ADNI MCI’s
ADNI PiB MCI’s
N = 65, 12 mo. follow-up
PiB(-)
Converters to AD
PiB(+)
18
3
47
Converters to AD 14
Summary
CSF A and PIB are closely related, neither is
related
l t d to
t cognition
iti
FDG-PET is related to cognition and tracks
change
h
att baseline
b
li and
d over time
ti
Statistically defined ROIs produce small sample
sizes in power calculations
FDG-PET predicts conversion from MCI to AD
PIB PET is
PIB-PET
i related
l t d to
t conversion
i
Still few subjects with PIB-PET longitudinal data
The Future: GO and ADNI2
18F
Amyloid imaging agents
C b
Can
be distributed
di t ib t d to
t almost
l
t every ADNI site
it
Standardization issues fundamentally identical
to FDG
Goals
Define A load in ADNI cohort especially VMCI
Relate to prior longitudinal change (ADNI1)
Examine different methods of quantitation
(
)
Prediction of decline/conversion (ADNI2)
Endophenotype for GWAS studies
[18F]-AV-45 (Florpiramine F18)
10 min data acquisition
AD
QuickTime™ and a
decompressor
are needed to see this picture.
[18F]-AV-45
18
F
O
O
O
N
Normal
Elderly
QuickTime™ and a
decompressor
are needed to see this picture.
NHCH3
Ki=3.1 nM
Dan Skovronsky, Avid Radiopharmaceuticals
[18F]3’-F-PIB and [11C]PIB in the same control
and
d AD subject
bj t
CONTROL
AD
[18F]3’-F-PIB
2.5
18
HO
F
S
N
NHCH3
90-120 min
Ki = 3 nM
[11C]PIB
HO
S
N
Ki = 2 nM
0.5
25
2.5
NH11CH3
40-90
40
90 min
05
0.5
Chet Mathis, U Pittsburgh
[18F]BAY94-9172
AD subjects
CONTROL
90-120 min
[18F]BAY94-9172
F]BAY94 9172
18
F
O
O
O
NHCH3
Ki = 6.7 nM
61 years (MMSE 30)
80 years (MMSE 26)
83 years (MMSE 20)
Rowe et al., Lancet Neurol. 2008
[18F]FDDNP and [11C]PIB in the Same Subjects
[11C]PIB
HO
S
NH11CH3
N
QuickTime™ and a
decompressor
are needed to see this picture.
p
[18F]FDDNP
]
NC
CH3
18
F
CN
N
CH3
Tolboom et al, J Nucl Med 2009
Florpiramine Clinical Trial Sites
QuickTime™ and a
decompressor
are needed to see this picture.
Dan Skovronsky, Avid Radiopharmaceuticals
Results: subjects < 70 years old
Dan Skovronsky, Avid Radiopharmaceuticals
Results: All Subjects
(N=89)
(N=55)
(N=60)
Dan Skovronsky, Avid Radiopharmaceuticals
Acknowledgements
Berkeley ADNI Group
Susan Landau
Cindee Madison
Connie Cheung
Beth Mormino
Danielle Harvey (UC Davis)
ADNI PET
Norm Foster
Danielle Harvey
Bob Koeppe
Chet Mathis
Eric Reiman
Ron Petersen, Paul Aisen, Cliff Jack, Les Shaw, John
Trojanowski, Laurel Beckett, Mike Weiner
and
The ADNI Investigators