Relation of education to brain size in normal aging

Subjects.

Subjects (table) were selected from participants of the Cardiovascular Health Study (CHS), an ongoing, multicenter, population-based observational study of 5,888 adults 65 years and older, including 2,495 men and 3,393 women.11-13 The major goal of the CHS is to identify risk factors related to the development and course of coronary heart disease and stroke in individuals living independently in the community. After providing informed consent, participants undergo extensive clinical evaluation (home interview and physical examination) and laboratory testing (including brain MRI, which is described later) at baseline, and annual follow-up assessment. Additional details of the CHS have been published.11

A detailed description of participant recruitment for the CHS has been published.12,13 For the current study, we identified from the CHS cohort a sample of 500 individuals recruited from two CHS sites (Pittsburgh and Hagerstown, PA) who gave written consent to participate in an ancillary investigation of cognitive functioning and aging (measures included the Mini-Mental State Examination [MMSE] and a neuropsychological test battery—these data will be the subject of a future report). (Saxton J, Ratcliff G, Newman A, et al., unpublished data).14 All available participants from these two sites in whom brain MRI was performed within 1 year of this cognitive testing were screened for inclusion in this study. We subsequently excluded from this cohort a total of 170 individuals for one or more of the following reasons: not right handed (individuals were determined to be right handed if they used their right hand to write, throw a ball, and brush their teeth),14 lifetime history of any psychiatric illness (including dementia) or of any illness or injury referable to the brain (per the CHS clinical evaluation described earlier), incomplete cognitive test data, incomplete MRI data (e.g., scan artifact, missing slices, etc.), or MR images with structural abnormalities (cortical infarct, n = 5; hydrocephalus, n = 1; tumor, n = 1; and markedly thickened calvarium, n = 1).15,16 The remaining 330 subjects were included in a previous study of sex differences in brain aging.17 For the current investigation we excluded 10 additional individuals with scores less than 24 on the MMSE (administered per the ancillary investigation of cognitive functioning described earlier) to ensure further the absence of cognitive impairment. A cutoff score of 24 on the MMSE is a well-accepted measure of significant cognitive impairment consistent with dementia in the elderly.18

The final sample consisted of 320 participants, 122 men and 198 women, ranging in age from 66 to 90 years (see the table). This sample was similar to the CHS population as a whole with regard to age (CHS mean ± SD, 72.77 ± 5.61 years), sex distribution (CHS, 59% female), and years of education (CHS mean ± SD, 12.35 ± 3.10 years). Years of education were determined by participants’ responses to a structured interview and were defined as the number of years of full-time education normally required to reach the reported final level (e.g., 10th grade = 10 years; bachelor’s degree = 16 years). Individuals with a general equivalency diploma (GED) were credited with 12 years. Of the 320 participants, 226 (71%) were taking medications for one or more medical conditions (see the table). No participant was taking medication known to affect brain size (e.g., steroids). Additional subject characteristics are given in the table.

Brain MRI technique.

As noted earlier, brain MRI was performed on all subjects as a result of their participation in the CHS. The standardized CHS brain MRI acquisition protocol has been described previously.19 MRI was performed on either a 1.5-T scanner (General Electric, Milwaukee, WI; n = 244) or a 0.35-T scanner (Toshiba, Rolling Meadows, IL; n = 76) at one of two CHS field centers (Pittsburgh and Hagerstown, PA, respectively). Head position was oriented in the scanner and was stabilized during the scanning procedure by the use of Velcro straps and foam head supports. To establish slice orientation, the first scanning sequence consisted of a T1-weighted sagittal series (repetition time [TR], 500 msec; echo time [TE] 20 msec; thickness, 5 mm; gap, 0 mm; and matrix, 128 × 256) centered at the midline to define the anterior commissure/posterior commissure (AC/PC) line. Then a second series of proton density-weighted images (TR, 3,000 msec; TE, 30 msec; flow compensated) and T2-weighted images (TR, 3,000 msec; TE, 100 msec; flow compensated) was obtained (thickness, 5 mm; gap, 0 mm; matrix, 256 × 192; number of excitations, 0.5 [one on the 0.35-T scanner]) oriented parallel to the AC/PC line, and extending from the vertex to the skull base. A third series consisting of T1-weighted (TR, 500 msec; TE, 20 msec) axial images was then obtained (thickness, 5 mm; gap, 0 mm; matrix, 256 × 192; number of excitations, 1) oriented parallel to the AC/PC line, and extending from the vertex to the skull base. Images were stored on nine-track magnetic tape.

