Sex Differences in Brain Aging: A Quantitative Magnetic Resonance Imaging Study

Background  Little is known about the effect of sex on age-related changes in brain structure.

Methods  Quantitative magnetic resonance imaging of the brain was performed in 330 elderly (age range, 66-96 years) volunteers living independently in the community, all of whom were participants in the Cardiovascular Health Study. Blinded measurements of global and regional brain size were made from T₁-weighted axial images by means of computer-assisted edge detection and trace methods. High measurement reliabilities were obtained.

Results  Age-specific changes in brain size were significantly greater in men than women for the peripheral (sulcal) cerebrospinal fluid volume, the lateral (sylvian) fissure cerebrospinal fluid volume, and the parieto-occipital region area. Main effects of age were observed for all the remaining brain regions examined (cerebral hemisphere volume, frontal region area, temporo-parietal region area, lateral ventricular volume, and third ventricle volume), but these effects were similar in men and women. Asymmetries in brain structures were not affected by aging in either sex.

Conclusions  Our results are generally consistent with the few published studies on sex differences in brain aging and suggest that, for at least some structures, aging effects may be more apparent in men than women. The neurobiological bases and functional correlates of these sex differences require further investigation.

BOTH POSTMORTEM (reviewed by Powers¹) and in vivo imaging (reviewed by Coffey²) studies have demonstrated that advancing age in humans is generally associated with decreased brain tissue size and increased brain cerebrospinal fluid (CSF) volume. Although sex differences have been described in the size, symmetry, and function of several brain structures,^3-8 only a small number of imaging studies have examined the effects of sex on brain aging in nonpatient samples of living humans (Table 1).^9-27 While the findings have been inconsistent, a few investigators have reported sex differences in the effects of age on some brain structures, and in most cases males showed greater aging changes than females.^{14,17,18,22,25,26} These studies are somewhat difficult to compare, however, given differences in subject samples (eg, sample size, age range, exclusion criteria), imaging and data acquisition protocols (eg, computed tomography vs magnetic resonance [MR] imaging), measurement technique, and statistical analyses (Table 1).

The present study used quantitative MR imaging morphometry to examine the effects of sex on age-related changes in the size of regional brain matter and CSF spaces in a large sample of elderly volunteers living independently in the community. We tested the hypothesis that such changes would be more dramatic in men than in women.

Subjects were selected from among participants in the Cardiovascular Health Study (CHS), an ongoing multicenter, population-based observational study of 5888 volunteers 65 years and older, including 2495 men and 3393 women.^28,29 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, subjects undergo extensive clinical evaluation (home interview and physical examination) and laboratory testing (including brain MR imaging [see below]) at baseline, and annual follow-up assessment. Additional details of the CHS have been published.²⁸

A detailed description of subject recruitment for the CHS has been published.^29,30 For the present study, we identified from the CHS cohort a sample of 500 subjects recruited from 2 CHS sites (Pittsburgh, Pa, and Hagerstown, Md) who gave written consent to participate in an ancillary investigation of cognitive functioning and aging (these data will be the subject of a future report). All available subjects from these 2 sites in whom brain MR imaging 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 subjects for 1 or more of the following reasons: not right-handed (subjects were determined to be right-handed if they used their right hand to write, throw a ball, and brush their teeth³¹); lifetime history of any psychiatric illness or of any illness or injury referable to the brain (per the CHS clinical evaluation described earlier); incomplete cognitive test data; incomplete MR imaging data (eg, scan artifact, missing slices); or MR images with structural abnormalities (cortical infarct, n=5; hydrocephalus, n=1; tumor, n=1; and markedly thickened calvarium, n=1).^32,33

The final sample consisted of 330 subjects, 129 men and 201 women, ranging in age from 66 to 96 years (Table 2). Our subjects were 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 education level (CHS mean ± SD, 12.35 ± 3.10 years). Of the subjects, 244 (74%) were taking medications for 1 or more of the following medical conditions: hypertension or ischemic heart disease (74 men [57.4]; 111 women [55.2]), peptic ulcer disease (18 men [13.9]; 21 women [10.4]), osteoarthritis (14 men [10.9]; 35 women [17.4]), hypercholesterolemia (9 men [6.9]; 30 women [14.9]), hypothyroidism (4 men [3.1]; 20 women [10]), infection (9 men [7.0]; 14 women [7.0]), diabetes mellitus (oral agent: 6 men [4.7]; 4 women [2.0]; insulin: 8 men [6.2]; 3 women [1.5]), postmenopausal hormone replacement (16 women [8.0]), gout (8 men [6.2]; 1 woman [0.5]), chronic obstructive pulmonary disease (2 men [1.6]; 8 women [4.0]), benign prostatic hypertrophy (5 men [3.9]), breast cancer in remission (3 women [1.5]), and hyperthyroidism (1 man [0.8]). No subject was taking medication known to affect brain size (eg, corticosteroids). Additional subject characteristics are given in Table 2.

