Characteristics of Alzheimer’s Disease and Mild Cognitive Impairment Influenced by the Time of Onset

Alzheimer’s disease (AD) is the most common neurocognitive disorder and is characterized by progressive declines in memory and cognition. Most individuals with AD develop symptoms after age 65, and the prevalence of AD is known to increase with age; aging is thus regarded as a major risk factor for AD. However, roughly 1–6% of AD patients present before age 65 [1]. Epidemiological studies have estimated the prevalence of early-onset dementia as approx. 48 to 76 per 100,000 individuals [2].

According to pathological research that focused on the subjects’ age at the onset of AD, early-onset AD (EOAD, which is defined as developing before the age of 65), and late-onset AD (LOAD, which develops after the age of 65) had the same histological characteristics [3, 4]. However, clinically, it has been pointed out that there are several differences between individuals with EOAD and those with LOAD, including differences in cognitive dysfunction. For example, cognitive dysfunctions including difficulty in speech, visuospatial dysfunction, and non-memory executive dysfunction are more prominent in EOAD [5, 6], whereas memory dysfunction is more prominent in LOAD [7]. Additionally, EOAD is associated with a faster progression of cognitive and functional declines compared to LOAD [6, 8, 9]. Recent neuroimaging studies have also shown that EOAD has stronger functional or structural changes than LOAD, particularly in the temporoparietal junction, posterior cingulate, and precuneus [8, 10]. On the other hand, it has been reported that changes in the medial temporal region are more pronounced in LOAD than in EOAD [11, 12].

Neuroimaging studies focusing on regional gray matter volume changes in early-onset mild cognitive impairment (EOMCI) revealed reductions in the entorhinal cortex, precuneus, and pars opercularis [13]. Differences in brain metabolic patterns between EOMCI and healthy subjects have also been evaluated. Compared to a healthy group, patients with EOMCI showed decreased metabolism in the posterior cingulate and parietal regions [14]. Regarding cognitive function, EOMCI patients had significantly higher scores on verbal recall and word fluency tests compared to patients with late-onset MCI (LOMCI), which occurs after age 65 [14]. Our literature searches have identified no studies that focused on patients with EOAD, LOAD, EOMCI, or LOMCI and evaluated them using a simple cognitive test, i.e., the Mini-Mental State Examination (MMSE).

In this study, we used magnetic resonance imaging (MRI) morphometry to measure brain morphology in individuals with EOAD, LOAD, EOMCI, or LOMCI and a healthy control (HC) group. We also verified the characteristic change patterns in the subjects’ cognition. EOAD and LOAD patients are known to have different clinical courses, symptoms, and neuroimaging findings, and we hypothesized that there are differences between EOMCI and LOMCI that are similar to the differences between EOAD and LOAD.

We first reviewed the cases of patients diagnosed with AD according to the US National Institute on Aging-Alzheimer’s Association criteria at our hospital [3]. Nineteen of the 46 patients developed AD before the age of 65 (early-onset). The diagnosis of mild cognitive impairment (MCI) was based on the following criteria: (1) meeting the diagnostic criteria for MCI published in the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5), (2) an MMSE score of 24–30 points, (3) clinical dementia rating of 0.5, and (4) the delayed free recall score for Story A in the Logical Memory II subtest of the Wechsler Memory Scale-Revised (WMS-R) is below the education-adjusted cutoff value [15, 16]. Total of 38 patients with MCI were enrolled in the study. Fifteen of the 38 patients developed MCI before the age of 65 years. For the HC group, we enrolled 23 cognitively intact, community-residing volunteers who responded to advertisements; all were ≥62 years old, and five were ≤65 years old.

All subjects underwent the MMSE for the evaluation of cognitive impairment and a 1.5 Tesla MRI scan. After each patient was informed in writing about the study, they were given the opportunity to withdraw from the study at any time. The study was explained to the healthy elderly in writing, and their written informed consent to participate in the study was obtained. This study was approved by the Ethics Committee of the University of Tsukuba Hospital, Japan (H29-314).

Patients underwent MRI as previously described [17]. For the regional gray matter volume analysis, the VBM8 toolbox (http://dbm.neuro.uni-jena.de/vbm/) was used to standardize individual three-dimensional (3D) T1 images. Each image was smoothed using a 12-mm full-width at half-maximum Gaussian kernel.

