[1] 20w Do cardiovascular risk factors explain the link between white matter hyperintensities and brain volumes in old age? A population-based study

[1] 251w Background and purpose: White matter hyperintensities (WMHs) and brain atrophy frequently coexist in older people. However, it is unclear whether the association between these two brain lesions is dependent on the aging process, a vascular mechanism or genetic susceptibility. It was therefore investigated whether the association between load of WMHs and brain atrophy measures is related to age, vascular risk factors (VRFs) or the APOE-e4 allele. Methods: This population-based study included 492 participants (age ≥60 years, 59.6% women) free of dementia and stroke. Data on demographics, VRFs and APOE genotypes were collected through interviews, clinical examination and laboratory tests. WMHs on magnetic resonance images were assessed using manual visual rating and automatic volumetric segmentation. Hippocampal and ventricular volumes were manually delineated, whereas total gray matter (GM) volume was measured by automatic segmentation. Data were analyzed with multivariate linear regression models. Results: More global WMHs, assessed using either a visual rating scale or a volumetric approach, were significantly associated with lower GM volume and higher ventricular volume; the associations remained significant after adjusting for age, VRFs and the APOE-e4 allele. In contrast, the association between global WMHs and hippocampal volume was no longer significant after adjusting for age, whereas adjustment for VRFs and APOE-e4 had no influential effect. Conclusion:The association of global WMHs with lower GM volume and higher ventricular volume is independent of age, VRFs and APOE-e4 allele, suggesting that the process of cerebral microvascular disease and neurodegeneration are associated independently of the normal aging process, vascular mechanisms or genetic susceptibility.

[1] 97w Brain aging significantly contributes to the transition from healthy to late-life cognitive and physical impairments [1]. Thus, detecting signs of brain aging in old adults and understanding the underlying mechanisms are critical from both clinical and public health perspectives [2]. Clinicopathological studies show that cerebral white matter hyperintensities (WMHs) detected on magnetic resonance imaging (MRI) are markers of cerebral small-vessel disease [3]. Postmortem imaging studies have also confirmed that the severity of regional brain atrophy (e.g. hippocampus) is strongly related to local tissue degeneration at autopsy [4], suggesting that brain atrophy is a reliable marker for neurodegeneration.

[2] 111w WMHs and brain atrophy are increasingly common as people age [5] and thus frequently coexist amongst older people. However, the association between WMHs and brain atrophy may be related not only to chronological age but also to common pathological processes such as vascular mechanisms and genetic susceptibility [6]. Indeed, age, vascular risk factors (VRFs) and APOE genotypes are all associated with both WMHs and brain atrophy in old age [7][8][9][10]. However, population-based research has yielded mixed results concerning the association between brain volumes and WMHs [6,11,12]. Few studies have simultaneously taken age, VRFs and genetic factors into consideration when assessing the relationship of WMHs to global and regional brain volumes [13,14].

[3] 47w In this population-based study of cognitively intact older people, an investigation of (i) whether WMHs are related to volumes of hippocampus, total gray matter (GM) and ventricles and (ii) whether the association between WMHs and brain volumes is related to age, VRFs and APOE genotypes was undertaken.

[1] 151w Table 1 shows the characteristics of the study participants. Of the 555 participants who undertook the MRI scan, 39 were excluded because the quality of the MRI was insufficient to assess WMHs or volumetric measures. Another 24 persons were excluded because of questionable dementia (n = 2), Parkinson's disease (n = 3), brain tumor (n = 4), stroke (n = 10) and arachnoid cyst (n = 5), leaving 492 participants for the current analyses. All participants in the analytical sample were cognitively intact (MMSE score ≥ 25). The mean age of participants was 70.3 (SD = 9.0) years and 59.6% were women. The analytical sample (n = 492) was younger than the remainder of the SNAC-K sample (n = 2871) (mean age 70.3 vs. 74.6 years, P < 0.01), included fewer women (59.6% vs. 65.8%, P < 0.01) and had fewer years of education (12.6 vs. 14.1 years, P < 0.01).

[2] 57w The Spearman correlation coefficient between the global WMH visual rating score and volume was 0.80 (P < 0.001). When the visual score and global WMH volume were compared by quartiles, the weighted j statistic was 0.60 (P < 0.001). The disagreements were distributed symmetrically by the cross-quartiles of global WMH visual score and WMH volume (Fig. 1).

[3] 122w In the univariate model, higher global WMH volume and visual score were significantly associated with lower GM volume and higher ventricular volume (P < 0.01, Table 2). The associations of global WMH volume with total GM and ventricular volumes were attenuated but still statistically significant after adjustment for age. Additional adjustment for sex, education, burden of VRFs and APOE-e4 allele did not substantially alter the results. Univariate analysis showed that higher global WMH volume and visual score were significantly associated with lower hippocampal volume (P < 0.01, Table 3). The association between global WMH volume and hippocampal volume was attenuated and became statistically non-significant after controlling for age. Additional adjustment for sex, education, VRFs and APOE-e4 allele did not alter the results.

