Dissociation between corpus callosum atrophy and white matter pathology in Alzheimer's disease

White matter hyperintensities in T2-weighted MRI scans are commonly found in the older population, increasing with age, cerebrovascular disease, and hypertension.^1-3 White matter hyperintensities also have been reported in AD.^3-7 An AD-specific increase of white matter hyperintensity compared with healthy age-matched control subjects is controversial, however.^7-10

One major white matter tract is the corpus callosum, which links cortical areas of both cerebral hemispheres. Several studies found significant atrophy of the corpus callosum in AD.^11-17 Little is known, however, about the contribution of white matter degeneration to corpus callosum atrophy in AD. Vermersch et al.¹⁴ reported a statistically significant correlation between corpus callosum size and the extent of white matter hyperintensity in AD. In contrast, we recently demonstrated a region-specific pattern of corpus callosum atrophy in AD that was correlated with dementia severity in the absence of significant white matter hyperintensity.¹²

In this study, we correlated total and regional values of white matter hyperintensity with total and regional corpus callosum areas in AD patients and healthy age-matched control subjects. We assessed possible differences in the contribution of aging to corpus callosum atrophy and white matter pathology between both groups and investigated the relation of corpus callosum atrophy and white matter hyperintensity to changes in cognitive function in AD. We hypothesized that healthy aging affects both corpus callosum size and white matter hyperintensity load on a regional basis, whereas a dissociation occurs between primary white matter degeneration and corpus callosum atrophy in AD.

Patients and methods. Patient selection. We investigated 20 clinically diagnosed AD patients who participated in a longitudinal study of Alzheimer's disease by the Laboratory of Neurosciences, National Institutes on Aging, National Institutes of Health. Sixteen patients were diagnosed as having probable and four patients as having possible AD, according to National Institute of Neurological and Cognitive Disorders and Stroke/Alzheimer's Disease and Related Disorders Association criteria.¹⁸ One patient had biopsy-proven AD. The degree of cognitive impairment was assessed using the Mini-Mental State Examination (MMSE).¹⁹ Eight patients with a MMSE score <10 were classified as severely demented; five patients with an MMSE score from ≥10 to <20 were moderately demented; and seven patients with a score ≥20 were mildly demented (table).

For comparison, 21 healthy volunteers were selected. All control subjects scored 29 or above on the MMSE. MRI and neuropsychological data from a subset of 20 subjects (9 AD patients and 11 control subjects) had been included in a previous study on corpus callosum atrophy in AD.¹²

Subjects with history or signs of hypertension, diabetes, or cardiac arrhythmia were excluded to minimize the effect of cardiovascular risk factors. Fasting blood glucose level was below 120 mg/mL in all subjects and electrocardiogram (ECG) was normal. All but four subjects were normotensive (systolic blood pressure below 140 mm Hg and diastolic blood pressure below 90 mm Hg). Four subjects, three AD patients and one control, showed borderline systolic hypertension, with systolic level not above 165 mm Hg and diastolic level below 90 mm Hg. Intracerebral pathology such as cerebral infarction, neoplasm, or normal pressure hydrocephalus was excluded for all subjects by MRI scanning. Overall brain atrophy in the AD patient group and extent of white matter hyperintensity were not exclusion criteria. Furthermore, significant comorbidity such as focal neurologic signs, hypothyroidism, and other pathological conditions that may influence cerebral structure and function directly or indirectly were excluded in all subjects by history, physical and neurologic examination, psychiatric evaluation, chest radiograph, ECG, EEG, brain MRI, and laboratory tests (complete blood count, sedimentation rate, electrolytes, glucose, blood urea nitrogen, creatinine, liver-associated enzymes, cholesterol, high-density lipoprotein, triglycerides, antinuclear antibodies, rheumatoid factor, Venereal Disease Research Laboratory, HIV, serum B₁₂, folate, thyroid function tests, and urinalysis).

