Magnetic Resonance Imaging Volume of the Angular Gyri Predicts Financial Skill Deficits in People with Amnestic Mild Cognitive Impairment

Authors

  • H. Randall Griffith PhD,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Christopher C. Stewart MS,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Luke E. Stoeckel PhD,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Ozioma C. Okonkwo PhD,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Jan A. den Hollander PhD,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Roy C. Martin PhD,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Katherine Belue BS,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Jacquelynn N. Copeland BS,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Lindy E. Harrell MD, PhD,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • John C. Brockington MD,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • David G. Clark MD,

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author
  • Daniel C. Marson JD, PhD

    1. From the Departments of *Neurology, Psychology, and #Cardiology, and Alzheimer's Disease Research Center, University of Alabama at Birmingham, Birmingham, Alabama; §Department of Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; Department of Neurology, School of Medicine, the Johns Hopkins University, Baltimore, Maryland; and **Birmingham Regional Veterans Affairs Medical Center, Birmingham, Alabama
    Search for more papers by this author

Address correspondence to H. Randall Griffith, ABPP-CN; SC 650, 1530 3rd Ave. S., Birmingham, AL 35294. E-mail: rlgriffith@uabmc.edu

Abstract

OBJECTIVES: To better understand how brain atrophy in amnestic mild cognitive impairment (MCI) as measured using magnetic resonance imaging (MRI) volumetrics could affect instrumental activities of daily living (IADLs) such as financial abilities.

DESIGN: Controlled, matched-sample, cross-sectional analysis regressing MRI volumetrics with financial performance measures.

SETTING: University medical and research center.

PARTICIPANTS: Thirty-eight people with MCI and 28 older adult controls.

MEASUREMENTS: MRI volumetric measurement of the hippocampi, angular gyri, precunei, and medial frontal lobes. Participants also completed neuropsychological tests and the Financial Capacity Instrument (FCI).

RESULTS: Correlations were performed between FCI scores and MRI volumes in the group with MCI. People with MCI performed significantly below controls on the FCI and had significantly smaller hippocampi. Among people with MCI, performance on the FCI was moderately correlated with angular gyri and precunei volumes. Regression models demonstrated that angular gyrus volumes were predictive of FCI scores. Tests of mediation showed that measures of arithmetic and possibly attention partially mediated the relationship between angular gyrus volume and FCI score.

CONCLUSION: Impaired financial abilities in amnestic MCI correspond with volume of the angular gyri as mediated by arithmetic knowledge. The findings suggest that early neuropathology within the lateral parietal region in MCI leads to a breakdown of cognitive abilities that affect everyday financial skills. The findings have implications for diagnosis and clinical care of people with MCI and AD.

Mild cognitive impairment (MCI) is a clinical construct that denotes the transitional phase between normal cognitive aging and Alzheimer's disease (AD).1 Diagnostic criteria indicate that people with MCI have “generally intact” activities of daily living,2 although investigations of instrumental activities of daily life (IADLs) in MCI have revealed mild impairments in informant-reported3 and laboratory-assessed IADLs.4,5 By definition, IADLs such as financial capacity,6 medication management, and driving are the skills that are necessary for independent living and autonomy of older adults in this society. Financial capacity comprises a broad range of conceptual, pragmatic, and judgment abilities, ranging from basic (e.g., identifying and counting coins and currency) to more-complex skills (e.g., paying bills, managing a bank statement, and exercising financial judgment).7 Declines in financial capacity are some of the earliest IADL changes noted in cognitive aging and dementia.8

Financial capacity deficits in the amnestic variant of MCI and in AD have been systematically investigated. People with amnestic MCI have been shown to be impaired on the Financial Capacity Instrument (FCI), a direct assessment measure of financial capacity, specifically on domains measuring financial concepts, bank statement management, financial judgment, and bill payment,4 all of which represent complex aspects of financial capacity. The neurocognitive predictors of financial abilities in people with amnestic MCI include measures of attention,9 visuomotor tracking,9 and written arithmetic ability.10

The presence of complex IADL deficits in MCI suggests that brain networks underlying the neurocognitive underpinnings of daily activities are becoming nonfunctional in the context of brain changes that occur in amnestic MCI. Brain changes are observable on positron emission tomography, structural MRI, and functional MRI in people with amnestic MCI, and these neuroimaging markers are often indicative of neuropathological changes related to AD.11 For instance, the importance of arithmetic to financial abilities suggests that pathology within the angular gyrus may be related to declines in financial capacity;12 attention and executive function as predictors of financial abilities could implicate pathology within the midline cortical regions such as the posterior cingulate and precuneus and medial superior frontal lobes.13–16 Although measures of episodic memory are not primary predictors of FCI performance in people with amnestic MCI,9,10 the prominence of hippocampal and medial temporal lobe pathology in these people17 suggests that atrophy of these regions may also play a role in cognitive and mental status declines affecting FCI performance.

