Demographic and clinical characteristics
Demographics, clinical characteristics and neuropsychological scores are listed in Table 1. The three groups were well matched in age (one‑way ANOVA, F = 0.491, P = 0.613), sex (Chi‑square test, χ2 = 0.878, P = 0.645), and frame-wise displacement (one‑way ANOVA, F = 0.768, P = 0.466). There were significant differences in MMSE score (one‑way ANOVA, F = 255.546, P< 0.001), years of education (one‑way ANOVA, F = 18.805, P< 0.001), and TIV (one‑way ANOVA, F = 3.884, P = 0.023).
Table 1 Demographics, clinical characteristics and neuropsychological assessment of the study sample
Characteristics
|
AD
(n=57)
|
aMCI
(n=57)
|
HC
(n=48)
|
p Value
|
Women (%)
|
31 (54)
|
35 (61)
|
30(63)
|
0.645b
|
Age (years)
|
67.74 (8.28)
|
66.33 (7.47)
|
66.75 (7.41)
|
0.613a
|
Years of education
|
6.14 (5.38)
|
9.53 (4.71)
|
11.69(3.48)
|
<0.001a
|
MMSE score
|
15.28 (5.22)
|
26.35 (1.58)
|
28.60(1.11)
|
<0.001a
|
Clinical Dementia Rating Scale
|
|
|
|
|
1 (%)
|
21 (37)
|
|
|
|
2 (%)
|
22 (39)
|
|
|
|
3 (%)
|
14 (25)
|
|
|
|
Total intracranial volume (mm3)
|
1314.27 (109.09)
|
1352.93
(99.70)
|
1368.67
(103.03)
|
0.023a
|
Frame-wise displacement (mm)
|
0.21(0.14)
|
0.22(0.16)
|
0.18(0.11)
|
0.466a
|
neuropsychological tests
|
|
|
|
|
AVLT (immediate)
|
2.01(1.62)
|
5.89(2)
|
9.04(1.76)
|
<0.001a
|
AVLT (delay)
|
0(0-0)
|
4.46(3.08)
|
10.13(2.77)
|
<0.001c
|
AVLT (recognize)
|
10(6-12)
|
12(11-14)
|
14(13.5-15)
|
<0.001c
|
VFT (Animals)
|
9(7-10)
|
13(10.5-14.5)
|
18.13(3.93)
|
<0.001c
|
DSpF
|
5(4-6)
|
7(6-8)
|
8(7-8)
|
<0.001c
|
DSpB
|
3(2-3)
|
4(3-5)
|
5(4-6)
|
<0.001c
|
cognitive functionscore
|
|
|
|
|
Episodic memoryd
|
-2.70(1.65)
|
0.27(1.58)
|
2.88(1.22)
|
<0.001a
|
Semantic memoryd
|
-0.85(0.63)
|
0.04(-0.43-0.32)
|
1.0(0.74)
|
<0.001c
|
Working memoryd
|
-1.48(1.63)
|
0.29(1.24)
|
1.42(1.32)
|
<0.001a
|
Datas with normal distribution are expressed as mean (SD) or as percentage.
Datas with non-normal distribution are expressed as median (interquartile range).
aGroup difference was tested by using one-way analysis of variance (ANOVA).
bGroup difference was tested by using Pearson’s Chi-square test.
cGroup difference was tested by using Kruskal-Wallis test.
dEach cognitive composite score was computed by converting raw scores on each component of neuropsychological tests to z scores, using the mean and SD for all participants in the study, and averaging the z scores to yield the composite score.
Abbreviations: AD, Alzheimer’s disease; aMCI, amnestic mild cognitive impairment; HC, healthy control; MMSE, Mini-Mental State Examination; SD, standard deviations; AVLT, auditory verbal learning test; VFT, verbal fluency test; DSpF, Digit Span Forward; DSpB, Digit Span Backward.
Correlations between GMV and education years
Significant negative correlations were observed between education years and the GMV of the dorsal anterior cingulate cortex (dACC) (see Fig. 1, Table 2.) in AD patients. No significant negative correlations were observed in aMCI patients or HCs, and no positive correlations were observed in any of the groups.
