Cerebral blood perfusion changes in multiple sclerosis Jens Wuerfel, Friedemann Paul, Frauke Zipp ��� Institute of Neuroimmunology, Charit�� ��� Universit��tsmedizin Berlin, Germany Available online 26 March 2007 Abstract The proximity of immune cell aggregations to the vasculature is a hallmark of multiple sclerosis. Furthermore, it is widely accepted that inflammation is able to modulate the microcirculation. Until recently, the detection of cerebral blood perfusion changes was technically challenging, and perfusion studies in multiple sclerosis patients yielded contradictory results. However, new developments in fast magnetic resonance imaging have enabled us to image the cerebral hemodynamics based on the dynamic tracking of a bolus of paramagnetic contrast agents (dynamic susceptibility contrast). This review discusses the technical principles, possible pitfalls, and potential for absolute quantification of cerebral blood volume and flow in a clinical setting. It also outlines recent findings on inflammation associated perfusion changes, which are inseparable from pathological considerations in multiple sclerosis. �� 2007 Elsevier B.V. All rights reserved. Keywords: Perfusion Magnetic resonance imaging Multiple sclerosis Inflammation Blood���brain barrier Dynamic susceptibility contrast 1. Introduction One of the crucial stages in the evolution of a multiple sclerosis lesion is considered to be the disruption of the blood brain barrier (BBB) [1,2], leading to edema in the CNS by accumulation of plasma fluids. This process is believed to be initiated by autoreactive CD4+ lymphocytes which migrate into the CNS and start an inflammatory response. Although BBB breakdown ��� imaged as focal enhancement in T1- weighted MRI aftergadolinium-DTPA (Gd-DTPA) injection ��� is the gold standard of lesion detection during the course of the disease, the deposition of contrast agent in the CNS has been shown to correlate only modestly with clinical disability. However, new imaging techniques, which allow more accuracy inmonitoringthe disease evolution,provide evidence of tissue damage in the normal appearing white matter (NAWM) preceding the appearance of new contrast-enhancing lesions. A decrease in magnetization transfer ratio (MTR) is described prior toenhancement, indicating a diminished ability for saturation exchange due to edema and inflammation [3,4]. Changes in the lipid spectra have been noted in magnetic resonance spectroscopy preceding lesions [5]. Similarly, subtle increases in the random water molecule motion reflecting alterations in tissue integrity were detected weeks before leakage of the BBB by means of diffusion weighted imaging [6]. However, local changes in blood flow, a major conse- quence of inflammation [7���9], have so far received little attention. Perfusion imaging has been performed in several other neurological diseases, such as stroke and different forms of dementia. Here, reduced perfusion parameters in several brain regions were linked to decreased metabolic activity secondary to neurodegeneration [10,11]. Although in multiple sclerosis only few perfusion studies have been published so far, findings of very early focal perfusion changes during the inflammatory phase of the disease [12], and of generally reduced grey matter perfusion in MS patients compared to healthy controls [13] might indeed influence our understanding of the underlying disease processes, and could be important for future therapeutic considerations. 2. In vivo perfusion measurements: Technical background and pitfalls Cerebral perfusion is defined as the volume of blood flowing through a given volume of tissue per unit of time [14]. It is comprised of three parameters: the cerebral blood flow Journal of the Neurological Sciences 259 (2007) 16���20 www.elsevier.com/locate/jns ��� Corresponding author. E-mail address: frauke.zipp@charite.de (F. Zipp). 0022-510X/$ - see front matter �� 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jns.2007.02.011

(CBF), the cerebral blood volume (CBV), and the mean transit time (MTT). Although the first reports on human cerebral flow measurements using magnetic resonance imaging date back as far as 1970 [15], perfusion measurements have remained technically challenging. In 1988, a transient change in contrast after injection of lanthanide chelates (Gd-DTPA) into healthy ratswas described and explained by susceptibility effects in the microenvironment [16]. Such microscopic gradients lead to fast dephasing of proton spins, and hence to a reduction in the transverse relaxation time (a decrease in T2/T2���), which can be detected as signal loss during the fast passage of a contrast agent through the tissue (first pass effect). This effect dominates over the simultaneous T1 relaxation enhancement and extends well beyond the intravascular compartment it also affects the surrounding tissue even though nondiffusible tracers, such as Gd-DTPA, remain strictly intravascular in the presence of an intact BBB [17]. Imaging the first pass effect of intravenously injected paramagnetic contrast agents with high temporal resolution is called Bolus Tracking or Dynamic Susceptibility Contrast