Baseline characteristics of the participants (n=48) compared between the two amyloid deposition groups (positive and negative) are shown in Table 4. There were 22 (45.8%) women and 26 (54.1%) men. Visual assessment revealed that 20 participants (42%) had PiB-PET positivity whereas the rest (58%) did not show cortical amyloid deposition. The amyloid positive and -negative groups did not differ by age, sex, or educational level. The mean age of the amyloid negative participants at baseline was 70.2 (SD 5.9) and 71.6 (SD 3.6) for amyloid positive participants.
There was significant difference in the percentage of APOE-ε4 carriers and non-carriers between the PiB-PET groups (p=0.01). Among the 20 PiB-PET positive participants, 10 (52.6%) were ε4 carriers, compared with 4 (14.3%) carriers among the 28 PiB-PET negative participants. There were 14 (29.8%) APOE-ε4 carriers in this study population, and among them 10 (71.4%) were PiB-PET positive. Among ε4 non-carriers, 9 (27.3%) were PiB-PET positive.
Significant difference was also found in NTB executive function score, which was lower in the PiB-positive group (mean -0.22, SD 0.45) compared with PiB-negative (mean 0.19, SD 0.62, P .02) group. The rest of the cognitive function scores, subjective cognitive function and medical history variables did not show any significant differences between the amyloid status groups.
Table 4: Baseline sociodemographic and clinical characteristics of the participants by
NTB processing speed Score 0.16 (0.97) -0.10 (0.79) 0.32
NTB executive function Score 0.16 (0.60) -0.22 (0.45) 0.02
MMSE score 27.0 (1.77) 26.0 (1.79) 0.87
Hypercholesterolemia, n (%) 19 (67.9%) 16 (80%) 0.35
Myocardial infarction, n (%) 3 (10.7%) 2 (10%) 0.94
Zung depression score 37.5 (8.91) 34.9 (10.0) 0.39
Numbers are means (SDs) unless otherwise specified. P values are shown from t-test for continuous variables and Pearson ꭓ2 test for categorical variables.
Table 5 shows results from the correlation analysis. Significant correlations were observed between amyloid deposition and lower executive function score (r=-0.33 P=0.02), and more APOE-ε4 carriers (r=0.41 P=0.00). Other variables did not have any significant correlations with amyloid status.
Table 5: Correlations between baseline population characteristics and amyloid status (positive vs negative)
Baseline population characteristics Coefficient P-value
Age 0.15 0.33
Sex -0.10 0.50
Education -0.15 0.32
APOE4 0.41 0.00
Cognitive test performance
NTB total score -0.12 0.42
NTB memory score 0.13 0.39
NTB processing speed Score -0.20 0.18
NTB executive function Score -0.33 0.02
MMSE score -0.01 0.94
Subjective cognitive performance
Prospective memory 0.15 0.33
Retrospective memory 0.08 0.60
Total memory 0.13 0.41
Medical history
Hypertension -0.02 0.89
Hypercholesterolemia 0.14 0.36
Myocardial infarction -0.01 0.94
Cardiac Failure -0.12 0.40
Angina Pectoris 0.20 0.17
Coronary Bypass -0.12 0.40
Angioplasty 0.13 0.38
Diabetes 0.01 0.95
Depression -0.10 0.49
Zung depression score -0.15 0.36
Coefficients and p-values are shown from nonparametric Spearman tests.
Results from the logistic regression analysis, presented in Table 6, showed that participants with higher executive function score were less likely to be amyloid positive (OR=0.11, 95%
CI=0.02 – 0.65, P = 0.02). Other variables did not show any significant associations with amyloid status.
Table 6: Associations between baseline population characteristics and amyloid status after adjustment for age, sex and APOE
Cognitive characteristics Odds ratio 95% Confindence Interval P-Value Cognitive test performance at baseline
NTB total score 0.62 0.14 – 2.69 0.52
NTB memory score 2.45 0.62 – 9.62 0.20
NTB complex memory score 1.46 0.43 – 4.99 0.55
NTB processing speed Score 0.86 0.40 – 1.84 0.70
NTB executive function Score 0.11 0.02 – 0.65 0.02
MMSE Score 1.21 0.78 – 1.87 0.40
Subjective cognitive performance
Prospective memory 1.03 0.86 – 1.25 0.74
Retrospective memory 1.00 0.85 – 1.18 1.00
Total memory 1.01 0.92 – 1.11 0.84
Change in cognitive test performance over time
NTB total score 0.11 0.01 – 1.50 0.10
NTB memory score 0.34 0.08 – 1.45 0.15
NTB processing speed Score 0.42 0.11 – 1.54 0.19
NTB executive function Score 0.86 0.12 – 6.11 0.88
NTB complex memory score 0.63 0.18 –2.28 0.48
MMSE Score 0.79 0.54 – 1.15 0.21
Odds ratios, 95% confidence intervals and p-values are shown from binary logistic regressions with amyloid status (positive vs negative) as the dependent variable. Analyses are adjusted for age, sex and APOE ε4 carrier status.
