Autobiographical memory (AM) refers to a form of declarative, long-term memory comprising knowledge of a person's own past. Changes in AM are typically observed in healthy aging, e.g., by a reduction of specificity and temporo-spatial or perceptual details (1). However, pronounced losses of AM, including important events of one's own past, form one of the clinical hallmarks of Alzheimer's disease (AD) (2) and may affect central components of the subjective sense of identity (3), which illustrates the particular importance of AM. These deficits can already be demonstrated in mild cognitive impairment (MCI), which is generally considered as the potential preclinical state of the condition and progressively makes AM inaccessible (4).

As part of declarative memory, AM comprises semantic facts and episodic events. Semantic AM contains general facts from different periods of lifetime, such as names or addresses, while episodic AM includes singular events that can be recalled with a richness of details and a feeling of re-experience when recalled (5, 6).

Neuroimaging studies identified numerous cerebral sites, including the frontal, temporal, and posterior regions, to be involved in declarative AM. As reviewed by Svoboda et al. (7) in their seminal work, this heterogeneity of brain regions may refer to a network including the medial and ventrolateral prefrontal cortices, medial and lateral temporal cortices, temporoparietal junction, retrosplenial/posterior cortices, and cerebellar structures. This heterogeneity of cerebral sites may support the different aspects of AM such as encoding/retrieval processes, emotional connotation, or semantic vs. episodic aspects. However, the 26 superordinate brain regions reviewed were derived from 24 positron emission tomography (PET) or functional MRI (fMRI) studies, but not examined in a single study (7).

In the present study, we sought to investigate these cerebral sites important for AM described by Svoboda et al. (7) for underlying subdimensions or cerebral networks. Following previous studies by our group (8, 9), this was done by calculating a factor analysis of the glucose metabolic rates obtained by PET in 129 subjects from a memory clinic. The factors identified were contrasted between patients with MCI and AD and the cognitively intact controls (cognitively healthy, HC) and subsequently correlated with clinical AM performance.

The present study yielded three major findings: (i) support for the hypothesis that changes in the networks of cerebral sites, including the medial and ventrolateral prefrontal cortices, medial and lateral temporal cortices, temporoparietal junction, retrosplenial/posterior cortices, and cerebellar structures, contribute to decline in MCI and early AD; (ii) evidence that these changes may also involve compensatory efforts in patients with MCI; and (iii) an indication that both semantic and episodic AM deficits refer to changes in the mesial temporal substructures.

Cortical activity obtained in a resting state was segregated into seven dimensions of cortical activation, with lower values on the factors frontal cortex, mesial temporal substructures, and occipital cortex, differentiating patients with MCI and early AD patients from cognitively intact control subjects. This applies to all of the three factors mentioned above, although changes of the occipital cortices were less frequently reported (9). According to Svoboda et al. (7), the frontal cortices play an important role in conscious experience of re-experiencing, memory reconstruction, and self-referential processing. Mesial temporal substructures, in particular the hippocampi, are important for the encoding and retrieval of “laboratory” episodic memories (27), while the occipital cortex is important for visuospatial processing and visual imagery, but seems to play a merely marginal role for AM retrieval (7). However, significant group differences also involved increased values on the factors posterior cingulum and left temporo-prefrontal areas in patients with MCI when contrasted with controls and both AD patients and controls, respectively. A decreased glucose uptake is generally considered as an early and robust indicator of functional deficits and atrophy processes in the respective brain areas. In contrast, an increased value may indicate compensatory efforts due to an unspecific and less efficient overactivation, which was observed in both FDG-PET and fMRI studies. Further evidence supporting this hypothesis comes from studies in healthy subjects that yielded an economization of cerebral activation under training of cognitive tasks (28, 29). As emphasized by Matura et al. (27), the medial prefrontal cortex and the posterior cingulate cortex support self-referential processing as a defining component of AM; hence, most neuroimaging studies on AM yielded an activation during AM retrieval. One may hypothesize that the patients investigated in the present study were still capable of recalling semantic and episodic facts due to an overactivation of the respective areas. An activation of the occipital cortices was reported in studies contrasting recent relative to remote autobiographical events (7). Along with this, a decreased glucose uptake was reported in a number of studies, including our own trial, with significantly higher values in “good performers” of a serial verbal learning task among both patients and controls.

Besides this, both semantic and episodic AM deficits were significantly correlated with changes in the mesial temporal substructures, but not to other factors. The role of the hippocampi, gyri parahippocampi, and the perirhinal and entorhinal cortices as core contributors to AM was emphasized by Svoboda et al. (7); atrophic changes of the respective structure are among the best-established findings in AD (30, 31). That both domains of AM were highly significantly associated with the mesial temporal substructures conforms with the view that both are highly interrelated and do not constitute independent forms of AM. This finding indicates a fundamental, at least gradual, overlap between SAM and EAM [see, for example, (32)] and confirms the hypothesis that EAM is integrated into a higher-level external framework of these general knowledge structures (33).

While we included data from 129 patients who were subsequently recruited in a large memory clinic, AM was only obtained from 22 patients. The sample size of 22 subjects in the AM analysis increased the risk of type II errors. However, the core significant findings of our study are consistent with those in the review of Svoboda et al. (7), indicating that our sample size was sufficient for large effects. In their review, Svoboda and her co-workers (7) considered findings from both fMRI and PET studies. Systematic effects of the imaging method used were not reported. From a clinical standpoint, FDG-PET is well-established in the diagnostic workup of MCI and mild AD as it provides greater sensitivity in the early recognition of the cerebral changes in MCI and mild AD (34, 35) since functional deficits may precede morphological changes but already involve decreased glucose metabolic rates. FDG-PET provides a measure of network indicating which structures are correlated within subjects across the group. With PET, we demonstrate that parts of the frontal cortex are more similar to each other in functional activity in healthy subjects than in patients, and this is the concept of the network evaluated. With fMRI, additional measures of within-subject correlations over time can be demonstrated, offering an alternative but not entirely dissimilar assessment of a network. Taken together, our preliminary findings indicate that a similar network of cerebral sites important for AM in otherwise healthy and young controls also applies to patients with MCI and mild AD. Changes within this network include both activity decrease and activity increase, which may refer to compensatory efforts. Both episodic and semantic AM were significantly correlated with changes in the mesial temporal substructures. This confirms the hypothesis (33) that personal semantic knowledge constitutes a framework for episodic AM; hence, both referred to the same structure.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the ethics committee of the medical faculty of Heidelberg University. The patients/participants provided their written informed consent to participate in this study.

CF, UH, MB, and JS: substantial contributions to the conception or design of the work or the acquisition and analysis or interpretation of data for the work. CF, JS, JP, CD, MB, and CH: drafting the work or revising it critically for important intellectual content. All authors provide approval for publication of the content and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

This research was funded by the Deutsche Forschungsgemeinschaft (471/5-1).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.