Elsevier

NeuroImage

Volume 60, Issue 3, 15 April 2012, Pages 1597-1607
NeuroImage

The spatial distribution of age-related white matter changes as a function of vascular risk factors—Results from the LADIS study

https://doi.org/10.1016/j.neuroimage.2012.01.106Get rights and content

Abstract

White matter hyperintensities (WMH) are a frequent finding on brain MRI of elderly subjects, and have been associated with various risk factors, as well as with development of cognitive and functional impairment. While an overall association between WMH load and risk factors is well described, possible spatially restricted vulnerability remains to be established.

The aim of this study was to investigate the spatial distribution of WMH in normally functioning elderly subjects. We introduce a voxel-based approach in which lesion probability is mapped as a function of clinical risk factors using logistic regression, and validate the method using simulated datasets. The method was then applied in a total of 605 participants of the LADIS study (age 74 ± 5 years, all with WMH), and the location of manually delineated WMH was investigated after spatial normalisation. Particularly strong and widespread associations were found for age, gender and hypertension. Different distribution patterns were found for men and women. Further, increased probability was found in association with self-reported alcohol and tobacco consumption, as well as in those with a history of migraine. It is concluded that the location of WMH is dependent on the risk factors involved pointing towards a regionally different pathogenesis and/or vulnerability of the white matter.

Highlights

► White matter changes are a frequent finding on brain MRI of elderly subjects. ► Overall severity is known to be associated with vascular risk factors. ► The spatial distribution of risk factor impact not well described. ► We apply voxel-based logistic regression to a large group of non-demented elderly. ► Particularly strong associations were found for age, gender and hypertension.

Introduction

White matter hyperintensities (WMH) are frequently seen on brain MRI of elderly subjects (de Leeuw et al., 2001). Although these lesions may be found in healthy subjects, their presence and severity has been associated with cognitive disorders, gait and balance disorders and depression (de Groot et al., 2000) (Guttmann et al., 2000, Whitman et al., 2001). Associations with measurements of atherosclerosis, i.e. intima–media thickness of the carotid arteries (Manolio et al., 1999), retinopathy (Wong et al., 2002) and changed endothelial marker profiles (Hickie et al., 2005, Ovbiagele and Saver, 2006) have further strengthened this concept.

The most important risk factors for WMH are increasing age (Breteler et al., 1994, Liao et al., 1997, Schmidt et al., 2000) and hypertension (Basile et al., 2006, De Leeuw, 2004, Jeerakathil et al., 2004, Longstreth et al., 1996). In the majority of studies, males are reported to have more WMH than females, but in general gender does not appear to be a strong risk factor (Longstreth et al., 1996).

Evidence of prior (cerebro)vascular or heart disease, increased homocysteine levels (Hassan et al., 2004, Vermeer et al., 2002) and lower forced expiratory volume (Longstreth et al., 1996) have also been reported as risk factors of WMH. Other risk factors, such as diabetes mellitus, hyperlipidemia, smoking, high body mass index, decreased vitamin B12, and alcohol have yielded inconsistent associations with WMH (Breteler et al., 1994, Hickie et al., 2005, Jeerakathil et al., 2004, Liao et al., 1997, Longstreth et al., 1996, Longstreth et al., 2005, Murray et al., 2005, Stenset et al., 2006).

Recent studies have emphasised the differences in risk factor profile between large and small vessel disease, and between different subtypes of small vessel disease (Jimenez-Conde et al., 2010, Khan et al., 2007). These studies indicate that while age and hypertension are strongly associated to the occurrence of WMH, factors such as hyperlipidemia and diabetes may relate less strongly or even inversely to WMH.

WMH arise in different locations through the brain. Traditionally, a distinction has been made between periventricular and subcortical WMH. It has been suggested that subcortical WMH could be more associated with hypertension and other vascular risk factors, while periventricular WMH have been considered a phenomenon related to more general age processes, including more benign structural changes in sub-ependymal white matter (Scheltens et al., 1993). However, later studies have found no differential association with vascular risk factors, and have suggested that the division between periventricular and subcortical regions/WMH is arbitrary (Barkhof and Scheltens, 2006, DeCarli et al., 2005). Therefore, the impact of WMH location is often assessed by dividing the brain into different brain regions, i.e. the frontal, parietal, occipital and temporal lobes, basal ganglia and infratentorial region.

The concept of a spatially varying vulnerability to WMH is important because it might reflect differences in pathophysiological mechanisms, and as such be of potential use to the design of clinical intervention. Furthermore it may help inform the interpretation of clinical MR scans showing WMH, and to establish a firmer basis for defining patterns related to normal versus pathological ageing.

