In this study, we explored functional and structural differences between patients with EF and RF using qEEG and GM volume. The qEEG analysis showed an increase in the theta power spectrum and a decrease in beta2 power in the EF group compared to the RF group. Additionally, the VBM analysis demonstrated decreased volumes in the left thalamus and the bilateral hippocampus in the EF but in the right frontal and temporal lobe in the RF group, when compared to the NCs. There was no statistically significant difference in GM volume reductions between the EF and the RF group.
The qEEG pattern observed in the EF group in our study was similar to the pattern observed for AD in numerous previous qEEG studies, which showed increased power in low frequency bands (delta and theta) and decreased power in high frequency bands (alpha and beta) [18–24]. A recent study suggested an increase in relative theta power as a first change in patients with AD . During the disease progression of AD, an early increase in theta and decrease in beta is followed by a decrease in alpha and an increase in delta power [25, 26]. Patients with MCI have also shown increases in theta power and decreases in alpha power when compared with normal elderly subjects [21, 27, 28]. Additionally, increased theta power and decreased parietal beta power may predict disease progression to AD in patients with MCI [29, 30]. Another study in non-demented and amyloid-positive subjects showed that higher delta and theta power were associated with clinical progression over time . The patients with amnestic MCI with EF in the present study showed increased theta and decreased beta power when compared with the RF group in our power spectrum and ROI source power analyses. Regarding differences between brain regions at the electrode and source level, we observed an increase in the low frequency band in the EF group in the frontal area, and a decrease in the beta band in the tempo-parietal as well as the frontal area. This is consistent with reported EEG patterns as a predictive factor of clinical progression to AD [28, 31] and may be associated with an anterior shift in band frequency source . Accordingly, disease progression to AD may be more easily predictable in patients with amnestic MCI with EF than in those with RF.
Additionally, the EEG connectivity pattern in patients with amnestic MCI with EF in our study seems similar to the AD pattern observed in previous studies. EEG coherence analyses in patients with AD showed a decrease in connectivity in all frequency bands, especially prominent in the alpha frequency band [21, 23, 24, 32, 33]. Furthermore, it has been shown that EEG coherence contributes to the discrimination of AD from normal aging  and progression to AD in patients with MCI . Recently, Musaeus et al. reported significant differences in coherence and iCoh in the theta and delta bands between groups with progressive and stable MCI, while higher theta coherence was associated with cognitive decline . In our present study, significant differences in iCoh between EF and RF were found in the theta and beta2 band. Patients with amnestic MCI with EF showed higher theta band connectivity in the frontal-frontal and frontal-temporal connections, but lower beta band connectivity in the temporal-cingulate connection, compared to those with RF. The higher theta coherence in patients with amnestic MCI with EF seems somewhat paradoxical under the hypothesis that EF rather than RF proceeds to AD. However, higher theta band coherence in MCI progressing to AD was also reported in a previous study , and one might speculate that this increased connectivity reflects a compensation effect in the early stage of AD. Additionally, our findings of increased theta connectivity in the anterior regions and decreased beta connectivity in the posterior regions in patients with amnestic MCI with EF may be related to the anterior shift of oscillation source in the progression to AD [28, 36, 37].
In our volumetric analysis, significant GM atrophy in the left thalamus and the bilateral hippocampi was observed in patients with amnestic MCI with EF, when compared to the NC group, whereas those with RF showed GM atrophy in the left thalamus, right superior frontal lobe, right superior temporal lobe, and right middle cingulum. These results may be explained by the fact that different brain regions are related to different memory functions. In earlier studies in animals and humans, the medial lobe structure, including the hippocampus, plays an important role in encoding new information and storing it in the neocortex [38, 39], while the dorsolateral prefrontal lobe is involved in retrieval through the frontal-subcortical neuronal circuit [40, 41]. Therefore, if amnestic MCI is divided into EF and RF, it can be assumed that the atrophy of the medial temporal lobe is more prominent in the group with EF, which is in accordance with our results. Additionally, these atrophic patterns including the volume of the hippocampus may be a good biomarker of AD, according to previous studies [42–45]. Although our VBM analysis demonstrated no significant differences in atrophic cortical regions between the EF and RF groups, when compared to the NCs, the pattern patients with EF showed was more similar to the brain atrophy observed in AD than that observed in patients with RF.