An important gap in neurodegenerative disease research is the identification of links between brain changes, cognitive change, and changes in IADLs; this gap exists despite advances in neuroimaging and in the assessment of IADLs. Only a few studies have attempted to use brain imaging techniques to examine how brain changes in AD and MCI can influence the everyday functional abilities of older adults.18–22 Unfortunately, such studies have generally relied on informant report of IADLs rather than direct-assessment measurements. The reliance on informant report is troubling, because studies have suggested that informant ratings of IADLs may be unreliable.23,24

Of the available neuroimaging technologies, structural MRI volumetrics is a potentially useful tool to examine IADL changes in MCI because it is sensitive to atrophy in people with MCI,25 is correlated with cognitive function in MCI,26 and has generally good correspondence with AD pathology at autopsy.11,27

The current study investigated MRI volumetrics of four structures suspected to play a role in financial abilities in persons with amnestic MCI: the medial prefrontal lobes, the hippocampi, the medial parietal lobe and precunei, and the angular gyri. As discussed below, these structures were chosen because they are brain regions known to be affected in MCI and AD and because of their prior association with cognitive abilities that could influence financial abilities. This study sought to identify which regions contribute to FCI performance in people with amnestic MCI and to explore the neurocognitive mechanism(s) that might mediate the relationships between brain atrophy and FCI performance in people with amnestic MCI.

METHODS

Participants

Sixty-six participants were included in this study, of whom 38 had amnestic MCI and 28 were age-matched healthy older adult controls. People with amnestic MCI were community-dwelling older adults who presented for clinical evaluation to the University of Alabama at Birmingham (UAB) Memory Disorders Clinic, a tertiary care outpatient clinic. They were subsequently recruited into the UAB Alzheimer's Disease Research Center (ADRC) and an associated study of functional change in amnestic MCI (Cognitive Observations in Seniors (COINS)). All participants underwent a research assessment that consisted of neurological evaluation, neuropsychological testing, capacity testing, and laboratory blood work. The diagnosis of amnestic MCI was made in a consensus diagnostic conference using original Mayo criteria.2 Informant or patient report of subjective memory impairment was determined through clinical examination and an informant interview conducted separately from the patient's ADRC examination. Neuropsychological tests used to fulfill criteria for amnestic MCI included the Wechsler Memory Scale—Revised Edition (WMS-R) Logical Memory,28 the California Verbal Learning Test II (CVLT-II),29 Dementia Rating Scale (DRS) Memory, and 10/36 Spatial Recall. The ADRC neuropsychologists determined impairment in relationship to appropriate age-based norms and clinical judgment regarding change from prior or premorbid level of function. Overall cognition was assessed through performance on the DRS-230 and the Mini-Mental State Examination (MMSE).31 Activities of daily living (for diagnostic purposes) were assessed according to informant report using a standardized interview format relevant to MCI and early AD.

At the time this study began, the UAB ADRC was using the original Mayo criteria for amnestic MCI.2 Retrospective examination of the frequency of cognitive impairment in this ADRC amnestic MCI cohort revealed that approximately two-thirds of the participants would probably meet criteria for amnestic MCI—multiple cognitive domains involved.32

Exclusion criteria for the amnestic MCI group included a diagnosis of exclusively nonamnestic MCI, evidence of another neurodegenerative disease, history of stroke, another chronic debilitating neurological illness (e.g., multiple sclerosis or cerebral palsy), severe organ disease, autoimmune disease, cancer (except skin cancer), alcoholism, or a terminal condition with less than 12 months to live. Participants were also excluded if they had untreated major depression or any other severe psychiatric disorder.

