Slowly crossing the blood–brain barrier takes time

Abstract

Dear Editor,

In their publication “Tracer kinetic model detecting heterogeneous blood–brain barrier permeability to water and contrast agent in Alzheimer's disease and dementia with Lewy bodies,” Dr. Xu and colleagues measure the brain tissue uptake of an magnetic resonance imaging (MRI) contrast agent over time in severely compromised patients with dementia and elderly controls.¹ From the measurements they simultaneously calculate the permeability of the blood–brain barrier (BBB) for gadolinium contrast agent and water molecules.

We feel this is an important and challenging attempt to assess BBB breakdown in neurodegenerative disease, as it is thought to be one of the key initiating mechanisms of dementia. We compliment the authors with this achievement, but would like to provide a different perspective on the approach and interpretation of the results.

Leakage of (Magnetic Resonance Imaging) MRI contrast agent from the blood circulation into the brain parenchyma is extremely subtle in neurodegenerative disease. Any MRI measurement in brain tissue is mainly determined by the signal from the gadolinium inside blood vessels. To properly measure the tiny amounts of leaking contrast agent requires long measurement times to let the contrast agent sufficiently accumulate in the brain parenchyma. A recent consensus article recommended acquiring images for at least 15 min and preferably longer.² The approach by Xu et al. only sampled 5 min. Including the first minutes of the multiple (re)circulations of the contrast agent means that for a substantial part of the measurement time, no (dynamic) equilibrium is reached for the distribution of the contrast agent over the blood and parenchyma. Moreover, these early (re)circulations reflect differences in the concentration time-course between the supplying artery and microvessels by delaying and broadening (i.e., dispersion) the concentration time-course profile.³ We are concerned that such hemodynamic effects may deteriorate the leakage estimation in such short-term measurements. The calculation of contrast agent leakage is based on differences between arterial and tissue time-courses, and the leakage part is determined predominantly from later signals when the circulation is in the steady-state and sufficient contrast agent has accumulated.

Investigators applied a tracer kinetic model to the MRI signal and considered three compartments (blood, interstitial, and intracellular space) and calculated various parameters, including the leakage of contrast agent and water exchange rate over the endothelium. The authors meticulously performed computer simulations in which single potentially superfluous parameters were omitted to determine the simplest model with the best fit, very much in line with previous work.⁴ The inclusion of water exchange effects has been intensively debated in the literature, but is a conceptually valuable addition in our opinion.⁵ However, authors did not consider an even much more simple model in which only the temporal change of the blood concentration traversing from the artery to the tissue capillaries is described. Brain tissue capillaries represent a fine deeply branched network which exerts resistance to the arterial blood stream and will alter (delay and dampen) the concentration (bolus shape) time-course. Such a dispersed concentration time-course can be erroneously conceived as gadolinium retention in tissue and be mimicked in tracer models by a positive leakage rate (Ktrans). Looking at the example Ktrans maps in Figure 2D, we notice conspicuous curvilinear structures depicting very high leakage rates. This could be fully explained from the change in time-course from arterial input, through the capillaries to intracerebral veins, which would mimic contrast agent retention by a strongly positive Ktrans value.

With ageing or neurodegenerative disease, not only brain parenchyma degenerates, but also the vasculature. In general, the peripheral resistance to cerebral blood flow increases, due to associated vascular pathology, vessel wall remodeling, tortuosity, and capillary rarefaction.⁶ With the regression of microvessels the vascular surface area available to the exchange of water molecules (also gadolinium) reduces with neurodegenerative disease. As exchange rates are proportional to the product of permeability and surface area (P × S),⁷ the combination of dispersed blood flow (described with higher Ktrans) and reduced water exchange (lower k) would alternatively explain the relationships seen in Figure 5 (second row). However, we are uncertain whether the reported observations are conceptually significant as the correlations were taken over all permeability pairs and all subjects groups, being prone to Simpson's paradox.⁸

In sum, we are concerned about the adequacy of the MRI measurement, the conceptual choices of the tracer kinetic modeling, and provide an alternative hemodynamic mechanism that could explain the presented results and needs to be considered. Simple hemodynamic modeling of the current data could show that. We do very much value the idea to research various cerebromicrovascular permeability aspects in neurodegenerative disease.

The authors declare no conflicts of interest. Author disclosures are available in the Supporting Information.