The two DLB groups were similar regarding clinical and demographical features, except age at onset (p<0.001), by definition (see Table 1). DLB groups were comparable for frequency of neuropsychiatric symptoms, namely depression (p=0.25), apathy (p=0.29), irritability (p=0.99), and anxiety (p=0.92). CSF p-tau load resulted higher in the LO-DLB than EO-DLB group (p=0.03). REM sleep behavior disorder was more frequent in the LO-DLB group (49% vs 21%, p=0.07).
Table 1- Clinical, CSF and demographical features in the two studied groups at the time of PET exam.
|
EO-DLB
|
LO-DLB
|
p-value
|
N
|
21
|
41
|
|
Age (years)
|
59.1 ± 3.8
|
74.8 ± 5.9
|
<0.001
|
Sex (female/male)
|
8/13
|
13/28
|
0.62
|
Disease Duration (years)
|
2.7± 1.7
|
2.3 ± 1.4
|
0.45
|
Education (years)
|
10.6 ± 3.3
|
10.7 ± 4.8
|
0.95
|
MMSE, total
|
19.8 ± 4.9
|
19.1 ± 4.8
|
0.65
|
CSF Ab 1-42 (z-score) †
|
-0.8±1.3
|
0.04±0.8
|
0.36
|
CSF total-Tau (z-score) †
|
-0.2±0.8
|
0.08±1.1
|
0.63
|
CSF total-Tau/Ab 1-42 (z-score) †
|
0.42±1.4
|
-0.08±0.9
|
0.44
|
CSF phosphorylated-Tau (z-score) †
|
-0.5±0.7
|
0.2±1.1
|
0.03
|
CSF phosphorylated-Tau/Ab 1-42 (z-score) †
|
-0.03±0.5
|
0.01±1.09
|
0.95
|
Neuropsychiatric Features, Number and (%)°
|
|
|
|
Depression
|
7 (54%)
|
12 (35%)
|
0.25
|
Apathy
|
8 (62%)
|
15 (44%)
|
0.29
|
Irritability
|
5 (39%)
|
13 (38%)
|
0.99
|
Anxiety
|
4 (31%)
|
12 (32%)
|
0.92
|
Core DLB criteria, Number (%)
|
|
|
|
Parkinsonism
|
13 (87%)
|
36 (88%)
|
0.91
|
Fluctuating Cognition
|
7 (50%)
|
20 (49%)
|
0.94
|
Visual Hallucinations
|
7 (50%)
|
26 (63%)
|
0.38
|
REM sleep behavior disorder
|
3 (21%)
|
20 (49%)
|
0.07
|
† CSF z-scores are calculated based on the distribution of the Milan and Geneva DLB groups. ° Data available for 47 out of 62 DLB participants. MMSE= Mini-Mental State Examination; CSF= cerebrospinal fluid; DLB = dementia with Lewy Bodies. Numbers in bold indicate significant results.
Group-analysis
SPM voxel-wise analysis showed a significant hypometabolism in middle and superior occipital lobe, angular gyrus, supramarginal gyrus, precuneus, inferior frontal gyrus and middle frontal gyrus, bilaterally (Figure 1). Regarding direct comparison of DLB groups, we found no differences in terms of brain hypometabolism in the DLB core regions (Table 2). However, differences emerged in extra-DLB typical regions. Namely, the EO-DLB group showed a more severe hypometabolism in the superior parietal cortex and PCC. Instead, the LO-DLB group presented a significant hypometabolism in the insula, supplementary motor area, and parahippocampal gyrus when compared to the EO-DLB group (Figure 1 and Table 2).