Control participants were 28 volunteers recruited from the community into the ADRC through newspaper advertisements or health fairs. These participants underwent ADRC neurological evaluation and neuropsychological testing to ensure the absence of medical and psychiatric conditions that could compromise cognition and were characterized as cognitively normal following ADRC diagnostic consensus conference. Controls in the ADRC also met the exclusion criteria listed above for participants with amnestic MCI.

All participants gave informed consent according to UAB institutional review board guidelines.

Financial Capacity Instrument

The FCI is a standardized psychometric instrument developed to directly assess the financial abilities and performance of older adults.4,33 The FCI assesses financial abilities based upon a three-level conceptual model of the financial capacity construct: specific financial abilities, broader financial activities, and overall financial capacity.4,33 The development, reliability, and validation of the instrument have been described elsewhere.33 A trained technician administered the FCI and scored performance on the items based upon standardized criteria. In this study, only the FCI domains of financial conceptual knowledge (Domain 2), bank statement management (Domain 5), financial judgment (Domain 6), and bill payment (Domain 7) were used because of prior evidence demonstrating that these domains discriminate best between healthy controls and people with amnestic MCI.4,9 Participants' scores on these domains were summed to create an abbreviated total FCI score. The prior experience of the participants with the financial skills measured using the FCI was determined according to self-report, informant report, or both. This information was then used to correct for lack of prior experience in any applicable financial tasks and domains after a previously described method.4

Neuropsychological Tests

A trained technician administered a battery of neuropsychological tests to each participant. The following measures were selected from this battery for inclusion in this study because of their prior association with FCI performance in MCI.9,10

  • Attention was measured using the Wechsler Memory Scale—Third Edition (WMS-III) Spatial Span subtest.34
  • Visuomotor tracking was measured using the Trail-Making Test Part A (Trails A).35
  • Arithmetic ability was measured using the Wide-Range Achievement Test, Third Edition (WRAT-3) Arithmetic subtest.36

The raw scores from these measures were used in the statistical analyses.

MRI Acquisition and Data Processing

MRI images were obtained using a Philips Intera 3 Tesla MRI system (Best, Netherlands) with a quadrature transmit–receive head coil to measure gray matter volumes from the regions of interest (ROIs). These imaging sequences consisted of multislice sagittal, axial, and coronal T1 fast field echo (FFE) scout sequences acquired using repetition time (TR)/echo time (TE)=11.1/4.6 ms, 256 × 128 resolution, and 10-mm slice thickness (for the purpose of image alignment) and multislice sagittal T1 FFE acquired using TR/TE=9.3/4.6 ms, 240 × 240 resolution, and 2-mm slice thickness.

After the MRI images were obtained, they were transferred to a workstation running SPM5 (http://www.fil.ion.ucl.ac.uk/spm/software/spm5). Tissue segmentation was derived in SPM5 and involved segmentation, normalization, and modulation of the brain and smoothing of the resulting gray and white matter images.37 The T1 image set was first skull stripped and the origin of each scan manually set to the anterior commissure. Next, a combined normalization–segmentation process was implemented on the scan using the prior probability templates provided with the SPM5 program. These tissue probability maps are modified versions of the Consortium for Brain Mapping Tissue Probabilistic Atlases. The results of this modulated segmentation–normalization process are gray matter, white matter, and cerebrospinal fluid images for each subject normalized to the same stereotactic space.37 Finally, gray and white matter images were smoothed using an 8-mm Gaussian kernel.

Voxel-based morphometry (VBM) was first used to explore the predictive ability of brain voxels within the MCI group in terms of FCI scores. Voxel-level data were then analyzed for clusters associated with FCI score, with a threshold of 10 voxels per cluster. Associations were then corrected for multiple comparisons, with significance set at P<.00025. Voxel clusters that predicted FCI score variance past this threshold were identified using the Montreal Neurological Institute (MNI) coordinate system.38

The ROIs for volumetric analyses were then identified using an automated pickatlas routine in SPM5. This method is based on an anatomical parcellation of the MNI MRI single-subject brain38 and has the advantage of requiring no trained user interface to reliably obtain volumetric masks from any of 45 cortical and subcortical ROIs (each hemisphere) after modulated MRI images are segmented and normalized in SPM5. Similar methods have been shown to have good reliability, with manual traces of the hippocampus in people with dementia.39 Based upon this method, bilateral volumes of the following ROIs were derived (Figure 1).

Figure 1.

