Resilience to cognitive impairment in the oldest-old: design of the EMIF-AD 90+ study

Table 2 summarizes the findings on the protective factors for cognitive impairment or dementia of the ten studies. A high level of education was found to be protective against dementia in the oldest-old and one study indicated that high cognitive activity, examined by looking at the time spend on reading, around age 90 years was related to resilience to dementia [4, 16‐18]. The influence of vascular comorbidities on cognition has been studied quite extensively in this age group. Most studies did not find an association between cholesterol levels and cognition in the oldest-old [15, 17, 19‐22]. Hypertension has mostly been found to be protective in the oldest-old, especially when hypertension is diagnosed after the age of 80 years [17, 19, 23‐27]. This is in contrast to studies that have shown a higher dementia risk in the presence of midlife hypertension [28]. In addition, although midlife diabetes mellitus has been related to dementia in younger subjects [29], the influence of diabetes mellitus on cognition might be less evident in the oldest-old [11, 30, 31]. The protective effect related to the absence of stroke seemed to persist in the oldest-old [18, 32] and one study on atrial fibrillation and dementia did not find an association [32]. The absence of depressive symptoms seemed to be associated with resilience to cognitive impairment, which is consistent with findings in younger subjects [14, 33, 34]. One study related sleep quality to cognition and reported a higher sleep quality in subjects without cognitive impairment, which is in line with results in younger subjects [35, 36]. With regard to the sensory system, visual and auditory impairments have been associated with worse cognitive functioning in the oldest-old [37, 38] and although olfactory impairment has been associated with incident dementia in a younger age group [39], no studies were found studying this in the oldest-old.

Data about physical performance and activity have been collected in the Leiden 85-plus study and The 90+ Study. Good physical performance, measured with handgrip strength, 4 m walk or standing balance tests, was associated with better cognitive functioning and lower dementia incidence in the oldest-old but high physical activity did not seem to influence dementia incidence [16, 40, 41].

With regard to genetics, the Apolipoprotein E (APOE) genotype, a major risk factor for AD in younger subjects, has been extensively studied in the oldest-old, with mixed results regarding the relation to cognition and dementia [42‐46]. The Danish 1905 birth cohort, PLAD and Vantaa 85+ Study also studied a number of other genotypes in the oldest-old and found some additional protective and risk genotypes which are described in Table 2.

Hallmarks of aging [47], such as inflammation and cellular senescence [48], have been scarcely studied in relation to cognition in the oldest-old. The Leiden 85-plus Study and The 90+ Study related inflammation markers to cognition and dementia but showed mixed results [49‐51]. In addition, telomere length measured in white blood cells were not associated with cognition, dementia prevalence or incident dementia [52].

Limited information is available about the relation of markers of neurodegeneration, such as amyloid β and tau measured in cerebrospinal fluid (CSF) and/or with a positron emission tomography (PET) scan with cognitive impairment in the oldest-old. Postmortem studies have shown that the prevalence of amyloid aggregation increases with age in cognitively healthy subjects but decreases in the oldest-old subjects with dementia [1]. A similar trend can be seen with regard to amyloid β measured in CSF or on an amyloid PET scan [53, 54]. In subjects without dementia, greater amyloid load has been associated with poorer cognitive functioning and a higher rate of incident dementia, although the number of oldest-old subjects in these studies was limited [55‐57]. There are a few studies that have related brain MRI measurements in the oldest-old to cognitive functioning. Less atrophy and fewer white matter hyperintensities were seen in subjects without dementia compared to subjects with dementia [58, 59] but white matter integrity was not related to cognition [60]. In younger subjects, neurophysiological measures on magnetoencephalography (MEG) have been related to dementia [61] but it is unknown whether this relationship persists in the oldest-old.

Data about the medical and family history, medication use, education and intoxications (alcohol use and smoking) were collected through a structured interview, in combination with information provided by the study partner (if available), general practitioner and/or medical specialist.

In Amsterdam, subjects were asked to complete six questionnaires. Activities of daily living (ADL) were evaluated by use of the Katz ADL [67]. Functional health and wellbeing were evaluated by the Short form-12 Health-related Quality of Life (SF-12 HRQoL) questionnaire [68] and by the Cognitive Complaints Index (CCI) [69]. Nutrition was evaluated by the Mini Nutritional Assessment (MNA-long version) [70]. Sleep disorders were evaluated by use of the Berlin Questionnaire which identifies the risk of sleep disordered breathing [71]. Cognitive activity during life, such as reading books and playing games, was assessed with the cognitive abilities questionnaire [72]. Subjects with cognitive impairment filled in the questionnaires together with a study partner. The GDS was filled in together with the researcher [66].

