In this study, the alterations in ERC morphological indexes (including thickness, volume and surface area) were first assessed among AD patients, aMCI-m, aMCI-s and HC participants. The ERC thickness, rather than the ERC volume and surface area showed a significant tendency in the conversion from aMCI to AD. Then, the AUC results demonstrated that combining ERC thickness and volume could better discriminate the four groups from each other than single ERC index alone. Furthermore, relative to the hippocampal volume, combination of ERC thickness and volume had better discriminating capacity between HC and aMCI-s, as well as aMCI-m and AD.
The ERC atrophy has been regarded as early potential biomarker in patients with MCI and AD[10, 26–30]. Our findings suggested that ERC thickness showed more significant changes than ERC surface area among AD, aMCI-m and aMCI-s, which was consistent with a previous study that showed AD appearing to have different effects on thickness and surface area[13]. To be more precise, both hemispheres of ERC thickness showed a decrease trend from HC to aMCI-s to aMCI-m to AD. These results add to the evidence that aMCI-m is more likely a transitional stage between aMCI-s and AD[5, 6, 17, 31]. More importantly, our AUC results further proved that ERC thickness had superiority over ERC surface area and volume in discriminating among the four groups. So, on the basis of previous studies, our study found that aMCI-s and aMCI-m showed different magnitudes of decreased cortical thickness in the bilateral ERC relative to the HC group. In that, ERC thickness may serve as a potential diagnosis index in patients with aMCI.
ERC thinning is sensitive to the early pathological process of AD, which may due to its own neurophysiological mechanism. In one aspect, early structural changes in AD are limited to specific laminae within ERC (layer II is particularly vulnerable)[32]. In the other aspect, the thickness of the cerebral cortex was calculated as the average distance between the gray/white boundary and the pial surface[33, 34]. It likely represents cytoarchitectural features or many components of the neuropil, such as intra-cortical axons, dendrites, synaptic elements, and glia, not limited to neuron numbers[13, 34–36]. It was reported that AD-related pathological alterations first resulted in synaptic neurodegeneration and then neuronal loss[37]. In addition, no significant neuronal loss in ERC was detectable in cognitively normal participants, while a very severe neuronal loss was seen in ERC in the very mild AD cases[38]. On account of above reasons, the ERC thickness, rather than the ERC surface area and volume appeared significant change even in the stage of aMCI-s.
Our results also revealed that ERC surface area was minimally affected in the conversion from aMCI to AD. To our surprise, aMCI-s group even showed an increase trend. This may be explained by compensation for ERC thinning[38]. The increase of ERC surface area autonomously compensates for ERC thinning in patients at earliest preclinical stage (e.g., aMCI-s), whereas the absence of such compensation mechanism in patients at late clinical stage (e.g., AD). However, the neurophysiological correlates of cortical surface area are less clear. The ERC surface area may relate to local subcortical factors, such as subjacent white matter volume or global factors, such as the head size[13]. So, regionally analyses of cortical surface area must take into account the global effects of head size and brain size[13, 34]. Previous work has shown that aging was related with reduced surface area, rather than AD[13, 39]. In line with this, we found that ERC surface area was relatively unchanged in aMCI and AD after adjusting for head size.
By definition, ERC volume was a product of thickness and surface area[13, 33, 39]. The ERC volumetric decrease was the result of a combination of the ERC thinning and ERC surface area change. No significant change in ERC surface area, even a slight increase in aMCI-s may weaken ERC volumetric reduction. That may explain why ERC volume atrophy was not significant in subtypes of aMCI.
As discussed above, the alteration of different ERC morphological indexs varied among the four groups. Combining multiple ERC indexs (e.g., volume, thickness, and surface area) may provide a complete understanding of progressive structural brain changes during the conversion of aMCI to AD. Moreover, the different morphological features had unique contributions to the classification of aMCI patients and healthy controls[34]. Thus, multi-parametric indexes may have the ability to detect subtle alteration in the progression of AD. The multivariate method, which combined certain indexes together, allows us to determine the relationships among different features beyond their individual values. Consistent with prior studies, the AUC results showed that the combination of ERC thickness and volume further improved discrimination among the four groups.
Memory impairment is the earliest and most prominent symptom of aMCI and AD[40–42]. Since the medial temporal lobe structures (ERC and HP) are specialized for memory functions, alteration of the medial temporal lobe, especially hippocampal volume loss, has been considered to be a key feature for early diagnosis of AD[43–45]. However, the ERC atrophy may be more closely associated with the pathologic processes of AD than HP atrophy[38]. Thus, the ERC atrophy could have an advantage over HP atrophy in discriminating among HC, subtypes of aMCI and AD.
The AUC results verified the above-mentioned assumption and showed that combination of ERC thickness and volume had a superior differential power than hippocampal volume for discriminating between HC and aMCI-s. The results also coincided with pathologic study stating that the pathology of AD starts in ERC, providing in vivo evidence for the Braak stages (Stages1 and 2 represent the entorhinal phase of the disease with minimal involvement of the hippocampus)[46]. In addition, combination of ERC thickness and volume had a better discriminating capacity than hippocampal volume between aMCI-m and AD. According to a longitudinal MRI study, atrophy rates in AD were significantly higher for ERC than for HP[29]. Considering that ERC was affected earlier and had higher atrophy rate than HP in AD[28, 29], the combination of ERC multiple morphometric indexs should reflect more comprehensive information during the process from aMCI to AD. Therefore, it is quite understandable that combination of ERC thickness and volume had an advantage regarding early diagnosis of aMCI and predicting conversion from aMCI to AD.
There were still some limitations in this study. First, although we use education level as a covariate for covariance analysis to reduce its impact on ERC evaluation, it is not as convincing as choosing subjects with similar education level, but because our study is based on real clinical data, there is no perfect control obtained. Second, although the present study revealed an ERC thickness decreasing tendency from HC to AD, however, this trend needs to be further confirmed by longitudinal studies.