Longitudinal Alteration of Intrinsic Brain Activity in the Striatum in Mild Cognitive Impairment

ADNI dataset

Data used in the preparation of this article were obtained in December 2015 from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.ucla.edu). The ADNI was launched in 2003 by the National Institute on Aging (NIA), the National Institute of Biomedical Imaging and Bioengineering (NIBIB), the Food and Drug Administration (FDA), private pharmaceutical companies and non-profit organizations, as a $60 million, 5- year public-private partnership. The primary goal of ADNI has been to test whether serial MRI, positron emission tomography, other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of MCI and AD. Determination of sensitive and specific markers of very early AD progression is intended to aid researchers and clinicians develop new treatments and monitor their effectiveness, as well as lessen the time and cost of clinical trials. The Principal Investigator of this initiative is Michael W. Weiner, MD, VA Medical Center and University of California – San Francisco. ADNI is the result of efforts of many co-investigators from a broad range of academic institutions and private corporations, and subjects have been recruited from over 50 sites across the U.S. and Canada. The initial goal of ADNI was to recruit 800 adults, ages 55 to 90, to participate in the research, approximately 200 cognitively normal older individuals to be followed for 3 years, 400 people with MCI to be followed for 3 years and 200 people with early AD to be followed for 2 years. For up-to-date information, see http://www.adni-info.org.

Participants

The data used in this study were obtained from ADNI2 or ADNIGO where rs-fMRI data were collected using 3T scanners. We identified 20 amnestic MCI subjects aged 65–90 years with consecutive annual rs-fMRI and cognitive data from baseline to 2-year follow-up. We thereafter identified 15 age, sex, and education-matched HC subjects with the same length of rs-fMRI and cognitive data (see Table 1). Of note, our decision on sample selection was based on a balance between a reasonable length of follow-up for a longitudinal study and a properly involved sample size. The diagnosis of amnestic MCI was made by a psychiatrist or neurologist at each study site and reviewed by a Central Review Committee, based on a subjective memory complaint and performance on neurocognitive testing, including the Logical Memory II subscale of the Wechsler Memory Scale-Revised (score ≤8, cut-off adjusted for education level), the Mini-Mental State Exam (score 24–30), and the Clinical Dementia Rating (global score = 0.5).

Measures of AD pathology and health information at baseline

Amyloid-beta (1–42) (Aβ_1-42) and pTau were derived from the cerebrospinal fluid aliquots, measured using the multiplex xMAP Luminex platform (Luminex Corp., Austin, Tex., USA) with immunoassay kit-based reagents (INNO-BIA AlzBio3; Innogenetics, Ghent, Belgium). An Aβ/pTau ratio was used as the ‘AD signature’ for which lower Aβ/pTau ratios indicated an increased burden of AD pathology [24]. Memory and executive function were measured using two composite scores developed based on a serial factor analyses, reported in previous studies [25, 26]. The composite memory index was based on the memory-related domains of the Mini Mental Status Examination, Alzheimer’s Disease Assessment Scale-Cognition subscale, Rey Auditory Verbal Learning Test, and Logical Memory test. The composite executive function index was based on the Wechsler Memory Scale- Revised Digit Span Test, Digit Span Backwards, Category Fluency, Trails A and B, and the Clock Drawing Test. Depressive symptoms were measured using Geriatric Depression Scale (GDS). Of note, two groups differed in their memory composite score, did not differ at the Aβ/pTau ratio, executive function, or GDS (see Table 1).

rs-fMRI data acquisition and preprocessing

In the current study, rs-fMRI data were collected yearly at baseline, one, and two-year follow-ups (three time points in total). All subjects were scanned on a 3.0 Tesla Phillips MRI. The rs-fMRI imaging data were obtained by using an echo-planar imaging (EPI) sequence (TR = 3000 ms, TE = 30 ms, slice thickness = 3.3 mm, matrix = 64×64, spatial resolution = 3×3×3 mm³, number of volumes = 140, number of slices = 48). The resting state functional imaging data were preprocessed using DPARSF [27].

ALFF analysis

After removing the linear trend, data was filtered using band pass (0.01–0.08 Hz) to do ALFF analysis. The ALFF values were calculated and extracted using Resting-State fMRI Data Analysis Toolkit (REST) [28]. Briefly, the time series of BOLD signal was converted to the frequency domain using the fast Fourier transform. The square root of the power spectrum was then calculated and averaged across 0.01–0.08 Hz for each voxel. The averaged square root was defined as the ALFF at the given voxel [18]. We also further decomposed the frequency range into the slow-4 (0.027–0.073 Hz) and slow-5 (0.01–0.027 Hz) bands [21], and relevant frequency-dependent ALFF values were calculated for each participant. To reduce the global effects across participants, the ALFF of each voxel was computed by dividing the global mean value [18, 29]. For the regions of interest (ROIs), the striatum was divided into the caudate and putamen. Bilateral caudate and putamen were defined and ROI masks were generated according to the Automated Anatomical Labeling atlas [30] (Fig. 1). Individual participant’s ALFF values in the putamen and caudate were then extracted to examine their association with Aβ/pTau ratio.

