Mild Cognitive Impairment (MCI), characterised by cognitive changes that do not affect everyday functioning, can be a transitional stage in the continuum from normal aging to dementia, and it is frequently considered a predementia phase of Alzheimer’s Dementia (AD) [1]. People with MCI can have impairment in one or more cognitive domains [2]: they may have memory impairment (amnestic MCI – aMCI) in association or not with deficits in other cognitive domains (aMCI single or multi-domain), or they may have difficulties in one or more cognitive domains other than memory (non-amnestic MCI – naMCI single or multi-domain).

An extended cognitive assessment based on standardized tests is necessary to properly diagnose MCI, and it must explore all cognitive domains: language, verbal and visuo-spatial short- and long-term memory, attentional-executive functions, visuo-constructional, and visuo-spatial abilities. A deficit in one or more of these domains, along with preserved independence in daily life, allows the clinician to diagnose MCI in patients who have a cognitive complaint (self-reported or noticed by others) [3]. Despite the fact that the neuropsychological evaluation is an objective, performance-based method to assess cognition, there is great variability both in literature and in clinical practice in defining MCI: some clinicians and researchers use a cut-off of 1.5/2 standard deviations (SD) below the age and education corrected normative data to identify MCI patients, while others employ a less stringent cut-off of 1.0 SD [4]. The clinical and cognitive characterization of MCI needs to be integrated with neuroimaging biomarkers to establish the underlying pathology and track disease progression over time.

Neuroimaging methods such as volumetric Magnetic Resonance Imaging (MRI) and Diffusion Tensor Imaging (DTI) have been widely used to investigate how performance in cognitive tests relates to brain structure, i.e., to study brain-behaviour associations. MRI volumetry allows to quantify brain volume, while DTI is used to estimate the structural integrity of brain connections by measuring diffusion of water molecules through tissues. This latter technique provides measures such as fractional anisotropy (FA), an index of microstructural integrity of cerebral white matter (WM), mean diffusivity (MD), related to the amount of water in the extracellular space, and radial diffusivity (RD), which indicates axonal damage [5].

Over the past years, several studies have attempted to explore the brain correlates of cognitive tests specifically in patients with MCI. In MCI population, global cognition, most frequently assessed with the Mini-Mental State Examination (MMSE) [6], has been positively associated with the grey matter (GM) density of the right inferior temporal gyrus and medial temporal structures (i.e., hippocampus) [7, 8]. Performance in tests assessing language such as verbal fluency tasks positively correlated with the GM density of structures such as the right middle frontal gyrus and left putamen [7, 9]. Neuroanatomical correlates of verbal long-term memory highlighted the role of medial temporal structures (hippocampus, parahippocampal gyrus, and entorhinal and perirhinal cortices) [8, 10, 11, 12], as well as the role of the WM integrity of the medial temporal lobe, fornix, and uncinate fasciculus [13, 14, 15]. Performance in tests assessing visuo-spatial long-term memory positively correlated with GM density of the right hippocampus [11] and with WM integrity of the fornix (measured by reduced RD) [15]. Attentional-executive functions correlated with GM density and WM microstructural integrity of frontal (i.e., superior frontal gyrus, middle frontal gyrus, and lateral and medial orbital gyri) and parietal (i.e., anterior cingulum) areas [16].

Most of these previous studies have focused on late-onset MCI (i.e., on patients with symptoms onset after the age of 65). However, it is well known that the effect of ageing on brain anatomy is substantial and may challenge the identification of the neuroanatomical correlates of cognitive tests in older people. The aim of this study was to explore the neuroanatomical correlates of cognitive tests in a group of young-onset MCI, i.e., patients whose cognitive complaints commenced before the age of 65. We studied the neuroanatomical correlates of cognitive tests frequently used for the neuropsychological assessment of patients with MCI exploring the following cognitive domains: language, visuo-constructional ability, memory, relational reasoning, and attentional-executive functions.

One hundred and four consecutive eligible subjects (43 male and 61 female) were recruited between 2019 and 2020; Table 1 provides the descriptive statistics of their demographic and cognitive data.