Table 2 Results of voxel-based regression analyses between neuroimaging parameters and years of education
Region
|
Number of voxels
|
t-score
|
MNI coordinates
|
x
|
y
|
z
|
Negative correlations between years of education and GMV
|
dorsal anterior cingulate cortex
|
291
|
-5.25
|
0
|
20
|
33
|
Negative correlations between years of education and ReHo
|
left anterior part of inferior temporal cortex
|
65
|
-6
|
-57
|
-12
|
-33
|
left posterior part of inferior temporal cortex
|
59
|
-4.69
|
-51
|
-30
|
-21
|
Negative correlations between years of education and left aITC rs-FC
|
right angular gyrus
|
32
|
-4.79
|
45
|
-66
|
42
|
left mid frontal cortex
|
27
|
-5.57
|
-39
|
15
|
57
|
right superior frontal cortex
|
16
|
-4.95
|
24
|
27
|
54
|
Coordinates (x,y,z) are given in MNI standard space. All regions listed are statistically significant at p<0.05 level (FDR corrected). Abbreviations: GMV, gray matter volume; ReHo, Regional Homogeneity; rs-FC, Resting-state Functional Connectivity; MNI, Montreal Neurological Institute; aITC, anterior part of inferior temporal cortex.
Correlations between ReHo and education years
Education years were significantly negatively correlated with the ReHo values of the left inferior temporal cortex (ITC) in AD patients (see Fig. 2, Table 2.). No significant negative correlations were observed in aMCI patients or HCs, and no positive correlations were observed in any of the groups.
Correlations between seed-based FC and education years
The dACC and ITC were then used as seeds for the whole brain connectivity analyses. Thus we obtained the FC maps and the FC maps were then correlated with education years. Significant negative correlations were observed between education years and the connectivity of the left anterior ITC (aITC) with the left mid frontal cortex, as well with the right superior frontal cortex and right angular gyrus (Fig. 3, Table 2.). No significant negative correlations were observed in the dACC and posterior ITC (pITC) based FC and no positive correlations were found.
Correlations between the values extracted from ROI and education years in AD stratified analysis
Table 3 displays the stratified analysis of the correlations between education years and MRI-based biomarkers of AD-related pathology (GMV, ReHo, and FC values of ROI) according to the disease severity (CDR scores). In the mild dementia group, all biomarkers we used were inversely associated with years of education (P< 0.01). This inverse relationship was attenuated and was not significant in patients with moderate or severe dementia, or in patients with aMCI or in HCs.
Table 3 Correlations between years of education and biomarkers of AD-related pathology: linear regression analysis
AD-related pathology biomarkers
|
Relationships with years of education
|
HC
|
aMCI
|
CDR1
(mild dementia)
|
CDR2
(moderate dementia)
|
CDR3
(severe dementia)
|
β
(SE)
|
p
|
B
|
β
(SE)
|
p
|
B
|
β
(SE)
|
p
|
B
|
β
(SE)
|
p
|
B
|
β
(SE)
|
p
|
B
|
GMV of dACC
|
-0.001
(0.002)
|
0.7
|
-0.062
|
-0.003
(0.002)
|
0.038
*
|
-0.258
|
-0.007
(0.002)
|
0.006**
|
-0.753
|
-0.006
(0.005)
|
0.233
|
-0.569
|
-0.005
(0.004)
|
0.239
|
-0.568
|
Reho of left aITC
|
0.008
(0.005)
|
0.108
|
0.261
