Dementia and Memory

Lars-Göran Nilsson, Nobuo Ohta, Lars-Göran Nilsson, Nobuo Ohta

• Publisher: Psychology Press
• Year: 2013
• ISBN:
  9781317916574
• Edition: 1
• Topic: Psychologie
• Subtopic: Geschichte & Theorie in der Psychologie

A negative effect of the ageing population is that more individuals are experiencing cognitive decline and some form of neurodegenerative disease. With the number of people experiencing dementia likely to double in the next 20 years, this change in society presents one of greatest challenges facing public health personnel in the 21 st century. The aim of this volume is to describe research that is in progress, and the major findings that have been obtained in the scientific study of dementia.

The chapters in the first section of the book focus upon early signs of dementia, and consider several approaches to finding early cognitive signs and biological markers of dementia. The second section considers whether dementia is inevitable for people who become very old, and features chapters on risk factors and proactive influences, cognitive reserve and intervention. Each chapter in the final section describes phenomena which are related to differences in function between memory systems, including anterograde memory in fronto-temporal dementia, and the role semantic memory and semantic cognition may play in developing an understanding of the development of the degenerative processes in dementia.

With contributions from world-class researchers in this area, the volume offers a concise overview of key findings in recent research on dementia and memory. It will be of great interest to researchers and advanced students of cognitive psychology, and to those working in related fields, such as gerontology, rehabilitation sciences, and allied health.

1
Introduction

Is Alzheimer’s disease predictable long in advance on the basis of memory performance?

Lars-Göran Nilsson

It is often stated in books and journals on aging that poor cognitive status is perhaps the single most disabling condition in old age (e.g., Sing-Manoux et al., 2012). It is also a well-cited fact that the number of elderly people is increasing in most countries around the world (Sundström, Rönnlund, Adolfsson, & Nilsson, 2013). One implication of this is that more and more people will experience cognitive decline and even some form of neurodegenerative disease. An estimated 36 million people suffered from dementia in 2010 – a number expected to nearly double every 20 years and hit 115 million by 2050. Such a rapid increase in dementia among the elderly over the next 30 years will have far-reaching economic implications for societies worldwide and will undoubtedly be one of the greatest challenges facing public health personnel in the 21st century (Alzheimer’s Disease International, 2009). Identifying factors that can delay the onset of dementia and approaches to promote further independent functioning is of great interest to both researchers and the general public.

There are serious attempts being made among researchers in this field to find biomarkers and early cognitive signs that might make it possible to identify individuals who are at risk and may show cognitive decline and perhaps will become demented much later. It has been proposed that outcomes such as dementia are the result of processes that go on for several decades (e.g., Launer, 2005). One consequence of this is of course that early detection of reliable biomarkers and cognitive performance profiles should make it possible to provide successful interventions to stop or postpone decline and dementia.

There is great variability in performance in cognitive aging. This variability has not been given the attention it deserves in research during the last three or four decades. This is somewhat surprising given that research on cognitive aging has flourished and increased in volume during this period. To illustrate this, it might be mentioned as a curiosity that, according to Web of Science, there were six papers published in this field in 1970, 14 papers in 1980, 83 in 1990, 1,797 papers in 2000, and 5,381 papers in 2010. A major reason for this development is of course the realization that there are many important issues to cover in this research. In addition to the large variability in cognitive performance mentioned, when does cognitive decline start, is dementia inevitable, and can markers be found early for persons who later will develop dementia are other issues that are intensively studied in many laboratories around the world, contributing to the great increase in published papers in this field. The present chapter will cover these issues in turn.

Large variability in cognitive performance

Generally, in reviews and in textbooks, it is presented as a fact that performance in cognitive tasks is decreasing in what is usually referred as normal aging. However, when analyzing data from a random sample, one can see that some participants are relatively stable across the life span, whereas others are declining in performance and still others may even improve. Rather than lumping all subjects together, it might be fruitful to take these differences for granted and try to form subgroups that can be characterized as differing in several variables, cognitive or noncognitive. Such a characterization can be done in many different ways and on the basis of several different factors.

This large variability in performance becomes very obvious in most studies published on successful aging. In one study from our own laboratory using a rather conservative definition of successful aging (Habib, Nyberg, & Nilsson, 2007), we demonstrated this very clearly. We defined individuals as successfully aged when they were 70 years of age or more and had a cognitive performance among the top 25% of all participants in the study with an age range of 35–80 years of age. In other studies on successful aging, less conservative criteria have usually been employed. For example, in the classical study on successful aging by Rowe and Kahn (1987), a more lenient criterion was used for defining successfully aged.