Image analysis and brain morphometry.

For the current study the brain images were transferred from magnetic tape to read/write magneto-optical disks. Data were analyzed on a workstation (Power Mac 8100; Apple, Cupertino, CA) with a high-resolution color graphic monitor. The measurements of regional brain size were made on the recalled T1-weighted axial images by one of two trained technicians blinded to all subject characteristics. Window center settings were first standardized to ensure precision in boundary detection.20 Structures were identified with the help of brain and MRI atlases21,22 and then measured with a combination of computer-assisted edge detection and manual tracing, using graphic analysis software (version 1.4, MedVision; Imnet/Evergreen Technologies, Castine, ME). The area (in square centimeters) within the outline was calculated automatically; volume (in milliliters) was determined by multiplying the area by the slice thickness and summing over the multiple slices in which the structure appeared (described later).

The following regions were defined for volume measurement17: Intracranial volume (ICV) was defined by the internal surface of the diploe23 and measured in every slice between the vertex and the superior border of the midbrain (approximately 12 to 15 slices per individual were measured). Intracranial size could not be measured reliably inferior to this level because of the presence of structures such as the globes and sinuses. As such, this measure is an underestimate of the true total intracranial volume. There was no significant correlation between age and intracranial volume.

Cerebral hemisphere volume was measured in every slice between the vertex and the skull base (approximately 18 to 20 slices per individual). Ventricular volumes were excluded from this measurement.

Lateral ventricle volume was measured in each slice on which lateral ventricles were present. We also measured the various subregions of the lateral ventricles including the body, the frontal horns, the posterior horns, and the temporal horns.

Third ventricle volume was measured in each slice beginning at the level of the foramen of Monro and extending inferiorly to the superior border of the midbrain (approximately three to four slices per individual).

Peripheral (sulcal) CSF volume was a calculated value derived by subtracting the cerebral hemisphere and ventricular volumes from the intracranial volume for each slice on which intracranial volume was measured. As such, this measure is an underestimate of the true total peripheral CSF volume.

Lateral (Sylvian) fissure CSF volume provided an indirect estimate of atrophy of the temporal lobe, as well as of the frontal and parietal lobes. The lateral fissures were measured in each slice on which they were present, beginning at the level of the foramen of Monro. When the lateral fissure communicated freely with the peripheral CSF, the anterior boundary of the fissure was defined by a horizontal line connecting the anterior tip of the temporal lobe to the medial temporal region.

It was not possible to subdivide the cerebral hemisphere reliably into its various lobes (i.e., frontal, temporal, parieto-occipital) because of difficulties in establishing boundaries for such subregions in the axial plane of orientation.24 Nevertheless, a regional brain morphometric analysis was possible on one of our axial slices. For this analysis, we followed the method of Pearlson et al.25 and chose a T1-weighted axial slice that passed through both the pineal gland and the foramen of Monro (hereafter designated as the region-of-interest [ROI] slice). This slice is approximately one slice above the AC/PC line and is especially suited to subregional analysis because it contains both gray and white matter, it is not dominated by CSF spaces, and it contains anatomic regions believed to be associated with performance on a number of neuropsychological tests.25 Using the boundary definitions of Pearlson et al.,25 the following four subregions were defined for area measurement on the ROI slice (ventricular areas were excluded from all regions; figure 1):

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Region of interest slice. Typical T1-weighted axial brain MRI at the level of the foramen of Monro, demonstrating the subdivisions of the region of interest slice. Reprinted with permission from Archives of Neurology 1998;55:169–179. © 1998, American Medical Association.17

• 1. Frontal region area: The posterior border of this region was defined by a horizontal line perpendicular to the posterior aspect of the genu of the corpus callosum.

• 2. Temporoparietal region area: This region was the area situated between the frontal lobes anteriorly and the parieto-occipital lobes posteriorly, and was bordered medially by the internal capsule.

• 3. Parieto-occipital region area: The anterior border of this region was defined by a horizontal line intersecting the anterior atria of the ventricles.

