Changes in volume with age—consistency and interpretation of observed effects

Neurobiology of Aging 26 (2005) 1271–1274 Commentary Changes in volume with age—consistency and interpretation of observed effects Terry L. Jernigan ∗ , Anthony C. Gamst Laboratory of Cognitive Imaging, University of California, San Diego and the VASDHS, 9500 Gilman Drive, La Jolla, CA 92093-0949, USA Received 6 May 2005; accepted 9 May 2005 These two MR morphometric studies by Walhovd et al. and Allen et al. add important information to the existing lit- erature about life-span changes in brain morphology. They are both relatively large studies with 73 and 87 individuals, respectively, and in both cases, the imaging and segmentation methods are carefully standardized and highly reliable. The methods themselves are quite different, of course, represent- ing relative extremes in the continuum of automation; and the authors have chosen to model their data using different statistical techniques. However, both studies have two major strengths in common—they included subjects over the wide age range from the early 20s to 88 years, and the investigators tested for evidence of nonlinearity in the age-functions they observed. Because these authors show the age-distribution of individual datapoints for many of the volumes they mea- sured, it is possible to compare directly the age-functions they observe. It should be noted that the data are presented somewhat differently in the two reports. Notably, the data reported in Allen et al. are shown graphically as plots of raw volumes with male and female subjects superimposed to show the gender differences. In this case the volumes exhibit the interindividual variability related to cranial volume (head size) and thus show larger scatter. In contrast, the datapoints shown graphically in Walhovd et al., are residual scores from which variability associated with an estimate of cranial vol- ume has been removed. Thus, these volumes show less scatter and exhibit no gender differences. Nevertheless, comparisons of the overall shapes of the age-functions are informative. Importantly, the inclusion of young adult subjects in these two studies served to clarify the impact of ongoing progres- sive volume changes that can be thought of as continuous with brain maturational effects. This is particularly clear for Corresponding author. Tel.: +1 858 622 5882; fax: +1 858 622 5890. E-mail address: tjernigan@ucsd.edu (T.L. Jernigan). 0197-4580/$ – see front matter © 2005 Published by Elsevier Inc. doi:10.1016/j.neurobiolaging.2005.05.016 the white matter volumes, known to continue to increase throughout childhood and into young adulthood. In these new studies the curvilinear form of these age-differences is clearly apparent, and the current results confirm that whether or not investigators have reported age-related volume reductions in white matter depends strongly on whether the ages of the sub- jects included were predominantly under 50 years, when the protracted adult myelination effects continue, or were over 60, after which these give way to the fairly precipitous losses observed in the present studies in the 70s and 80s. Combined data published previously in separate reports on brain matu- ration [5] and aging [2] are presented graphically in the figure below for comparison. They also exhibit a strongly curvilin- ear age-function. The quadratic function (shown as a dashed line) improves the fit markedly relative to the linear fit (solid line), though the addition of a cubic term adds little (dash-dot line). The volumes in the figure are presented as standardized residuals (removing variability associated with volume of the supratentorial cranial vault) to facilitate comparison to those shown in Walhovd et al. Taking into account differences in data presentation, the results across the three studies, using different morphometry methods, are very similar (Fig. 1). The hippocampus data presented in the two new studies reported here, also showing a curvilinear age-function, are especially important, as the nonlinearity was less apparent in earlier studies. The evidence that hippocampal volumes increase significantly, perhaps until 40 years of age, has important implications for the interpretation of previous stud- ies, and indeed for any evaluation of age-related volume loss in hippocampus. Previously, investigators have, either explic- itly or implicitly, made the assumption that volume loss in this structure is best measured by comparison of volumes in elderly subjects with those of young adults. However, the present findings suggest that such comparisons may conflate mid-life increases with late-life decreases. The implication