Differences in the MMSE subscales among the five diagnostic groups were analyzed by an analysis of covariance controlling for educational background and sex. We evaluated equality of the error variances by using Levene’s test of equality of error variances. The least significant difference post hoc tests were used for multiple comparisons. SPSS Statistics for Windows 23.0 software (SPSS Japan, Tokyo) was used for the statistical analysis.

First, differences in regional gray matter volume among the five groups were examined using a Statistical Parametric Mapping 12 multiple regression model while controlling for educational background and sex. Next, we compared the differences between healthy participants and each disease group, respectively. Statistical values were considered significant if the voxel-level threshold was p < 0.001 (uncorrected), and the cluster size threshold was p < 0.05 (after correction by family-wise error [FWE]) for multiple comparisons.

The subjects’ characteristics are summarized by group in Table 1. Some patients with AD could not complete the WMS-R test. Significant differences in age, MMSE total score, and age at onset were observed among the subjects in the five groups (HC, EOAD, LOAD, EOMCI, and LOMCI).

First, we evaluated equality of the error variances, and the all of significance values for MMSE subscale were less than 0.05, indicating that the equal variances assumption were violated. Then, we reanalyzed the group differences by non-parametric ANCOVA controlling for educational background and sex. Except for the naming score subitem, significant differences were observed among the five diagnostic groups in the subitems of the MMSE (Table 2). The comparison of the EOAD and LOAD groups’ results on the MMSE subitems revealed that the EOAD group had significantly lower scores for orientation for place than the LOAD group. In the comparison of the EOMCI and LOMCI data, the recall score in the LOMCI group was significantly lower. Our comparisons of the EOAD and LOAD groups and the EOMCI and LOMCI groups revealed no other significant differences in these results.

Regarding differences in gray matter volume between the LOAD and HC group, significant decreases were observed in diffuse areas. We thus performed the analysis again by changing the statistical power to a more rigorous level (voxel-level threshold p < 0.01 [FWE-corrected], cluster size threshold p < 0.005 [uncorrected]), and there were significant decreases in the bilateral temporal region, left insula, and orbital surface of the frontal lobe in the LOAD group compared to the HC group (Fig. 1a). In the EOAD patients, significant decreases were observed in diffuse areas at first defined level and observed in bilateral temporal regions, posterior cingulate gyrus, bilateral parietal regions, and right frontal region, compared to the HCs at a rigorous level (Fig. 1b). No significant volume difference was observed between the EOAD and LOAD groups. The LOMCI group showed significant decreases in gray matter volume in the bilateral medial, right temporal region, left insula, bilateral frontal orbital surface, and left prefrontal region compared to the healthy subjects at the first defined level (Fig. 2a). In addition, compared to the HCs, the patients with EOMCI exhibited significant decreases in the posterior cingulate gyrus, bilateral prefrontal cortex, left precuneus, bilateral prefrontal cortices, and left thalamus, although at a trend level (voxel-level threshold p < 0.02 [uncorrected], cluster size threshold p < 0.05 [uncorrected]) (Fig. 2b).

The LOAD group had significantly higher scores for place orientation compared to the EOAD group, and the EOMCI group had significantly higher scores than the LOMCI group for the “three-word delayed recall” task. Regarding gray matter volume, we observed that the hippocampal volume was maintained in the EOMCI and EOAD groups compared to the LOAD and LOMCI groups. To our knowledge, this is the first report to use the MMSE to assess different patterns of cognitive impairment in EOAD and LOAD and MCI.

As is well known, various cognitive dysfunctions gradually progress in AD. Among these, orientation for time is known to be impaired in AD from the early stage of the disease [18]. Our present analyses demonstrated that time disorientation in the LOAD group was significantly lower than in the healthy group, which is consistent with previous reports. However, our findings revealed that the EOAD group had declines in both time orientation and place orientation. Several research reports have reported that EOAD causes functional impairments other than memory, particularly visual-spatial cognitive impairment [5‒7]. We speculate that this visuospatial cognitive dysfunction may have caused the disorder in place orientation.