• 4. Intracranial area (IA): This region was defined by the inner surface of the diploe (as stated earlier).

Extensive reliability studies of our measurement techniques have indicated that area/volume measurements of these regions are highly reliable.23 On the basis of a randomly selected sample of 10 brains from the current study,17 intraclass correlation coefficients for interrater reliability of the two raters ranged from 0.85 (for small regions such as the third ventricle) to 0.99 (for large regions such as the cerebral hemisphere). Similarly, intraclass correlation coefficients for intrarater reliability ranged from 0.84 to 0.99.

Preliminary analyses.

Using exploratory methods, the data were examined for outliers and extreme values by means of box plots and normal quantile–quantile plots. These analyses demonstrated no need for transformation. In addition, multiple linear regressions, using the initial subject variables model (described later), were conducted on untransformed and logarithmically transformed outcome variables. The residuals from these regressions were examined by means of deviation plots and normal quantile–quantile plots, again to assess whether the outcome variables needed transformation. The results of these analyses also indicated that the untransformed data best fit the assumptions of normal-theory linear regression.

Regression analyses.

The outcome variables consisted of the cerebral volumes, the left/right differences for the relevant cerebral volumes, the cerebral areas from the ROI slice, and the left/right differences for these cerebral areas. To assess the effect of education on these brain volumes (areas), the variability in volumes (areas) was first adjusted for relevant subject variables and confounder variables using the following procedure.

An initial subject variables model contained four predictor variables known to be associated with regional brain size18: IV (for volume data) or IA (for area data), sex, age, and the age-by-sex interaction. This initial model was reduced to a final subject model by a backward, bidirectional, stepwise regression procedure using Mallow’s Cp as the criterion for removal of any of the four variables. Mallow’s Cp assesses the competing demands of the loss of precision with too many predictors and the gain in bias with too few. All variables in the equation had to contribute to the minimization of Cp. Any variable whose presence in the equation failed to decrease the Cp was eliminated. We next adjusted for the effects of two potential confounder variables—MR scanner assignment and presence of hypertension/ischemic heart disease (see Discussion). These two variables were appended to the final subject model, and that model was then reduced by the same Cp method to a final model, with the constraint that no predictor in the final model could be eliminated at this stage. The effect of education was then assessed by appending it to the final model and testing it for significance. Hence the effect of education on a given brain volume (area) was always adjusted for the presence of relevant subject and confounding variables.

Regression diagnostics.

Regression diagnostics were conducted to determine whether there were data points in the sample sufficiently anomalous to alter the regression results. The regression was checked for consistency (Did any participant’s data drastically alter the precision of the overall regression?), leverage (Did any participant’s data drastically alter the precision of the regression coefficients?), influence (Did any participant’s data drastically alter the estimate of the regression coefficients?), and normality (Did any participant’s data drastically alter the normal distribution of the regression residuals?). These diagnostic analyses revealed no anomalous data.

Methodological limitations.

Our findings are subject to certain potential limitations. Although cross-sectional studies of age effects allow for relatively efficient and rapid acquisition of large amounts of data, they are subject to secular effects such as birth cohort.17 This effect refers to the possibility that brain size, like cranial size, may exhibit systematic changes over successive birth cohorts in the general population. If such trends actually exist in the population at large and if they are not secondary to secular trends associated with correlates such as cranial size (note in the current study that intracranial volume was not correlated with age), then an assessment of the true effects of aging per se on brain volume will require longitudinal investigation.

A second issue relates to the health status of our participants. Our sample represents a group that may be somewhat healthier than the entire elderly population because of selection criteria for the CHS and the current study.13,14 As such, our findings may not be applicable to the entire population of seniors. There is also heterogeneity of health status within our participants, in that 29% were free of major systemic illness whereas 71% had at least some mild physical disease, corresponding to the distinction between successful and usual aging.26 Such differences in health status could account for differences in brain aging, and indeed systemic disease such as hypertension has been found to be associated with changes in brain structure.19,27,28 However, as noted earlier in the statistical analysis section, the presence of hypertension/ischemic heart disease was not confounded with the main effects of education in our study. Finally, it is possible that at least some individuals in our sample may have experienced undetected cognitive decline despite our screening and exclusionary criteria. Still, our use of an MMSE cutoff score of 24 is widely accepted in the literature18 and represents an appropriate strategy for selection of a sample that is representative of the community-resident “usual aging” population.