[4] 61w In the univariate model, a higher visual score for subcortical WMHs and PVHs was significantly associated with smaller total GM and hippocampal volumes and with larger ventricular volumes. After controlling for age, the associations with subcortical WMHs and PVHs remained statistically significant for GM volume but not for hippocampal and ventricular volumes. The results remained unchanged in the fully adjusted model.

[5] 27w There was no interaction of global WMH, subcortical WMHs and PVHs with age, sex, education, VRFs and APOE-e4 allele on volumes of total GM, hippocampus or ventricles.

[1] 92w The main findings from this population-based study of people aged ≥60 years are as follows: (i) a larger load of global WMH was linked to a smaller total GM volume and a larger ventricular volume, independent of age, vascular burden and APOE genotype; (ii) the association between WMH load and smaller hippocampal volume disappeared after controlling for age; and (iii) subcortical WMHs and PVHs were associated with lower GM volume, even after controlling for age, VRFs and APOE-e4 allele, but their associations with ventricular and hippocampal volumes disappeared after controlling for age.

[2] 117w The association of global WMH with smaller GM and larger ventricular volume is consistent with previous reports from population-based studies of older people [12][13][14]23,24]. Our research extends past research by showing that this association is independent of demographics, vascular burden and APOE genotype. The Rotterdam Scan Study and the Three-City Dijon MRI study found that more WMHs were related to a smaller hippocampal volume [13,25]. By contrast, the Austrian Stroke Prevention Study and the Honolulu Asia Aging Study found no independent association between WMHs and hippocampal volume [26,27], which is in line with our data. Different approaches in assessing WMHs and hippocampal volume and differences in characteristics of the study sample might contribute to the mixed results.

[3] 51w A good agreement was found between volumetric and visual rating approaches in assessing global load of WMH, which is consistent with previous research [23,28,29]. Our results also demonstrated that both approaches are equally valid in studying the associations between WMHs and both global (e.g. ventricles) and regional (e.g. hippocampus) brain volume.

[4] 228w The mechanisms underlying the association of WMHs with smaller GM volume and ventricular enlargement remain unclear. Evidence suggests that, in the process of normal brain aging, decrease of GM volume is secondary to changes in subcortical WMHs, possibly through retrograde neuronal degeneration of fibers traversing these lesions [13], whereas ventricular enlargement may arise from compensation or mollification owing to loss of brain parenchyma (e.g. white matter) [24]. Our findings suggest that the relationship of WMHs with smaller GM volume and larger ventricular volume is unlikely due to shared influences of aging, vascular mechanisms or genetic susceptibility (e.g. APOE-e4 allele) to cerebrovascular and neurodegenerative lesions. Instead, other mechanisms such as impaired cerebral blood flow and loss of white matter integrity might be involved. Low cerebral blood flow is associated with both WMHs and smaller GM volume and thus may mediate the association between WMHs and GM volume [30,31]. Furthermore, microstructural changes in white matter, which are related to WMHs, may also contribute to alterations in GM and ventricles by causing ischaemic damage to axons, oligodendrocytes and other glial cells [6,31]. The absence of evidence supporting the link between WMHs and hippocampal volume may be due to fewer direct connections between hippocampus and subcortical regions [22]. Additionally, WMHs, as a measure of global brain microvascular damage, may be less sensitive to lesions in local brain regions such as the hippocampus.

[5] 200w Our study was based on a relatively large sample of older people living in the community. Furthermore, the consistency of both volumetric and visual rating approaches for global WMH in studying the association between WMHs and brain volume was verified. However, some limitations related to the cross-sectional design deserve mention. First, the temporal relationship between WMHs and brain atrophy could not be determined. Moreover, the cross-sectional association may be subject to bias caused by selective survival. For example, selective survival could result from earlier mortality in people with lower brain volumetric measures and/or more severe WMHs, but evidence suggests that this is not the case [32]. In summary, this study suggests that global WMH and brain volume are associated, independent of age, VRFs and the APOE-e4 allele, whereas the association between WMHs and hippocampal volume is accounted for by chronological age. Although microvascular and neurodegenerative lesions in the brain represent distinct pathologies, both lesions may be related, at least partly, to the process of global brain aging. Because cerebral microvascular disease, reduced GM and ventricular enlargement have been linked to cognitive consequences, further research is needed to clarify the independent and joint effects of the two pathologies on cognitive dysfunction.

[1] 123w Participants were derived from the Swedish National Study on Aging and Care in Kungsholmen (SNAC-K), a multidisciplinary study of aging and health. People aged 60+ years living either at home or in institutions in the Kungsholmen area, Stockholm, Sweden, were identified. The sample included four younger-age groups with a 6-year interval (60, 66, 72 and 78 years) and seven older-age groups with a 3-year interval (81, 84, 87, 90, 93, 96 and 99+ years). Of all 4590 eligible persons, 3363 (73.3%) were examined at baseline (March 2001-June 2004). Of these, 2204 persons from the non-institutionalized, non-disabled and nondemented participants in SNAC-K were invited to undertake a structural MRI scan during September 2001 and October 2003, and 555 (25.2%) agreed to undergo the scan.

