Sie sind auf Seite 1von 27

Biomarker Values Predictive of Cognitive Decline in MCI: Baseline or Progression?

Hiroko H. Dodge, PhD


Associate Professor of Neurology Director, Biostatistics and Data Core, Layton Aging and Alzheimers Disease Center Oregon Health and Science University Portland, OR Adjunct Research Associate Professor Michigan Alzheimers Disease Center Ann, Arbor, MI

Nothing to disclose

Predictive ability of biomarkers for cognitive declines: Why important?


Clinical trials would be improved by identifying the biomarkers most strongly associated with cognitive and functional declines at each stage of AD. Identifying biomarkers associated with subtle declines in cognitive functions among asymptomatic subjects is especially critical as research efforts move towards early identification of high risk subjects and prevention of progression.

Require Longitudinal Data Analyses

Statistical Approaches: Examples


CHANGE POINT MODELS Silbert, Dodge et. al. Trajectory of white matter
hyperintensity burden preceding mild cognitive impairment. Neurology 2012;79:741-747 (PMC3421153) Buracchio, Dodge et. al. The trajectory of gait speed preceding MCI. Archives of Neurology. Archives of Neurology, 2010; 67:980-986 (PMC2921227) Dodge, Ganguli, et al. Terminal decline and practice effects in non-demented older adults. Neurology, 2011:77;722-730. (PMC3164394)

Statistical Approaches: Examples


APPLICATION OF MIXED EFFECTS MODEL Erten-Lyons, Dodge et. al. Neuropathological basis of ageassociated brain atrophy. JAMA Neurology 70:616-622, 2013.

Dodge, Ganguli et. al. Cohort Effects in Age-Associated Cognitive Trajectories. Journal of Gerontology: Medical Sciences. In press.

Statistical Approaches: Examples


LATENT TRAJECTORY ANALYSIS Dodge, Kaye et. al. Inhome walking speeds and variability trajectories associated with Mild Cognitive Impairment. Neurology, 2012: 78(24):1946-1952. (PMC3369505)

Biomarkers vs. cognitive outcomes


Cognitive Functions
Individual Specific Deviations from population mean trajectory (random components)

Population mean/average trajectory

time (age)

Variability explained by Age, education, (reserve factors)

Baseline biomarker values

Rate of progression in biomarker values

Aim/Data
To examine which components ---baseline values or biomarker progressions-explains more variability in cognitive declines in memory and executive functions. DATA: the Alzheimers Disease Neuroimaging Initiative (ADNI 1). 526 subjects with valid data in at least one of our variables of interest were used in this study.

Cognitive Outcomes
Trajectory (slope) of cognitive functions:

1. ADNI-memory (ADNI-Mem) and 2. ADNI-executive (ADNI-Exe).


(Crane et al., 2012; Gibbons et al., 2012)

The scores are psychometrically optimized composite scores of memory and executive function, derived from items from ADNI neuropsychological tests

Approach: 2-stage
Stage 1. Individual-specific slope of the longitudinal trajectory of each biomarker was estimated using mixed effects models.

Longitudinal trajectories of biomarkers model = 00 + 0 + 10 + 1 + ,

where (0 , 1 ) ~ 2 (0, ). ~

0, and

Approach
Stage 2. Estimated individual-specific slope of each biomarker and observed baseline values were used as predictors of cognitive declines using mixed effects models.
= + 10 1*<0.5+ + 2 1*0.5+ ( 0.5) + ; = 0 + 1 Age at baseline + 2 Sex + 3 Education + 4 Apoe4 + 5 0 + 0 ; 2 = 20 + 21 Sex + 22 Apoe4 + 23 Changes in diagnosis + 24 0 + 25 1 + 2 ;

Approach
Variability in cognitive declines explained by subject-specific baseline biomarker values was compared with variability explained by biomarker progressions.

Results

Outcome: memory

Normal Group
Standardized effect size

Biomarker % of variability _bl: baseline values explained by _prog: progression biomarkers ttau_bl (1 sd=51) -5.40% ttau_prog (1 sd=0.21) -3.00% Abeta42_bl (1 sd=56) -7.70% Abeta42_prog (1 sd=0.14) -14.80%

-0.01
0.02

0
0.02

A positive percentage indicates that the corresponding predictor explains the variation in cognitive decline, while a negative percentage indicates that inclusion of the predictor adds more estimation error instead of improving model fitting.

Outcome: memory

Normal Group Standardized effect size

Biomarker % of variability _bl: baseline values explained by _prog: progression biomarkers FDG-PET_bl (1 sd=0.15) -0.60% FDG-PET_prog (1 sd=0.18) 1.50%

0.04
0.1

Outcome: memory

Normal Group

Biomarker _bl: baseline values _prog: progression Log_wmh/icv_bl (1 sd=1.57) Log_wmh/icv_prog (1 sd=0.05) Hpcv/icv_bl (1 sd=3.7E-04) Hpcv/icv_prog ( 1 sd=0.07) Ventricles/icv_bl (1 sd=5.8E-03) Ventricles/icv_prog (1 sd=0.11) Total brain/icv_bl (1 sd=0.04) Total brain/icv_prog (1 sd=0.09)