All subjects or the holders of their Durable Power of Attorney signed consent forms to undergo National Institute on Aging and neuropsychological assessment for clinical investigation and research. The study was approved by the National Institute on Aging Institutional Review Board.

MRI. Eighteen T2-weighted and 18 proton-weighted axial slices (slice thickness 6 mm, repeat time [TR]/echo time [TE] 2000/80 and 2000/20, respectively) and 32 proton-weighted coronal slices (slice thickness, 6 mm; TR/TE, 2000/20) were obtained using a 0.5-T tomograph (Picker Instruments, Cleveland, OH). Furthermore, all subjects were investigated with a 90-slice T1-weighted sagittally oriented volumetric sequence (slice thickness 2 mm; in plane resolution 1 by 1 mm; TR/TE, 20/6; flip angle, 45°).

Callosal area measurements. Areas of the total corpus callosum and five callosal subregions were measured by one investigator blinded to the subject's diagnosis in the sagittal T1-weighted MRI slice that best represented the midsagittal section, according to a described method¹² using the Analyze software for region of interest measurement (Biomedical Imaging Resource, Mayo Foundation, Rochester, MN) on a Sun workstation (Sun Microsystems, Mountain View, CA). The number of pixels within each region was summed automatically and multiplied by pixel area to obtain absolute values (mm²) for the areas of total corpus callosum and of five subregions (labeled C1 to C5 in rostral-occipital direction). Areas of regions C1 and C2 were summed to obtain the anterior corpus callosum area, whereas the posterior corpus callosum area was defined by the sum of subregions C3 to C5.

The intraclass correlation coefficient for total corpus callosum was 0.96 for interrater and 0.98 for intrarater reliability; for regional corpus callosum measurements, intrarater reliability ranged from 0.98 for subregion C1 to 0.75 for subregion C3.

White matter hyperintensity grading. Periventricular and deep white matter hyperintensities were graded by one investigator on all T2- and proton-weighted axial slices for each subject using a slight modification of a rating scale reported by Scheltens et al.²⁰ Signal changes on the corpus callosum were added as an extra item to this scale. For each subject, the total white matter hyperintensity load was defined as the sum of all five subscores: periventricular hyperintensity score, deep white matter score in the four lobes (white matter hyperintensity), corpus callosum score, basal ganglia hyperintensity, and infratentorial hyperintensity scores. The white matter hyperintensity subscores for the temporal, parietal, and occipital lobes were summed to obtain the posterior deep white matter score.

High levels of interrater reliability have been reported for the white matter hyperintensity and periventricular hyperintensity scores of this scale.^10,14,20

Statistics. Patient and control groups were compared in age using Student's t-test, in gender distribution with the chi-square test, and for white matter hyperintensity scores using the Mann-Whitney U test. Explorative data analysis revealed that corpus callosum areas were normally distributed in both groups without any outliers and that variance was not different between groups. Differences in total corpus callosum between AD patients and control subjects were assessed with Student's t-test. Group differences in the distribution of callosal areas were tested using repeated measures analysis of variance (ANOVA), with groups as the between-subjects factor and the five callosal subregions as the within-subject factor. A significant group by subregion interaction was followed up by pairwise single effect analysis using Student's t-test.

The correlation between total corpus callosum and white matter hyperintensity scores was assessed using Spearman's rank correlation. Regional interactions between corpus callosum areas and white matter hyperintensity scores were assessed with a multiple regression model. First, the frontal and posterior corpus callosum areas as dependent variables were predicted by the total white matter hyperintensity load and the frontal and posterior white matter hyperintensity subscores. In the second step, age was introduced as an additional independent variable. In a further linear regression model, corpus callosum areas and white matter hyperintensity subscores as dependent variables were predicted by age and in the AD group by the MMSE score as the measurement of overall cognitive impairment.

All analyses were done using SPSS (Chicago, IL) for Windows release 7.5.1. p Values below 0.05 were considered significant.