 Regions of interest as derived using SPM for volumetric analysis. Regions are labeled.

Superior frontal gyri, medial: This region was defined as the medial aspect of the frontal cortex and included all of the superior frontal gyrus to a posterior border of the supplementary motor area and cingulate, with an inferior border at the line of the anterior–posterior commisure, at which point the orbital portion of the superior frontal gyri occurs.38

Hippocampi: This region consisted of the gray matter in the region of the temporal ventricular horns, including the dentate gyrus, uncus, and hippocampus proper, limited caudally by the parahippocampal ramus of the collateral fissure.38

Precunei: This region is defined as the medial cortical surface of both hemispheres with posterior boundaries of the parietooccipital sulcus, dorsally by the marginal sulcus, and ventrally by the subparietal sulcus.38

Angular gyri: This region was defined anteriorly by the angular sulcus and caudally at the occipitoparietal line.38

All ROIs were derived for both hemispheres. Given that asymmetries were not observed in these structures at the group level, hemispheric ROI volumes were summed to obtain bilateral ROI volumes, a procedure that also effectively reduced the number of statistical tests performed. The influence of sex and age on MRI volumes was adjusted for using multiple regression analyses based on the control participants.40 The regression equations were then applied to the volumes of participants with MCI, and the predicted variance was removed from the observed volumes. The result is a residual score that removes variance due to body size and aging. The z-score transformations of adjusted volumes were then computed based on the control group's mean adjusted volumes.

Statistical Analyses

Demographic data were compared using independent-sample t-tests or chi-square tests (for categorical data). Group differences on neuropsychological, FCI, and MRI measures were tested using independent-sample t-tests. Within the MCI patient group, Pearson correlations were first run to examine the associations between neuropsychological test scores, MRI volumes, and FCI performance (approximating a 5:1 subject-to-variable ratio). VBM analyses were then conducted as described above. A forward stepwise multiple regression (probability of F to enter=0.05; probability of F to remove=0.10) was then performed to determine which of the four MRI volumes best predicted total FCI score. This stepwise multiple regression was subsequently repeated with an initial predictor block entered to account for overall cognition (MMSE score) and demographic variables before the stepwise block. Finally, models of mediation using MRI volume(s) as the predictor variable(s), FCI score as the dependent variable, and neuropsychological variables as the mediator(s) were investigated using a method described previously.41 A two-tailed alpha level of .05 was adopted for all analyses.

RESULTS

Demographics, Dementia Staging, and Clinical Characteristics

Demographic information, dementia staging, and clinical characteristics are presented in Table 1. As expected, the matched samples showed no significant differences in terms of age, education, race, or sex. Staging using the Clinical Dementia Rating (CDR)42 scale was different: all people with MCI had CDR scores of 0.5, and all controls had CDR scores of 0.0. Almost 40% of the people with MCI were taking a cholinesterase inhibitor for memory loss.

Table 1. Demographics and Study Measures in Controls and Patients with Mild Cognitive Impairment (MCI)
VariableControls
n=28
Amnestic MCI
n=38
P-ValueEffect
Size d
  • *

    Abbreviated Financial Capacity Instrument (FCI) (four domains) maximum score=133.

Age, mean ± SD71.6 ± 5.470.8 ± 6.4.550.15
Male/female9/1913/25.86
Caucasian/African American24/430/8.48
Education, years, mean ± SD14.8 ± 2.614.2 ± 2.6.360.23
Mini-Mental State Examination score, mean ± SD29.4 ± 1.128.2 ± 1.4.0010.96
Wechsler Memory Scale—Third Edition Spatial Span total score, mean ± SD14.0 ± 2.712.3 ± 2.9.0170.62
Trail Making Test Part A, seconds, mean ± SD30.5 ± 9.043.0 ± 19.3.0020.79
Wide-Range Achievement Test, Third Edition, Arithmetic score, mean ± SD40.3 ± 5.137.8 ± 4.8.070.49
Total FCI score, mean ± SD*124.4 ± 4.7107.2 ± 15.9.0011.38
Clinical Dementia Rating staging, 0.0/0.528/00/38.001
Cholinesterase inhibitor, yes/no15/23

Neuropsychological Performance

Neuropsychological test performance is also shown in Table 1. Controls performed significantly higher than people with MCI on the MMSE, Spatial Span, and Trails A. There was a trend observed (P=.07) toward lower performance on WRAT-3 Arithmetic in people with amnestic MCI.