Table 2 – Comparisons of brain hypometabolism (Z scores) between EO-DLB and LO-DLB groups
|
EO-DLB
|
LO-DLB
|
p-value
|
ROIs
|
|
|
|
Angular Gyrus
|
2.25±1.36
|
2.02±0.9
|
0.49
|
Calcarine Cortex
|
0.82±0.77
|
0.94±0.58
|
0.48
|
Middle Frontal Gyrus
|
0.51±0.91
|
0.53±0.81
|
0.95
|
Fusiform Gyrus
|
-0.37±0.59
|
-0.16±0.65
|
0.23
|
Superior Temporal Gyrus
|
0.006±0.96
|
0.37±82
|
0.12
|
Inferior Temporal Gyrus
|
0.28±0.82
|
0.32±0.71
|
0.84
|
Lingual Gyrus
|
0.49±0.75
|
0.57±0.60
|
0.66
|
Inferior Occipital Lobe
|
1.44±0.68
|
1.29±0.89
|
0.52
|
Middle Occipital Lobe
|
1.66±0.88
|
1.45±0.82
|
0.35
|
Superior Occipital lobe
|
1.16±0.71
|
0.91±0.78
|
0.12
|
Superior Parietal Cortex
|
1.22±0.91
|
0.84±0.86
|
0.04
|
Posterior Cingulate Cortex
|
0.81±0.98
|
0.76±0.72
|
0.03
|
Insula
|
-1.14±0.87
|
-0.63±0.64
|
0.01
|
Supplementary Motor Area
|
-0.23±0.73
|
-0.12±0.72
|
0.03
|
Parahippocampal Gyrus
|
-1.47±0.83
|
-0.97±0.78
|
0.01
|
CIS
|
0.43±0.03
|
0.42±0.03
|
0.30
|
CIS = cingulate island sign; EO-DLB= early-onset dementia with Lewy bodies; LO-DLB= late-onset dementia with Lewy bodies. Significant p-values are reported in bold.
Metabolic connectivity analysis
Overall, comparable whole-brain connectivity alterations were found in the two clinical groups as compared to HC (Figure 2). We showed decreases and increase in connectivity, the latter was more evident in the posterior regions of the EO-DLB group. Some discrepancies were also evident in the pathological nodes. Namely, the EO-DLB showed pathological nodes in precuneus, inferior and superior parietal cortex, angular gyrus, inferior temporal cortex and middle occipital cortex, the LO-DLB, in the lingual gyrus, inferior occipital cortex, fusiform, middle temporal cortex, and cerebellar vermis.
The RSN metabolic connectivity in EO-DLB, LO-DLB and HC as reference, revealed: i) for the EO-DLB, reduced connectivity in the frontal components of the EXN, aDMN and pDMN and in the LIN, a loss of connectivity in thalamus and pons and an increased connectivity between ventral putamen and amygdala; ii) for the LO-DLB, a loss of connectivity in the ATN, involving the dorsolateral prefrontal cortex (Figure 3).
Spatial Distribution of Neurotransmitter Systems
The hypometabolism z-score maps were spatially correlated with AChT (r=-0.368, p <0.001), GABAa (r=0.331, p=0.004), SERT (r=-0.256, p<0.001), and DAT (r=-0.220, p<0.001) neurotransmitter maps in the whole DLB group.
As for the EO-DLB group, we found a significant association between hypometabolism z-score maps and the distribution of AChT (r=-0.414, p <0.001), GABAa (r=0.348, p=0.005), SERT (r=-0.326, p<0.001), DAT (r=-0.292, p<0.001) and 5HT4 (r=-0.133, p=0.04) maps.
As for the LO-DLB group, hypometabolism z-score maps were significantly associated with the distribution of AChT (r=-0.345, p<0.001), GABAa (r=0.333, p=0.004), SERT (r=-0.22, p<0.001), and DAT (r=-184, p=0.002) maps (Figure 4).
The resulting negative correlation coefficients indicate that the hypometabolism characterizing each group was present in structures specific for each neurotransmitter.
The post-hoc direct comparison between EO and LO-DLB Fisher’s z scores showed differences only in the SERT map (p=0.029), and the largest hypometabolism cluster driving the relationship between hypometabolism and SERT neurotransmitter maps was localized in the in PCC [MNI XYZ coordinates= 8 -66 42, T=7.45, FWE-corrected p<0.001] (Figure 4).