The imaging results showing the structural neuroanatomical correlates of cognitive tests are summarized in Table 2 and displayed in Fig. 1 and Supplementary Fig. 1. Table 3 shows the corresponding t Statistic values and correlation coefficients related to the voxels in which the p value was maximum, along with their coordinates in MNI space. For all the explored tests, the correlations survived multiple comparison correction considering that we had explored two different imaging modalities (VBM and TBSS), except for the one between GM and WM and semantic fluency.

Fig. 1.

Correlations between imaging parameters and cognitive tests. On the left side of each column are displayed significant correlations between cognitive tests and grey matter density (in red-yellow positive correlations, in blue-light blue negative correlations). On the right side of each column are displayed significant correlations between cognitive tests and parameters of white matter integrity (in red-yellow positive correlations with fractional anisotropy and in blue-light blue negative correlations with mean diffusivity, except for the Stroop test, where blue-light blue represents negative correlations with fractional anisotropy, and the TMT-B, where red-yellow represent positive correlations with mean diffusivity). Results are displayed at p 0.05, FWE corrected for multiple comparisons using TFCE. MMSE, Mini-Mental State Examination; FAB, Frontal Assessment Battery; FCSRT, Free and Cued Selective Reminding Test; FWE, Family-wise error; RCPM, Raven’s Coloured Progressive Matrices; ROCF, Rey-Osterrieth Complex Figure; TMT, Trail Making Test.

We investigated the structural neuroanatomical correlates of cognitive tests in a group of young-onset MCI (i.e., onset of cognitive complaints before the age of 65) patients, to minimise the effect of ageing on the study of brain-behaviour associations.

The identified associations between cognitive scores and focal grey matter atrophy were in line with the lateralization of verbal and visuo-spatial abilities [43]. In fact, the grey matter correlates of tests such as verbal fluency (i.e., phonemic, semantic, and alternate fluency) and verbal long-term memory (i.e., FCSRT) were mostly located in structures of the left hemisphere, whereas the correlates of tests of visuo-constructional ability (i.e., ROCF – copy), visuo-spatial long-term memory (i.e., Corsi Supraspan), and non-verbal relational reasoning (i.e., RCPM) were more lateralized in the right hemisphere.

The neuroanatomical correlates of the MMSE were sparse in the brain’s grey matter, highlighting the fact that this is a test meant to examine more than one cognitive domain. Interestingly, in our sample the variability in the MMSE mostly related to the atrophy of occipital and temporal areas, which had been shown before, albeit not exclusively [7]. The lack of correlates in frontal areas confirms that the MMSE has low sensibility in revealing frontal dysfunction [44], and this is especially evident in our group in which we controlled for the effect of aging by focusing on young-onset sample and including age among the covariates of no interest [45].

Grey matter neuroanatomical correlates of language ability tests especially involved subcortical structures, in line with previous literature [9], confirming the role of basal ganglia in language tasks [46].

Performance in the FCSRT, which measures verbal long-term memory ability, focally correlated with the integrity of the posterior part of the left hippocampus, in line with previous studies exploring the correlates of other auditory verbal learning tests in older MCI subjects [8, 10], as well as in healthy subjects [47]. The fact that in our young-onset MCI group the correlation between the immediate recall of the FCSRT and the grey matter density precisely delineated the posterior part of the left hippocampus is remarkable and novel. It confirms the role of the posterior part of hippocampus in the registration process [48, 49], and provides further evidence for dissociable memory networks along the anterior–posterior axis of the hippocampus [50]. Notably, there were no white matter correlates of the FCSRT.

Another test that had focal grey matter correlates only was the ROCF, whose variability was significantly associated with grey matter density of the right temporo-parietal junction/angular gyrus, in line with the known involvement of this structure in visuo-spatial abilities.

In contrast to a previous study [11] which found few neuroanatomical structural correlates of visuo-spatial long-term memory limited to the right hippocampus, our study highlighted the involvement of several cortical and subcortical structures in addition to the right hippocampus in this cognitive domain, still mainly localized in the right hemisphere. This suggests that the test we used (i.e., Corsi Supraspan) is complex and involves processes related to movement monitoring and visual perception, and not only long-term memory abilities.