Financial Performance

Comparison of FCI score performance is also shown in Table 1. Variance was not equal between groups (Levene's Test=21.86, P<.001). A t value was thus applied that does not assume equality of variances. Comparison of total FCI score demonstrated that controls performed significantly better than people with amnestic MCI.

VBM Analysis

The regression of brain voxels onto FCI scores showed a positive association with one cluster of voxels, with maxima located at MNI coordinates −54, −48, and 44 (uncorrected P=.01) (posterior parietal cortex in the area of the angular gyrus), although this association became a trend and did not reach significance when the multiple-comparison correction was applied (corrected P=.08) (Figure 2). A second cluster of voxels with maxima located at 18, 12, and −24 (right ventral frontal region) showed a trend before correction (P=.05), although after correction, this finding did not approach significance (corrected P=.26).

Figure 2.

 Results of voxel-based morphometry analysis of Financial Capacity Instrument scores in people with amnestic mild cognitive impairment. (A) The cluster with maxima at coordinates −54, −48, and 44 on the glass brain projection (P=.08 corrected) (the second cluster was not significant after correction). (B) The maxima of the above coordinates projected onto T1 magnetic resonance imaging scans in each dimension for comparison.

MRI Volumes

MRI volumes from the four ROIs are displayed in Table 2. The hippocampi were significantly smaller bilaterally in people with amnestic MCI than in controls. A trend was observed (P=.06) toward smaller medial frontal lobe volume in people with amnestic MCI. The percentage of people with MCI with adjusted z-score volumes more than 1 standard deviation (SD) below the mean was greatest in the hippocampi (47.4%), followed by angular gyri (34.2%) and then the medial frontal lobe (28.9%) and precunei (28.9%).

Table 2. Volumes of Regions of Interest in Controls and Patients with Mild Cognitive Impairment (MCI)
Measurecm3, Mean ± Standard
Deviation
Difference of
Mean Z Scores
P -Value*Patients with MCI >1
Standard Deviation Below
the Mean, %
Effect Size d
Controls
n=28
Amnestic MCI
n=38
  • *

    Statistical test performed on Z score adjusted values.

Medial frontal lobes11.5 ± 1.610.9 ± 1.6−0.50.0628.90.37
Precunei16.9 ± 2.016.3 ± 2.2−0.38.1828.90.27
Angular gyri7.7 ± 0.97.5 ± 1.0−0.35.2434.20.25
Hippocampi5.8 ± 0.55.3 ± 0.7−0.98.00147.40.75

Neuropsychological Test, MRI Volume, and FCI Correlations

Table 3 shows the correlations between the neuropsychological tests, MRI volumes, and total FCI score as obtained within the MCI group. MRI volumes were highly correlated with each other (significant correlation coefficients (r) ranging from 0.48 to 0.80) and also showed correlations with neuropsychological variables. Medial frontal lobe volume was correlated with WRAT-3 Arithmetic. Angular gyrus volume was correlated with WMS-III Spatial Span and WRAT-3 Arithmetic. Precuneus volume was correlated with WMS-III Spatial Span, Trails A, and WRAT-3 Arithmetic. The hippocampi volume showed no correlations with neuropsychological test scores within the amnestic MCI group.

Table 3. Correlations Between Neuropsychological Tests, Magnetic Resonance Imaging Volume Regions of Interest, and Financial Capacity Instrument Score in Patients with Amnestic Mild Cognitive Impairment (MCI)
 Pearson Correlation Coefficient (Cohen's d)
Spatial SpanTrails AWRATMFLAngular GyrusPrecuneusHippocampus
  1. P< * .05 , ** .01.
    Abbreviated Financial Capacity Instrument (FCI) used (four domains).
    DRS=Dementia Rating Scale-2; MFL=medial superior frontal lobe; Spatial Span=WMS-III Spatial Span; Trails A=Trail Making Test Part A; WRAT=Wide-Range Achievement Test Third Edition Arithmetic subtest.

