Freund & Pettman, U.K. Reviews in the Neurosciences 21, 187-221 (2010) Structural Brain Changes in Aging: Courses, Causes and Cognitive Consequences Anders M. Fjell and Kristine B. Walhovd Center for the Study of Human Cognition, Department of Psychology, University of Oslo, Norway SYNOPSIS that between 25% and 100% of the differences between young and old participants in selected The structure of the brain is constantly cognitive functions can be explained by group changing from birth throughout the lifetime, differences in structural brain characteristics. meaning that normal aging, free from dementia, is associated with structural brain changes. This paper reviews recent evidence from magnetic KEY WORDS resonance imaging (MRI) studies about age- related changes in the brain. The main magnetic resonance imaging (MRI), diffusion conclusions are that (1) the brain shrinks in tensor imaging (DTI), cognition, cross-sectional, volume and the ventricular system expands in longitudinal, neuropsychology, cerebral cortex healthy aging. However, the pattern of changes is highly heterogeneous, with the largest changes seen in the frontal and temporal cortex, and in INTRODUCTION the putamen, thalamus, and accumbens. With modern approaches to analysis of MRI data, At least two reasons exist to study the brains of changes in cortical thickness and subcortical healthy elderly people: First, most people volume can be tracked over periods as short as experience changes in specific cognitive abilities one year, with annual reductions of between during aging /1/, especially related to performance 0.5% and 1.0% in most brain areas. (2) The on speeded tasks /2/, executive function /3/, and volumetric brain reductions in healthy aging are episodic memory /4,5/ (but see /6/). Such cognitive likely only to a minor extent related to neuronal changes are likely partly caused by age changes in loss. Rather, shrinkage of neurons, reductions of macrostructural brain properties. Magnetic synaptic spines, and lower numbers of synapses Resonance Imaging (MRI) can be used to quantify probably account for the reductions in grey the volume or thickness of specific brain structures matter. In addition, the length of myelinated in vivo, yielding a window into the human brain axons is greatly reduced, up to almost 50%. during aging. Thus, by comparing the brains of (3) Reductions in specific cognitive abilities—for young and elderly participants, and the brains of instance processing speed, executive functions, the same individuals scanned repeatedly as they and episodic memory—are seen in healthy aging. get older, we may achieve a better understanding Such reductions are to a substantial degree of the neuro-biological foundation for age-related mediated by neuroanatomical changes, meaning cognitive changes, and how changes in the brain may lead to changes in cognitive function. This Accepted: April 24, 2010 knowledge may illuminate why some healthy people experience higher rates of cognitive decline Address for correspondence: than others, and ultimately, whether ways can be Anders M. Fjell found to effectively counteract this process. Department of Psychology, University of Oslo The second reason for studying the brains of PB. 1094 Blindern, 0317 Oslo, Norway. healthy elderly people is that we must understand E-mail: [email protected] how the brain changes in normal aging to be able VOLUME 21, NO. 3, 2010 187 188 A.M. FJELL AND K.B. WALHOVD to identify age-related pathology, especially 0.5% /17,18/, enabling the identification and Alzheimer‘s disease (AD). In what ways is the tracking of brain changes over short periods (See atrophy in AD qualitatively different from that Figure 1 and Figure 2). seen in healthy aging? Recent research has shown that most brain areas that are atrophic in AD are also affected by healthy aging, although to a lesser COURSE OF STRUCTURAL BRAIN CHANGES extent /7/. Thus, attempts to differentiate AD from IN AGING healthy aging at a very early stage can either be targeted at identifying accelerated atrophy in To date, more than 50 cross-sectional MR-studies specific brain structures above atrophy seen in have tested the effects of age on the volume or normal aging /8,9/, especially in the hippocampus thickness of various brain structures (for a table of and the entorhinal cortex /10-12/, or at identifying 31 cross-sectional studies reporting correlations patterns of change across several brain regions that between age and subcortical volumes, see /19/). may indicate a pattern of atrophy characteristic of Methodological differences related to scan quality, AD and not characteristic of normal aging /13-15/. of 31 cross-sectional studies reporting correlations This paper has three major themes. First, we participant