As noted earlier, brain MR imaging was performed in all subjects as a result of their participation in the CHS. The standardized CHS brain MR imaging acquisition protocol has been previously described.³⁴ Magnetic resonance imaging was performed on either a 1.5-T scanner (General Electric, Milwaukee, Wis) (n=248) or a 0.35-T scanner (Toshiba) (n=82) at 1 of 2 CHS field centers (Pittsburgh and Hagerstown, 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 milliseconds; echo time [TE], 20 milliseconds; 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 (TR, 3000 milliseconds; TE, 30 milliseconds; flow compensated) and T₂-weighted (TR, 3000 milliseconds; TE, 100 milliseconds; flow compensated) images was obtained (thickness, 5 mm; gap, 0 mm; matrix, 256×192; number of excitations, one-half [1 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 milliseconds; TE, 20 milliseconds) 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 vertex to skull base. Images were stored on 9-track magnetic tape.

For the present 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, Calif) with high-resolution color graphic monitor. The measurements of regional brain size were made on the recalled T₁-weighted axial images by 1 of 2 trained technicians blinded to all subject characteristics. Window center settings were first standardized to ensure precision in boundary detection.³⁵ Structures were identified with the help of brain and MR imaging atlases^36,37 and then measured with a combination of computer-assisted edge detection and manual tracing, using graphic analysis software (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 measurement.

Intracranial volume (IV) was defined by the internal surface of the diploe¹⁶ and measured in every slice between the vertex and the superior border of the midbrain (approximately 12-15 slices per subject were measured). Intracranial size could not be reliably measured 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 18-20 slices per subject). 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 3-4 slices per subject).

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 reliably subdivide the cerebral hemisphere into its various lobes (ie, frontal, temporal, parieto-occipital) because of difficulties in establishing boundaries for such subregions in the axial plane of orientation.² Nevertheless, a regional brain morphometric analysis was possible on 1 of our axial slices. For this analysis, we followed the method of Pearlson et al³⁸ and chose a T₁-weighted axial slice that passed through both the pineal gland and foramen of Monro (hereafter designated the "region-of-interest [ROI] slice"). This slice is approximately 1 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 anatomical regions believed to be associated with performance on a number of neuropsychological tests.³⁸ Using the boundary definitions of Pearlson et al,³⁸ the following 4 subregions were defined for area measurement on the ROI slice (ventricular areas were excluded from all regions) (Figure 1).

Frontal region area: The posterior border of this region was defined by a horizontal line intersecting the anteriormost aspect of the lateral ventricles.

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.

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

Intracranial area (IA): This area was defined by the inner surface of the diploe (per above).

Extensive reliability studies of our measurement techniques have indicated that area/volume measurements of these regions are highly reliable.¹⁶ On the basis of a randomly selected sample of 10 brains from the current study, intraclass correlation coefficients for interrater reliability of the 2 raters ranged from 0.85 (for small regions such as the third ventricle) to 0.99 (for large regions such as the cerebral hemisphere). Intraclass correlation coefficients for intrarater reliability ranged from 0.84 to 0.99.

By exploratory methods, the data were examined for outliers and extreme values by means of box plots and normal quantile-quantile plots. Transformations of the outcome variables—in particular, cube root transformations for the volume data, square root transformations for the ROI data, and logarithmic transformations for both—were reviewed. These analyses demonstrated no need for transformation.

Regressions, using the full model given below, 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.

Our analysis treated intracranial size as a covariate. An alternative approach is to use percentage size based on the ratio of brain structure size to intracranial size. We rejected this approach for 2 reasons. First, the ratio approach implicitly assumes that brain size is perfectly correlated with intracranial size. Although the 2 are highly correlated, we found the assumption of perfect correlation untenable. Second, the ratio approach creates outcome variables that are necessarily bounded between 0 and 1. Such variables may have distributions poorly suited for linear regression analysis.

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. There were 4 predictor variables. The first predictor in the regression equation was either IV (for the 2 sets of volume data) or IA (for the 2 sets of area data), as appropriate. The second predictor was sex, with the effect coded as 1 for men and −1 for women. The third was age, centered at 75 years (roughly the mean age of the sample) to eliminate colinearity with the age×sex interaction. The fourth predictor was the age×sex interaction, created by multiplying the (centered) age variable by the sex variable.