Results. The AD and control group were matched for age (p ≤ 0.82) and gender distribution (p ≤ 0.71). The groups did not differ significantly in total white matter load nor in any white matter hyperintensity subscore. Only the periventricular scores were significantly higher in the AD patients compared with control subjects (p ≤ 0.02).

Absolute mean total corpus callosum area (mm²) was significantly smaller in the AD group compared with control subjects (p ≤ 0.001). Furthermore, a repeated measures ANOVA revealed a statistically significant difference in the distribution of callosal area between AD and control groups (p ≤ 0.001). To elucidate the regional pattern of corpus callosum atrophy, areas of the five callosal subregions were compared separately between groups. Areas of the two most rostral (C1 and C2; p ≤ 0.001 and 0.005, respectively) and the most occipital (C5; p ≤ 0.001) subregions were significantly decreased in the AD group. Conversely, mean areas of the two callosal subregions representing the posterior part of the callosal body (C3 and C4) were not significantly different between AD and control groups (p ≤ 0.2 and 0.07, respectively).

Area of total corpus callosum as correlated significantly with age in control subjects (standardized beta = -0.5; p ≤ 0.003) and with age and the MMSE score in the AD patients (standardized beta = -0.46 and 0.44; p ≤ 0.05). Regarding regional correlations, we found a significant correlation between anterior corpus callosum area and age in control subjects (standardized beta = -0.47 and 0.57; p ≤ 0.02). In contrast, posterior corpus callosum area was not correlated significantly with age nor MMSE score in both groups. Total white matter hyperintensity load was correlated with age in AD (standardized beta = 0.5; p ≤ 0.03) and control subjects (standardized beta = 0.64; p ≤ 0.002) but not with the MMSE score.

The white matter hyperintensity scores were not significantly correlated with total corpus callosum in any group. In control subjects, however, the correlation between total corpus callosum and the total white matter hyperintensity load tended toward significance (Spearman's rho = -0.4; p ≤ 0.08). Prediction of anterior callosal area by total white matter hyperintensity load and frontal and posterior white matter hyperintensity subscores revealed a significant positive correlation between anterior callosal area and frontal lobe white matter hyperintensity score in AD (standardized beta = 0.49; p ≤ 0.03) and a significant negative correlation in control subjects (standardized beta = -0.48; p ≤ 0.03; figure). Neither total nor posterior white matter hyperintensity scores contributed a significant explanatory power to the multiple linear regression models in both groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure. Frontal lobe white matter hyperintensities score plotted against anterior corpus callosum area in controls (n = 21).

Introduction of age into the multiple regression model explained the correlation between anterior callosal area and frontal white matter hyperintensity score in control subjects but not in AD. The posterior area of the corpus callosum was not correlated with total or regional white matter hyperintensity scores in both groups.

The regional pattern of corpus callosum atrophy in AD compared with control subjects remained essentially stable after normalization of the corpus callosum areas to the total intracranial volume. Normalized total corpus callosum area was reduced in AD (p ≤ 0.001), predominantly in regions C1, C2, and C5 (p ≤ 0.001, 0.06 and 0.001, respectively), with a relative sparing of regions C3 and C4. In AD, normalized total callosal area and normalized frontal callosal area were correlated with the MMSE score (Spearman's rho = 0.47 and 0.53, respectively; p ≤ 0.05) but not with age. Furthermore, normalized frontal callosal area was not correlated with frontal white matter hyperintensity load in AD (p ≤ 0.2). In control subjects, normalized frontal corpus callosum area was correlated with age (rho = -0.5; p ≤ 0.03). The correlation between normalized frontal corpus callosum area and frontal white matter hyperintensity load, however, did not reach significance (standardized beta = -0.43; p ≤ 0.06).