In Amsterdam, the study partner was asked to complete five questionnaires: the AD8 (an 8-question test for the study partner to assess mild dementia) [73], the Amsterdam instrumental Activities of Daily Living (iADL) scale (a study partner based tool aimed at detecting iADL problems in early dementia) [74, 75], the Neuropsychiatric Inventory Questionnaire (NPI-Q, to assess the severity of behavioral symptoms in the subject and the distress these symptoms cause in the study partner) [76], the Mayo Sleep Questionnaire (MSQ, to screen for the presence of Rapid Eye Movement (REM) sleep disorders) [77], and finally the CCI [69].

In Manchester, subjects were asked to complete the SF-12 HRQoL questionnaire [68], the Physical Activity Scale for the Elderly (PASE) [78], the CCI [69] and the cognitive abilities questionnaire [72]. The study partner was asked to complete the AD8 [73], the Functional Activities Questionnaire (FAQ) [79] and the CCI [69].

The neuropsychological assessment took approximately one and a half hours during which several cognitive domains were tested. Table 4 gives an overview of the different cognitive tests that were administered, which domain they examine and at which site they were performed.

In Amsterdam, data on waist and hip circumference (cm), and hand grip strength (kg), as well as a standard neurologic screening examination were recorded during the first home visit. Hand grip strength was measured to estimate muscle strength and was performed with a hand dynamometer (Jamar hand dynamometer; Sammons Preston, Inc., Bolingbrook, IL., USA) [80]. In addition, a ‘Diagnostick’ was used to determine whether the subject had atrial fibrillation by measuring one derivative of an electrocardiogram [81]. At the end of the first home visit, the subject was asked to wear an accelerometer (DynaPort MoveMonitor, McRoberts B.V., The Hague, The Netherlands) for seven days to measure physical activity and sleep quality.

During the hospital visit in Amsterdam, continuous blood pressure measurements were performed non-invasively using a digital photoplethysmogram on the right middle finger (Nexfin®, BMEYE, Amsterdam, The Netherlands), resulting in beat-to-beat BP data. The Short Physical Performance Battery (SPPB) included balance tests, a 4 m walk to measure walking speed and the chair stand test [82]. Body composition, including the Body Mass Index (BMI), was measured using a Bioelectrical Impedance Analysis (BIA; InBody 770; Biospace Co., Ltd., Seoul, Korea).

In Manchester, waist and hip circumference (cm), hand grip strength (kg), BMI, resting blood pressure, heart rate, ankle/brachial pressure index [83] and a 4 min walking test were recorded at the clinical research facility.

Measurements of the sensory system were only performed in Amsterdam. With regard to visual functioning, best corrected visual acuity was tested with an Early Treatment Diabetic Retinopathy Study (ETDRS) chart. Intra-Ocular Pressure (IOP) and refraction data of all subjects were obtained, and all subjects underwent slit lamp examination and indirect fundoscopy. Pupils were dilated using tropicamide 0.5% and phenylephrine 5%. Peripapillary Retinal Nerve Fiber Layer (pRNFL) thickness and macular (layer) thickness were measured with Spectral Domain Optical Coherence Tomography (SD-OCT, Heidelberg Spectralis) using Heidelberg’s build-in software [84]. With enhanced depth imaging, the choroid was imaged and its thickness was (manually) measured. With fundus photography (Topcon TRC 50DX type IA), we acquired digital fundus images (50°). From these, seven Retinal Vascular Parameters (RVPs) were obtained using Singapore I Vessel Assessment (SIVA, version 3.0) [85].

For the auditory function, we used the digits-in-noise (DIN) test [86]. The DIN test is a speech-in-noise test using digit triplets as speech material. The digit triplets are presented against a constant level of stationary background noise. The test uses an adaptive procedure to determine the signal-to-noise ratio at which a listener understands 50% of the digit triplets correctly (i.e. the speech reception threshold (SRT) in noise). Olfactory function was measured using “Sniffin’ Sticks” (Burghart, Wedel, Germany). The test consists of pen-like odor dispensing devices with odors that are considered to be familiar. The smell test in the present study contained the odor identification part of the test [87].

In both centers, blood samples were collected according to the biobanking pre-analytical Standard Operating Procedures (SOPs) of the Biomarkers for Alzheimer’s disease and Parkinson’s disease (BIOMARKAPD) project [88]. Blood samples were collected for DNA and RNA analysis, inflammation markers, proteomics, neurodegenerative markers (amyloid β, tau, neurofilament light), routine blood analysis (i.e. lipids and glucose), vitamin status (B12 and folic acid) and, in Amsterdam only, for Peripheral Blood Mononuclear Cells (PBMCs). Planned DNA analysis includes Single Nucleotide Polymorphisms (SNP) analysis of known genetic risk factors for AD or amyloid pathology [89‐92]. DNA and RNA isolation will be performed by EMIF-AD partners. Remaining samples will be stored for future biomarker identification and validation studies.

