Impaired Cerebral Perfusion in Multiple Sclerosis: Relevance of Endothelial Factors

Cerebral perfusion imaging by quantitative MRI exogenous contrast agent

Over the past 100 years, several techniques have been devised to measure CBF. From the very early invasive measurements, human CBF assessment has been complemented by the efforts to develop quantification techniques such as radioactive 133 Xe, positron emission tomography, or single-photon emission computed tomography.^30–36 Despite their low spatial resolution, these techniques led to the proof of hyperperfusion during acute inflammation and hypoperfusion in the quiescent phases in animal models of MS. Nevertheless, the ultimate goal of a totally noninvasive method that enables the mapping of CBF with high temporal and spatial resolution over the wide range of relevant blood flows has not been attained. The general model used for perfusion quantification is based on the principles of nondiffusible tracer kinetics and relies on the assumption that with intact BBB, the tracer remains intravascular. Considering a bolus of contrast tracer injected, its concentration within a given volume of interest (VOI) at a later time t can be defined according to the following functions: (1) transport function, reflecting the distribution of transit times through the voxel, which is dependent on the vascular structure and flow; (2) residue function, which is the fraction of tracer still present in the VOI at time t following an ideal instantaneous unit bolus injection; and (3) arterial input function (AIF) corresponding to the concentration of contrast agent in the feeding vessel to the VOI at time t. The Gd-DTPA concentration is measured through its effect on magnetic resonance (MR) signal intensity. It is assumed as/to be necessary that the contrast agent concentration time curve corresponds to changes in regional MR signal intensity. The MRI perfusion sequences used are termed DSC and DCE. With exogenous contrast medium, the DSC sequence shows a decrease in signal intensity on T2-weighted images, whereas DCE sequence evidence increases in signal intensity on T1-weighted images. At present, the so-called AIF^{26,35,37–39} is central to quantification of CBF. The DSC acquisition achieves perfusion parameters such as CBF, CBV, MTT, and time to peak (TTP), which are always relative and indirect perfusion parameters.

The advantages of DSC are few minutes of acquisition time, possible quantitative measurement (through deconvolution of the measured tissue curves by an AIF and not under 8 mL/min/100 g), useful for emergency setting, and has no age limitation (children). During clinical practice, the DSC perfusion map (CBF, CBV, MTT, TTP) does not permit, as discussed above, quantitative perfusion values but demonstrates critical hemodynamic disturbances in a number of clinical setting. The main clinical applications of DSC are acute stroke assessment, tumor grading, differential diagnosis among chronic cerebrovascular diseases, infections, demyelinating lesions, and tumors. ²⁷ The evaluation of mismatch between CBV and CBF in ischemia areas can assess areas of reversible ischemia vs irreversible infarction: decreased CBF associated with normal or increased CBV suggests reversible ischemia, whereas reduced CBF and CBV are indicative of irreversible infarction. This semiquantitative perfusion map associated with other MRI acquisitions such as diffusion-weighted MRI (DWI) and MR angiography can optimize the diagnosis of acute stroke in an emergency setting without ionized radiation (perfusion computed tomography [CT]). The CBV semiquantitative evaluation achieves differential diagnosis between tumor recurrence vs radiation necrosis or infection vs tumors or tumor-like MS lesions vs tumors. Disadvantages of DSC consist of under- and/or overestimation of relative cerebral blood volume (rCBV) due to high magnetic susceptibility caused by hemoglobin catabolic molecules (hemorrhagic areas) or due to parenchymal contrast medium leakage caused by high BBB permeability. In enhancing high-grade gliomas with substantial BBB breakdown, typical Gd-based contrast agents extravasate, reducing T2* DSC-MRI signal intensity loss by signal intensity increase in regions where T1 effects are significant. ⁴⁰ The DSC perfusion parameters cannot be corrected when the hemorrhagic areas are present, whereas parenchymal contrast medium leakage can be avoided by preload contrast medium administration before DSC acquisition. The DSC leads to hemodynamic evaluation, although it does not supply information about BBB permeability. On the contrary, DCE furnishes information about not only perfusion but also permeability parameters.

