Reduced Thickness of the Anterior Cingulate Cortex as a Predictor of Amnestic-Mild Cognitive Impairment Conversion to Alzheimer’s Disease with Psychosis

Participants

In this study, participants with aMCI were recruited from a memory impairment clinic at Pusan National University Hospital in South Korea between June 2015 and March 2019. A diagnosis of aMCI was made using the following Petersen’s criteria [16]: 1) subjective memory complaints by patients or caregivers; 2) objective memory impairment based on the word list delayed recall test (Seoul Verbal Learning Test-delayed recall, SVLT-delayed recall) [17] with a cut-off score of 1.5 SD under population mean standardized for age and sex; 3) normal activities of daily living (ADL), based on clinical findings and the Seoul Instrumental Activities of Daily Living (S-IADL) [18]; and 4) the absence of dementia. The exclusion criteria for all participants were as follows: 1) significant psychiatric conditions (e.g., delirium, bipolar disorder, major depressive disorder, schizophrenia and other psychotic disorders, substance abuse); 2) significant cerebrovascular conditions (e.g., severe white matter hyperintensities with Fazekas scale 3, multiple (more than five) lacunes, hemorrhages, or tumors); 3) AD or other neurodegenerative diseases (e.g., dementia with Lewy bodies, frontotemporal dementia, Parkinson’s disease); 4) significant medical conditions (e.g., severe organ failure, metabolic or hematologic disorders, clinically significant abnormal laboratory findings); and 5) history of taking antipsychotic drugs within the past 1 year. This study was approved by the Institutional Review Board of the Pusan National University Hospital, and both the subjects and their caregivers were asked to sign a consent form after being informed of the purpose and procedures of this study. All methods were performed following the relevant guidelines and regulations.

Study design

This is a prospective observational study on the association between the ACC and HC with conversion from aMCI to AD + P. At baseline, all participants underwent a standardized dementia screening consisting of the following: a patient history from either the patient or an informant, medical and neurological examinations, the Korean version of the Neuropsychiatric Inventory (K-NPI) [19], the clinical dementia rating score-sum of box (CDR-SOB) [20] for evaluation of dementia severity, the Seoul Neuropsychological Screening Battery (SNSB) [17] for comprehensive neuropsychological assessment, S-IADL [18] for complex activities of daily living, and brain magnetic resonance imaging (MRI) for cortical thickness or volume.

In this study, the primary outcome was time-to-progression from aMCI to incident AD + P. The diagnosis of AD + P consisted of two stages: AD was first diagnosed, and then the presence or absence of psychosis was evaluated. First, to determine the presence of AD, follow-ups were conducted by annual visits to our memory clinic, in which a full patient history taking, a neurologic examination, CDR-SOB, SNSB, and S-IADL were repeated. All patients diagnosed with AD routinely visited our memory clinic every 3 months to determine the presence of incident psychosis. AD subjects were censored when AD + P was diagnosed. Patients without AD + P (MCI stable or AD-P) were censored at the time of their most recent follow-up. To determine the presence of psychosis, clinical interviews with psychiatrists were conducted and scales such as the K-NPI [19] were used at baseline and at every 3-month follow-up. Two psychiatrists with expertise in dementia research made the clinical diagnosis of AD + P after reviewing all available raw data, such as clinical interviews and K-NPI. The K-NPI is a valid and reliable tool for assessing the severity and frequency of behavioral and psychiatric symptoms of dementia, including delusions, hallucinations, agitation, dysphoria, anxiety, euphoria, apathy, disinhibition, irritability, aberrant motor behavior, and neurovegetative changes, including nighttime behavior (e.g., sleep disorders) and eating changes (e.g., appetite disorders).

Diagnosis of AD with psychosis

AD was diagnosed using the National Institute of Neurological and Communication Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association criteria for probable AD [21]. A diagnosis of psychosis was made according to the diagnostic criteria for psychosis of AD proposed by Jeste and Finkel [22]: 1) diagnosis of AD dementia; 2) occurrence of psychosis after diagnosis of dementia; 3) persistent or intermittent delusions, hallucinations, or both for at least 1 month; 4) severe psychotic symptoms are enough to disrupt functioning; and 5) exclusion of schizophrenia and related psychotic disorders (e.g., schizoaffective disorder, delusional disorder, psychotic mood disorder), delirium, and other causes of psychotic symptoms (e.g., substance use and other general medical conditions).

MRI data acquisition and image analysis through Freesurfer

Each participant underwent a structural T1-weighted MRI scan at the baseline evaluation. All images were acquired at the Pusan National University Hospital Imaging Center on a Siemens 3T Trio TIM scanner (Erlangen, Germany). For each participant’s cortical thickness or volume estimations, 3D magnetization-prepared rapid gradient echo sequence was acquired using the following parameters: repetition time = 1800 ms, echo time = 2.07 ms, flip angle = 12°, acquisition matrix = 256×256, field of view (FOV) = 250×250 mm², slice thickness = 1 mm, and total number of slices = 256. All image acquisitions had the same slice orientation, parallel to the anterior and posterior commissure lines. The movement was restricted to expandable foam cushions. Scans with movement or any other image (reconstruction) artifacts were excluded. The Freesurfer version 5.1 software package (http://surfer.nmr.mgh.harvard.edu/) was used to analyze the cortical thickness or volume on 3-D T1-weighted images [23]. The regions of interest were the subregions of the ACC (rostral ACC, rACC; caudal ACC, cACC) and HC included in the Desikan-Killiany Atlas [24]. We visually examined all images to ensure segmentation accuracy.

