Abnormal FDG-PET Patterns and Amyloid Positivity Predict Personality Changes in Cognitively Preserved Individuals

1. Introduction

Personality, long regarded as a relatively stable psychological structure, has recently attracted renewed interest in aging neuroscience due to its sensitivity to biological and cognitive changes. The Five-Factor Model (Big Five), empirically validated across many cultures, describes the personality construct according to five fundamental dimensions: neuroticism, extraversion, openness, agreeableness, and conscientiousness [1]. Recent longitudinal studies have shown that even in the absence of cognitive deficits, levels of extraversion, openness, and conscientiousness tend to decline with age, while neuroticism follows a slightly upward trajectory [2,3]. These changes are not solely explained by psychosocial adjustments but could also reflect the evolution of brain aging processes. A growing body of evidence has revealed significant associations between certain personality traits and cognitive decline in normal aging. High levels of neuroticism and low levels of conscientiousness are associated with increased cognitive vulnerability, while extraversion and openness may offer some degree of protection [4,5,6,7]. A recent systematic review by Troisi and colleagues [8] also supported the idea that higher levels of conscientiousness and openness may limit the age-related cognitive decrement in healthy controls. Lower agreeableness at baseline was related to increased risk for cognitive decrement but this association was less robust [9].

The relationship between changes in personality assessed with the Five-Factor Model and subsequent development of cognitive impairment in cognitively preserved elderly persons is still disputed. In the Baltimore Longitudinal Study of Aging, no evidence for preclinical change in personality before the onset of mild cognitive impairment or dementia was identified [10]. In another longitudinal study of a genetically enriched cohort of healthy controls, Caselli and collaborators [11] found that an increase of neuroticism and decrease of openness levels characterize cases that transit to mild cognitive impairment (MCI).

Some neuroimaging studies have examined how personality traits may relate to Alzheimer disease (AD) and vascular signature biomarkers, such as loss of brain volumes, cerebrospinal fluid (CSF) Aß and tau levels, positron emission tomography (PET) amyloid deposits and tau pathology, cerebral glucose metabolism, as well as white matter hyperintensities (WMH). In early studies, higher neuroticism was associated with smaller regional volumes and greater decreases in volume with increasing age in non-demented individuals. In contrast, higher conscientiousness was related to larger regional volumes and less decline with advancing age [12]. The positive role of high levels of conscientiousness and openness on AD-signature magnetic resonance imaging (MRI) quantitative markers was also detected in a more recent study of our group [13]. In the same line, increasing neuroticism and decreasing extraversion, openness and conscientiousness were associated with lower cerebrospinal fluid (CSF) Aβ1-42 but not with tau and ptau-181 concentrations in mixed cohorts [14]. Several cross-sectional investigations focused on the associations between NEO-PI factors and PET amyloid and tau burden as well as fluorodeoxyglucose (FDG)-PET metabolic patterns prior to cognitive decline. Higher levels of extraversion and conscientiousness were related to lower accumulation of brain beta-amyloid deposition in cognitively normal elderly individuals [15,16]. In the PREVENT-AD cohort including non-demented individuals with a family burden of AD, lower neuroticism and higher measures of openness and extraversion were related to lower PET amyloid and tau burden [17]. Increased neuroticism was also associated with pathological tau accumulation in regions known to be vulnerable to AD pathophysiology in cognitively normal individuals [18,19]. Consistently, higher neuroticism and lower scores on extraversion and conscientiousness were significantly associated with decreased glucose metabolism in brain regions typically affected by AD neuropathological processes in non-demented individuals [20]. However, negative or ambiguous data were also reported for both amyloid burden and WMH [21,22].