Third, the cognitive reserve hypothesis relies on the assumption that education level truly reflects baseline level of overall intellectual functioning. Although this assumption is reasonable, assignment of education level in some cases may be arbitrary (e.g., those with a GED were assigned 12 years) and the education level reached by a person may be determined by factors other than cognitive potential (e.g., economic, military service, sex roles, etc). This assumption may also be limited by cohort effects wherein older individuals are less likely to have completed higher education. Thus, future studies might consider examining measures of intellectual capacity in addition to education (e.g., occupation, special hobbies such as chess, test data such as vocabulary score). Finally, because our study was limited somewhat by fewer participants with low education levels, additional research is needed with larger numbers of individuals with low levels of education.

Fourth, our data do not permit a discrimination between a direct effect of education on brain structure and an effect that is mediated by differential survival of more highly educated individuals in the community. One formulation of the reserve hypothesis would suggest that individuals with brain atrophy and low education may be more likely to become demented than either low-educated persons without atrophy or high-educated persons with atrophy, leading to their underrepresentation in a community-dwelling cohort. A higher risk of death would have a similar effect. If such reasons for differential dropout from our community resident cohort were the explanation for our findings, then we would expect to have seen an increasing relationship between brain atrophy and education with age (i.e., an age-by-education interaction), but no such relationship was found. Still, because individuals with brain atrophy and low education may have reduced survival relative both to low-educated persons without atrophy and to high-educated persons with atrophy, prospective data are required to determine whether education is truly associated with age-related brain atrophy.

Fifth, there exists the possibility of a type I error in our findings given that multiple brain regions were tested and the p value for the education effect was only less than 0.03. Thus, our analyses must be considered exploratory, and our finding of an education effect requires confirmation by additional “hypothesis testing” research.

Finally, the measurements of regional brain size in our study are subject to certain limitations. First, because of limitations inherent in the CHS MRI acquisition protocol, our analyses of brain size were restricted to the axial plane (three-dimensional reconstruction was not possible without dramatic loss of resolution).24 Second, accurate delineation of regional boundaries can be affected by several sources of technical error, including improper window center settings, magnetic field inhomogeneity (resulting in spatial distortion of objects and object pixel nonuniformity), and differences in MRI technical variables.20 The effects of these variables were minimized in this study by use of a set of procedures that has been shown to optimize the accuracy of MRI size measurements.20,23 Third, field strength differences between the two scanners could affect estimates of brain size.24 As noted earlier in the statistical analysis section, scanner assignment was not confounded with the main effects of education in our study.

Relation of education to age-specific cerebral atrophy.

We found that the age-specific increase in peripheral CSF volume—a marker of diffuse cortical atrophy—was significantly greater in persons with higher education. For example, among elderly persons of similar age, sex, and intracranial size, those with 16 years of education had approximately 8 to 10% (depending on age) more peripheral CSF volume than did those with only 4 years of education (see figure 2). These observations are consistent with the reserve hypothesis, although the magnitude of the education effect is relatively small (e.g., effect size, 0.015; r = 0.12; see figure 2); education accounts for only 1.44% of the variance in peripheral CSF volume). Thus, the biological impact of this education effect requires elucidation.

In the only other imaging study of normal individuals of which we are aware, Kidron et al.9 did not find a relation between education and peripheral CSF volumes of frontal, temporal, or parietal regions on MRI (1.5 T) in 20 elderly volunteers (mean age ± SD, 73.5 ± 5.4 years) from the community. In an MRI (1.5-T) study of 60 patients with probable AD disease, Mori et al.29 likewise found no relation between education and peripheral CSF volume (calculated as the difference between intracranial volume and brain parenchymal volume). The negative findings from these two studies may have been a result of low power due to relatively small sample sizes, or to differences in measurement technique.