Few studies however, have investigated these relationships between risk factors and WMH localisation. The WMH distribution in the presence of hypertension has been studied, but is still unclear. In one study, hypertensive subjects had a higher load of periventricular WMH and fronto-temporal WMH in comparison to normotensive subjects (Wiseman et al., 2004). On the contrary, other investigators showed that hypertension is correlated with subcortical WMH (Murray et al., 2005) or found no difference in WMH pattern at all (Enzinger et al., 2006). One previous study considered the distribution of WMH using a truly voxel-based approach (Enzinger et al., 2006). However, this study focused on the differential distribution of WMH, when classified according to a visual rating scale, and considered only a limited set of vascular risk factors. To our knowledge, other risk factors than age and hypertension have not been studied in detail with respect to WMH location. Voxel-wise statistical analysis was used in another previous study (Holland et al., 2008) to compare the distribution of WMH in patients with dementia or amyloid angiopathy vs. controls. Furthermore, one study considered the impact of low cardiac output on the distribution of WMH, but did not analyse the distribution by voxel-wise statistical scores (Jefferson et al., 2011)

Therefore the aim of this study is to characterise the spatial distribution of lesion probability as a function of individual or groups of risk factors using a voxel-wise analysis. While this approach avoids the bias associated with predefined regions of interest, it is also complicated by the binary distribution of the outcome variable (i.e., either absence or presence of a lesion), which precludes the use of standard linear regression methods. The few studies that have attempted lesion-probability mapping (LPM) using a linear approach, therefore have combined this with permutation-based inference (Enzinger et al., 2006). In the present study we present another approach, using direct fitting of a logistic function describing the lesion probability in each voxel. In this study we present data from a large cohort of normally functioning elderly subjects with white matter abnormalities, and use the logistic regression method to characterise the influence of risk factors on the specific risk of having a lesion at a given location.

Section snippets

Study population

Data were drawn from the multi-centre, multinational Leukoaraiosis and Disability (LADIS) study. The LADIS project studies the role of WMH as an independent predictor of the transition to disability in initially non-disabled elderly. The rationale and design of the LADIS study have been described elsewhere (Pantoni et al., 2005). In short, 639 elderly subjects who had no or only mild disability in their instrumental activities of daily living (IADL) were enrolled. Subjects presented with

Results

A total of 605 subjects could be included in the analysis, while 34 had to be excluded due to missing FLAIR or T2-weighted scan, inadequate volume coverage or image contrast. The subjects included had a median age of 73.8 years (range of 64.1–85.4), and a gender distribution of 282 men/323 women. The 11 recruiting centres were represented evenly in the data set, contributing from 8 to 11% of the subjects each. The MRI was performed an average of 11.4 ± 28.4 (median 7) days after inclusion and the

Discussion

In the present work we used a method for probability mapping based on direct non-linear fitting of the relationship between risk factors and local probability for white matter hyperintensity (WMH). The study further confirmed that the occurrence of WMH is strongly related to age in most white matter areas, in particular those close to the ventricles. Meanwhile, other factors, such as gender, hypertension and consumption of alcohol and tobacco showed associations of similar extent but with

Conclusion

In conclusion we have demonstrated significant association patterns between WMH and a number of risk factors, using a rigorous voxel-based method. Gender, hypertension and smoking were among the risk factors giving the most distinct patterns. We have demonstrated the capability of the method to produce normative spatial probability maps for WMH as a function of given risk factors. We suggest that future studies analyse the association with risk factors and/or functional consequences in more

Disclosure statement

None of the authors state any actual or potential conflicts of interest with people or organisations within 3 years of beginning the present work.

Acknowledgments

This study was supported by the European Union (grant QLRT-2000-00446, Impact of age-related brain white matter changes on transition to disability in the elderly “Leukoaraiosis And DISability”). We would also like to acknowledge the Danish Velux Foundation for their financial support to the Danish group.

References (39)

  • J.C. de Groot et al.

    Cerebral white matter lesions and cognitive function: the Rotterdam Scan Study

    Ann. Neurol.

    (2000)
  • F. De Leeuw

    Interaction between hypertension, apoe, and cerebral white matter lesions

    Stroke

    (2004)
  • F.E. de Leeuw et al.

    Prevalence of cerebral white matter lesions in elderly people: a population based magnetic resonance imaging study. The Rotterdam Scan Study

    J. Neurol. Neurosurg. Psychiatry

    (2001)
  • C. DeCarli et al.

    Anatomical mapping of white matter hyperintensities (WMH): exploring the relationships between periventricular WMH, deep WMH, and total WMH burden

    Stroke

    (2005)
  • C. Enzinger et al.

    Lesion probability maps of white matter hyperintensities in elderly individuals

    J. Neurol.

    (2006)
  • F. Fazekas et al.

    MR signal abnormalities at 1.5 T in Alzheimer's dementia and normal aging

    AJR

    (1987)
  • A.S. Glantz et al.
  • C.R. Guttmann et al.

    White matter abnormalities in mobility-impaired older persons

    Neurology

    (2000)
  • A. Hassan et al.

    Homocysteine is a risk factor for cerebral small vessel disease, acting via endothelial dysfunction

    Brain

    (2004)
  • Cited by (0)

    1

    A list of collaborators of the LADIS study is presented at the end of the paper.

    View full text