Besides their cross-sectional design, there are two major limitations of these studies. First, personality factors were considered as stable constructs in the course of brain aging. However, several lines of evidence indicate that NEO-PI factor changes occur in the preclinical phases of AD [11,23]. Studying the association between AD imaging markers at baseline and subsequent changes in personality factors may provide arguments related to the reverse causality hypothesis (i.e. personality changes could be the consequence of incipient AD and cerebrovascular pathology). Second, most of the previous studies used quantitative PET markers that are rarely used in busy, routine clinical settings. Our study adopts a 4.5-year longitudinal design focusing on changes in personality traits, assessed by the NEO-PI-R—an established instrument for evaluating the five major personality factors—among older adults without any cognitive impairment at baseline. This assessment is paired with multimodal neuroimaging data, including amyloid PET, FDG-PET, and structural MRI scans. Amyloid positivity and abnormal PET patterns were assessed visually by fully trained specialists blind to cognitive data in order to be close to a real-life situation. The main objective was to explore whether changes in personality traits upon follow-up may be related to the early presence of AD-related abnormal PET and MRI patterns and subtle cognitive changes that may be detected in the course of normal aging. We hypothesize that subsequent changes in personality traits, and in particular neuroticism, openness and agreeableness are associated with greater amyloid burden, cerebral hypometabolism, increased cerebrovascular burden and medial temporal atrophy (MTA) at baseline.

2. Materials and Methods

2.1. Participants

The study was received the approval of the local Ethics Committee, and all participants gave written informed consent prior to inclusion. Individuals were selected from an ongoing cohort study on cognitively intact elders, as described in detail previously [24,25,27]. Cases with three neurocognitive assessments at baseline, 18 months and 55 months, brain amyloid and FDG-PET as well as 3T MRI at baseline and APOE status were considered. Exclusion criteria included psychiatric or neurologic disorders, sustained head injury, history of major medical disorders (neoplasm or cardiac illness], alcohol or drug abuse, regular use of neuroleptics, antidepressants or psychostimulants and contraindications to PET or MRI. To control for the confounding effect of acute vascular events on MRI findings, individuals with subtle cardiovascular symptoms, hypertension (non-treated) and a history of stroke or transient ischemic episodes were also excluded from the present study. The final sample included 58 cognitively intact controls: 34 women and 24 men, mean age: 72.8 ± 3.6 (mean ± SD) ranging from 69 to 87 years.

2.2. Personality Assessment

Personality features and dimensions were assessed at baseline and upon 55-month follow-up using the French version of the NEOPI-R [1]. Participants were asked to complete the 240-item self-report version of the NEOPI-R questionnaire using a five-point Likert agreement scale. The NEOPI-R assesses the following five personality factors: Neuroticism is the tendency to feel negative emotions including anxiety, hostility, and anger; Extraversion encapsulates the proneness towards positive emotions and feelings such as warmth and enthusiasm; Openness, the personal inclination to experience and the appreciation of new situations and thoughts with a curious, imaginative and creative attitude; Agreeableness, characterized by trustful, cooperative and altruistic tendencies, and finally Conscientiousness, is the predisposition to be reliable, resolute and well organized, and unwilling to deviate from rules and moral principles.

2.3. Neuropsychological Assessment

At baseline, all individuals were evaluated with an extensive neuropsychological battery, including the Mini-Mental State Examination (MMSE) [28], the Hospital Anxiety and Depression Scale (HAD) [29], and the Lawton Instrumental Activities of Daily Living (IADL [30]). Cognitive assessment included (a) attention (Digit-Symbol-Coding [31], Trail Making Test A [32], (b) working memory (verbal: Digit Span Forward [33], visuo-spatial: Visual Memory Span (Corsi) [34], (c) episodic memory (verbal: RI-48 Cued Recall Test [35], visual: Shapes Test [36], (d) executive functions (Trail Making Test B [32], Wisconsin Card Sorting Test and Phonemic Verbal Fluency Test [37], (e) language (Boston Naming [38], (f) visual gnosis (Ghent Overlapping Figures), (g) praxis: ideomotor [39], reflexive [40], and constructional (Consortium to Establish a Registry for Alzheimer’s Disease (CERAD), Figures copy [41]. All individuals were also evaluated with the Clinical Dementia Rating scale (CDR) [42]. In agreement with the criteria of Petersen et al. [43], participants with a CDR of 0.5 but no dementia and a score exceeding 1.5 standard deviations below the age-appropriate mean in any of the cognitive tests were classified as MCI and were excluded. Participants with neither dementia nor MCI in the last follow-up were classified as cognitively healthy controls.