Trails A−0.39* (0.85)     
WRAT0.54** (1.28)−0.49** (1.12)    
MFL0.22 (0.45)−0.30 (0.63)0.40* (0.87)   
Angular gyrus0.36* (0.77)−0.32 (0.68)0.34* (0.72)0.68** (1.86)  
Precuneus0.38* (0.82)−0.42** (0.93)0.36* (0.77)0.62** (1.58)0.80** (2.67) 
Hippocampus0.21 (0.43)−0.16 (0.32)0.22 (0.45)0.52** (1.22)0.50** (1.16)0.48** (1.09)
FCI0.58** (1.42)−0.56** (1.35)0.74** (2.20)0.32 (0.68)0.44** (0.98)0.31 (0.65)0.25 (0.52)

The FCI score showed correlations with all three of the neuropsychological variables (r ranging from −0.56 to 0.74). The FCI score was correlated with the angular gyri ROI (Figure 3) but not with the other three ROIs.

Figure 3.

 Scatterplot demonstrating the correlation between Financial Capacity Instrument (FCI) score and adjusted angular gyrus volume.

Regression Models

Two stepwise regressions were computed to determine the key predictors of FCI score. The first stepwise regression was computed to determine the key MRI volume predictor of FCI score, with all four ROIs considered for the regression. The regression entered the angular gyri ROI and explained 19% of the variance in FCI scores (F (1, 36)=8.44, P=.006).

Because demographic variables and overall mental status could help explain variability in FCI scores within this patient group, a regression analysis of FCI score was performed that force-entered MMSE score, age, education, and sex in one block and subsequently entered the four MRI volumes in a second stepwise block. After the initial block was force-entered, the angular gyri was the only ROI that met entry criteria for the model, uniquely accounting for 12% change in the variance in addition to the covariate block (F (1, 32)=7.61, P=.01).

Tests of Mediation

The WMS-III Spatial Span and WRAT-3 Arithmetic measures met the initial assumptions for mediation.43 As detailed above, angular gyrus volume was correlated with neuropsychological measures, and the neuropsychological measures were in turn correlated with FCI score.

The first mediator model tested whether WMS-III Spatial Span score mediates the relationship between angular gyrus volume and FCI score. The Sobel equation was computed with the unstandardized coefficients of the correlation of angular gyrus volume with Spatial Span (B=0.88, SE=0.38) and the correlation between Spatial Span and FCI score, when angular gyrus volume is also accounted for (B=2.63, standard error (SE)=0.78).41 The test statistic (1.93) just missed significance (P=.05); the beta for the relationship between angular gyrus volume and FCI score changed from 0.44 to 0.26 (P=.07) after accounting for Spatial Span score, suggesting a partial mediation.

The second mediator model tested whether WRAT-3 Arithmetic score mediates the relationship between angular gyrus volume and FCI score. The Sobel equation was computed with the unstandardized coefficients of the correlation of angular gyrus volume with Arithmetic (B=1.36, SE=0.64) and the correlation between Arithmetic and FCI score when angular gyrus volume is also accounted for (B=2.18, SE=0.38). The test statistic (1.99) was significant (P=.05), indicating that Arithmetic score partially accounts for the relationship between angular gyrus volume and FCI score; the beta for the relationship between angular gyrus volume and FCI score changed from 0.44 to 0.17 (P=.18) after accounting for Arithmetic score, indicating a partial mediation.

DISCUSSION

There are three main findings of interest from this study. First, MRI volume of the angular gyri, one of the posterior cortical regions known to be implicated in the early neuropathology of AD, was moderately correlated with a direct assessment measure of financial abilities in people with amnestic MCI. Second, angular gyrus volume was shown to be a unique predictor of financial abilities after accounting for overall mental status and demographic variables. Third, arithmetic ability was shown to partially mediate the relationship between angular gyrus volume and financial abilities. These findings have implications for understanding the pathway by which early declines in financial abilities can occur in people with amnestic MCI.

To the authors' knowledge, the current study is one of the first to link brain measures in areas other than the hippocampus with IADLs in people with amnestic MCI, demonstrating that structural brain measures are sensitive not only to cognitive and clinical impairments, but also to emerging deficits in cognitively demanding IADLs such as financial capacity. People with amnestic MCI are known to experience brain volume changes that have implications for cognitive function and prognosis. Hippocampal atrophy in amnestic MCI corresponds to memory impairment26 and disease progression25 and is predictive of risk of later “conversion” to AD,44 although volumetric abnormalities also occur in other cortical regions in people with MCI, such as the posterior cingulate45 and angular gyrus.46,47 The data from the current study generally correspond with these prior findings; hippocampal volumes were significantly smaller in people with amnestic MCI than in controls. Although the other ROIs did not reach significance, they were smaller in people with amnestic MCI than controls in all instances, and in the case of the angular gyrus, more than one-third of the people with MCI showed emerging volumetric abnormalities (values more than 1 SD below the mean).