We also confirmed that the EOMCI group’s scores on the three-word delayed recall task were significantly higher than those of the LOMCI group. This result was consistent with research showing that memory impairment is not noticeable in EOAD [14]. These points suggested that to the differences between EOMCI and LOMCI have characteristics that are similar to the differences between EOAD and LOAD.

Prior neuroimaging research has indicated that (i) structural changes in EOAD are more widespread than those in LOAD, particularly in the temporoparietal junction, posterior cingulate cortex, and precuneus [8, 10], and (ii) structural changes in the medial temporal region are more pronounced in LOAD than in EOAD [11, 12]. Our present investigation demonstrated that the volume of bilateral medial temporal regions was preserved in the patients with early-onset cognitive impairment (EOAD+EOMCI). Although studies focusing on the regional gray matter volume in EOMCI have reported the preservation of medial temporal regions [13], other studies observed gray matter volume reductions in the medial temporal cortex, lateral temporal cortex, and frontal cortex of individuals with amyloid-positive EOMCI [19]. However, that study included EOMCI patients whose symptom onset occurred before the age of 65 but were ≥65 years old at the time of the study. Further replication studies with age-matched subjects are expected to clarify this point.

There are several limitations to this study. First, because the diagnostic groups were defined by the time of the onset of symptoms, age was not controlled for in the comparative study. However, another study indicated that the amygdala, hippocampus, and entorhinal cortex are not easily affected by age [20]. In our present study, although no atrophy of hippocampal volume was observed in the young-onset cognitive impairment group compared to the slightly older healthy elderly group, it is presumed that there is no need to consider the effects of age-related changes. Second, the sample size in the present investigation was relatively small (84 patients, 23 controls). In addition, apolipoprotein E (APOE) polymorphisms were not evaluated. The APOE genotype is known to influence the age of cognitive impairment onset and amyloid-β accumulation [21, 22]. Therefore, EOMCI with the APOE-4 allele is considered to have a stronger reduction in gray matter volume than EOMCI without the APOE-4 allele. Further studies with larger sample sizes and APOE genotyping are necessary to test our present findings. Third, MMSE is very convenient cognitive test. However, the MMSE has very limited sensitivity to detect subtle cognitive impairment and is not suitable to evaluate cognitive function quantitatively. In this study, we used the MMSE and showed the cognitive changes in early-onset dementia, which past cognitive studies had detected, but we did not estimate the severity of cognitive dysfunction in disorders. Based on the results of this preliminary study, it is necessary to conduct future studies using tests specifically designed to assess specific cognitive functions. Last, recent study about the diagnosis of AD stated that AD has to be defined by its unique neuropathologic findings, such as amyloid and tau deposition [23]. However, we diagnosed the participants by clinical symptoms. Further AD studies diagnosed by specific biomarkers might shed light on the differences between early-onset and late-onset dementia.

In conclusion, earlier research revealed that individuals with early-onset cognitive impairment exhibit milder memory impairment and more severe visuospatial cognitive dysfunction compared to individuals with late-onset cognitive impairment. These points are consistent with our present results. Additionally, neuroimaging studies that focused on EOAD detected atrophy of the neocortex, but not the hippocampus, compared to LOAD. Our present analyses revealed that the hippocampal volume was preserved in EOMCI and EOAD compared to LOAD and LOMCI. These findings suggest that the distinct pattern of differences in neuropsychological changes between EOAD and LOAD is common to MCI, which is at an intermediate level between healthy cognition and AD. Our findings may indicate diagnostic differences between early-onset and late-onset dementia at the clinical level.

After the study was explained to each patient, his or her written informed consent for participation was obtained. This study was approved by the Medical Ethics Committee of the University of Tsukuba Hospital, Japan (reference no. H29-314).

The authors have no conflicts of interest to report.

This study was not supported by any sponsor or funder.

H.S., M.O., A.K., and T.A. designed the study. M.O., A.K., Y.N., T.T., M.T., K.N., and T.A. acquired the data. H.S., M.O., and A.K. analyzed the data and wrote the article. K.N. and T.A. revised the article. All authors contributed to and have approved the final manuscript.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author (M.O.) upon reasonable request.