Brain metabolism and cerebrospinal fluid biomarkers profile of non-amnestic mild cognitive impairment in comparison to amnestic mild cognitive impairment and normal older subjects

Participants

Community-dwelling older adults (≥60 years old) were recruited through community announcements and invited to participate in meetings hosted in recreation centers throughout the city of São Paulo, Brazil. Subjects reporting any cognitive complaints were evaluated by a neurologist (FHGP). Cognitive complaints needed to be reported by the subjects and were confirmed by a collateral source, usually a relative or spouse. To be included in the study, subjects had to have completed 4 years or more of formal education, or what constitutes “primary education” in Brazil. Older volunteers without cognitive complaints were also recruited as members of the control group (CG). All participants underwent a complete neurological and psychiatric evaluation, a comprehensive neuropsychological test battery, and an assessment for symptoms of depression and anxiety. After each initial assessment, a final diagnosis was established by a consensus of neurologists with expertise in cognitive and behavioral neurology.

The revised Petersen et al. [13] criteria were used to diagnose individuals with MCI. Classification of aMCI or naMCI was determined according to the presence or absence of impairment in memory tasks (which defines aMCI), or other cognitive domains in the neuropsychological assessment [3, 13]. We considered a cognitive function or domain to be impaired if the test Z scores on that function were greater than 1.5 below the appropriate mean for age and education on one test, or if the Z scores were between 1.0 and 1.5 below the appropriate mean on more than one test of the same cognitive domain [14]. MCI patients were classified into aMCI or naMCI according to the criteria described. The domains impaired in the naMCI group were: executive/attention (n = 25), executive/attention and language (n = 5), and executive/attention and visuospatial function (n = 2).

The cognitive battery used to screen subjects included the Mini-Mental State Examination (MMSE) [15, 16] and the Brief Cognitive Battery [17] followed by the clock drawing test [18]. This battery has been shown to have good accuracy for the diagnoses of early dementia [19]. The following neuropsychological tests were used to define naMCI and aMCI: memory tests (visual reproduction and logical memory subtests of the Wechsler Memory Scale (revised), delayed recall of the Rey–Osterrieth Complex Figure, and the Rey Auditory Verbal Learning Test); constructive abilities (Block Design subtest of the Wechsler Adult Intelligence Scale, and copy of the Rey–Osterrieth Complex Figure); visual perception (matrix reasoning); and attention and executive functions (Trail Making Test (Parts A and B) and the Stroop Test). Application, scoring, and interpretation of the results obtained for all tests were performed according to each reference guide. Scores were adjusted according to appropriate age and education norms. Additional details of the evaluation protocol used, and neuropsychological tests administered, are reported in de Gobbi Porto et al. [20].

Exclusion criteria included: volunteers with clinically relevant psychiatric symptoms meeting DSM-IV criteria (the Geriatric Depression Scale (GDS) [21] and the Geriatric Anxiety Inventory [22] were administered to all MCI patients to evaluate the intensity of depressive and anxiety symptoms, respectively; patients with scores >5 on the GDS were excluded); any uncompensated clinical comorbidity, such as cardiac failure or anemia; history or presence of signs of neurologic diseases, such as Parkinson’s disease, epilepsy, inflammatory disease or stroke, with the exception of migraine; presence of any drug abuse (especially alcoholism); cognitive decline intense enough to interfere with daily activities according to clinical judgment or a score of ≥5 on the Functional Activities Questionnaire [23]; cognitive decline consistent with a diagnostic of dementia according to clinical judgment; diabetes mellitus without adequate glycemic control in the last 2 weeks (excluded because of possible interference with the [¹⁸F]FDG-PET); and presence of neoplastic or significant vascular lesions on the MRI scan, according to the judgment of a neuroradiologist (EAAFF) and one of the authors (AMNC). Individuals were not excluded based on the presence of white matter hyperintensities (WMH) on the MRI scan. The presence of WMH was an exclusionary criterion only if associated with focal neurological signs or gait impairment in the neurologic examination. No patient was using cholinesterase inhibitors or memantine. Antidepressant use was not strictly exclusionary; patients using antidepressants were allowed to participate if they were on a stable dose for at least 3 months and did not have symptoms of an active psychiatric disease at the time of screening.

This research project was approved by the ethics committee of the Hospital das Clínicas da Faculdade de Medicina da Universidade de São Paulo, and was in agreement with the provisions of the Declaration of Helsinki. All participants signed a term of consent.

CSF analysis

Of the 65 MCI patients enrolled, 54 (aMCI, 28 patients; naMCI, 26 patients) agreed to have a lumbar puncture to assess amyloid beta peptide (Aβ), tau, and phosphorylated tau (p-tau) protein levels in the CSF. Lumbar punctures were performed after an 8-hour fast, always between 8:00 and 10:00 a.m. A 22-gauge spinal needle was inserted between the fourth or fifth lumbar vertebral body by a trained neurologist. Approximately 10 ml CSF were taken. All of the samples were collected in polypropylene tubes, briefly centrifuged, and stored at a temperature of −80 °C. Total tau, p-tau, and Aβ were determined quantitatively using a commercial sandwich enzyme-linked immunosorbent assay (Innotest hTAUAg Innotest _amyloid 1–42; Innogenetics, Ghent, Belgium).