The regression models were the same for all outcome variables. Each outcome variable was first regressed against the full model consisting of IV (or IA, as appropriate), sex, age, and age×sex, using the hierarchical method in the order given. In this approach, the significance of a predictor is adjusted for all predictors preceding it in the list, but not adjusted for any predictors following it. In all tests, the significance level was set at .05. If age × sex was found significant, the full model was accepted, regardless of the significance values of any of the preceding predictors, and testing was stopped. If age × sex was not significant, then it was eliminated from the equation and the regression was run again. If age was significant, then this model was accepted and testing stopped. Otherwise, age was eliminated and the regression was run again. These iterations were repeated until a significant effect was found or no predictors were left. The regression coefficients from the final accepted model were then used to interpret the results.

The regional cerebral measures are shown in Table 2. Results are discussed first for measures that showed an effect of sex on age-related changes in volume (ie, an age×sex interaction), and then for measures that showed only a main effect of age without an interaction.

Significant age×sex interactions were found for the peripheral CSF volume, the lateral fissure CSF volume, and the parieto-occipital region area. For each of these regions, men showed greater age-related changes than did women. Table 3 illustrates these interactions for persons with an average intracranial size, from ages 65 to 95 years. For the peripheral CSF volume, the regression coefficient was 2.11 for men but only 0.06 for women (P<.03). At age 65 years, men had a mean peripheral CSF volume about 5.70 mL smaller than that of women, but at age 95 years, men had a mean peripheral CSF volume about 55.67 mL larger than that of women (Table 3). For the lateral fissure CSF volume, the regression coefficient was 0.23 for men but only 0.10 for women (P<.04). At age 65 years, men had a mean lateral fissure volume about 0.80 mL larger than that of women, but at age 95 years, this difference increased to 4.86 mL (Table 3). For the parieto-occipital region area, the regression coefficient was −0.31 for men but only −0.09 for women (P<.03). At age 65 years, men had a mean parieto-occipital region area about 2.15 cm² larger than that of women, but at age 95 years, men had a mean parieto-occipital region area about 4.54 cm² smaller than that of women (Table 3).

Age was significantly related to each of the remaining brain matter and CSF regions measured. Increased age was associated with decreased cerebral hemisphere volume (coefficient = −2.79, P<.001), frontal region area (coefficient = −0.13, P<.001), and temporoparietal region area (coefficient = −0.13, P<.001). Increased age was also associated with increased volumes of the lateral ventricles (coefficient = 0.95, P<.001) and the third ventricle (coefficient = 0.05, P<.001).

To examine potential laterality differences in the effects of sex on age-related changes in regional cerebral size, left − right differences were analyzed by means of the same hierarchical regression model described above. There were no age × sex interactions and no main effects of age or sex for any of the regions. A main effect was found for intracranial area, but for the frontal region only. Increasing intracranial area was associated with an increased left − right difference in frontal region area (coefficient = −0.01, P<.01), a result of greater increases of the right side than of the left.

We found that age-specific changes in brain size were significantly greater in men than women for the peripheral (sulcal) CSF volume, the lateral (sylvian) fissure CSF volume, and the parieto-occipital region area. Main effects of age were observed for all the remaining brain regions examined, but these effects were similar in men and women. Asymmetries in brain structures were not affected by aging in either sex. Our blinded measures of these brain regions were highly reliable, and our estimates of their age-specific sizes agree closely with previous reports, including those that used more sophisticated voxel-by-voxel techniques.² Our results shed light on some of the conflicting findings in the literature (discussed later) and extend these observations to a large sample of elderly persons living independently in the community.

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. 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 (in the present study, cranial size 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 subjects. First, our sample represents a group that may be somewhat healthier than the entire population because of selection criteria for the CHS and the current study.³⁰ As such, our findings may not be applicable to the entire population of seniors. Second, there is heterogeneity of health status within our subjects, in that 26% were also free of major systemic illness while 74% had at least some mild physical disease, corresponding to the distinction between successful and usual aging.³⁹ 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.^40,41 The prevalence of this condition was generally similar among the men and women in our study, however. Furthermore, studies in subjects free of major medical illness have reported sex differences in age-related changes in brain structure similar to our present findings.^22,25 Still, it is possible that sex differences in the prevalence of systemic diseases may account for some of the sex differences observed in structural brain aging.