Magnetic Resonance Imaging (DSC-MRI). It is the most commonly used method to investigate hemodynamic changes. Maps of CBV, CBF and MTT are calculated on the assumption of an approximately linear relationship between the concentration of the contrast agent in the tissue and the change in T2��� relaxation rate. CBV is determined by integrating the tissue concentration time curve of each voxel, and CBF by the deconvolution of this curve with an arterial input function (AIF: the concentration of contrast agent in the feeding vessel to the volume of interest (VOI) at a given time) [18]. MTT, the average time required for a given particle of tracer to pass through the tissue, can be calculated by the central volume theorem MTT=CBV/CBF [19]. Due to the rapid bolus transit, fast imaging is required in order to capture the first pass of the bolus (e.g. repetition time (TR) b1.5 s). Rapid injection of the contrast agent followed by a saline flush is imperative to obtain a sharp input bolus to the tissue (e.g. N3 ml/s). Today, the availability of fast imaging techniques such as epi-planar imaging (EPI) allows us to obtain sequential images of the wash in and wash out phase of a tracer during the few seconds of bolus transit through the tissue. Until recently, its use in clinical studies was restricted due to a number of technical limitations, such as a relatively poor spatial resolution leading to partial volume effects [20], limited volume coverage, and a low signal-to-noise ratio (SNR). Increased gradient performance and higher field strengths made perfusion MRI feasible even in a routine clinical setting. However, some caveats should be taken into consideration. Since the tracer concentration cannot be measured directly by MRI, it must be measured indirectly through its effect on signal intensity. Therefore the concen- tration of the contrast agent in various tissue compartments must be known or estimated in order to derive hemodynamic parameters from dynamic tracer analyses. As vascular structure has a strong influence on flow, an accurate acquisition of the AIF is essential [21]. The AIF depends not only on the shape of the injected bolus but also on the cardiac output, the vascular geometry, and the cerebral vascular resistance [22,23]. Accordingly, the site of the AIF estimation should be chosen as close to the region of interest as possible. For the quantification of absolute CBF, the AIF must be deconvolved. Several model-dependent and model- independent deconvolution methods have been assessed and discussed in the past [24���28], and were recently reviewed by Ostergaard [18]. The selection of an approach sufficiently reflecting the in vivo situation is necessary for the correct prediction of absolute CBF values [29]. In some pathologies, such as multiple sclerosis, a cons- picuous artefact can be observed in case of permeable capillary walls due to the defective BBB, e.g. in contrast-enhancing lesions. Here, the above mentioned assumption of a negligible effect of the contrast agent influx on T1 do not hold true. Leakage into the tissue leads to an artificially low estimate of CBV. Haselhorst et al. described an algorithm which can be applied to correct for the leakage artefact by fitting a model function for the plasma concentration to the measured data [30]. 3. Cerebral perfusion changes precede blood���brain barrier breakdown in MS Early measurements of global perfusion in multiple sclerosis using radioactive 133 Xe [31], positron emission tomography (PET) [32], or single photon emission computed tomography (SPECT) techniques [33] suffered from a very low spatial resolution. Findings of a generally lower perfusion in MS patients compared to healthy individuals were contradicted by the demonstration of increased perfusion during the time of acute inflammation in an animal model of MS [34]. With the advent of modern MR imaging and postproces- sing techniques, it became possible to differentiate altera- tions in cerebral perfusion more accurately. Rashid et al. reported a hypoperfusion in the grey matter tissue, particularly in the thalamus and the caudate nuclei, and an increase in white matter perfusion in comparison to healthy controls, applying a novel continuous arterial spin labelling (CASL) technique and a statistical parametric mapping analysis (SPM) [13]. Haselhorst et al. were among the first to investigate different MS lesion types in a cross-sectional study [30]. After application of an extended BBB leakage correction to the data, the authors found a significantly higher CBV in acute Gd-DTPA enhancing plaques and a lower CBV in T1 hypointense lesions in comparison to the NAWM and T1 isointense tissue. Similar results were shown in a study by Ge et al. [35]. Our own group presented the first longitudinal data on perfusion changes during the genesis of MS lesions [12]. We confirmed that acute plaques are increased in both CBF and CBV prior to contrast enhancement, and that decreased perfusion persists in those plaques that remained T1 hypointense after several weeks. To our surprise, hyperperfusion developed weeks 17 J. Wuerfel et al. / Journal of the Neurological Sciences 259 (2007) 16���20