The assessment of cognitive changes over time was made using a continuous cognitive score, taking into account both the increase and decrease of performances in a given test across the three assessments, as described previously. Cognitive trajectories were defined after sum- ming the number of cognitive tests at follow-up with performances at least 0.5 standard deviation higher or lower compared with the first evaluation (Z-scores). Change in cognition between inclusion and last follow-up was defined as the sum of the continuous cognitive scores at two follow- ups, as previously described [44].

2.4. MR Imaging Acquisition

Imaging data were acquired on a clinical routine whole-body 3T MR scanner (Tim Trio; Siemens, Erlangen, Germany). The structural 3D T1 sequence was performed with the following funda-mental parameters: 256x256 matrix; 176 sections; 1x1x1 mm3; TE, 2.3 ms; TR, 2300 ms. Additional sequences (T2WI, susceptibility-weighted imaging, diffusion tensor imaging) were used to exclude incidental brain lesions.

2.5. Amyloid PET Imaging

Fifty-four 18F-Florbetapir- (Amyvid) and four 18F-Flutemetanol-PET (Vizamyl) data were acquired on 2 different instruments (Siemens BiographTM mCT scanner and GE Healthcare Discovery PET/CT 710 scanner) of varying resolution and following different platform-specific acquisition protocols. The 18F-Florbetapir images were acquired 50 to 70 minutes after injection and the 18F-Flutemetanol PET images 90 to 120 minutes after injection. PET images were reconstructed using the parameters recommended by the ADNI protocol aimed at increasing data uniformity across the multicenter acquisitions. More information on the different imaging protocols can be found on the ADNI web site (http://adni.loni.usc.edu/methods/).

2.6. FDG PET Imaging

PET/CT data acquisition was performed on a Siemens BiographTM mCT or Vision scanner according to the guidelines of the European Association of Nuclear Medicine (EANM) [45]. The PET acquisition was started approximately 30 minutes after injection of 200 MBq of 18F-FDG. The PET emission study (20 min, one-bed position) was conducted followed immediately by the CT study used for attenuation correction. Ultra-law dose brain CT imaging was performed under standard conditions (120 kVp, 20 mAs, 128 × 0.6 collimation, a pitch of 1 and 1 s per rotation).

2.7. Visual MR Assessment

The visual analysis of brain MRI images was performed by an independent, board-certified specialist in neuroradiology (SH) blind to the neuropsychological data. MTA was assessed using the established score [46], ranging from 0 (no atrophy) to 4 (significant atrophy). WMH load was assessed with the Fazekas score [47], ranging from 0 (no white matter lesions) to 3 (confluent white matter lesions). The number of CMB was assessed based on the SWI or SWAN sequences. Only lesions were considered which were probable CMB, and the corresponding phase images were also analyzed to discriminate probable CMB versus micro-calcifications [48]. The total number of CMB was evaluated.

2.8. Visual Amyloid PET Analysis

The visual analysis of amyloid PET images was conducted by an independent, board-certified specialist in nuclear medicine (VG) blind to the neuropsychological data, following the tracer-specific standardized operating procedures approved by the European Medicinal Agency. Specifically, regional positivity was assessed for each scan, specifying if uptake was identified in the lateral frontal, parietal, posterior cingulate and precuneus, anterior cingulate, temporal lateral and striatal regions in either of the two hemispheres [49].

2.9. Visual FDG PET Analysis

FDG PET reading was performed by visual analysis of the output of an automated voxel-wise comparison with a reference database, as recommended in guidelines [45] and previously described in detail [50]. Images were classified as normal when no significant deviations from the normal distribution were observed and pathological when significant regional reductions of glucose metabolism were documented.