Furthermore, these data suggest that angular gyrus volume and the neural networks that it is a part of is a specific neurocognitive substrate by which financial capacity is influenced. The first evidence supporting the conclusion is that, despite high intercorrelations between the volumetric ROIs in this sample, other brain volumes including the hippocampus were meagerly correlated with FCI performance, whereas angular gyrus volume showed a reasonably robust correlation. An exploratory VBM analysis showed a strong trend for a cluster of voxels located in the posterior parietal cortex in the area of the angular gyrus to predict FCI scores. Additionally, regression demonstrated that the volumetric predictor of FCI in the sample of people with MCI was angular gyrus volume. The third evidence in support of a specific relationship between angular gyrus volume and FCI performance is the more-stringent regression model that accounted for overall mental status, age, education, and sex (factors that would plausibly influence performance on financial abilities). Despite adjustment for these covariates, angular gyrus volume remained a key predictor of FCI performance, uniquely accounting for 12% of variance in scores. It could be argued that the relationship between MRI volume and financial abilities and IADLs is simply a proxy for the effects of global cognitive change on IADLs, although the data suggest that there is a specific relationship between financial abilities and angular gyrus volume in people with amnestic MCI.

Another important finding is the demonstration that arithmetic abilities, and to some extent attention, mediate the relationship between angular gyrus volume and financial abilities in people with amnestic MCI. The people with amnestic MCI in this study were more impaired than matched controls on performance on the FCI, consistent with prior reports.4,9 The Arithmetic score was the primary neurocognitive predictor of overall financial abilities in a prior study using a larger sample of people with amnestic MCI and the full FCI instrument.10 The findings of the current study further confirm the importance of mathematical ability in financial abilities by demonstrating the link with brain volumetrics of the angular gyri. Prior studies have demonstrated a role for the angular gyri in calculation abilities,48–50 and a recent study linked regional metabolic abnormalities of the angular gyri with dyscalculia in AD.12 The data from the current study add to the literature by demonstrating an association between angular gyrus volume and calculation ability in people with amnestic MCI and extending this finding to everyday financial tasks. Attention is also possibly a mediator of the relationship between angular gyrus volume and FCI in this study. In a prior study, an attention composite score that included Spatial Span was predictive of FCI domain scores in people with MCI.9 Posterior lateral cortical regions such as the angular gyri have been implicated in attention deficits in people with AD.13

Brain structures that show changes in early AD and MCI, such as the angular gyrus and posterior cingulate, are components of larger neural networks that are responsible for neurocognitive processes.51,52 In turn, changes in these neurocognitive processes probably result in impairment in complex daily activities. One such critical network that has received increasing attention is the default mode network, in which abnormalities have been detected in people with AD53 and MCI.54 This network has been implicated in executive function,55 memory,56 and goal-directed behavior57 and as such is probably a primary neural system mediating cognitive functions subserving complex daily activities. Thus, the current findings are an initial effort to identify brain structures that underlie neural network degradation associated with changes in complex daily activities.

One potential matter to address is the amount of overlap between the FCI and WRAT-3 Arithmetic score. Both measures are strongly correlated (0.74), and the Arithmetic measure predicted 55% of the variance in FCI scores. The financial domains measured using this version of the FCI were financial conceptual knowledge, bank statement management, financial judgment, and bill payment. Of these, only the financial conceptual knowledge domain directly involves calculations (figuring percentages on tax). The other domains assess abilities such as demonstrating knowledge regarding financial affairs (bank statements, bills, taxes), locating information on bank statements and bills, organizing bills, and exercising financial judgment in fraud scenarios. An exploratory analysis of correlations with domain- and task-level performance on the FCI found that angular gyrus volume is correlated with conceptual tasks and pragmatic tasks requiring calculation (data not shown). Application of written arithmetic problems such as those measured using the WRAT-3 requires not only calculation abilities, but also conceptual knowledge regarding calculations (division, fractions, percentages), visual scanning, and working memory (e.g., see the correlation between Arithmetic scores and Spatial Span in Table 3). It has previously been reported that performance on WRAT-3 Arithmetic was impaired in people with AD, who showed a variety of errors such as operation substitution (addition rather than subtraction) that indicate possible degradation of semantic knowledge regarding calculations, as well as digit substitution and other inattentive errors.58 One possible implication is that the angular gyri are important not only for performance of calculations, as well as reading and writing,50,59 but also for knowledge of mathematic operations and financial concepts. Such a supposition would require further study.