Table 7 shows the analysis of cognitive changes over time in relation with baseline amyloid status. Compared with individuals with negative PiB-PET status at baseline, those with positive PiB-PET status showed significantly poorer performance over time for NTB total score (estimate = -0.10, 95% CI= -0.20 – -0.01, P= 0.04), and NTB memory score (estimate = -0.18, 95% CI= -0.34 – -0.03, P= 0.02). However, there were no significant associations between baseline amyloid status and change in NTB executive function score (estimate = -0.02, 95%
CI= -0.13 – -0.09, P= 0.72) or NTB processing speed score (estimate = -0.06, 95% CI=-0.23 – 0.10, P=0.46).
Table 7: Associations between baseline amyloid status and changes in cognition over time Variables Estimate 95% Confidence Interval P-Value NTB total score
Estimates, 95% confidence intervals and p-values are shown from mixed-effects regression models with maximum likelihood estimation analyzing change in cognitive scores as a function of baseline amyloid status, randomization group, time, group × time interaction, and amyloid status x time interaction. The amyloid status x time interaction term showed the association between amyloid status and change in cognition, while the amyloid status term showed the association between amyloid status and cognition at baseline.
A graphical representation of change in cognitive performance over time for the amyloid positive and amyloid negative groups is shown in Figures 1-4. The figures show estimated means of cognitive scores at baseline, 1 year and 2 years, with higher scores indicating better cognitive performance. Error bars are CIs. The differences between amyloid groups regarding cognitive change from baseline to 2 years were assessed with the mixed-effects regression models (p values in figures shown for the amyloid status x time interaction).
6. DISCUSSION
6.1 Prevalence of amyloid positivity and relation to APOE ε4 carrier status
This study investigated the associations between brain amyloid status on PiB-PET scans and clinical characteristics in a population of individuals at increased risk for dementia but without dementia or substantial cognitive impairment. In this population, 20 (42%) of the 48 participants were amyloid positive. The percentage of the Aβ+ participants in this study surpassed the proportion of Aβ+ individuals in healthy older general populations. The approximate percentage for this age group has been reported to be around 20-30% (Chételat et al. 2013, Rowe et al. 2010). In previous studies, prevalence of amyloid positive status has been quite varied for populations without dementia. For the particular age group of the FINGER study, there have been studies with amyloid positive individuals ranging from 19-33% of the total population (Roberts et al. 2018). Another study having an age range of 55-90 years found 41% of individuals with positive amyloid status (Insel et al. 2016). Differences between studies may be due to the choice of age groups, or differences in how amyloid status was determined (e.g. the inclusion of various amyloid-PET tracers or CSF results). Also, FINGER participants were not an unselected healthy older general population. They were recruited for an intervention study aiming to prevent cognitive decline in a population with risk factors for dementia.
The percentage of APOE-ε4 allele carriers in this study (29.8%) was within the range previously reported for the Finnish population (Abondio et al. 2019). Results from this study revealed that 71.4% of the APOE ε4 carriers and 27.3% of the non-carriers had brain amyloid accumulation.
Another study in cognitively unimpaired individuals from the ADNI cohort has reported that 58% of the APOE ε4 carriers were amyloid positive, compared with 34% among non-carriers (Insel et al. 2016). In the population-based Mayo Clinic Study of Aging, prevalence of amyloid positivity among ε4 carriers without cognitive impairment ranged between 18.5-53.1%
depending on age group (from 50 to 89 years) (Roberts et al. 2018). An earlier large meta-analysis reported between 28.6-67.8% amyloid positive individuals among ε4 carriers with normal cognition for the age ranging between 60-80 years (Jansen et al. 2015). This increased to 23.7-71.5% if subjective cognitive impairment (SCI) was present, and 55.9-82.1% if MCI was present (Jansen et al. 2015). Prevalence of amyloid positivity in the FINGER PET sub-study of at-risk individuals thus seems to be in between values reported for cognitively normal general populations and populations with SCI/MCI.
APOE-ε4 carrier status was significantly related to higher likelihood for amyloid deposition.
This is in line with previous studies reporting links between APOE genotype and brain amyloid accumulation. The APOE-ε4 allele has a strong association with the risk of Alzheimer’s disease dementia (ADD), and also impacts the age during the onset of ADD (Morris et al. 2010, Suri et al. 2013). This robust relation of APOE with Aβ+ status makes individuals with ε4-related genetic vulnerability a significant target population for future treatment related trials (Boehm-Cagan & Michaelson 2014, Liao et al. 2014).
6.2 Amyloid status and cognition
In cross-sectional analyses at baseline, a significant difference was observed in NTB executive function score between amyloid positive and negative groups. The amyloid-negative group performed better in comparison to the amyloid-positive group in this cognitive domain, even after adjusting for age, sex or APOE genotype. Rest of the cognitive measures (NTB total score, memory and processing speed, and MMSE score) did not differ by amyloid status. Amyloid status was also not associated with subjective memory complaints.
