Age dependency of risk factors for cognitive decline

Study sample

Data were derived from the on-going Longitudinal Aging Study Amsterdam (LASA) [15]. This is a longitudinal, population-based study in the Netherlands focusing on trajectories of physical, psychological, social and cognitive functioning in subjects aged 55 years and older. In 1992–1993 a random sample of men and women aged 55–85 years, stratified for age and sex, from three geographic areas of the Netherlands (Amsterdam, Zwolle and Oss) was included. Follow-up measurements were conducted about every 3 years. Data collection included a main and medical interview conducted in the homes of the subjects. The main interview was done by trained and supervised interviewers and the medical interview was performed by trained nurses. All subjects gave informed consent and the study was approved by the Ethical Review Board of the VU University Medical Center (VUmc), Amsterdam, the Netherlands and conducted according to the principles of the Helsinki declaration.

At the start of the study in 1992–1993, 3107 subjects were enrolled. To select cognitively healthy subjects at baseline, subjects with an age and education corrected MMSE lower than 27 points were excluded (this cut-off is based on the lowest 10th percentile of the MMSE in the Maastricht Aging Study (MAAS) [16], leaving 2527 subjects at baseline. In 1995–1996, 2545 subjects were re-examined. See for further details about the following cycles of this LASA cohort and for the sample size per risk factor Additional file 1: Table S1 and S2.

Measurements

Statistical analyses

Determination of the best-fitted spline regression model

For the longitudinal associations between the cognitive outcome measures and age in the total group, a linear spline regression model showed a better fit then a cubic spline regression model. A model with two knots placed at the ages 70 and 80 years showed the best fit for the three outcome measures, dividing the sample in three age groups: ≤70 years (young-old subjects), > 70–80 years (old subjects) and > 80 years (oldest-old subjects) (Fig. 1). The spline regression model showed a better fit then a linear regression model without splines for all the different associations.

Differences between age groups at baseline

Most characteristics differed between age groups (Table 1). Years of education, follow-up time, scores on the cognitive tests, total and LDL cholesterol levels, alcohol use, smoking and physical activity all decreased with age. The level of homocysteine and inflammation markers and the presence of cardiovascular comorbidities and depressive symptoms increased with age.

Association of risk factors with cognitive decline in the total sample

Age dependency of risk factors

The association of LDL cholesterol, homocysteine, hypertension, history of stroke, depressive symptoms, IL-6, ACT, alcohol use and smoking with cognitive decline differed between the age groups (Table 3 and Fig. 2). The presence of APOE ε4, total and HDL cholesterol level, a history of DM or MI, CRP level and physical activity did not show an age effect. In general, the regression coefficient changed from a negative association in the young-old and old subjects to a positive association in the oldest-old subjects. This means that on top of the decline in MMSE and 15WT as visualized in Fig. 1, the presence of these risk factors was associated with more decline in MMSE or 15WT in the young-old and old subjects and less decline in MMSE or 15WT in the oldest-old subjects. If we determined the association of the age-dependent risk factors with cognitive decline per age group, we found that hypertension, high IL-6 levels, and alcohol use were significantly associated with less cognitive decline in the oldest-old subjects (Table 3 and Additional file 1: Tables S5–S6). Smoking was significantly associated with more memory decline in the young-old subjects and high LDL cholesterol with more MMSE decline in the young-old subjects (Table 3 and Additional file 1: Table S6).

