Arjen Lindenholz

126 CHAPTER 5 in the African American subpopulation. 20 This difference between the studies could be caused by several factors, such as differences in study population (community-based population in the United States versus Dutch participants with an ischemic stroke or TIA). In addition, in contrast to these studies, we used the 3DMPIR-TSE sequence at 7T MRI, which has the advantage of a higher signal-to-noise and contrast-to-noise ratio, enabling optimal cerebrospinal fluid and blood suppression and clear delineation of both inner and outer vessel wall boundaries, reducing the risk of misclassification of vessel wall lesions. 16,21,28 At least three hypotheses can be put forward to explain the lack of an association that we found between some of the accepted risk factors for cardiovascular disease (e.g., smoking) and intracranial vessel wall lesions. First, theremay be a real difference between risk factors for extracranial and intracranial atherosclerosis, a hypothesis that is supported by previously published studies. 8,34 Second, the intracranialvesselwalllesionsthatwescoredmaynotonlyrepresentatherosclerotic plaques but also vessel wall thickness fluctuation due to aging, natural variation or partial volume effects, thereby underestimating any association between the risk factors and ‘true’ atherosclerotic lesions. So far, there has been limited validation of in vivo intracranial vessel wall MRI sequence with histologic findings. 35 However, ex vivo vessel wall MRI studies have shown that lesions found with vessel wall MRI do represent intracranial atherosclerosis. 36 Third, it may be that certain risk factors are only associated with ICAS in specific ethnic populations, or with more advanced stages of ICAS, in which point luminal narrowing begins to develop. A limitation that should be considered is the relatively small number of study participants and a potential selection bias towards less-affected participants, which is inevitable in this type of MRI study. Severely disabled patients or patients with severe cardiovascular diseases treated with stent placement, pacemakers or implantable cardioverter defibrillators were not able to participate because of safety reasons. The small sample size and consequent ‘selection’ of less- affected participants in theory might have led to a reversed causality with an underestimation of the true association between risk factors and vessel wall lesion burden. In addition, as we based our results on the multivariable models only, we have run three of such multivariable models (adjusted models in Table 2, 3, and 4 ). The application of Bonferroni correction to control for the family-wise error rate would leave increasing age, diabetes mellitus, and systolic blood pressure as the main predictors of abnormalities. Furthermore, relating vascular risk factors to the number of intracranial vessel wall lesions is challenging. Vascular risk factors are dependent on several aspects, such as severity of exposure status, lifestyle behavior, medication adherence, and genetic susceptibility. Therefore, it is not possible to include all potential confounders, effect modifiers and residual biases into one model. Studies with a larger number of patients that include