[2] 26w SNAC-K was approved by the Ethics Committee at Karolinska Institutet and by the Regional Ethical Review Board in Stockholm, Sweden. All participants provided written informed consent.

[3] 57w Participants were scanned on a 1.5-T system (Philips Intera, Eindhoven, Netherlands) [15]. The MRI protocol included an axial 3D T1-weighted fast field echo, an axial proton density/T2 turbo spin echo run twice with a shift of 3 mm to fill the gaps, and an axial turbo fluid attenuated inversion recovery (FLAIR) sequence, as fully reported elsewhere [15,16].

[4] 83w The WMH load on FLAIR images was visually assessed using a modified rating scale [17]. WMHs in subcortical regions were scored on a scale from 0 to 6 based on the number and size of the hyperintense signals. Periventricular hyperintensities (PVHs) were scored from 0 to 2. Global WMH was computed as the sum of the scores for subcortical WMHs and PVHs. All visual assessments were made by a single clinical neuroradiologist (AL). The j statistic for intra-rater reliability for WMHs was 0.60.

[5] 117w Global WMH volume was also measured using the Lesion Segmentation Toolbox (LST) [18] in the Statistical Parametric Mapping 8 software (SPM8, Welcome Trust Centre for Neuroimaging, FIL, London, UK). LST uses both the FLAIR and T1-weighted images to automatically detect and segment the WMH. All WMH maps were visually scrutinized and manually corrected for higher volumetric precision in MRIcroN (operator CHE). Correction included addition of missed WMH voxels and removal of false positives (e.g. MRI artifacts considered as WMHs) and was applied to the whole brain except the pons, where it was difficult to visually disentangle lesions and MRI artifacts. After 1 month, 10 randomly selected images were reassessed for WMHs. The intraclass correlation coefficient was 0.99.

[6] 48w The hippocampal and ventricular volumes were manually assessed, as fully described elsewhere [15]. The intraclass correlation coefficients were >0.93 for volumetric assessments of hippocampus and ventricles [15]. Total GM and intracranial volume (ICV) were automatically calculated following a standard procedure in SPM8 [19]. All segmentations were visually verified.

[7] 222w Volumetric measures were adjusted by dividing them by the individual ICV and multiplying by the mean ICV from the sample [22]. The load of global WMH, subcortical WMHs and PVHs was categorized into quartiles. The Spearman rank test was first employed to assess the correlation of the visual rating and automated volumetric measures of global WMH. Next, the weighted Cohen's j coefficient was calculated to assess the agreement of quartiles of visual rating and volumetric measures of global WMH. The burden of VRFs was determined by summing the number of VRFs. Multivariate general linear regression was used to estimate b coefficients and the 95% confidence interval (CI) of brain volume associated with WMHs. Ventricular volume was log-transformed because of its skewed distribution. Interactions were tested by simultaneously including individual variables and their cross-product terms in the same model. Results from three models are reported: model 1 was a univariate model; model 2 included age; and model 3 included age, sex, education, burden of VRFs and APOE genotype. Exploratory analyses were also performed by entering the VRFs one by one to examine the possible effects of the seven individual factors, but the results were not substantially changed by any of the individual factors (data not shown). SAS 9.2 statistical software for Windows (SAS Institute Inc., Cary, NC, USA) was used for all analyses.

[1] 68w Data on demographics, medical history (e.g. hypertension and diabetes), lifestyle (e.g. smoking and alcohol) and use of medications (e.g. antihypertensive and blood-glucose-lowering drugs) were collected through interviews and clinical examination [16]. Educational level was defined as maximum years of formal schooling. Medical drugs were classified according to the Anatomical Therapeutic Chemical Classification System. Body mass index (BMI) was calculated as measured weight (kg) divided by height (m) squared.

[2] 123w Smoking status was categorized as current versus noncurrent smoking. Alcohol intake was classified as none or occasional, light to moderate, or heavy [16]. Leisure-time physical activity was classified as inactivity versus activity (health-enhancing and fitness-enhancing activity) [20]. Using a random-zero sphygmomanometer, arterial blood pressure was measured twice on the right arm (5-min interval) whilst in a seated position. The definition of hypertension was mean blood pressure (two readings) ≥140/90 mmHg or the use of antihypertensive agents. Diabetes was ascertained according to self-reported history, use of oral bloodglucose-lowering drugs, insulin injection or HbA1c > 5.3% [21]. High cholesterol was defined as total cholesterol ≥6.5 mmol/l. APOE allelic status was determined following standard procedures. Global cognitive function was assessed with the Mini-Mental State Examination (MMSE).