The measurements of regional brain size in our study are subject to certain limitations. First, because of limitations inherent in the CHS MR imaging acquisition protocol, our analyses of brain size were restricted to the axial plane (3-dimensional reconstruction was not possible without dramatic loss of resolution). The axial plane does not permit optimal boundary delineation of many brain regions, and as such our anatomic definitions were arbitrary and frequently underrepresentative of the true size of the structure. In particular, our estimates of regional brain size were based on single-slice area measurements (the ROI slice) rather than multislice volume measures, which are more valid estimates of true brain size.² 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 MR imaging technical variables.³⁵ 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 MR imaging size measurements.^16,35 Third, field strength differences between the 2 scanners could affect estimates of brain size.² To test whether such differences could have confounded the relations between brain size and the predictor variables, scanner assignment was entered as a covariate in the regression analyses (entered after sex). Scanner assignment was not confounded with any of the age × sex interactions or the age main effects.

We found that the age-related increase in peripheral CSF volume, a marker of cortical atrophy, was significantly greater in elderly men than women. For example, from ages 65 to 95 years, men (of average IV) had an increase in peripheral CSF volume of approximately 32% compared with less than a 1% increase in women (Table 3). Gur et al¹⁴ also found that the ratio of sulcal CSF volume to IV was greater for elderly (55 years and older) subjects and for men. Similarly, Blatter et al²³ found higher correlations between age and "subarachnoid" CSF volume (adjusted for IV) in men (r=0.653) than in women (r=0.545), although these correlations were not statistically compared. Other studies that examined peripheral CSF volume have found no sex effects on age-related increases.^20,25,26

We found no sex differences in the age-related decrease in cerebral hemisphere volume (ie, there was no age × sex interaction) and no age effects (neither main nor interaction with sex) on the left − right difference in cerebral hemisphere volume. Similar negative findings have been reported.^{11,14,16,19,20,25,26} Although Condon et al¹⁰ found that men, but not women, exhibited a significant correlation between age and the ratio of total brain volume to IV, these correlations were not statistically compared. Similarly, Blatter et al²³ observed higher correlations in men than women between age and the ratio of total brain volume to IV (r=−0.675 vs r =−0.539, respectively), but again these correlations were not statistically compared. Murphy et al²⁵ reported that men had a significantly greater age-related decrease in the ratio of cerebral hemisphere volume to IV than did women.

Our finding of a significant sex effect on the age-related increase in peripheral CSF volume, in the absence of a sex effect on age-related volume loss of cerebral hemisphere brain matter, is consistent with the observations of Gur et al.¹⁴ Taken together, these reports suggest that while peripheral CSF volume may show a greater age-related increase in men than women (likely as a result of cortical atrophy), such sex differences in cortical atrophy may not be apparent statistically when the cortex is averaged in with a relatively larger structure, such as the cerebral hemisphere. We are not aware of any studies that have examined sex effects on age-related tissue loss in the cortex per se.

We found that the age-associated increase in lateral fissure CSF volume, a marker of frontotemporal (and, to a lesser extent, parietal) atrophy, was significantly greater in men than women. For example, from ages 65 to 95 years, men (of average IV) had an increase in lateral fissure volume of approximately 80%, while women had an increase of only approximately 37% (Table 3). Although Sullivan et al²⁰ found no sex differences in the age-related increase in sylvian fissure volume, they used computed tomographic scanning and relatively thicker brain slices (10 mm).

In contrast to the results with lateral fissure volume, we found no sex effects on the age-related decrease in temporoparietal region area or frontal region area. Thus, in our study, lateral fissure CSF volume was a more revealing marker of atrophy in these regions than was their direct measurement from the ROI slice. The literature is conflicting with regard to the effects of sex on age-related changes in temporal lobe size (Table 1). Cowell et al²² and Murphy et al²⁵ found that men exhibited greater age-related decreases in the ratio of temporal lobe volume to IV than did women. Similarly, Golomb et al¹⁸ found that age-related hippocampal atrophy was more common in men than women, and Raz et al²⁶ observed greater age-related inferior temporal volume loss in men than women. In contrast, Murphy et al²⁵ actually observed greater temporal lobe atrophy in women than men. Despite differences in which sex is more affected, the published results suggest that sex may impact the age-related volume loss of the temporal lobe region. These findings may provide a neuroanatomical substrate for the sex differences noted earlier in age-related verbal memory impairment.^42-44

The literature is also conflicting with regard to the effects of sex on age-related changes in frontal lobe size (Table 1). Cowell et al²² and Murphy et al²⁵ both observed greater age-related frontal lobe volume loss in men than in women. In contrast, others have found no sex effects.^16,19-22,26 These discrepant results may reflect differences between studies in samples and brain measurement techniques (ie, quantitative vs qualitative measures, area measures from a single slice vs volume measures from multiple slices).