Tracer kinetic modeling of DCE sequence is able to measure 4 parameters involved in a 2-compartment exchange model: fractional plasma volume, fractional interstitial volume, plasma flow, and K-trans (permeability-surface index). To measure the above perfusion parameters, mathematical models such as Toft and Patlak are used. The Toft model calculates bidirectional exchange flow (intravascular-interstitial space) and therefore BBB permeability.^41,42 As shown by Heye et al, the Toft model is able to measure K-trans even with mild BBB permeability changes. There are, however, some critical aspects to be considered when measuring K-Trans: (1) capillary surface area (capillary density and size, and microvessel pathology) and (2) temporal resolution (the latter being longer than the tissue blood compartment transit time).^42–46

K-trans is the result of the leakage of contrast medium across the BBB related to plasmatic contrast concentration and is tightly linked to CBF. Using the Patlak method, based on 2-compartment analysis, only unidirectional transport of contrast agent across the BBB from the blood plasma into the extravascular has been considered.^42,43,45

However, for both models, AIF, high plasmatic volume, and MRI parameters (acquisition time, precontrast longitudinal relaxation time T1, dose of contrast medium, and field stability) need to be set in order to achieve accuracy in measuring BBB permeability. Indeed, the characteristic of acquisition parameters such as acquisition time has to be calibrated in function of the disease to be studied. Depending on the brain disease, the choice of the models and acquisition time improve the BBB permeability measure.

To demonstrate late permeability phenomena and not underestimate the permeability of small lesions or normal-appearing white matter (NAWM), the Toft method requires a long acquisition time. In animal models, diffuse signal enhancement in parts of lesions has been observed up to 5 hours after injection. This observation suggests that other mechanisms may become manifest several hours after injection.^42,45 In humans, during clinical practice, acquisition time has to be as long as possible to capture slow interstitial uptake when BBB is not so altered as to permit a contrast leakage few minutes after the contrast intravenous injection (tumors, infective processes, active MS lesions, etc.). Indeed, in diseases such as dementia, small vessel disease, MS, and so on, NAWM may demonstrate altered permeability values only if the acquisition time is long enough.^46–48 To summarize, it is known that the demonstration of new MS lesions during MR follow-up and MS lesion load does not always correlate to disability outcomes ⁴⁹ or that annual progression of atrophy is not always associated with benign MS.^50,51 Volumetric MRI, MR spectroscopy, and quantification of magnetic transfer ratio (MTR) in MS have contributed to demonstrate the consequence of a long-lasting neuroinflammation rather than MS prognosis or MS pathophysiology. On the contrary, quantification of BBB permeability could demonstrate the existence of ongoing subclinical disease activity even in the absence of relapses or MRI activity. The DCE perfusion analysis focused on BBB permeability may predict conversion from optic neuritis to MS 2 years earlier and, by differentiating MS from optic neuritis, may constitute an early prognostic factor in the pathogenesis of MS. ⁵¹

Cerebral perfusion imaging by quantitative MRI endogen contrast agent

Arterial spin labeling is a quantitative cerebral perfusion measurement technique that harnesses blood water as an endogenous contrast agent and thus, uniquely, is entirely noninvasive. ²⁹ Radiofrequency pulses “labeled” hydrogen proton spins in arterial blood water. These act as an endogenous tracer and CBF maps are acquired: the tagged spins reaching the region of interest are proportional to CBF. In other terms, the CBF depends on changes in tissue longitudinal relaxation (T1) caused by the exchange of water between blood and tissue. The subtraction between the control image (protons not labeled) and the labeled image results in a perfusion-weighted image and is then quantified into the CBF map.

As in the physiological literature, in ASL maps, the quantification of CBF (based on the time delay between labeling and imaging) is expressed in mL/100g per minute.

Arterial spin labeling sequence can be summarized in 3 phases: labeling, postlabeling, and read-out of signal. It is crucial to synchronize the time of image acquisition and the arrival of labeled bolus to the target tissue named postlabeling delay (PLD). The arrival of labeled hydrogen proton spin bolus is influenced by arterial transit time (ATT), which is the velocity of blood.