Discussion. In this study, we explored the contribution of primary white matter degeneration to region-specific corpus callosum atrophy in healthy older subjects and AD patients. We found atrophy of the corpus callosum-predominantly in the callosal rostrum and splenium-in AD patients with a relative wide range of white matter hyperintensity, which is consistent with our previous findings of a regional pattern of corpus callosum atrophy in the absence of white matter pathology.¹² This pattern remained stable after normalization to total intracranial volume. In addition, we showed that extent of overall cognitive impairment assessed by the MMSE score predicted the size of the total and anterior corpus callosum in the AD group, independently of age. This underscores the possible role of the corpus callosum as a marker for the progression of the disease. We found no correlations between corpus callosum areas and white matter hyperintensity scores in AD. The significant positive correlation between frontal corpus callosum area and frontal white matter hyperintensity load tends toward an unexpected direction. There is no hypothesis that would explain an increase of corpus callosum area parallel to an increase of white matter pathology in AD. Many studies show that disorders involving primarily white matter degeneration (e.g., multiple sclerosis and vascular dementia) lead to corpus callosum atrophy and not to an increase of corpus callosum size^21-24. Therefore, this result probably does not reflect a mutual dependency of frontal corpus callosum and white matter hyperintensity load in AD but the influence of confounding factors. This interpretation is further supported by the lack of a significant correlation between both parameters after normalization.

In contrast to Vermersch et al.,¹⁴ we found no contribution of white matter hyperintensity load to corpus callosum atrophy in AD. Two factors may explain this discrepancy. First, the subjects in the Vermersch study were older than ours. Because there is evidence that late-onset AD subjects are more likely to exhibit AD-specific white matter pathology compared with early onset AD subjects,^10,25 the contribution of white matter pathology to corpus callosum atrophy my increase with advanced age. Second, the AD group of the previous study included subjects with extensive white matter lesions, even large confluent areas. This may result from inclusion of subjects with significant vascular comorbidity confounding AD pathology.

In healthy control subjects, we showed that age predicted the size of the anterior corpus callosum, although not that of the posterior corpus callosum. This is consistent with previous reports that showed predominantly rostral atrophy of the corpus callosum in healthy older subjects, compared with young control subjects.^15,17 The total white matter hyperintensity load increased also with age in control subjects. Furthermore, we report a reduction of frontal corpus callosum area parallel to increase of frontal lobe white matter hyperintensities in control subjects that was explained by age and remained essentially stable after normalization to head size. This correlation was not explained by an overall correlation with the total white matter hyperintensity load in the multiple regression model. The finding of an age-dependent region-specific correlation between frontal corpus atrophy and frontal white matter hyperintensity load is interesting, considering evidence that white matter hyperintensity load is significantly predictive for frontal lobe cortical metabolic decline and frontal lobe-mediated cognitive dysfunction in healthy older subjects.^26,27

Several studies suggest that the frontal lobe may be disproportionately sensitive to changes that occur with age^28,29. We showed that age specifically correlated with the area of frontal corpus callosum that predominantly contains projections from frontal cortical areas.^30,31 Age further correlated with white matter pathology. As a result, white matter degeneration and region-specific corpus callosum atrophy occur together in healthy older subjects. In contrast, in AD patients we showed a dissociation between white matter pathology and corpus callosum atrophy, with the corpus callosum size largely dependent on the degree of dementia. Our findings underscore the hypothesis that corpus callosum atrophy can serve as a marker for a disease-specific pathology in AD, independent of primary white matter degeneration. This is consistent with neuropathologic studies^32,33 that show that AD specifically affects the intracortical projecting neurons, including the callosally projecting neurons,^34,35 whose loss may lead to corpus callosum atrophy as an indicator of neocortical disconnection.

Acknowledgments

The authors thank Prof. H. Hippius, MD, for his generous support. Part of the presented material originates from the doctoral thesis of S.J. Teipel (Ludwig-Maximilian University, Munich, Germany; in preparation).