Statistical analysis

During the follow-up period, aMCI patients were classified into three groups according to the onset of AD or AD + P: 1) aMCI stable (non-conversion to AD or AD + P), 2) AD converters (conversion to AD), and 3) AD + P converters (conversion to AD + P). Comparisons among the three groups in baseline demographic and clinical characteristics were analyzed using analysis of variance with Bonferroni correction for multiple comparisons for continuous variables and chi-square test for discrete variables. Analysis of covariance (ANCOVA) was used to investigate the differences in the thickness (or volume) of the ACC or HC among the three groups with Bonferroni adjustment for multiple comparisons after controlling for demographic or clinical variables. Bivariate survival analyses were conducted with baseline demographic and clinical characteristics to determine predictors associated with the risk of incident AD + P in patients with aMCI. Significant predictors from bivariate analyses were included as covariates in the multivariate Cox regression analysis. Survival analysis with time to incident AD + P as an event variable was calculated with the multivariable Cox proportional hazards models using each sub-regional ACC thickness or HC volume as a continuous variable with the enter method with an adjustment for other covariates. Separate Cox proportional hazard models were used for each brain area as a predictor. Statistical analyses were performed using the SPSS software (version 20.0, IBM, Inc., Armonk, NY, USA). Statistical significance was defined as p < 0.05.

Demographic and clinical characteristics

A total of 194 patients with aMCI were recruited for the study at baseline. During the longitudinal follow-up, 10 aMCI patients were excluded because they did not complete at least one follow-up. Five MCI patients progressed to other types of dementia (two vascular dementia, two Lewy body dementia, and one frontotemporal lobar degeneration) and were excluded because they were more likely to have pathophysiological mechanisms different from AD + P. Three aMCI patients progressed to aMCI with psychosis and were also excluded to clearly distinguish AD from schizophrenia, delusional disorders, and psychotic mood disorders. One patient with AD who developed psychotic symptoms after steroid use was excluded. One patient with aMCI was excluded because the magnetic resonance imaging (MRI) data contained too much noise to be analyzed using the Freesurfer pipeline. Two patients with AD had psychotic symptoms but their symptoms did not affect their quality of life, thus they were excluded from the analysis. This left the final group of 172 MCI patients (mean age = 72.83±7.93 years; females = 86 (50.0%); mean education = 5.85±4.72 years; CDR-SOB = 1.67±1.48; total brain volume = 1342.82±145.89 cm³). After an average follow-up period of 3.72 (±2.66) years, of all aMCI patients, 94 patients (54.6%) progressed to AD (AD converters) and 18 patients (10.4%) progressed to AD + P converters. See Fig. 1 for the flow diagram.

OPEN IN VIEWER

Fig. 1

Flow diagram used to exclude participants from study and reach final sample size.

Table 1 describes the demographic and clinical differences among the three groups (MCI stable, AD converters, and AD + P converters). Significant differences in CDR-SOB were detected among the groups (p < 0.05). Patients with AD converters had significantly higher CDR-SOB scores than those in the MCI stable group. There were no significant differences in age, sex, education level, follow-up period, and total brain volume among the groups.

Cingulate subregional thickness and HC volume among three MCI groups

The differences in the cingulate sub-regional thickness and HC volume among the three aMCI groups (aMCI stable, AD converters, and AD + P converters) are shown in Table 2. As there were significant differences among the groups on CDR-SOB, the following ANCOVA models were conducted with the addition of the covariate. A significant difference in the left cACC and both left and right HC volumes across the three groups were detected (p < 0.05). Using post hoc tests (Bonferroni correction), it was found that AD + P converters had thinner left cACC compared to aMCI stable or AD converters (1.91±0.49 versus 2.33±0.47 or 2.19±0.48, p < 0.05). Compared to the aMCI stable group, AD converters showed smaller HC volume on both sides (left HC, 2.70±0.48 versus 2.94±0.47, p < 0.05; right HC, 2.78±0.47 versus 3.03±0.43, p < 0.05). In other cingulate areas, the differences among groups were not statistically significant, but AD + P converters showed a decreasing trend in both rACC (left rACC, p = 0.073; right rACC, p = 0.097) and right cACC (p = 0.083).

Association of subregional ACC thickness and HC volume with risk of incident AD + P

Bivariate Cox regression analyses were conducted to determine which clinical characteristics were associated with an increased risk of incident AD. As there were no clinical characteristics associated with the risk of AD + P in bivariate analysis, the following Cox proportional hazard models examining the association of sub-regional ACC thickness and HC volume with the risk of incident AD + P were conducted without the addition of covariates.

Cox proportional hazard analyses showed that the risk of AD + P was associated with sub-regional ACC thickness but not HC volume (Tables 3 4). Univariate analyses showed that reduced cortical thickness in the left cACC (HR [95%CI], 0.224 [0.087–0.575], p = 0.002) and right cACC (HR [95%CI], 0.318 [0.132–0.768], p = 0.011) were associated with a higher risk of incident AD + P. This association remained significant when adjusted for both HC volume (left cACC, HR [95%CI], 0.242 [0.093–0.627], p = 0.004; right cACC, HR [95%CI], 0.338 [0.139–0.823], p = 0.017), suggesting that the effect of subregional cACC thickness on the risk of incident AD was independent of HC volume. In the HC, however, no association with the risk of incident AD was found in the univariate analysis (Table 4).