|
-0.003
(0.003)
|
0.372
|
-0.127
|
-0.018
(0.004)
|
<0.001***
|
-0.739
|
-0.038
(0.012)
|
0.006**
|
-1.485
|
-0.007
(0.005)
|
0.247
|
-0.62
|
Reho of left pITC
|
0.003
(0.006)
|
0.649
|
0.076
|
-0.008
(0.004)
|
0.037
*
|
-0.284
|
-0.025
(0.006)
|
0.002**
|
-0.831
|
-0.003
(0.014)
|
0.829
|
-0.101
|
-0.011
(0.01)
|
0.305
|
-0.586
|
left aITC rs-FC
|
right angular gyrus
|
0.002
(0.01)
|
0.814
|
0.038
|
-0.004
(0.009)
|
0.662
|
-0.061
|
-0.034
(0.01)
|
0.004**
|
-0.72
|
-0.041
(0.026)
|
0.127
|
-0.899
|
-0.036
(0.012)
|
0.017
*
|
-1.043
|
left mid frontal cortex
|
0.013
(0.008)
|
0.098
|
0.271
|
0.002
(0.008)
|
0.763
|
0.039
|
-0.031
(0.008)
|
0.002**
|
-0.717
|
-0.035
(0.019)
|
0.081
|
-0.873
|
-0.032
(0.014)
|
0.05
|
-0.987
|
right superior frontal cortex
|
0.002
(0.009)
|
0.835
|
0.032
|
-0.002
(0.007)
|
0.786
|
-0.038
|
-0.029
(0.009)
|
0.007**
|
-0.658
|
-0.05
(0.021)
|
0.029*
|
-1.244
|
-0.039
(0.016)
|
0.039
*
|
-1
|
Structural MR (GMV): Adjusted for age, sex, MMSE, and total intracranial volume (TIV).
Resting-state fMRI (Reho and rs-FC): Adjusted for age, sex, MMSE, and head motion parameters.
Abbreviations: CDR, Clinical Dementia Rating; SE, standard error; GMV, gray matter volume; ReHo, Regional Homogeneity; rs-FC, Resting-state Functional Connectivity; dACC, dorsal anterior cingulate cortex; aITC, anterior part of inferior temporal cortex; pITC, posterior part of inferior temporal cortex.
β stand for unstandardized coefficients, B stand for standardized coefficients (Beta).
P< 0.05*, P< 0.01**, P< 0.001***
Values extracted from ROI compared between CR+ and CR− groups
We dichotomized patients (aMCI and AD) into CR+ and CR− groups with a cutoff point of 8.5 years of education (the median of education years within the entire sample). The demographics and neuropsychological characteristics of CR+ and CR− groups are depicted in Table S1. We compared GMV, ReHo, and FC values between CR+ and CR− groups at each stage of AD dementia. In CDR1 and CDR2 patients, CR+ showed greater gray matter volume atrophy of the dACC and lower ReHo values of left ITC. However, in CDR3 and aMCI patients, there were no significant differences.
Table S1 Demographicsof the study samples subjects split by Diagnosis and CR group
Characteristics
|
aMCI
|
CDR1
(mild dementia)
|
CDR2
(moderate dementia)
|
CDR3
(severe dementia)
|
CR-
(n=22)
|
CR+
(n=35)
|
CR-
(n=12)
|
CR+
(n=9)
|
CR-
(n=15)
|
CR+
(n=7)
|
CR-
(n=7)
|
CR+
(n=7)
|
Age (years)
|
64.79±
6.57
|
68.46±
8.22
|
66.58±
5.66
|
70.22±
7.68
|
69.27±
7.36
|
62.86±
9.33
|
63.71±
7.34
|
72.14±
12.13
|
Women (%)
|
9(41)
|
17(49)
|
6(50)
|
3(33)
|
5(33)
|
3(43)
|
2(29)
|
3(43)
|
MMSE score
|
26.18±
1.65
|
26.58±
1.47
|
19.83±
3.19
|
20.67±
2.12
|
†12(11-14)*
|
17.43±
1.72*
|
7.86±
5.11
|
10.86±
1.46
|
Totalintracranial volume (mm3)
|
1353.17±97.12
|
1352.59±
105.25
|
1341.59±
95.19
|
1314.34±
112.45
|
1312.50±
98.90
|
1290.83±
123.87
|
1322.10±
123.65
|
1286.79±
145.08
|
GMV of dACC
|
0.51±
0.05
|
0.48±
0.06
|
0.51±
0.04*
|
0.45±
0.03*
|
0.50±
0.04*
|
0.44±
0.05*
|
0.47±
0.06
|
0.43±
0.04
|
Reho of
left aITC
|
0.61±
0.14
|
0.56±
0.10
|
0.66±
0.10*
|
0.47±
0.07*
|
0.62±
0.12*
|
0.51±
0.11*
|
0.56±
0.06
|
0.50±
0.05
|
Reho of
left pITC