Finally, the associations that we found between grey matter density and attentional-executive functions confirmed a prominent role of fronto-temporo-parietal areas [16]. Of notice, the correlates of the TMT-B also emphasised the involvement of occipital regions given that this is a visual attention task. The correlates of the FAB, which tests verbal abstraction abilities, Luria sequence learning, Go/No-Go, and other non-visual tasks, were mainly located in bilateral dorso-lateral and peri-insular prefrontal cortex.

In our sample, the white matter neuroanatomical correlates were widespread rather than focal and involved both the hemispheres: for almost all cognitive domains, we found a relationship between the performance and several white matter tract fibers, including the superior longitudinal fasciculus, inferior longitudinal fasciculus, uncinate fasciculus, inferior fronto-occipital fasciculus, forceps minor and major, cingulum, corpus callosum, and anterior thalamic radiation. Interestingly, there were differences in the neuroanatomical correlates of selective and divided attention: selective attention investigated with the Cancellation Test was more related to grey matter atrophy, while divided attention examined by TMT-B was also related to white matter integrity. Working-memory too, both verbal and visuo-spatial, was more related to white matter integrity than to grey matter density, whereas visuo-constructional ability and verbal long-term memory were related to grey matter atrophy only.

The major limitation of this work is that we did not have a group of healthy controls, nor did we have a group of elderly MCIs to compare neuroanatomical correlates of our young-onset group. A further possible limitation of the study is that some of the included MCI subjects were older than 65 at the moment of recruitment, but the fact that they had developed symptoms before the age of 65 allowed us to still consider them as subjects with young-onset MCI. Despite of these limitations, this is the first study focused on young-onset MCI which investigates the neuroanatomical correlates of the different cognitive tests included in an extended neuropsychological battery simultaneously and could provide new information about such an early stage of cognitive impairment.

In our young-onset MCI group, we found that the neuroanatomical correlates of tests of verbal and visuo-spatial abilities reflected the well-known lateralization in the left and right hemispheres, respectively. These linear correlates were cleaner and more focal compared to previous studies in older MCI group, possibly reflecting the influence of ageing on these correlations. This is especially true for the FCSRT testing verbal memory, which was selectively associated with atrophy of the posterior left hippocampus, and for the copy of ROCF, testing visuo-constructional ability, which was selectively associated with atrophy of the right temporo-parietal junction. Most of the other tests were also linearly correlated with the variability of white matter microstructural integrity, and such associations involved several white matter tracts in both hemispheres. These correlations were more widespread and less focal than in previous studies conducted in older MCI subjects, suggesting that white matter integrity may have a role in the development of MCI especially in its young onset.

We showed that most of the cognitive tests (except the FCSRT and ROCF) involve several underlying processes that rely on the functioning of grey matter areas as well as on the integrity of several white matter tracts. This furthers a view of brain functioning based on networks and connections rather than on the functioning of focal cortical areas in isolation [5].

The data sets generated and/or analyzed during the current study are not publicly available due to regulatory requirements but are available from the corresponding author on reasonable request.

CC, FG, MAM, AM, AC and GZ designed the research study. CC, DB, GV, CG, DS, SS and MT performed the research. EB provided help with FSL software. CC and EB analyzed the data. CC, AC and GZ wrote the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The study was approved by the Local Ethics committee (832/2018/SPER/AOUMO) and all patients signed the informant consent.

We are grateful to our patients and their families.

The study was supported by the grant Dipartimenti di Eccellenza 2018-2022, MIUR, Italy, to the Department of Biomedical, Metabolic and Neural Sciences, University of Modena. The study was also funded by the European Union ERC, UnaWireD, project number 101042625. However, views and opinions expressed are those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council Executive Agency. Neither the European Union nor the granting authority can be held responsible for them.

The authors declare no conflict of interest. Giovanna Zamboni is the member of Editorial Board of this journal. Also, Giovanna Zamboni is serving as one of the Guest editors of this journal. We declare that Giovanna Zamboni had no involvement in the peer review of this article and has no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to Gernot Riedel.