In the study by Habib et al. (2007), we used a technique referred to as Q-mode factor analysis (Reyment & Jöreskog, 1996) to identify individuals in the sample who were at the top end of the cognitive performance ranking. Stephenson (1935) developed the Q-mode factor analysis originally to study subjectivity, but the technique has later been adapted to study heterogeneity in a multivariate data-set by means of classification of cases on the basis of objective criteria. In Habib et al. (2007), we used 138 cognitive and noncognitive variables in the Betula dataset to identify the two endpoints normal aging and successful aging. The technique reduces the number of variables into a smaller number of latent factors that can show cognitive success. For example, we predicted that cognitive performance, education, and health would be good predictors for successful aging.

The data used for this enterprise emanate from the longitudinal, prospective Betula project (Nilsson et al., 1997; Nilsson et al., 2004). This project started in 1988 in Umeå, a city of about 110,000 inhabitants in northern Sweden. Participants have been sampled randomly from the population in Umeå. All participants were 35, 40, 45, 50, 55, 60, 65, 70, 75, and 80 when tested the first time to form a narrow-age sample. To date, there are six different, independent samples that have been formed this way for a total of 4,700 participants in the whole dataset. In the first sample (S1), tested for the first time at test occasion 1 (T1) 1988–1990, there were 100 participants in each of the 10 age cohorts. The second sample (S2) also included 100 participants in each age cohort, and they were tested the first time at T2 (1993–1995). S3 also included 100 participants in each of the younger cohorts, with a bit fewer participants in the three oldest cohorts. Participants in S3 were also tested the first time at T2. Slightly more than 85% of the S1 participants also took part in the testing at T2. A majority (10%) of those in S1 who did not participate at T2 had died during the five-year interval between T1 and T2. Three percent of the participants in S1 had moved far away from the Betula venue in Umeå and the economical situation did not allow us to travel to their new living environment for testing. There were also 2% of the S1 participants, who did not return for testing, because of various reasons. The most common cause given among the younger participants was that they were too busy with other commitments to participate. Among the oldest participants, the most common cause for not participating was poor health.

At T3 (1998–2000), S1, S2, and S3 participants were called back for testing. The return rate from T2 to T3 was about the same for all three samples, i.e., about 85%, and the reasons for not returning for testing were the same as previously. A new sample (S4) was also called to T3. These were 35, 40, 45,… and 90 years-old when tested at T3. There were 50 persons in each age cohort.

At T4 (2003–2004), S1 and S3 were called back to testing, with a return rate of slightly below 85%. A new sample (S5) was also invited to testing at T4. They were in the ages of 35, 40, 45,… 90 years with 50 persons in each age cohort.

At T5 (2008–2010), S1 and S3 were called back and there was a return rate around 80%. A new sample (S6) was again invited for testing. They were 25, 30, 35,… and 80 years of age when tested. There were 30 persons in each age cohort of this new sample.

A sixth wave of data collection (T6) will be conducted during 2013–2014. At this occasion, S1, S3, and S6 will be called back for testing. No new sample will be included.

The reason for bringing in a new sample at each new test occasion is to allow possibilities for estimating the size of potential test-retest effects. This effort has proven to be very valuable since it has provided data to show that cross-sectional and longitudinal data are very different throughout the adult life span (e.g., Rönnlund et al., 2005). This particular issue will be discussed more later.

At each test occasion (T1–T5), participants came to the Betula lab twice. At the first occasion, a thorough health examination including blood sampling and saliva sampling was conducted. The participants were also interviewed about health history by a nurse, and they were given several questionnaires on social life and demographic variables to complete. The second time they came to the lab, they were presented with an extensive cognitive test battery. The tests included in the battery have been presented in detail elsewhere (Nilsson et al., 1997; Nilsson et al., 2004; Nilsson et al., 2006). In brief, the tests in the battery covered cognitive domains such as episodic memory, semantic memory, working memory, visuospatial ability, speed of processing, and decision making.

Brain imaging was also conducted at three occasions. At T3, structural magnetic resonance imaging (sMRI) was conducted on 139 participants. At T4 and T5, 60 participants were given both structural and functional MRI. At T5, another 386 participants were given both structural and functional MRI. These latter participants will be called back at T6 for a longitudinal data collection on structural MRI including diffusion tensor imaging (DTI) and two protocols for functional MRI, one hippocampus protocol, and one frontal lobe protocol. Resting state was also examined in the MRI session.

