Introduction

The blood-brain barrier (BBB) is important for maintaining homeostasis in the brain¹. When the BBB is disrupted, unwanted biomolecules can freely cross the barrier and leak into the brain parenchyma. The commonly used imaging method to quantify BBB permeability is dynamic contrast-enhanced MRI (DCE-MRI)². During DCE-MRI, a contrast agent is injected into the vascular system, while continuously acquiring T1-weighted images preferably with a high temporal resolution and a long measurement period (>25min). However, subtle BBB leakage is a slow process and does not require this high sampling rate³. The proposed method consists of a limited number of post-contrast T1-weighted images that are sparsely sampled over time, leaving more time in between for other MRI scans. This is clinically attractive, since it can easily be added to MRI protocols without significantly increasing the scan time. Furthermore, the post-contrast T1-weighted images have better signal-to-noise ratios and the BBB leakage map has a higher spatial resolution, facilitating the detection of BBB leakage in small brain structures. The aim of this study was to explore the feasibility of the novel protocol, by applying it to patients with a brain tumor.

Method

Subjects: Two patients (males, age 66 and 60 years) with a brain tumor lesion (WHO II grade glial tumor in the right hemisphere and recurrent craniopharyngioma) were included.
MRI acquisition: Imaging was performed on an 3T MRI system (Achieva TX, Philips Healthcare, Best, the Netherlands) using a 32-element head coil. The protocol for the novel BBB method consisted of one pre-contrast and four post-contrast T1-weighted sequences (Turbo field echo with inversion pulse; TR/TI/TE=6.7/1060/3.0ms; field of view (FOV): 240x240x171mm³; reconstructed voxel size: 0.36x0.36x1.0mm³; Compressed SENSE reduction factor: 4.5; acquisition time=2:19min). The post-contrast scans were acquired at 2, 10, 20 and 30 minutes after contrast agent injection (gadobutrol, dose 0.1mmol Gd/kg, infusion rate: 3mL/s). Especially, the latter scans are important for detecting subtle BBB leakage rates, since they allow the contrast agent more time to accumulate in the extravascular extracellular space³. A fast DCE sequence (saturation recovery gradient recalled sequence; 90° non-selective saturation prepulse with a time delay of 120ms; TR/TE=5.3/2.5ms; FOV: 156x156x55mm³; reconstructed voxel size of 1.0x1.0x5.0mm³; time interval: 2.7s; 30 volumes) was applied during bolus injection to measure the arrival and initial blood circulation of the contrast agent and to obtain the first part of the vascular input function (VIF).
Image analysis: Whole cerebrum white and gray matter regions were obtained by automated segmentation (Freesurfer, v6.0.0⁴), with manual adjustment. The tumor tissue was manually delineated under supervision of an experienced neuroradiologist. Motion correction was performed by registering the post-contrast T1-weighted and the fast DCE images to the pre-contrast T1-weighted image. For conversion of the signal intensity changes to concentration, a pre-contrast T1 relaxation time map (T10) was estimated for the native tissue. The patient-specific VIF was manually retrieved from the superior sagittal sinus. The graphical Patlak approach was used for voxel-wise pharmacokinetic modeling of the concentration in tissue and plasma to obtain the BBB leakage rate (K_i)⁵. Histograms of the K_i values were calculated from the noisy K_i-maps in gray matter, white matter, and tumor tissue. Hereafter, the histogram approach was applied to extract the detectable leakage volume fraction (V_L) and the corresponding noise-level-corrected K_i⁶.

Results

The image acquisition method used for quantification of BBB permeability is shown in figure 1, in which free time frames (white) are shown, which were used to perform other relevant MRI sequences in between. This figure also shows an example of concentration curves over time. Histograms without noise removal are depicted in figure 2. Table 1 shows the V_L, the noise-level-corrected K_i values and the standard deviation of the subtracted noise, calculated for white matter, gray matter and tumor tissue. The K_i and the V_L in the strongly enhanced craniopharyngioma were higher than in the low-grade tumor. In both subjects, higher V_L and K_i were measured for gray matter compared to white matter. The K_i-maps for both patients are shown in figure 3.

Discussion

The study shows that, despite the limited temporal resolution after the initial administration of the contrast agent, the time-efficient protocol was able to measure strong leakage in hyperpermeable tumor tissue as well as subtle BBB leakage in healthy appearing tissue. The K_i values are in the same order as found in literature². Moreover, the usability was also supported by obtaining high tissue permeability in a craniopharyngioma, where strong leakage was expected, and low tissue permeability in a low-grade glioma. The Patlak model tends to underestimate the true K_i values of high leaking tissue, indicating that the BBB leakage in the hyperpermeable tumor is even higher than estimated⁷. The new DCE-MRI protocol creates 3 time-slots of 6-8 minutes in between for other MRI sequences, of which the effect of contrast agent presence should be evaluated in more detail. Currently, this technique is being applied in more brain tumor patients.

Conclusion

The proposed MRI protocol with long-time interleaved sampling for detection of subtle BBB leakage is promising and clinically attractive due to its short total scanning time. This method needs to be further evaluated in a larger group of patients with various levels of BBB disruption.

Acknowledgements

No acknowledgement found.