The thesis used data from the exploratory PET sub-study of the Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER). FINGER trial was a multi-center lifestyle intervention among 1260 older at-risk individuals. It was the first randomized controlled trial showing the possibility to prevent cognitive impairment with a multi-domain lifestyle intervention (Winblad et al. 2016). The two-year intervention involved nutritional guidance; exercise; cognitive training and social activity; and management of metabolic and vascular risk factors. Control group participants were provided with regular health recommendations. The principal result of the intervention was cognitive scores which was assessed through a modified Neuropsychological Test Battery (mNTB) (Ngandu 2014).
4.1 Selection criteria for FINGER participants
The individuals were selected among the survey participants of the Finnish type 2 diabetes prevention program in 2004 or 2007 (Saaristo et al. 2007) or the National FINRISK study in 1972, 1977, 1982, 1987, 1992, 1997, 2002 or 2007 (Ngandu 2014). These cross-sectional observational surveys were organized to study risk factors of chronic non-communicable diseases. The rate of participation in these surveys was ranging from 70% to above 96%
(Vartiainen 2010). The invitation to participate in the FINGER study was sent to the people aged 60–77 years old in 2009, and having Cardiovascular Risk Factors, Aging and Incidence of Dementia (CAIDE) Dementia Risk Score (Kivipelto 2006) of 6 points or more at the start of the study. CAIDE scoring encompasses easily measurable factors (sex, age, education, physical inactivity, hypercholesterolemia, hypertension as well as obesity) which are related to the risk of dementia. The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) neuropsychological battery (Moms et al. 1989) was used to evaluate the cognition of the participants during screening appointments. Everyone had to fulfil at least one of these criteria to be included in FINGER: (a) Mini-mental state examination (MMSE) ≤26/30 points; or (b) Word List Recall ≤75%; or (c) Word List Learning task (10 words × 3) ≤19 words. Participants having cognitive scores at average level or a little less than anticipated for age (according to Finnish population norms) were selected by these criteria (Ngandu 2014). Factors influencing safe participation in the intervention (especially physical exercise part), major depression, current malignant diseases, MMSE < 20, dementia/substantial cognitive impairment in accordance with a clinical interview, hearing or communicative ability, severe loss of vision, symptomatic cardiovascular disease, re-vascularisation within a year, factors hindering
co-operation as evaluated by the study physician, in addition to coincidental engagement in any other intervention trial comprised the exclusion criteria. Randomization was done in sections of 4 participants (2 individuals randomly assigned to either intervention or control group) by the study nurse with computer software at each study site (Ngandu 2014).
4.2 The FINGER PiB-PET exploratory study
During the FINGER trial, brain 11C-PiB-PET imaging was available at the Turku PET Centre.
The PiB-PET exploratory study was conducted only in the FINGER cohort in Turku. When PET resources became available and if there were no contraindications, the most recently recruited FINGER trial participants were selected for the PET sub-study. In total, 48 individuals (22 female, 26 male, mean age 71.4 years, SD 5.2 years) from the Turku cohort of the FINGER primary study sample were able to engage in this PET sub-study. The evaluation methods for amyloid deposition and methods for the PiB-PET scan have been previously described in detail (Kemppainen 2018). Two experienced readers analysed the scans visually. The assessment was visually positive or negative following consensus agreement. Participants categorised as PiB positive presented cortical retention of 11C-PiB-PET in at least one cortical region classically affected in AD. In contrast, individuals categorised as PiB negative had nonspecific 11C-PiB-PET retention in white matter. With the automated ROI analysis, the scans were quantitatively assessed as well. A composite PiB retention score was determined as the average of the prefrontal, parietal, precuneus, anterior cingulate, posterior cingulate, and lateral temporal ROIs amyloid deposition (Kemppainen 2018). In this thesis, data on brain amyloid status collected in connection to the baseline FINGER visit were used.