Risk factor	MMSE			15WT immediate recall			15WT delayed recall
	≤70	> 70–80	> 80	≤70	> 70–80	> 80	≤70	> 70–80	> 80
APOE ε4	0.72	−4.24	−10.17	−0.65	−3.85	−5.32	0.18	−4.15	−8.52
Total cholesterol	−3.37	−0.10	1.26	−1.08	− 1.09	1.53	0.97	−1.53	1.27
LDL cholesterol	−5.40*	0.66	1.83*	−0.18	−1.92*	1.96*	0.68	−2.62	1.41
HDL cholesterol	7.35	−2.17	−1.07	− 1.84	2.64	0.50	6.21	1.45	2.97
Homocysteine	0.39	−0.40	−0.92	− 0.39	− 0.21	0.13	− 0.76*	−0.08	0.13*
Hypertension	0.04	−2.14*	6.52*	−5.65*	−0.22	5.06*	−4.04	3.54	0.43
DM	−0.95	−0.46	−9.32	− 0.11	−3.49	−0.29	−1.68	−5.44	4.90
MI	2.42	−1.59	2.23	0.41	1.93	5.64	4.90	0.62	10.14
Stroke	1.25	−9.16*	9.16*	2.60	−0.34	6.70	0.79	−1.58	−4.30
Depressive symptoms	−0.07	0.02	−0.06	−0.21	0.00	0.07	−0.11*	−0.23	0.21*
CRP	−0.18	0.18	0.19	−0.21	0.03	0.19	−0.19	−0.11	0.16
IL-6	0.02*	−0.10*	1.31*	0.27	−0.11	0.64	0.31	−0.03	0.88
ACT	0.00	0.01	0.06	−0.01	− 0.01	0.03	− 0.01*	−0.04	0.03*
Alcohola: minimalb	2.82*	−0.26	7.93*	3.19	−3.09	−0.31	5.91	−3.03	0.25
Alcohola: moderatec	0.78	−0.95	3.48	2.85	−3.51	1.42	3.19	−0.47	−0.69
Smoking	−0.30	−1.05	−5.15	−5.84*	0.53*	−1.13	−4.52	−2.88	−0.61
Physical activity	0.00	0.00	0.02	−0.01	0.01	0.00	0.01	0.01	0.00

Shown are beta’s (multiplied by 100) of the associations of a risk factor with cognitive decline within each age group. They show the extra cognitive decline (next to the overall cognitive decline as visualized in Fig. 1) per age group in the presence of a risk factor. A negative beta indicates that a unit increase in the risk factor is associated with more cognitive decline. Bold beta’s indicate a significant (p < 0.05) association with cognitive decline in that age group (in Additional file 1: Tables S5–S7 we present the standard errors and p-values corresponding to the beta’s in this table per age group)

15WT 15 Words Test, ACT a1-antichymotrypsin, IL-6 interleukin-6, LDL Low-Density Lipoprotein, MMSE Mini-Mental State Examination

*Association of risk factor with MMSE or 15WT decline is significantly different between these two age groups. In case of three *: difference is significant between ≤70 and > 80 years old group and between > 70–80 and > 80 years old group. Beta’s are determined by linear mixed models in combination with splines and adjusted for sex and education

^aNo alcohol use is reference group. ^bWomen:1 drink/day, men: 1–2 drinks/day. ^cWomen: > 1 drink/day, men: > 2 drink/day

Post-hoc analyses with hypertension defined by the measured high blood pressure only or the use of antihypertensive medication only, yielded similar results (Additional file 1: Table S3).

Selection during follow-up

Subjects who were retained in the later LASA cycles had a lower age, higher level of education, higher scores on the cognitive outcome measures, higher cholesterol levels, less comorbidities, lower levels of inflammation markers, higher alcohol use, lower levels of smoking and had a higher level of physical activity at baseline compared to subjects who dropped out during follow-up (Additional file 1: Table S4).

APOE genotype

Our finding that the APOE ε4 genotype increases the risk for cognitive decline regardless of age is at odds with some earlier studies that reported a decrease of risk for dementia with age [11, 26] but is consistent with others [27]. Differences between studies may be explained by differences in selection of subjects (normal cognition or MCI) and outcome measure (progression to dementia or cognitive decline).

Cardiovascular factors

Most cardiovascular factors were associated with less decline at higher age than at younger age, which is in line with previous studies [12, 28]. High cholesterol in late-life can be an indicator of a better nutritional status and a better overall health and therefore associated with less cognitive decline [28, 29]. Additionally, cholesterol synthesis is thought to decrease with aging, but only in AD patients and not in subjects with a normal cognition [30]. The association of low cholesterol with more cognitive decline in the oldest-old subjects might therefore be an expression of underlying AD pathology.

Hypertension may prevent cognitive decline at old age by ensuring the cerebral blood flow [31, 32]. On the other hand, low blood pressure can be a consequence of neurodegenerative disease and therefore an early sign of dementia onset, although this is an aspect which one can also expect in the young-old subjects [33, 34].