In Amsterdam, four millimeter skin biopsies were taken from the inner upper medial arm and will be stained for senescence markers p16, p53 and telomere associated foci.

In Amsterdam, up to 20 mL CSF was obtained by lumbar puncture in Sarstedt polypropylene syringes using a Spinocan 25 Gauge needle in one of the intervertebral spaces between L3 and S1. A half mL CSF was immediately processed for leukocyte count, erythrocyte count, glucose, and total protein. The remaining CSF was mixed and centrifuged at 1300–2000 × g at 4 °C for ten minutes. Supernatants were stored in aliquots of 0.25–0.5 mL and frozen within two hours at − 80 °C and stored for future biomarker discovery studies. The processing and storing of CSF was performed according to the BIOMARKAPD SOP [88]. Amyloid β 1–42, total tau and phosphorylated tau 181 will be analyzed in a single batch. Remaining samples will be stored for future biomarker identification and validation studies.

At both sites, a duplex ultrasound scan of the carotid artery was performed. In Amsterdam, the right carotid artery was scanned to assess intima media thickness and distension using ArtLab software [93, 94]. In Manchester, left and right carotid arteries were scanned to determine velocity, vessel thickness, stenosis and plaques, rated according to the North American Symptomatic Carotid Endarterectomy Trial guidelines [95].

Subjects underwent locally optimized brain MRI protocols including 3D-T1, fluid attenuated inversion recovery (FLAIR), susceptibility weighted imaging (SWI), diffusion tensor imaging (DTI) and resting state functional MRI (rs-fMRI). MRI scans were performed on Philips 3 T Achieva scanners. Additionally, in Manchester regional cerebral blood flow was measured by arterial spin labelling [96], but no DTI scan was acquired in Manchester. In Amsterdam, if a subject could not undergo the MRI scan, we considered a CT scan (Philips Ingenuity TF or Gemini TF camera). Scans will be analyzed locally and centrally by EMIF-AD partners using the Neugrid infrastructure if applicable (see Additional file 1).

[¹⁸F] Flutemetamol, a specific fibrillary amyloid β radiotracer, was used for the amyloid PET scans. In Amsterdam, [¹⁸F] flutemetamol was produced by General Electric (GE) Healthcare at the Cyclotron Research Center of the University of Liège (Liège, Belgium) and PET scans were performed using a Philips Ingenuity TF PET-MRI scanner (Philips Medical Systems, Cleveland, Ohio, USA) or, in case of a PET-CT scan, the Philips Ingenuity TF (Philips Medical Systems, Best, the Netherlands) or Gemini TF scanner (Philips Medical Systems, Best, the Netherlands). In Manchester, [¹⁸F] flutemetamol was produced at the Wolfson Molecular Imaging Centre (WMIC)‘s Good Manufacturing Practice radiochemistry facility using GE Healthcare’s FASTlab and cassettes and PET scans were performed using a High Resolution Research Tomograph (HRRT; Siemens/CTI, Knoxville, TN). In both centers, the emission scan was performed in two parts. First a 30 min dynamic emission scan was started simultaneously with a bolus intravenous injection of 185 MBq [¹⁸F] flutemetamol. The second part of the scan was performed from 90 to 110 min post injection. In Amsterdam, immediately before each part of the PET scan a T1-weighted gradient echo pulse MRI or low dose CT scan was obtained. This MRI or CT scan was used for attenuation correction of the PET scan. In Manchester, two seven minute transmission scans, one before the first emission scan and the other after the second emission scan, using a ¹³⁷Cs point source were acquired for subsequent attenuation and scatter correction.

All [¹⁸F] flutemetamol scans were read visually as positive or negative. Additionally, we determined time activity curves for each region of interest with cerebellum grey matter as input function [97]. The dynamic data were analyzed on a voxel-by-voxel level using the Simplified Reference Tissue Model 2 (SRTM2) [98, 99]. Finally, we investigated tracer uptake by using a simplified method: the standardized uptake value ratio (SUVr, target to grey matter cerebellum SUV over 90–110 min pi) [100]. Variability in acquisition of amyloid PET scans were reduced by harmonizing acquisition protocols and will be reduced by adding it to the analyses as a covariate.

In Amsterdam, MEG was performed using a 306 channel whole-head system (Elekta Neuromag Oy, Helsinki, Finland). Eyes-closed and eyes-open resting-state MEG data were recorded with subjects in supine position inside a magnetically shielded room. We will use transformed time series [101] to extract spectral properties (relative band power and peak frequency) [102], and estimates of functional connectivity between brain regions, and metrics that characterize the topology of the functional brain networks [103, 104]. These analyses will be applied using Elekta’s beamformer software, and both in-house developed Matlab tools and BrainWave software (http://​home.​kpn.​nl/​stam7883/​brainwave.​html).