MRI acquisition

After the initial work-up, all patients underwent a 3.0 Tesla MRI scan to exclude structural diseases and also for co-registration with [¹⁸F]FDG-PET images. Following the procedure outlined in Fazekas et al. [24], a neuroradiologist (EAAFF) performed a visual semi-quantitative assessment of a WMH scale, using the axial fluid-attenuated inversion recovery (FLAIR) sequence. The following MRI sequences were acquired: sagittal 3D T1, axial T2 FSE, axial fluid-attenuated inversion recovery (FLAIR), coronal T2 SPIR, and diffusion.

[¹⁸F]FDG-PET imaging acquisition

To ensure blood glucose levels lower than 180 mg/ml, participants fasted for at least 4 hours prior to the intravenous injection of 370 MBq [¹⁸F]FDG in a peripheral vein. Following the tracer injection, patients rested with eyes open and ears unplugged for 60 minutes in a calm, silent and slightly darkened room. Acquisition of the [¹⁸F]FDG-PET data was run for 15 minutes (matrix = 256 × 256, zoom = 2.5, pixel size = 1.04 mm) using a Siemens Biograph PET-CT scanner (CTI/Siemens, Knoxville, TN, USA). Images were reconstructed with the ordered subset expectation maximization method (with six interactions and 16 subsets) and then smoothed with a 5 mm Gaussian filter. Data were also corrected for scattering, attenuation, and decay. Attenuation correction was performed using single helical computed tomography.

[¹⁸F]FDG-PET imaging processing and analysis with SPM8

The following procedures were validated previously [25] and are used as a standard for PET and single photon emission tomography (SPECT) explorative analyses. All PET images of the participants were co-registered with the their MRI images (volumetric T1 sequence) and spatially normalized in SPM8 software (Wellcome Department of Cognitive Neurology, Functional Imaging Laboratory, London, UK) into a standard stereotactic space, based on the SPM8/Montreal Neurologic Institute space, using a 12-parameter linear affine normalization and a further nonlinear iteration algorithm. This was performed using an SPM8 template for [¹⁸F]FDG-PET. Each of the scans was also individually smoothed with a Gaussian kernel to reduce the impact of misregistration into template space and to improve the signal-to-noise ratio. Images were then smoothed with a 4 mm full-width Gaussian kernel at half-maximum Gaussian filter. To ensure the analysis only included voxels mapping cerebral tissue, a default threshold of 0.8 of the mean uptake inside the brain was selected. Global uptake differences between brain scans were adjusted using the “proportional scaling” statistical parametric mapping (SPM) option. Once adjusted, radioactive counts for each participant were normalized with the average global counts of each group (i.e., the “global means” normalization approach). The relevant peak voxels were identified in terms of coordinates according to Talairach and Tournoux with the help of the Talairach Client software (Research Imaging Institute, University of Texas Health Science Center, San Antonio, TX, USA), and after conversion from the SPM/Montreal Neurological Institute space [26, 27].

Statistical analysis

Clinical and CSF data analyses were performed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). PET data were analyzed on a voxel-by-voxel basis using the SPM8 software (Wellcome Department of Cognitive Neurology) in conjunction with MATLAB R2009a (The Mathworks In., Natick, MA, USA). An analysis of variance (ANOVA) test was used to search for regional brain glucose metabolism (rBGM) differences across the three groups (naMCI, aMCI, and CG).

Post-hoc analyses with nonpaired t tests were used to examine differences between each pair of groups. SPM8 maps were generated with p <0.001 and the threshold for significance at the voxel level was set at p = 0.001 (Z score = 3.09) with a minimum extension of 10 voxels in the corresponding cluster. This is a classical threshold for PET and functional MRI studies [25], and has been used previously in studies related to MCI [9, 11, 28]. The initial exploratory analyses with SPM maps generated a t statistic for each voxel, thus constituting statistical parametric maps.

To explore the homogeneity of rBGM alterations between groups in particular areas, the higher Z scores within each map were identified, and a volumetric region of interest (ROI) in the corresponding cluster of voxels was generated. Subsequently, numeric values representing [¹⁸F]FDG uptake measures in that cluster for each individual (after the whole normalization process) were obtained with the toolbox MarsBar for SPM [29] under the option “explore design/files and factors”. With this approach, regions of particular interest—voxels in the precuneus—could be better explored.

A directed analysis of ROIs was also completed with the SPM8 software (small volume correction (SVC)). Areas in the temporoparietal association cortex were chosen based on previous reports of regions with typical metabolic impairment in early AD [30, 31]. The following areas were analyzed: posterior cingulate cortex, precuneus, angular gyrus, middle temporal gyrus, and inferior temporal gyrus.