% of variability explained by biomarkers -4.20%


-3.00% -0.20% -4.70% -1.80% -0.70% -0.30% -3.00%

Standardized effect size

0.03
-0.29 0.04 0 0.01 -0.06 0.02 0.02

Outcome: memory Biomarker _bl: baseline _prog: progression

Normal Group % of variability explained by biomarkers Standardized effect size

Precuneus Thickness/icv_bl (1 sd=2.3E-07) Precuneus Thickness/icv_prog (1 sd=0.10) Medial Temporal Thickn./icv_bl (1 sd=2.3E-07) Medial Temporal Thickn./icv_prog (1 sd=0.10)

-2.36%

0.01

-0.48%

-0.01

-3.62%

-0.01

-0.05%

0.1

Outcome: memory Biomarker _bl: baseline values _prog: progression ttau_bl (1 sd=51) ttau_prog (1 sd=0.21) Abeta42_bl (1 sd=56) Abeta42_prog (1 sd=0.14) FDG-PET_bl (1 sd=0.15) FDG-PET_prog (1 sd=0.18)

Among MCI* % of variability explained by biomarkers -1.90% 0.30% Standardized effect size 0 -0.04

5.10%
10.30%

-0.15
-0.48

12.20%
12.70%

0.09
0.08

Outcome: memory Biomarker _bl: baseline values _prog: progression Log_wmh/icv_bl (1 sd=1.57) Log_wmh/icv_prog (1 sd=0.05) Hpcv/icv_bl (1 sd=3.7E-04) Hpcv/icv_prog ( 1 sd=0.07) Ventricles/icv_bl (1 sd=5.8E-03) Ventricles/icv_prog (1 sd=0.11) Total brain/icv_bl (1 sd=0.04) Total brain/icv_prog (1 sd=0.09)

Among MCI % of variability explained by biomarkers

Standardized effect size

0.50%
0.10% 9.00% 19.80% 8.70% 39.40% 2.40%

-0.03
0.01 0.07 0.08 -0.04 -0.12 0.06

16.00%

0.05

Outcome: memory Biomarker _bl: baseline _prog: progression

Among MCI
% of variability explained by biomarkers Standardized effect size

Precuneus Thickness/icv_bl (1 sd=2.3E-07) Precuneus Thickness/icv_prog (1 sd=0.10) Medial Temporal Thickness/icv_bl (1 sd=2.3E-07) Medial Temporal Thickness/icv_prog (1 sd=0.10)

3.49%

0.09

5.38%

0.07

4.36%

0.09

25.52%

0.11

Outcome: memory Biomarker _bl: baseline values _prog: progression ttau_bl (1 sd=51) ttau_prog (1 sd=0.21) Abeta42_bl (1 sd=56) Abeta42_prog (1 sd=0.14) FDG-PET_bl (1 sd=0.15) FDG-PET_prog (1 sd=0.18)

Among AD % of variability explained by biomarkers -7.90% -17.80% Standardized effect size -0.19 -0.08

-8.60%
6.60%

-0.05
-0.04

30.00%
84.00%

0.1
0.12

Outcome: memory Biomarker _bl: baseline values _prog: progression Log_wmh/icv_bl (1 sd=1.57) Log_wmh/icv_prog (1 sd=0.05) Hpcv/icv_bl (1 sd=3.7E-04) Hpcv/icv_prog ( 1 sd=0.07) Ventricles/icv_bl (1 sd=5.8E-03) Ventricles/icv_prog (1 sd=0.11) Total brain/icv_bl (1 sd=0.04) Total brain/icv_prog (1 sd=0.09)

Among AD
% of variability explained by biomarkers
Standardized effect size

-4.70%
3.00% 4.70% 26.00% 4.20% 63.80% -4.50% 26.00%

0.1
0.1 -0.07 0.15 0.11 -0.18 0.05 0.17

Outcome: memory

Among AD % of variability Standardized effect size explained by biomarkers 5.14% 5.13% 0.07 0.08

Biomarker _bl: baseline _prog: progression

Precuneus Thickness/icv_bl (1 sd=2.3E-07) Precuneus Thickness/icv_prog (1 sd=0.10) Medial Temporal Thickness/icv_bl (1 sd=2.3E-07) Medial Temporal Thickness/icv_prog (1 sd=0.10)

6.14%

0.09

64.65%

0.17

Conclusions

A number of studies have shown support for the hypothetical AD progression model developed by Jack et al (Jack et al., 2013; Jack et al., 2010). Our study results also coincides with the model; Across diagnostic groups, the percentages of variability in cognitive declines explained by functional (FDG-PET) or structural (brain morphometric) biomarkers (either their baseline values or progressions) increased significantly as disease progressed from normal to AD.

For most biomarkers, biomarker progressions were more predictive of (associated with) memory decline than baseline values. This suggests that clinical trials which require recruiting at risk subjects could be improved by using progression rather than baseline values in biomarkers to enrich the study subjects. Future studies are warranted to estimate the incremental effectiveness of improving clinical trial statistical power by using biomarker progression criteria

Special Thanks to:


Michigan ADC Roger Albin, MD Robert Koeppe, PhD Oregon Health & Science University ADC Jeffrey Kaye, MD Lisa Silbert, MD UC Davis Daniel Harvey, PhD University of Pittsburgh Mary Ganguli, MD Funding Sources

Michigan ADC pilot grant R13 AG030995, Friday Harbor Advanced Psychometric Work Shop 2011
MCI Symposium