With regard to cerebral parenchymal size, we found no association of education with the age-specific decrease in cerebral hemisphere volume. These results replicate our earlier findings in a different sample of healthy adults (n = 66, 30 to 91 years old) in which education was unrelated to total cerebral hemisphere volume on MRI (1.5 T).23 Similar negative findings have been reported by the three other studies that have examined this issue in normal individuals.30-32

Our finding of a significant association of education with the age-related increase in peripheral CSF volume stands in contrast to the absence of an education association with age-related volume loss of cerebral hemisphere brain matter. Taken together, these findings suggest that although peripheral CSF volume may show a greater age-related increase in those persons with higher education, such associations of education with cortical atrophy may not be apparent due to statistical measurement error. For example, in our study the standard error of measurement for the cerebral hemispheres (100.0 mL/320^1/2) was roughly twice that for the peripheral CSF (49.9 mL/320^1/2). Thus, for changes in cerebral hemisphere volume and peripheral CSF volume to be statistically detectable, the change in cerebral volume must be twice as large as that for CSF volume. Therefore, a statistically detectable change in peripheral CSF volume may not yield a correspondingly detectable change in hemispheric volume. Finally, although studies suggest that cortical gray matter may be more sensitive than subcortical white matter to volume loss with aging,24 we are not aware of any studies that have examined the effects of education on age-related tissue loss in the cortex per se.

Relations of education to age-specific differences in regional cerebral size.

We found no relation between education and the age-specific decrease in sizes of the frontal, temporoparietal, or parieto-occipital regions. These results replicate our earlier findings in a different sample of healthy adults (n = 66, 30 to 91 years old) in which education was unrelated to total volumes of the frontal or temporal lobes on MRI (1.5 T).23 Similar negative results were also reported by Kidron et al.,9 who measured peripheral CSF volumes as an indirect marker for frontal, temporal, and parietal region parenchymal atrophy in their MRI (1.5 T) study of 20 elderly community volunteers (described earlier). In contrast, two studies of young adults have reported a relation between education and frontal region size.30,31 Finally, in their MRI study of patients with probable AD, Kidron et al.9 found a relation between education and parietal region peripheral CSF volume (an indirect marker of parenchymal atrophy), but not frontal or temporal region CSF volume. These discrepant results across studies may reflect methodological differences in samples (i.e., age, sex ratio, diagnosis) and brain measurement techniques (i.e., region definition, slice location and orientation, area measures from a single slice versus volume measures from multiple slices).

With regard to ventricular volumes, no relation was found between education and the age-specific increase in volumes of the lateral ventricles or the third ventricle. These results replicate our earlier findings in a different sample of healthy adults (n = 66, 30 to 91 years old) in which education was unrelated to total volumes of the lateral or third ventricles on MRI (1.5 T).23 We are not aware of any other imaging studies that have examined the effects of education on the age-related increase in ventricular CSF volume. Likewise, we are not aware of any studies that have examined the effects of education on the age-related volume loss of structures that form the borders of the lateral ventricles (i.e., caudate nuclei) or the third ventricle (i.e., the thalamus). This analysis would be relevant because age-related ventricular enlargement is presumed to occur as a result of shrinkage of these periventricular brain structures.

Education, brain aging, and the reserve hypothesis.

Brain morphologic characteristics in humans appear to be sensitive to the effects of education as well as age and sex, and converging data suggest that these variables may interact over the life span to influence brain size. Our finding of a greater age-specific increase in peripheral CSF volume in normal elderly persons with higher education is consistent with the reserve hypothesis that such individuals are afforded greater protection from any clinical manifestations of cortical atrophy. Of course, detailed cognitive testing is required to define the level of cognitive functioning of our sample and its relationship to cortical atrophy, and these data will be the subject of a future report. Despite these considerations, our results are the first to provide direct neurobiological support for the reserve hypothesis as it relates to age-related changes in brain structure of healthy elderly individuals, and they are consistent with the few imaging studies of AD demonstrating greater regional brain atrophy9 and hypoperfusion10 in those patients with more education.

The mechanism by which education may be related to preserved cognitive functioning in the setting of cortical atrophy is unknown but may be suggested by the absence of any association between education and age-associated increases in ventricular volume. This observation suggests that education is not associated with relatively greater age-related atrophy of those striatal structures (e.g., caudate nucleus) that form the lateral walls of the lateral ventricles. Preserved striatal structure may imply preserved integrity of frontosubcortical circuits critical to executive cognitive functioning, which in turn would afford the individual a greater cognitive “buffer” against any clinical manifestations of brain aging or cortical atrophy.