In the Habib et al. (2007) study, we used data from Betula participants in S2 and S3 aged 50–90 years in the Q-mode factor analyses. The conservative criterion we used (70 years of age or older performing among the 25% best of all YYY participants). This very strict criterion gave us 17 participants who were classified as successfully aged. Among these participants, there was one person who was 85 years of age when tested at T2 (1993–1995). All participants in the Habib study were tested at T2 and retested at T3. A qualification for being classified as successfully aged was that they would remain in the top 25% category at both tests.

Some of those noncognitive variables characterizing those who we classified as successfully aged at T2 and remained in this category at T3 were education as the far most dominating variable, good health (especially health as expressed in subjective ratings), and, somewhat unexpectedly, natural teeth. With respect to this latter variable, we thought first that this was an expression of socioeconomic status. And it can easily be that such a component is involved in the effect. However, we also discovered in the literature that a Japanese research group was doing research on mice and rats on this particular topic. The Japanese colleagues offered two different hypotheses to explain their data showing less efficient water maze learning in animals for which they had pulled out teeth. One hypothesis was one of innervation. There is slight damage to the brain whenever a tooth is pulled out. The other hypothesis was that lack of teeth causes poor chewing and thereby less blood flow in the brain. In one recent Betula study, we aimed at applying these hypotheses to the study of own teeth on human cognition (Hansson et al., 2013). There is a slight support for the blood-flow hypothesis, such that the lack of own teeth has an effect on cognitive decline. However, there is no relationship between number of own teeth and dementia.

It is obvious that the same method could be employed to examine those elderly with the lowest cognitive performance to try to establish which variables characterize these persons. We have not yet researched this. However, in a recent study (Josefsson, de Luna, Pudas, Nilsson, & Nyberg, 2012), we aimed at establishing criteria for finding three groups of participants who differ cognitively: One group that is maintaining the cognitive performance at a stable, high level across a long time – 15 years across four test occasions in the Betula design. We might refer to this group as maintainers and among some of which might be classified as successfully aged. A second group includes participants who remain relatively stable or decrease slightly at the end of the period of 15 years. We might refer to this group as average. The third group we are referring to as decliners. They decline in performance across the same time period.

When does cognitive decline start?

On the basis of cross-sectional comparisons between young and old persons, it has been claimed that cognitive aging start as early as 20 years of age (e.g., Salt-house, 2009). The cross-sectional data on which such conclusions are made typically show an almost linear decrease in performance from 20 to 80 years or more (e.g., Park et al., 2002). In fact, such data have been around in the literature in this field since the 1930s (Jones & Conrad, 1933) and in many publications in more recent years (e.g., Salthouse, 2005; Salthouse, Atkindson, & Berish, 2003; Schroeder & Salthouse, 2004). In sharp contrast to these claims are arguments from other authors saying that cognitive performance remains stable with relatively little decline up to 60 or even 70 years of age (e.g., Aartsen, Smiths, van Tilburg, Knopscheer, & Deeg, 2002; Albert & Heaton, 1988; Plassman, Welsh, Helms, Brandt, Page, & Breitner, 1995; Rönnlund et al., 2005). These latter data are typically based on longitudinal data.

As described in a recent paper by Nilsson (2012), this contrast between cross-sectional and longitudinal data in a series of papers by Rönnlund and his colleagues on the basis of the Betula Study (Rönnlund et al., 2005), is crucial for understanding the development of memory across age and for understanding how cognitive decline develops across age.

In several recent studies (Rönnlund et al., 2005; Rönnlund & Nilsson, 2006; Rönnlund, Lövdén, & Nilsson, 2008; Nyberg et al., 2010), we have examined the results obtained by means of a cross-sectional design and a longitudinal design. Needless to say, the data from these two designs look quite different. An example of this on the basis of data from Rönnlund et al. (2005) is given in Figures 1.1 and 1.2.

As can be seen from the cross-sectional data presented in Figure 1.1, the decrease in performance for episodic memory from the age of 35 years to the age of 85 years is approximately linear. This linear pattern has also been observed in several other cross-sectional studies (e.g., Schaie, 1996; Park et al., 2002), and in some cases, the decrease in performance has been observed to start as early as at 20 years of age. Salthouse (2009) has argued that this gives a true picture of the development of episodic memory in adulthood and old age. Nilsson, Sternäng, Rönnlund, and Nyberg (2009) have argued that this cross-sectional pattern gives an overestimation of the age decrement in performance due to a confounding between maturational change and cohort-related influences. Raz and Lindenberger (2011) and Schaie (2009) have arrived at similar conclusions.

Based on the same participants in the Betula Study, longitudinal data for the same episodic memory tasks are presented in Figure 1.2. As can be seen, the data pattern is radically different as compared to the cross-sectional data presented in Figure 1.1. The longitudinal time frame of these data is 10 years measured at three points of measurement. In the longi...