Executive function can be described as a group of cognitive competencies that are essential to design, supervise and perform a chain of goal-driven complex activities (Royall et al. 2002). A poor result on executive tests is regarded to depict a range of cognitive impairments such as poor initiation, organization, or planning (Lezak et al. 1995). Vascular and metabolic abnormalities have been associated with a decline in executive function in the general population (Knopman et al. 2009, Schuur et al. 2010). Although executive impairment is regarded as more characteristic for vascular cognitive impairment, it is increasingly recognized as present in the early phases of AD as well (Levit et al. 2020). It is possible that the presence of vascular and lifestyle risk factors in the FINGER participants may explain the observed association between amyloid status and executive functioning.
In longitudinal analyses in the FINGER PET sub-study, amyloid positivity at baseline also had a significant effect on cognitive changes over time. Mixed‐effect models indicated that, compared with amyloid negative participants, the amyloid positive group had poorer performance over time in the NTB total score and NTB memory domain. There were no differences in these NTB scores at baseline between the amyloid groups. Decline in memory performance is one of the core clinical characteristics of amnestic MCI and one of the initial symptoms of AD (Kemppainen 2018). Following the inclusion criteria for the principal
FINGER study, participants with somewhat poorer memory scores than the normal estimate for age were included. Hence, it can be suggested that the mild memory impairment over time is due to the amyloid accumulation.
Although amyloid positive individuals had poorer executive function at baseline compared with amyloid negative individuals, the trajectories for change in executive function over time were not significantly different between the groups. Amyloid status also had no significant impact on the change in NTB processing speed.
It is not yet fully clear to what extent the effects of brain amyloid accumulation on cognition are domain-specific, especially in asymptomatic at risk or disease stages. Studies have previously reported conflicting results when it comes to investigating the association between cognition and amyloid status in cognitively normal older populations. While some of the studies found an association with various cognitive domains or episodic memory (Petersen et al. 2016, Pietrzak et al. 2015), others reported no significant relations (Oh et al. 2012, Rowe et al. 2010, Wirth et al. 2013). For instance, Petersen et al. 2016 study did find an association between cognitive impairment and non-memory domains by Aβ status. From population characteristics to sample size, many factors could cause inconsistency. Although age and educational level are often the most powerful predictors of cognitive scores, some associations with amyloid accumulation were seen even after adjusting these factors (Pike et al. 2011).
A 2013 Hedden et al. meta-analysis study comprising 34 cohorts of cognitively normal aging populations reported relations between Aβ-positivity and subnormal performance in cognitive performance such as global cognition, memory, and executive function. It is probable that cognitive scores are partially related to the amyloid status from the whole spectrum (Hedden et al. 2013). Contrarily, one study found no association between Aβ deposition with cognitive decline in PD (Melzer et al. 2019). Another study found significant results for poor visuospatial memory performance but no relation with cognitive complaints and verbal memory performance (Konijnenberg et al. 2019).
6.3 Amyloid status and other population characteristics
No significant relation of age, sex and education with the amyloid status was found in this study.
The age-specific prevalence of brain amyloid load in cognitively normal individuals has been previously reported to complement the age-specific prevalence of ADD when the interval is
20-30 years (Jansen et al. 2015). It is possible that the narrower age range of FINGER participants may explain the lack of association between age and amyloid status. However, the results of this study are in line with other neuropathological studies reporting no difference in positive amyloid status by sex (Barnes et al. 2005, Jansen et al. 2015).
Although the participants in this study had vascular and lifestyle risk factors for dementia, there were no statistically significant associations between clinical characteristics such as cardiovascular conditions, diabetes or depression and amyloid deposition. It might be due to the limited sample size of the study leading to loss of some of the associations. Furthermore, the study did not take into account medication for diabetes mellitus, hypercholesterolemia or hypertension as a modification of this interconnection between blood pressure and cholesterol and amyloid accumulation has been demonstrated (Glodzik et al. 2016). Even though autopsy reports showed that DM patients had decreased amyloid accumulation in the brain, studies have demonstrated an association of increased risk of dementia among DM individuals (Ahtiluoto et al. 2010). Mechanisms are not yet fully clear, which is why it is important to further study early amyloid pathology in relation to vascular/metabolic conditions.
6.4 Strengths and limitations of the study
The FINGER trial has successfully recruited a well-characterized, general elderly Finnish population with detailed data on amyloid status, cognitive and clinical characteristics. These individuals had various risk factors to develop dementia but without substantial cognitive impairment, providing the opportunity to study amyloid positivity in a very early at-risk stage.
The main limitation of the present study is the small sample size which has limited the statistical power. The selection of the participants also is one of the limitations of this study. The PET sub-study participants were not necessarily diverse from the rest of FINGER participants except for slightly younger age (Kemppainen et al. 2018). However, they were participating in a lifestyle intervention trial. Usually, more health-conscious and educated individuals participate in such studies which might have affected the results. AD is also more likely to develop with increasing age. Older adults (age above 77 years) were not included in the FINGER trial. The inclusion of younger, healthier groups might have an influence in the results of this study. In addition, although the terms ‘negative and positive amyloid status’ have been used in this study, there is currently no generally accepted cut-off for amyloid pathology on PET scans, and research on this topic is still ongoing. The trial duration of 2 years was too short for the at-risk
participants to develop dementia, and dementia status will have to be determined during ongoing extended follow-ups of the participants.