Skip to main content

Table 3 Effect of tau, amyloid, and Lewy body pathology on regional brain volumes

From: Association of Alzheimer’s and Lewy body disease pathology with basal forebrain volume and cognitive impairment

Model

P(M)

P(M|data)

BFM

BF10

Error %

Evidence for H1/H0

Posterior basal forebrain volume

    

Null model

0.25

0.003

0.01

1.0

  

Lewy + Braak

0.083

0.68

36.2

664.6

3.7

Extreme for H1

Lewy + Braak + CERAD

0.25

0.28

0.90

90.5

3.4

Very strong for H1

Lewy + CERAD

0.083

0.02

0.18

19.6

3.9

Strong for H1

Lewy

0.083

0.01

0.10

10.7

6.9

Strong for H1

Braak

0.083

0.006

0.05

5.8

2.3

Moderate for H1

Braak + CERAD

0.083

5.6*10− 4

0.005

0.55

3.1

Weak for H0

CERAD

0.083

4.6*10− 4

0.004

0.45

3.1

Weak for H0

Anterior basal forebrain volume

     

Null model

0.25

0.64

4.3

1.0

  

CERAD

0.083

0.22

4.5

1.03

3.7

Weak for H1

Braak

0.083

0.04

0.57

0.20

1.4

Moderate for H0

CERAD + Lewy

0.083

0.04

0.49

0.17

5.6

Moderate for H0

Lewy

0.083

0.03

0.37

0.13

2.2

Moderate for H0

Braak + CERAD

0.083

0.01

0.14

0.05

1.8

Very strong for H0

Braak + Lewy

0.083

0.009

0.11

0.04

7.5

Very strong for H0

Braak + CERAD + Lewy

0.25

0.007

0.03

0.01

2.5

Very strong for H0

Hippocampus volume

    

Null model

0.25

4.8*10− 6

1.2*10− 5

1.0

  

Braak

0.083

0.67

16.3

422,422

 

Extreme for H1

Braak + CERAD

0.083

0.19

1.7

116,617

 

Extreme for H1

Braak + Lewy

0.083

0.09

0.68

53,193

 

Extreme for H1

Braak + CERAD + Lewy

0.25

0.06

0.14

11,497

 

Extreme for H1

CERAD

0.083

9.0*10− 4

0.007

565.2

 

Extreme for H1

CERAD + Lewy

0.083

9.6*10− 5

7.0*10− 4

60.2

 

Very strong for H1

Lewy

0.083

4.8*10− 7

3.5*10− 6

0.30

 

Moderate for H0

  1. Results from Bayesian ANCOVAs across all groups with the respective regional volume measure as dependent variable and the three neuropathological staging systems (Braak stage, CERAD score, presence of Lewy body pathology) as predictors including covariates for age at MRI, sex, years of education, time between MRI and autopsy, and site. Covariates are included in all models (including the null model)
  2. BF10, Bayes factor in favour of H1 over H0; BFM, degree to which the data have changed the prior model odds; error %, numerical stability of BF10 over 10 000 Markov Chain Monte Carlo iterations; H0, null hypothesis; H1, alternative hypothesis; P(M), model’s prior probability; P(M|data), model’s posterior probability after observing the data