In fact, arterial occlusive pattern or arterial-venous fistula may increase or decrease ATT, respectively, and may affect CBF analysis. The critical aspects of ASL are to optimize the PLD and to consider ATT (different among neonates, children, and adults), low S/N ratio (only 16% of signal intensity is useful to create the CBF map after the subtraction of image control and labeled image), high signal intensity due to large vessels, and magnetization transfer of energy to stationary tissues. The advantages of ASL perfusion are the absolute quantification of CBF, the short sequence acquisition time (few minutes), and the absence of contrast medium administration and ionizing radiation. ²⁹ As far as perfusion studies using MRI are concerned, the blood oxygenation level–dependent (BOLD) signal needs to be mentioned. The BOLD signal is the standard technique used to generate images in fMRI studies and relies on the close relationship between CBF and NVU function.

Both the BOLD and ASL reflect different aspects of the dynamic variation of cerebral perfusion. The first is based on the reduction of signal intensity (T2-weighted images) due to magnetic susceptibility to deoxyhemoglobin (neuronal activity correlated to O₂ consumption), whereas the latter obtains increased signal intensity (T1-weighted images) from excited blood protons (CBF) in the target tissue. The BOLD signal depends also on CBF, although it does not express it directly, whereas ASL gives absolute CBF quantification. In other terms, BOLD reflects the neurovascular coupling, ie, the temporal and regional linkage between neural activity and metabolic rate of oxygen consumption, CBF, and CBV.

The BOLD signal can be quantified by mathematical and mechanistic modeling tests and may represent a putative biomarker of brain perfusion variations.^52–54

Brain perfusion in MS

In patients with MS, cerebral hypoperfusion was first observed about 30 years ago and it has become evident now that reduced cerebral perfusion, likely due to NVU malfunction and involving both the gray and white matter, is an integral feature of MS pathology.^2,55 These findings may influence our understanding of the underlying disease processes and be important for future therapeutic considerations. The relationship between MS lesions and the cerebral vasculature has long been recognized. Histopathologic evidence of vascular occlusion was described in the 1930s by Putnam, ⁷⁴ suggesting that vascular inflammation precedes demyelination and could be a primary event in the evolution of the disease. Later, Wakefield et al ⁷⁵ demonstrated fibrin deposition and thrombosis of vessels in the absence of cellular infiltration. Occlusive changes have also been confirmed in the retinal venules of patients who later developed MS. ⁷⁶ Nevertheless, it is still unclear whether abnormal perfusion is a precursor of lesions or occurs independently of lesion development through a different mechanism. Among the mechanisms responsible for reduced perfusion in MS, vasoactive substances have been suspected to play a leading role. In animal models, the demyelinating lesion has been demonstrated at a vascular watershed such as the white-gray matter border, after 1 week from activation of innate immune mechanisms and not centered on the demyelinating injection site of lipopolysaccharide. This pathological pattern suggests that the immunological answer has been realized in an area where the vascular susceptibility to hypoxia is higher. In the same experiment, major oxygen administration can partially reduce superoxide and NO formation and cellular energy default by interrupting the hypoxia mechanism. In the animals that underwent inspired oxygen application, the demyelination process was reduced. If a relationship between demyelinating lesions and critical vascular watershed has been demonstrated, a vascular phase underpinning the disease progression has to be considered. ⁷⁷ It is known that micro- and macro-circulation changes (eg, venous thrombosis, stroke) lead to enhanced BBB permeability and increased paracellular fluid (shown by white matter [WM] vasogenic edema on T2-weighted images) until critical oxygen availability (reduced ADC area on diffusion-weighted images). Enhanced BBB permeability, indirectly measured by MRI (ie, edema and reduced brain perfusion), is linked to (1) deregulation of brain vessels involving interactions of vasoactive substances both at a receptor and at a signal transduction level and (2) endothelial transcellular or paracellular changes and/or endothelial TJ disruption. These events may represent general examples of either an effective pathological feature or NVU properties able to respond to injury. ⁷⁸ Given the tight relationship between CBF and BBB permeability, as well as the tight relationship between brain vasoreactivity and function, it could be important to know brain hypoperfusion onset in MS disease. To know the onset of cerebral hypoperfusion in MS might clarify whether CBF reduction is the cause or the consequence of MS pathophysiologic process. Moreover, brain hypoperfusion or forerunning BBB permeability could identify translation from quiescent to acute phase in relapsing-remitting (RR)-MS and/or from RR-MS to progressive form. Indeed, RR-MS and secondary progressive MS (SP-MS) groups, showing a different correlation between a brain perfusion parameter such as CCT and MS duration disease as well as endothelin 1 (ET1) plasmatic concentration, could be considered as 2 different pathologic entities. ⁴⁴ To try to understand when the neuroprotective process failed (MS progressive form) or is limited and partially reversed (MS RR form) could, in our view, be a helpful, alternative point of view and drive a different research hypothesis.