Our analysis of left − right difference in temporo-parietal and frontal region areas showed no age effect (neither main effect nor interaction with sex). Similar negative results have been reported.^{16,20,22,25,26} In contrast, Cowell et al²² observed that the right greater than left asymmetry of frontal lobe volume to IV was larger in older women than in younger women, whereas in men no such group differences were seen.

We found that age-related decreases in parieto-occipital region area were greater for men than women—for example, from ages 65 to 95 years, men (of average IA) lost approximately 15% of their parieto-occipital lobe area, while women lost only 4% (Table 3). Using a somewhat different definition of this brain region, Cowell et al²² did not find any sex effect on the age-related decrease in the ratio of the posterior cerebral hemisphere volume to IV. Murphy et al²⁵ likewise found no sex differences in the age-related decrease in parieto-occipital region volume to IV, although they actually observed worse atrophy in women for the ratio of parietal lobe volume to IV. Similarly, Raz et al¹⁹ reported that women exhibited greater age-related volume loss in the visual cortex than did men. These widely divergent findings indicate a need for additional research. Our analysis of left−right difference in parieto-occipital lobe area disclosed no age effect (neither main effect nor interaction with sex). Similar negative results have been reported.^16,20,22,25

With regard to ventricular volumes, we found no sex effects on the age-related increase in lateral ventricular CSF volume or third ventricular CSF volume. Similar negative findings have been reported by the majority of studies that have examined the lateral ventricles^{11,14,16,17,19,20,25,27} or the third ventricle.^14,16,17,20 Since age-related ventricular enlargement is presumed to occur as a result of shrinkage of periventricular brain matter, our results are also consistent with other studies that found no effect of sex on the age-related volume loss of structures that form the borders of the lateral ventricles (ie, the caudate nuclei)^12,25 or the third ventricle (ie, the thalamus).²⁵ In contrast, Grant et al⁹ reported that men, but not women, exhibited a significant age-related increase in lateral ventricular volume, although this apparent sex difference was not tested. Likewise, Blatter et al²³ observed higher correlations in men than women between age and lateral ventricle volume (adjusted for IV) (r=0.444 vs r=0.218, respectively) and between age and third ventricle volume (adjusted for IV) (r=0.634 vs r =0.406, respectively), but again these correlations were not statistically compared. Kaye et al¹⁷ reported that the precipitous age-related increases in lateral ventricular volume began about a decade earlier in men than women. Finally, Murphy et al²⁵ found that women actually had a greater age-related increase in the ratio of third ventricle volume to IV than did men.

Our analysis of left − right difference in lateral ventricle volume showed no age effect (neither main effect nor interaction with sex). Similar negative findings were noted by Murphy et al²⁵ for the ratio of right − left lateral ventricle volume to IV. However, Gur et al¹⁴ found that the ratio of ventricular CSF volume to IV was more pronounced in the left hemisphere than in the right, a difference they attributed primarily to elderly men.

In summary, brain morphologic characteristics in humans appear to be sensitive to the effects of both age and sex, and converging data suggest that these 2 variables may interact over the life span to influence brain size. These data should provide a useful context within which to interpret changes in regional brain structure associated with "abnormal" aging. The neurobiological bases for these sex differences in brain aging are not known. Neuroendocrinological differences between sexes have been proposed as a possible explanation given that gonadal corticosteroids affect brain development and aging, and that age affects both the function and regional distribution of androgen and estrogen systems in the brain.²² Still, most studies of human brain aging at the cellular level have not examined sex effects.

The behavioral effects in humans of these sex differences in brain aging are likewise unknown. These findings may provide a neuroanatomical substrate for the sexually dimorphic effects of age on cerebral blood flow and metabolism,^3-5,8 and it is possible that sex differences in brain aging could interact with a superimposed pathological process to produce sex differences in brain disorders such as Alzheimer disease.²⁵ In this regard, sex differences in brain aging are consistent with observed sex differences in some aspects of cognitive aging.^42-44 Correlative neuropsychological investigations are currently under way in our laboratory to determine the potential functional significance of differences between the sexes in brain aging.

Accepted for publication July 21, 1997.

This study was supported in part by the Allegheny-Singer Research Institute, Pittsburgh, Pa, the Mental Illness Research Association, Detroit, Mich, and the National Institutes of Health, Bethesda, Md (grant MH 46643).

We acknowledge the assistance of Theresa Cunningham, Lorrie Cain, Mike Dormnod, Sandy Giconi, Bonnie Lind, Linda Wilkins, and John Yee, MA.

Reprints: C. Edward Coffey, MD, Department of Psychiatry, Henry Ford Health System, 1 Ford Pl, Detroit, MI 48202 (e-mail: ecoffey1@hfhs.org).