Depressive symptoms

Earlier studies have shown that depressive symptoms are an important risk factor for cognitive decline and dementia, also among the oldest-old subjects [7, 35]. We replicated this finding for the total sample but also found that the association of depressive symptoms with memory decline was less in the oldest-old group than in the young-old group. This may be explained by the fact that older subjects score higher on the CES-D questionnaire for reasons other than depression, such as somatic morbidity [36]. In line with this explanation, earlier cross-domain latent growth models on LASA data demonstrated that delayed recall was associated with increasing levels of depressed affect, but not with depressive somatic symptoms [37].

Inflammation markers

Inflammation has been described as an important mechanism underlying cognitive decline but most of these studies were performed in younger subjects [38]. We did not find an association of inflammation markers with cognitive decline in the total sample but noted differences between age groups showing that higher IL-6 and ACT levels were associated with less cognitive decline in subjects aged 80 years and older compared to younger subjects. Potentially, higher inflammation markers in older subjects are a sign of a better inflammatory response and therefore related to better overall health and cognitive functioning.

Lifestyle factors

Minimal and moderate alcohol use were positively associated with decline in MMSE and memory functioning compared to no alcohol use, which is in line with earlier studies [3, 8]. We found that the positive association of alcohol use with cognition was strongest in subjects aged 80 years and older. In this age group, no alcohol use is frequently related to poor physical functioning. Therefore, the negative association of no alcohol use with cognitive decline is potentially an indirect effect [39].

In accordance with our findings, a meta-analysis in 2015 showed that the negative association of smoking with cognitive decline is decreasing with age [40]. Survival bias and the presence of competing risks are probably the most important phenomena to explain this finding [40–42].

In contrast to a meta-analysis of observational studies, we did not find an association between physical activity and cognitive decline [43]. However, a meta-analysis of intervention studies on the effect of aerobic exercise on cognitive decline in cognitively normal subjects did not find an effect [44]. We found that in the oldest age group less physical activity was associated with more cognitive decline, but because the over age interaction effect was not statistically significant, this finding should be interpreted cautiously.

Cognitive outcome measures

We used different cognitive tests as outcome measure (MMSE, 15WT total and delayed recall) and in the total sample, most variables showed similar findings for the different outcome measures and the course of the tests was very similar with age (Table 2 and Fig. 1). However, we also found different results for the different outcomes, which may be explained by the fact that the tests measure different disease processes; memory decline is presumed to be an early marker of Alzheimer’s disease and decline in MMSE can be caused by a broader range of diseases [45].

Strengths and limitations

This is the first study that analyzes the influence of age on the association of different types of risk factors with cognitive decline in one prospective cohort study. Earlier studies have indicated age differences but never studied the various risk factors in one cohort. Additionally, the use of a nonlinear analyze technique, is an important added value of this study to earlier literature. A limitation of this study is that we did not have data about the presence of cardiovascular risk factors, such as hypertension and high cholesterol, before age 55 years. Therefore, we could not discriminate between a high blood pressure and cholesterol emerging at high age or already present at younger ages. While we tested many risk factors with different outcomes at the same time, this also increased the risk of false-positive findings. However, we decided not to correct for multiple testing given the exploratory nature of the study and the increased risk of missing important findings when applying Bonferroni adjustments (type II errors) [46, 47]. For simplification purposes, we describe our results in relation to cognitive decline, although we only used the MMSE and 15WT to assess cognition. It therefore needs to be noted that these results cannot automatically be extrapolated to other forms of cognition. In addition, the 15WT is a test for verbal episodic memory and does not assess other types of memory such as visual and semantic memory.

Scientific and clinical implications

With this study we showed that age needs to be taken into consideration when studying risk factors for cognitive decline. It is not only needed to add age as a confounder but especially as an effect modifier to analyses as it changes the relation between a risk factor and cognitive decline. From a clinical perspective, these results suggest that different preventive strategies might be effective in young-old versus oldest-old subjects. Antihypertensive and cholesterol lowering medication might not be appropriate for the oldest-old subjects who develop hypertension and hypercholesterolemia at a high age.