4.3 Cognitive and other characteristics relevant for this project 4.3.1 Cognitive tests
Trained study psychologists carried out a series of standardized neuropsychological tests, i.e. a modified and extended form of the neuropsychological test battery, NTB (Harrison 2007) for a thorough cognitive evaluation. The total composite mNTB score involving 14 tests was the principal outcome measure. Other cognitive measures were determined as follows: the memory domain involved Logical Memory immediate recall (assessment range, 0–25 points) and delayed recall (assessment range, 0–25 points) of the Wechsler Memory Scale-Revised (WMS-R) (Wechsler 1997); Visual Paired Associates immediate recall (assessment range, 0-18 points)
as well as delayed recall (assessment range, 0–6 points); and Word List Learning (assessment range, 0–30 points) and Delayed Recall (assessment range, 0–10 points) of the CERAD test battery (Moms et al., 1989). Category Fluency Test (Moms et al. 1989), Concept Shifting Test (Van Der Elst 2006a) (condition C), Digit Span (Wechsler 1997), a minimized 40-stimulus adaptation of the earliest Stroop test (Golden 1998) (interference range 3–2) and Trail Making Test (Reitan 1958) (shifting score B-A) were included in the executive function domain. The processing speed domain involved Stroop test (condition 2), Letter Digit Substitution Test (Van Der Elst 2006b) and Concept Shifting Test (condition A).
The present study used data on NTB cognitive test measures from the baseline, 1 year and 2 year FINGER visits. MMSE data from baseline was also used.
4.3.2 Subjective cognitive complaints
A questionnaire regarding subjective cognitive complaints was provided to the participants (Prospective and Retrospective Memory Questionnaire, PRMQ) (Vaskivuo et al. 2019). The reason for this assessment was that self-reported experience of cognitive problems, or subjective cognitive decline in the aging population has been recognized as a potential indicator of non-normative cognitive change (Rabin et al. 2015). People with self-experienced impairment or decline in one or several aspects of cognitive function that are generally informant-corroborated but without showing any decline in daily activities have a higher risk of progressing to dementia (Rabin et al. 2015).
Data on subjective cognitive complaints from the baseline FINGER visit were used in this thesis.
4.3.3 Zung depression scale
The Zung scale is a self-rated depression scale developed by Zung in the 1960s (Zung 1965).
Individuals with potential depressive disorders may be identified through this scale. Although it has been widely used in research, the cut-off value for clinically significant depression is still controversial. Researchers may also use different cut off values for different countries due to geographical and cultural variations affecting the answers provided by the individuals. In this thesis, the continuous Zung score indicating level of depressive symptoms was used (data from the baseline FINGER visit).
4.3.4 Medical history
History of cardiovascular conditions (hypertension, hypercholesterolemia, myocardial infarction, cardiac failure, angina pectoris) or procedures (coronary bypass, angioplasty), diabetes, and depression was self-reported by the participants during the screening/baseline visit with the study physician.
4.3.5 APOE genotype
APOE genotype was assessed from venous blood samples using a method that has been described in detail previously (polymerase chain reaction and TaqMan genotyping assays for two single-nucleotide polymorphisms, rs429358 and rs7412, on the Applied Biosystems 7500 platform) (Solomon et al. 2018). APOE genotype was categorized into ε4 allele carriers and non-carriers. No separate group was defined for ε4ε4 homozygous individuals due to the very small number of participants with this genotype.
4.4 Statistical analyses
The statistical tests were conducted with IBM SPSS (version 25) and STATA (version 15) statistical software. All p-values were calculated with 2-tailed tests, and values of p < 0.05 were considered statistically significant.
For cross-sectional analyses at baseline, the difference in categorical variables (sex, APOE-ε4, etc.) between the amyloid-positive and amyloid-negative groups were analyzed with the Pearson χ2 test. The 2-sample t-test was used to test between-group differences in continuous variables such as age, years of education, and cognitive scores. Nonparametric Spearman tests were used for correlation coefficients between amyloid status and other variables. Logistic regression models with amyloid status (positive versus negative) as dependent variable, and clinical measures as independent variables were also conducted. The logistic regression models were adjusted for age, sex, and APOE-ε4 carrier status, and results are reported as odds ratios (ORs) with 95% confidence intervals (95% CIs).
The longitudinal relationships between PiB-PET status at baseline and change in cognitive performance over time were assessed with mixed effects regression models with maximum likelihood estimation (xtmixed command in Stata). In these models, change in cognitive scores during the 2-year trial were analyzed as a function of baseline amyloid status, randomization group, time, group × time interaction, and amyloid x time interaction. Results from these
models are reported as estimates, 95% CI for estimates, and p-values for associations between (1) baseline amyloid status and baseline cognition, and (2) baseline amyloid status and change in cognition over time.