|
0.80±
0.14
|
0.73±
0.11
|
0.89±
0.13*
|
0.69±
0.11*
|
0.80±
0.15*
|
0.66±
0.10*
|
0.73±
0.11
|
0.64±
0.09
|
left aITC rs-FC
|
right angular gyrus
|
0.37±
0.31
|
†0.34(0.17-0.44)
|
0.45±
0.19*
|
0.12±
0.18*
|
0.45±
0.17*
|
0.18±
0.23*
|
†0.20(0.16-0.56)
|
0.13±
0.13
|
left mid frontal cortex
|
0.28±
0.20
|
†0.20(-0.01-0.4)
|
0.43±
0.17*
|
0.14±
0.19*
|
0.40±
0.17*
|
0.19±
0.18*
|
0.28±
0.19
|
0.09±
0.16
|
right superior frontal cortex
|
0.38±
0.32
|
0.33±
0.19
|
0.40±
0.17*
|
0.09±
0.17*
|
0.43±
0.18*
|
0.25±
0.19*
|
0.42±
0.22*
|
0.16±
0.19*
|
Datas with normal distribution are shown as mean (SD), determined with independent-sample t test.
†Datas with no normal distribution are shown as mean (interquartile range), determined with the Wilcoxon rank test and the Fisher exact test.
* indicated statistically significant (P < 0.05).
Abbreviations: GMV, gray matter volume; ReHo, Regional Homogeneity; rs-FC, Resting-state Functional Connectivity; dACC, dorsal anterior cingulate cortex; aITC, anterior part of inferior temporal cortex; pITC, posterior part of inferior temporal cortex.
Correlations between the values extracted from ROI and cognitive function
Pearson correlation analyses were used to examine the relationship between ROI-based neuroimaging variables and clinical cognitive function scores. In HC group, GMV of dACC was significantly negatively correlated with semantic memory (r = -0.407, p = 0.006), but not in the episodic and working memory (see Table 4.). In AD group, GMV of dACC was significantly positively correlated with semantic memory (r = 0.453, p = 0.001) and episodic memory (r = 0.334, p = 0.015). ReHo values of the left pITC were significantly positively correlated with episodic memory (r = 0.391, p = 0.004), and the rs-FC values of the left aITC connected with the right angular gyrus were also significantly positively correlated with episodic memory (r =0.312, p = 0.023). The rs-FC values of the left aITC connect with the left middle frontal cortex were significantly positively correlated with episodic memory (r =0.338, p = 0.013) and semantic memory (r =0.277, p = 0.045) (see Table 5.). There were no significant correlations between ROI-based neuroimaging variables and clinical cognitive function scores in aMCI group (see Table S2.).
Table 4 Correlations between ROI-based neuroimaging variables and clinical cognitive function scores in HC group
Cognitive system
|
Episodic memory
|
Semantic memory
|
Working memory
|
r
|
p
|
r
|
p
|
r
|
p
|
aGMV of dACC
|
-0.237
|
0.122
|
-0.407
|
0.006**
|
0.107
|
0.488
|
bReho of left aITC
|
-0.089
|
0.564
|
-0.152
|
0.326
|
-0.06
|
0.7
|
bReho of left pITC
|
-0.241
|
0.114
|
-0.039
|
0.801
|
0.13
|
0.399
|
Left aITC rs-FCb
|
|
|
|
|
|
|
right angular gyrus
|
0.039
|
0.802
|
0.014
|
0.929
|
0.132
|
0.393
|
Left middle frontal cortex
|
0.109
|
0.479
|
0.088
|
0.571
|
0.027
|
0.862
|
right superior frontal cortex
|
0.03
|
0.848
|
-0.065
|
0.674
|
0.052
|
0.738
|
The relationship between neuroimaging variables and clinical cognitive function scores were assessed by using Pearson correlation analysis.
aAdjusted for age, sex, education, and TIV.
bAdjusted for age, sex, education, and head motion parameters.