2.10. APOE Epsilon 4 Status

APOE epsilon 4 status was assessed as described earlier [27]. Subjects were divided according to whether they were a carrier of the APOE epsilon 4 allele (4/3 versus 3/3, 3/2 carriers).

2.11. Statistics

Sex differences in demographic, personality and imaging data were assessed with Chi-2, Mann-Whitney u test and unpaired t-test depending on the variable distribution. A Benjamini-Hochberg correction for multiple comparisons was applied. Simple and multiple linear regression models were used to identify predictors of the change in personality factor scores (dependent variable), including both binary (sex, APOE epsilon 4 presence, amyloid PET positivity, abnormal FGD-PET in visual inspection), ordinal (education, MTA, CMB, Fazekas score at baseline) and continuous (age, MMSE baseline score, continuous cognitive score) data. All of these variables were considered in LASSO (least absolute shrinkage and selection operator) linear regression models that were used as a variable selection method to detect independent variables associated with each of the five NEOPI-R personality factors score change (dependent variables) avoiding for multiple comparison biases. The significance level was set at P < 0.05. All analyses were performed with Stata release 18.5 (College Station, Texas, USA).

3. Results

The final sample included 58 individuals (34 females, age range: 69-76 years). Demographic, clinical and imaging data are summarized in Table 1. There was a significant difference in education levels in favor of men. In contrast, the cognitive decrement upon follow-up was significantly more pronounced in men compared to women. The mean MMSE score was of 28.6 (1.2) confirming the cognitive preservation of our cohort. Total agreeableness scores were significantly higher in women both at baseline and follow-up. No significant sex-related differences were found in respect to the other personality factors. Based on LASSO selection of variables, linear regression models were built for each NEO-PI factor score change. In univariate models, the decrease of openness was related to abnormal FDG PET patterns (regression coefficient: -4.89 [-9.18, -0.59], p = 0.026). This association persisted in multivariable models. This single variable explained 15% of the variance of openness decrease. The stronger association between AD signature markers and personality factors was found for agreeableness in both univariate and multivariable models (Table 2). In univariate models, lower MMSE scores (regression coefficient: 3.76 [0.77, 6.75], p = 0.015), and presence of amyloid positivity (regression coefficient: -12.37 [-20.91, -3.82], p = 0.005) at baseline were associated with a decrease of agreeableness scores. In multivariable models, these two parameters persisted as significant predictors of agreeableness decrease explaining 11% and 8% of its variance respectively (Table 3). Of importance, no relationship was found between the other personality factors and AD signature biomarkers. In the same line, vascular imaging markers were unrelated to the evolution of personality factors in this cohort.

4. Discussion

Our data reveal that amyloid positivity as well as abnormal FDG-PET at baseline may determine subtle changes in NEO-PI personality factors in normal aging. In this cohort of highly educated elderly persons, the decrease of openness upon a 4.5-year follow-up was related to abnormal AD-signature FDG patterns at baseline. Initial MMSE scores and PET amyloid positivity predict the decrease of agreeableness scores. Of importance, these associations were detected using LASSO linear regression models after controlling for demographic factors and other AD- and vascular signature biomarkers.

At first glance, the present series displayed a subtle cognitive decrement over time as illustrated by the small decrease of the continuous cognitive scores mainly in men. No case satisfied the MCI criteria upon follow-up remaining within the range of normal aging. However, a closer analysis of AD-signature biomarkers shows that more than 20% of cases had significant amyloid burden in visual inspection or abnormal FDG-patterns compatible with the initial phases of AD process. In the same line, mild of moderate mesial temporal lobe atrophy was found in 76% of cases. In respect to vascular burden, moderate to severe Fazekas scores were found in 15.5% of cases. Taking together, these descriptive data indicate that a substantial percentage of our cases display initial AD-compatible neurodegeneration and, in a lesser degree, vascular burden at a preclinical phase.