This study represents one of the first to directly link structural brain measures in amnestic MCI with a direct-assessment measure of financial abilities. A grossly neglected area of research has been investigations of the neuroanatomical basis of changes in IADLs in older adults using neuroimaging. Only a few such studies have been reported, and none to the authors' knowledge with a clinically diagnosed population of people with amnestic MCI. In one study, MRI volumes of cortical gray matter and the hippocampus were predictive of contemporaneous clinician ratings of IADLs based on caregiver interview in healthy individuals and people with AD.18 Another study using an informant-based rating of IADLs (Informant Questionnaire on Cognitive Decline in the Elderly) examined the correspondence between these ratings and hippocampal volume and white matter hyperintensities in a mixed group of English and Spanish speakers, some of whom were cognitively impaired based on psychometrically defined thresholds.21 The difference between the current study and these other reports is the focus on direct assessment of a specific IADL, investigation of several ROIs, and study of a group of people with clinically diagnosed amnestic MCI. Additionally, the focus on MRI volumes within the MCI group probably influenced the lack of findings in the hippocampus, because both prior studies included imaging data from cognitively normal individuals in their statistical analyses. Other authors have indicated that hippocampal volumes in people with AD may fail to correspond with behavioral measures as an artifact of the extensiveness of hippocampal atrophy and more widespread effects on other brain regions.60 Such factors may be at play in people with MCI as well.

The clinical and future research implications of the current findings are grounded in the importance of understanding IADL changes for clinicians and families. Decline of IADLs is the phenomenological “face” of dementia to people with AD and their families, because increasing oversight and engagement of family members is typically required in the care of these people. Moreover, changes in IADLs such as financial capacity have public health significance, because such changes can be important markers of disease progression in MCI and are a crucial diagnostic criterion for clinically detecting conversion to dementia. Although it has been increasingly appreciated that cognitive changes in MCI can lead to changes in IADLs, it is not yet understood how neurodegeneration of important neural pathways lead to deficits and declines in IADLs. Such knowledge would be important for researchers to develop comprehensive models of IADL declines and will assist in early identification and care, as well as inform diagnosis and support instrument development.

Limitations of the current study should be acknowledged. Its approach to volumetric analysis may not correspond directly with volumetrics derived from manually traced volumes, and results obtained using such methods may not be directly comparable. Other findings have suggested that SPM-based approaches have high correspondence with manual tracing.39 Persons with amnestic MCI in this study are undergoing longitudinal follow-up as part of their participation in the UAB ADRC. Because the current article is based on a cross-sectional study, the ultimate clinical and neuropathological status of these participants is unknown. ADRC's observed rates of conversion to AD from amnestic MCI are consistent with the conventional annualized 15% rate of risk,61 indicating that a majority of the participants in this study will probably clinically manifest AD in the future. Correspondence of findings in a sample of people with mild AD would also be desirable.

In conclusion, findings supporting a relationship between MRI brain volumes and IADL deficits in people with amnestic MCI are presented. Financial abilities showed a robust relationship with volume of the angular gyri, and a measure of arithmetic abilities and possibly also attention partially mediated this relationship. These findings hold promise for further investigation of IADL declines in amnestic MCI and mild AD using neuroimaging techniques to increase knowledge of how brain atrophy and corresponding pathology and cognitive change may place these people at high risk for changes in complex daily activities such as financial capacity.

ACKNOWLEDGMENTS

Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper.

This study was funded by grants from the National Institute on Aging (ADRC—1P50 AG16582-10, Marson, PI; 1R01 AG021927-05, Marson, PI) and Alzheimer's of Central Alabama (Griffith, PI).

Author Contributions: All of the authors contributed to manuscript design, data analysis, and preparation and writing of the manuscript.

Sponsor's Role: None.

Ancillary