P< 0.05*, P< 0.01**, P< 0.001***
Abbreviations: HC, healthy control; GMV, gray matter volume; ReHo, Regional Homogeneity; rs-FC, resting-state functional connectivity; dACC, dorsal anterior cingulate cortex; aITC, anterior part of inferior temporal cortex; pITC, posterior part of inferior temporal cortex; TIV, total intracranial volume; ROI, region of interest.
Table 5 Correlations between ROI-based neuroimaging variables and clinical cognitive function scores in AD group
Cognitive system
|
Episodic memory
|
Semantic memory
|
Working memory
|
r
|
p
|
r
|
p
|
r
|
p
|
aGMV of dACC
|
0.334
|
0.015*
|
0.453
|
0.001**
|
0.205
|
0.141
|
bReho of left aITC
|
0.172
|
0.217
|
0.049
|
0.728
|
0.021
|
0.884
|
bReho of left pITC
|
0.391
|
0.004**
|
0.2
|
0.151
|
0.102
|
0.466
|
Left aITC rs-FCb
|
|
|
|
|
|
|
right angular gyrus
|
0.312
|
0.023*
|
0.082
|
0.562
|
0.089
|
0.525
|
Left middle frontal cortex
|
0.338
|
0.013*
|
0.277
|
0.045*
|
0.084
|
0.552
|
right superior frontal cortex
|
0.108
|
0.44
|
-0.035
|
0.805
|
-0.128
|
0.36
|
The relationship between neuroimaging variables and clinical cognitive function scores were assessed by using Pearson correlation analysis.
aAdjusted for age, sex, education, and TIV.
bAdjusted for age, sex, education, and head motion parameters.
P< 0.05*, P< 0.01**, P< 0.001***
Abbreviations: AD, Alzheimer’s disease; GMV, gray matter volume; ReHo, Regional Homogeneity; rs-FC, resting-state functional connectivity; dACC, dorsal anterior cingulate cortex; aITC, anterior part of inferior temporal cortex; pITC, posterior part of inferior temporal cortex; TIV, total intracranial volume; ROI, region of interest.
Table S2 Correlations between ROI-based neuroimaging variables and clinical cognitive function scores in aMCI group
Cognitive system
|
Episodic memory
|
Semantic memory
|
Working memory
|
r
|
p
|
r
|
p
|
r
|
p
|
aGMV of dACC
|
0.138
|
0.323
|
0.08
|
0.569
|
0.259
|
0.061
|
bReho of left aITC
|
-0.043
|
0.758
|
-0.106
|
0.449
|
0.006
|
0.968
|
bReho of left pITC
|
-0.132
|
0.345
|
0.038
|
0.785
|
-0.087
|
0.537
|
Left aITC rs-FCb
|
|
|
|
|
|
|
right angular gyrus
|
-0.139
|
0.321
|
0.05
|
0.724
|
-0.062
|
0.658
|
Left middle frontal cortex
|
-0.259
|
0.061
|
0.016
|
0.91
|
-0.184
|
0.186
|
right superior frontal cortex
|
0.026
|
0.856
|
-0.156
|
0.266
|
-0.099
|
0.483
|
aAdjusted for age, sex, education, and TIV.
bAdjusted for age, sex, education, and head motion parameters.
P< 0.05*, P< 0.01**, P< 0.001***
Abbreviations: aMCI, amnestic mild cognitive impairment; GMV, gray matter volume; ReHo, Regional Homogeneity; rs-FC, resting-state functional connectivity; dACC, dorsal anterior cingulate cortex; aITC, anterior part of inferior temporal cortex; pITC, posterior part of inferior temporal cortex; TIV, total intracranial volume; ROI, region of interest.