The association between AD-patterns of FDG hypometabolism and decrease of openness scores upon-follow up suggests that this personality factor is early affected by the progression of AD process in healthy controls. This observation should be interpreted in conjunction with a series of previous reports on the positive effects of openness in the preservation of hippocampal and inferior parietal lobe volumes [13,51,52,53], as well as white matter tracts connecting posterior and anterior brain regions and dorsolateral prefrontal cortex [54,55]. Increased functional connectivity within mesocortical networks has been also described in open people [56]. Higher measures of openness were related to lower PET amyloid and tau burden and higher CSF [14,17]. In the same line, openness has been associated with reversion from mild cognitive impairment (MCI) to normal cognition [57]. Altogether, these findings support the idea that elderly persons with high levels of openness are more resistant to AD process, yet this protection may decrease quickly once the AD-PET biomarkers become detectable.

One main finding of the present study is the strong association between the decrease of agreeableness and early amyloid positivity in healthy controls. As a psychological construct, agreeableness refers to the tendency of establishing interpersonal relationships without aggressiveness searching for social acceptance. Agreeable persons are more prone to avoid conflicts, adapt themselves to other’s commitments and adopt easily a majoritarian viewpoint. Traditionally, high agreeableness in adult lifespan is thought to be a positive trait of personality that promotes subjective well-being [58], better outcome in mental health treatments [59], less disengagement coping [60] and less sexual aggressive behavior [61]. A previous study showed that higher levels of agreeableness had a modest but significant protective effect on cognition [62]. To date, most of the cross-sectional studies failed to identify a protective role for this personality factor in respect to AD lesions and vascular burden in old age. In contrast, our previous work has shown that high agreeableness and conscientiousness may paradoxically be associated with more pronounced atrophy in mesial temporal lobe and regions of the mentalizing network (mPFC, amygdala, precuneus), highlighting the heterogeneity of the relationship between this personality factor and brain integrity [44,52]. In conjunction with these observations, the present results imply that agreeableness decreases rapidly in controls with early amyloid positivity and lower MMSE scores at baseline.

From a clinical viewpoint, our findings support the idea of the reverse causality: personality patterns at baseline may impact on the emergence of AD pathology but they may also change rapidly as a function of the presence of early AD-related PET abnormalities. The association between lower MMSE scores at baseline and personality changes should be interpreted with caution. First, the MMSE score is a cognitive screener sensitive to several confounders that remains within the narrow window of normal range in our sample. Second, the continuous cognitive score that explored the cognitive evolution of our controls taking into account the entire neuropsychological battery was not associated with personality changes upon follow-up. It is thus likely that high performers among elderly controls are more exposed to a progressive decrease of their agreeableness not related to subsequent cognitive decline.

Several limitations should be considered when interpreting these data. First, our cases are not representative of the whole spectrum of brain aging. Despite their recruitment in the community, they display no or very mild vascular morbidities, relatively high level of education and very low levels of neuroticism compared to age-adjusted normative values. We cannot thus exclude that the observed associations may be no longer present when cases with mixed demographic profiles or pathologies are considered. Second, the use of visual ratings of AD positivity and FDG-PET patterns instead of standardized continuous metrics (such as centiloids) as in our previous studies decreases the statistical power but guarantees that our observations may be reproducible in routine clinical settings. However, our set of AD biomarkers does not include PET-tau documentation, CSF specimens and, most importantly, plasma AD biomarkers. We cannot thus comment on their possible association with the evolution of NEO-PI personality factors in the elderly. Lastly, the combination of all significant predictors allows for explaining a modest percentage of personality factor changes. Although substantial given the marked heterogeneity of normal aging and relatively small sample size, this percentage indicates the presence of additional predictors that may impact on the association between personality factors and AD-signature and vascular biomarkers. In this respect, the inclusion of mixed cases with significant vascular pathology and higher levels of neuroticism with in vivo assessment of tau pathology, CSF and AD plasma biomarkers in larger elderly cohorts would be a useful complement, in the efforts to get better insight into the biological determinants of personality changes in old age.