Other data analyses

Data analyses were conducted using SPSS 22.0. MCI-HC group comparisons on sample characteristics were made using independent t-tests for continuous variables or χ² tests for categorical variables. Generalized Estimating Equation (GEE) (Models (1) to (3)) or Generalized Linear Model (GLM) (Models (4) to (6)) models with an identity link and linear scale response were used. For GEE analysis including within-subject factor (i.e., time or frequency band), Auto-regression (AR)-1 working correlation matrix was used to reflect the characteristics of correlations between times or the correlation between frequency bands.

The change over all three points (from baseline to two-year follow-up) of striatum within-group (model equation: y =β₀ + β₁Time + ɛ) wherein y refers to the ALFF value (full band, slow-4, or slow-5) for the region of the striatum under analysis at each time point, and “Time” was a considered a continuous variable.

Of note, models (4) to (6) uses change of striatal ALFF (full or frequency-depended band) or cognitive as the outcome (lower scores indicate greater decline) rather than model (1) to (3) using repeated yearly measures of striatal ALFF or cognition. All tests with False Discovery Rate (FDR)-adjusted p value at two-tailed and less than 0.05 were considered significant. Of note, FDR adjustment was applied to address multiple comparisons between main ROIs (bilateral caudate and putamen), frequency bands (slow-4 and slow-5), and cognitive domains (memory and executive function).

Longitudinal changes of striatal ALFF

We first examined within-group 2-year change of the striatum (analytical Model (1)). The full band ALFF decreased yearly by 0.03 unit (SE = 0.01, Wald’s χ² = 12.51, FDR-adjusted p = 0.002) in R-putamen in the MCI group; conversely, there was no change in the putamen or caudate in the HC group. Since we did not find any significant decline in bilateral caudate, the further analysis was only focused on putamen.

Next, we examined between-group putaminal change (analytical Model (2)). The ALFF of the R-putamen declined 0.03 unit faster yearly in MCI group, compared to the HC group (see Table 2 “full band”). Figure 2 displays the longitudinal changes of bilateral putamen between MCI and HC groups.

Association between longitudinal changes of putamenial ALFF and cognitive trajectory

Applying analytical Model (3), controlling for age, education, GDS, group, baseline Aβ/pTau ratios, and time, decline of ALFF in R-putamen was significantly associated with the decline of memory (B = 1.31, SE = 0.32, Wald’s χ² = 16.94, FDR-adjusted p < 0.001). Longitudinal changes of putamenial ALFF were not associated with the change of executive function over time (FDR-adjusted p > 0.10).

Association between change of putamenial ALFF and baseline Aβ/pTau ratio

Applying analytical Model (4), controlling for age, baseline Aβ/pTau ratio had a significant main effect on change of ALFF in R-putamen (B = 0.01, SE = 0.004, Wald’s χ² = 5.11, FDR-adjusted p = 0.048) (see Fig. 3A), but not in L-putamen (B = –0.005, SE = 0.008, Wald’s χ² = 0.52, FDR-adjusted p = 0.47). Thus, a one unit difference (reduction) in Aβ/pTau ratio resulted in a 0.01 decrease in ALFFs at two-year follow-up in the R-putamen. There was no main effect of group, or interaction effect of group and baseline Aβ/pTau ratio, suggesting that the association between AD pathology and decline of ALFF was similar across MCI and HC groups.

Frequency band-dependent effects

We examined the within-group change of the putamen for each subdivided band (slow-4 and -5) (analytical Model (1)). For MCI group, slow-4 band of the bilateral putamen (B: –0.02 to –0.03, Wald’s χ²: 5.37 to 12.12, FDR-adjusted p: 0.001 to 0.028), as well as slow-5 band of the R-putamen (B = –0.03, SE = 0.008, Wald’s χ² = 11.06, FDR-adjusted p < 0.001) declined significantly at two-year follow-up. There was no change for the HC group.

We examined between-group change of the putamen for each subdivided band (slow-4 and -5) (analytical Model (2)). ALFF in slow-4 band, but not slow-5, of the R-putamen declined significantly more at two-year follow-up in the MCI group, compared to the HC group (see Table 2).

Next, we examined the influence of frequency band on the relationship between baseline Aβ/pTau ratio and change of the putamen in ALFFs (analytical Model (5)) using the entire sample. Compared to the slow-5 band, there was a more pronounced relationship between Aβ/pTau ratio and ALFF change of the R-putamen in the slow-4 band (β₅ in Model (5): B = 0.004, SE = 0.002, p = 0.044). This result obtained controlling for age and diagnostic group (see Fig. 3B).

Finally, we did not find the influence of frequency band on the relationship between striatal and cognitive changes (analytical Model (6)).