Brain perfusion in MS: cytotoxic or vasogenic edema

Cerebral hypoperfusion has been demonstrated in secondary progressive form (SP-MS) as well as in RR-MS patients, although ischemic threshold is not usually reached.

However, MRI demonstrated cytotoxic signs inside MS lesions in diffusion-weighted images, suggesting a pathologic vascular aspect, although there are more than one reasons able to explain it.

The reduction of apparent diffusion coefficient (ADC), cytotoxic sign, in MS acute demyelinating lesions could be due to (1) hypoxic changes leading to ischemic precondition expressed by hypoxia markers in oligodendrocytes such as hypoxia-inducible factor 1 and heat shock protein 70, (2) intramyelinic edema or myelin vacuolization and/or myelin breakdown reducing water movement in the extracellular space because fiber tract organization is reduced, and (3) the presence of a hypercellular inflammatory infiltrate and/or iron-laden macrophages. The most relevant thing to note is that reduction of ADC precedes the enhancement leakage in MS acute lesions demonstrating BBB breakdown. ⁷⁹

Cerebral ischemic stroke is a pertinent model for understanding CBF/biochemical changes. Several studies relative to the ischemic process have demonstrated that the same molecules support different actions (protection or toxicity) depending on the time frame of the pathophysiologic pathway. Some examples are as follows: N-methyl-d-aspartate (NMDA) receptors induce acute excitotoxicity ⁸⁰ in an early phase of ischemic lesion, whereas in the late frame without NMDA signaling, chronic neuronal remodeling cannot take place. ⁸⁰ High Mobility Group 1 (HMGB1) is an expression of acute cerebral ischemia, necrosis, and core infarct, ⁸¹ while in the late stages reactive astrocytes release HMGB1 and promote angiogenesis and synaptic plasticity.^82,83 Matrix metalloproteinases degrade and damage neurovascular substrates in the acute phase, although proteases are themselves critically important for neurovascular remodeling during the recovery phase.^84,85

Nevertheless, the molecular interaction and their exact time frames in regulating vessel reactivity and pathologic or regenerative process are not yet well known in stroke, MS, or other neurodegenerative diseases (Figure 2).

OPEN IN VIEWER

Figure 2.

MRI perfusion in MS disease and endothelial factors. Different plasmatic titration of endothelial factors modulate the brain perfusion and model the space/time NVU modification: vasodilation (increased CBV, CBF, TTP and MTT) leading to hyperperfusion in active demyelinating lesion region of interest ROI (1), whereas a reduced brain perfusion is demonstrated (reduced TTP) in chronic and stable condition in normal-appearing WM and stable demyelinating lesion. BBB indicates blood-brain barrier; CBF, cerebral blood flow; CBV, cerebral blood volume; HMGB1, High Mobility Group 1; MTT, mean transit time; MRI, magnetic resonance imaging: MS: multiple sclerosis; NMDA, N-methyl-d-aspartate; NVU, neurovascular unit; TTP, time to peak.