recruitment and screening, number of will present what is known about the effects of participants, brain structure measurement, and healthy aging on brain morphometry, and some statistical choices make it challenging to directly factors that may modulate these, including genetic compare results across studies. The general variations. A short review of the results obtained scientific consensus is that age influences total with a newer MRI technique, DTI, will also be brain volume, but there are large differences given, as this is a very promising measure related between structures in how strong the effects are. to white matter (WM) integrity. We will also Some structures are found to decline substantially discuss how the morphometric effects seen in in old age, while others appear better preserved. In healthy aging are different from the atrophy seen in addition, different brain structures show different AD. Second, we will discuss possible neuro- age trajectories. Some brain areas are declining biological foundations for the morphometric linearly from early in life, whereas others continue changes. What happens at the molecular level to increase in volume well into middle adulthood when a brain structure is reduced by e.g. 25% before eventually beginning to deteriorate in the during the course of the adult lifetime? The final later part of life. theme of this paper is the cognitive consequences Most studies of age-effects on the brain are of the macrostructural brain changes. We will cross-sectional, and only age-differences, not age- review the studies that have directly targeted the changes, can be observed in such studies. cognitive correlates of age-related brain changes, Especially, the issue of cohort effects is and discuss to what extent morphometry can be challenging when the age-span sampled often used to explain the decrements in specific exceeds 50 years. Thus, the few longitudinal cognitive functions often seen among healthy studies that exist are exceptionally important. Still, elderly. major methodological problems are also associated The advent of MRI yielded an opportunity to with longitudinal designs. First, scanner replace- study macrostructural characteristics of the human ments and upgrades make it very difficult to follow brain in vivo for the first time. Incredible the same group of participants over longer periods, developments in MR scanners and software have e.g. 10 or 20 years. Thus, only exceptionally do made MRI a tool of increasing value in the study longitudinal MR-studies span more than 5 years. of the effects on the brain of healthy aging and Second, selective drop-out constitutes a challenge, age-related degenerative disorders. For instance, and re-recruitment at each follow up may be the thickness of the cerebral cortex can now be necessary to counteract this. Finally, because measured with sub-millimeter accuracy /16/, and longitudinal studies require much larger efforts and changes in cortical thickness between two time finances, the samples sizes are usually smaller than points can be measured with an error as low as in cross-sectional studies. Even with these caveats REVIEWS IN THE NEUROSCIENCES BRAIN CHANGES IN AGING 189 Fig. 1: Steps in processing of MRI scans. The figure Fig. 2: Calculation of cortical thickness and illustrates the main processing steps in a freely comparisons among different brains. The figure available software package for calculation of illustrates output from processing of MRI scans, brain volumes and cortical thickness based on how different brains are registered and magnetic resonance imaging (MRI) scans compared, and the calculation of cortical (FreeSurfer, see surfer.nmr.mgh.harvard.edu/). thickness at each point (vertex) of the brain Figure made by Inge K Amlien. surface. Figure made by Inge K Amlien . in mind, longitudinal studies are very important GM loss in the cortex appears to be somewhat additions to the cross-sectional studies. We will greater than in subcortical structures /19,24,32/. start by giving an overview of the extant cross- When specific structures have been studied in detail, sectional MRI-aging literature, before comparing the results have unfortunately diverged substantially these results to the conclusions obtained by across studies, and differences in how a structure is longitudinal designs. defined and how the scans were segmented comp- licate comparisons. Adding to this problem, in several studies only a few structures are segmented, CROSS-SECTIONAL STUDIES OF AGE- making it difficult to assess the relative age-effect of EFFECTS ON THE BRAIN different structures. Fortunately, recent studies have addressed the issues of comparisons between The consensus from cross-sectional studies is different brain structures and samples