Demographic characteristics
Baseline and follow-up demographic data are summarized in Table 1 and e-1. In individuals included in the baseline and follow-up studies, age, sex, and education were not different between the ε4- and ε4+ groups. The ε4+ group showed higher proportions of Aβ-positivity and clinical dementia and worse global cognition measured at baseline than the ε4- group. However, all demographic characteristics and global cognition were not different in each group stratified by Aβ-positivity. Compared to baseline, global cognition worsened at follow-up in both the ε4- and ε4+ groups. The number of ε4 carriers was greater in patients with dementia than that in CU and MCI patients.
Table 1. Baseline demographic characteristics of 272 participants who completed the baseline study
|
Aβ±
|
|
Aβ-
|
|
Aβ+
|
|
ε4-
|
ε4+
|
|
ε4-
|
ε4+
|
|
ε4-
|
ε4+
|
N
|
195
|
77
|
|
134
|
24
|
|
61
|
53
|
Baseline age (years)
|
70.4 ± 10.3
|
70.0 ± 8.6
|
|
68.4 ± 10.3
|
65.5 ± 8.3
|
|
74.7 ± 8.7
|
72.1 ± 7.9
|
Females (%)
|
127 (65%)
|
51 (66%)
|
|
90 (67%)
|
16 (67%)
|
|
37 (61%)
|
35 (66%)
|
Education (years)
|
11.1 ± 4.9
|
11.2 ± 5.0
|
|
11.0 ± 4.9
|
11.4 ± 4.2
|
|
11.2 ± 4.8
|
11.1 ± 5.3
|
Duration (years)
|
2.6 ± 1.5
|
3.1 ± 1.4
|
|
2.3 ± 1.5
|
2.4 ± 1.3
|
|
3.0 ± 1.5
|
3.2 ± 1.4
|
Aβ-positivity (%)
|
61 (31%)
|
53 (69%)a
|
|
0 (0%)
|
0 (0%)
|
|
61 (100%)
|
53 (100%)
|
ε2/2 : ε2/3 : ε3/3
|
1 : 37 : 157
|
n.a.
|
|
1 : 31 : 102
|
n.a.
|
|
0 : 6 : 55
|
n.a.
|
ε2/4 : ε3/4 : ε4/4
|
n.a.
|
2 : 60 : 15
|
|
n.a.
|
2:21:1
|
|
n.a.
|
0 : 39 : 14
|
Baseline diagnosis
CU/MCI/DEM (%)
|
79/75/41
(41/38/21%)
|
17/30/30a
(22/39/39%)
|
|
72/49/13
(54/37/10%)
|
15/7/2
(63/29/8%)
|
|
7/26/28
(11/43/46%)
|
2/23/28
(4/43/53%)
|
MMSE
|
25.4 ± 4.7
|
23.5 ± 5.3a
|
|
26.8 ± 3.2
|
26.7 ± 2.5
|
|
22.2 ± 5.7
|
22.1 ± 5.7
|
CDR-SB
|
1.6 ± 2.3
|
2.7 ± 2.5a
|
|
0.9 ± 1.5
|
0.8 ± 1.4
|
|
3.1 ± 3.0
|
3.6 ± 2.4
|
Data are presented as mean ± SD. aP < 0.05 for the comparisons between the ε4- and ε4+.
Abbreviations: CU = cognitively unimpaired, MCI = mild cognitive impairment, DEM = dementia, Aβ+/- = Aβ-positivity, ApoE = apolipoprotein-E, MMSE = Mini-Mental State Examination, CDR-SB = Clinical Dementia Rating sum-of-boxes
Baseline Aβ and tau burden
In all 272 individuals, the ApoE ε4+ group exhibited greater Aβ burden in the global cortex, prefrontal, parietal, lateral temporal, parahippocampal, and cingulate cortices and hippocampus than the ε4- group, and all regions survived correcting for multiple comparisons. Conversely, in Aβ+ individuals, Aβ burden was greater in the ε4- group in the global cortex, sensorimotor, superior parietal, occipital, and insula cortices than in the ε4+ group, although all regions did not survive correcting for multiple comparisons (Fig. 1A). Surface-based statistics showed similar results as VOI-based comparisons (Fig. 1B).
In all individuals, greater tau burden was observed in the ε4+ group in the lateral and medial temporal, cingulate, and insula cortices, and all regions survived multiple comparisons (Fig. 1A). In Aβ+ individuals, the ε4+ group showed greater tau burden in the medial temporal regions, and only the amygdala and hippocampus survived correcting for multiple comparisons. When the baseline global cortical Aβ burden was included as an additional covariate in the model, ApoE ε4+ group exhibited greater tau burden in all medial temporal regions when compared to ε4- group (Additional file 1: Fig. S1A). Surface-based statistics showed greater tau burden in the medial temporal and anterior cingulate regions in the ε4+ group than in the ε4- group, however, none of the regions survived after correcting for multiple comparisons (Fig. 1B).
Increased baseline Aβ burden in the hippocampus was associated with the number of ε4 alleles. Likewise, tau burden in the medial temporal regions showed an association with ε4 allele in a dose-dependent manner (Fig. 4A). In Aβ+ individuals, increased tau burden in the hippocampus and amygdala was associated with the ε4 allele in a dose-dependent manner (Fig. 4B).
Longitudinal changes in Aβ and tau burden
Examples of baseline Aβ and tau burden and their changes at follow-up are demonstrated in Fig. 2. In all 187 individuals, the ε4+ group exhibited a higher Aβ accumulation rate than the ε4- group in the global cortex, superior parietal, occipital, lateral temporal, parahippocampal cortices and amygdala, however, none of the regions survived correcting for multiple comparisons (Fig. 3A). A surface-based comparison showed a higher Aβ accumulation rate in the ε4+ group in the diffuse cortical areas than in the ε4- group, and small regions in the lateral temporal cortex survived correcting for multiple comparisons (Fig. 3B). In Aβ+ individuals, there was no difference in the Aβ accumulation rate between the two groups.
In all individuals, tau accumulation rate in the ε4+ group was higher in the global cortex, prefrontal, parietal, occipital, lateral and medial temporal, posterior cingulate, and insula cortices compared to the ε4- group. Except for the prefrontal, superior parietal, and posterior cingulate cortices, all regions survived correcting for multiple comparisons (Fig. 3A). In addition, the increase in tau accumulation rate is associated with the number of ApoE ε4 allele (Fig. 4B). Similar to the VOI-based results, a surface-based comparison of the annual increase in tau showed a higher tau accumulation rate, particularly in the diffuse parietotemporal cortex in the ε4+ group (Fig. 3B). In Aβ+ individuals, the ε4+ group exhibited a greater annual increase in tau burden in the middle temporal and hippocampus, although none of the regions survived correcting for multiple comparisons (Fig. 3A). Surface-based statistics also showed a higher tau accumulation rate in the small regions in the basal and lateral temporal and sensorimotor cortices (Fig. 3B).
Even after the inclusion of baseline global cortical Aβ burden as an additional covariate in the model, statistical results for the VOI-based comparison of tau accumulation rates between the two ApoE groups were almost similar (Additional file 1: Fig. S1B).