Brain perfusion and neurovascular signaling

In the nervous system, the coupling between brain vessels (CBF) and neurons is thought to be mediated by vasoactive mediators, among which are NO/ADMA and endothelins. Under normal brain perfusion, NO is constitutively produced by endothelial nitric oxide synthase (eNOS) at nanomolar concentrations, the enzyme activity being calcium and calmodulin dependent. ⁸⁶ In an inflammatory environment such as the MS brain, the inducible form of NOS (iNOS) is upregulated, producing micromolar concentration of NO and nitrogen reactive species. Nitric oxide has 2 major effects on cerebral vessels, both of which may be involved in the pathogenesis of MS lesions, namely, vasodilation and alteration of the BBB. Vasodilation by itself may facilitate inflammation by decreasing the velocity of blood flow, thereby aiding leukocyte transmigration, the latter facilitated by NO-induced BBB breakdown. In addition, NO may cause conduction block, perhaps by impairing the function of sodium channels; demyelinated axons are particularly vulnerable to this effect. Furthermore, raised concentrations of NO and related reactive species may impair synaptic transmission, which, in addition to compromising transmission in motor and sensory pathways, may contribute to the loss of function in patients with MS. ⁸⁶ Increased NO levels could drive the overproduction of its own inhibitor (ADMA) or counterbalancing vasoactive peptides such as ET1 and endothelin 3 (ET3), again produced by endothelial cells. ET1 vasoconstriction is mediated by A receptors: high ET1 concentration and high ET1 A receptor affinity promote a severe and prolonged vasoconstriction. However, concurrently, ET1-ETB receptor stimulation may induce vasodilation. ET3, in comparison with ET1, has a high affinity for B-receptor proportional to its concentration (ET3 [nM] vs ET1 [μM]).^87,88 Considering the prevalent brain endothelial origin of ET3 and its high affinity to B receptor (antagonist action of ET1 on receptor type A), ET3 could potentially have a role in neurovascular uncoupling due to endothelial dysfunction in patients with MS. Other modulating factors involved in NO pathway include ADMA, an endogenous inhibitor of NOS. Indeed, elevation of ADMA has been demonstrated in patients with MS, although it not yet known whether its role is inflammatory or vasoactive or likely both. Even if high ADMA titration is associated with impaired vasodilation, no correlation with brain perfusion parameters (such as cerebral circulation time [CCT]) and plasmatic ADMA levels has been found.^18,89,90 To identify biomarkers of chronic cerebral hypoperfusion inducing NVU changes and their putative brain injury and clinical disability is still a challenge (Figure 3). Cerebral hypoperfusion has been demonstrated in many pathologies (MS, small vessel disease, diabetes, Alzheimer disease, Binswanger disease, etc.).^91–93 In MS, plasmatic NO, ET1, and ADMA are increased, ⁹⁴ and it is generally suggested that this may contribute to cerebral hypoperfusion which, in turn, could contribute to the disease pathogenesis. ⁴ Our hypothesis is, rather, that ADMA (by inhibiting both eNOS and iNOS) and ET1 (by opposing the NO-induced vasodilation) could express a compensatory response at least in early disease phase. ¹⁸ In other terms, ET1 and ADMA overproduction, inducing brain vasoconstriction, could be an adaptive response to vascular-derived insults. On the contrary, in the later phase, their persistently high levels can cause functional/structural abnormalities of the brain microvasculature by inducing reduced vessel compliance. In our recent work, we have demonstrated that in SP-MS patients but not in RR-MS, a significant correlation between ET1 and CCT exists. In fact, in RR-MS the high levels ET1 may act to prevent astrocyte hypoxic stress and promote neuroprotective process.^13,20,94 A different correlation between ET1 and uncoupling CBF/BBB may explain and focus the different prognosis between RR and SP/PP MS forms. Indeed, other behavior is shown in SP-MS patients: ET1 level is correlated with increasing CCT and is not related to disease duration, suggesting a complete progressive perfusion imbalance in SP-MS patients at the disease onset.

OPEN IN VIEWER

Figure 3.

BBB and stroke pathophysiology. In different phases (early or late) of ischemic stroke, the same biochemical factors have opposite roles: neurotoxic vs neuroprotective effects. ADMA indicates asymmetric dimethylarginine; BBB, blood-brain barrier; CBF, cerebral blood flow; CBV, cerebral blood volume; ET1, endothelin 1; ET3, endothelin 3; HMGB1, High Mobility Group 1; MTT, mean transit time; NMDA, N-methyl-d-aspartate; NO, nitric oxide; TTP, time to peak.