Detection of IV-gadolinium Leakage from the Cortical Veins into the CSF Using MR Fingerprinting

Abstract

Purpose:

It has been reported that leakage of intravenously administered gadolinium-based contrast agents (IV-GBCAs) into the cerebrospinal fluid (CSF) from the cortical veins even in healthy subjects can be detected using a highly sensitive pulse sequence such as heavily T₂-weighted 3D fluid-attenuated inversion recovery and 3D-real inversion recovery (IR). The purpose of this study was to evaluate the feasibility of MR fingerprinting to detect GBCA leakage from the cortical veins after IV-GBCA.

Materials:

Fourteen patients with suspected endolymphatic hydrops (EH) who received a single dose of IV-GBCA (39–79 years old) were included. The real IR images as well as MR fingerprinting images were obtained at 4 h after IV-GBCA. T₁ and T₂ values were obtained using MR fingerprinting and analyzed in ROIs covering intense GBCA leakage, and non-leakage areas of the CSF as determined on real IR images. The scan time for real IR imaging was 10 min and that for MR fingerprinting was 41 s.

Results:

The mean T₁ value of the ROI in the area of GBCA leakage was 2422 ± 261 ms and that in the non-leakage area was 3851 ± 235 ms (P < 0.01). There was no overlap between the T₁ values in the area of GBCA leakage and those in the non-leakage area.

The mean T₂ value in the area of GBCA leakage was 319 ± 90 ms and that in the non-leakage area was 670 ± 166 ms (P < 0.01). There was some overlap between the T₂ values in the area of GBCA leakage and those in the non-leakage area.

Conclusion:

Leaked GBCA from the cortical veins into the surrounding CSF can be detected using MR fingerprinting obtained in <1 min.

Introduction

We have previously reported that intravenously administered gadolinium-based contrast agents (IV-GBCAs) leak into the cerebrospinal fluid (CSF) from the cortical veins even in subjects without blood brain barrier (BBB) disruption.^1–3 This leakage cannot be detected with regular T₁-weighted images, but with heavily T₂-weighted 3D fluid-attenuated inversion recovery (3D-FLAIR) or 3D-real inversion recovery (3D-real IR) images.^4–6 GBCAs in the CSF can enter the brain parenchyma, probably through the perivascular space in humans^1,7–10 as well as animals.¹¹ This “glymphatic” route is speculated to be the path by which GBCA enters the brain, and ultimately accumulates.^10,12,13 In addition to the cortical veins, the leakage points for GBCAs into the CSF have been suggested to include the peripheral part of the cranial nerves, the circumventricular organs, and the spinal ganglions.^2,14 In our clinical practice, we routinely obtain 3D-real IR images with a longer TR for the evaluation of endolymphatic hydrops (EH) of the inner ear at 4 h after a single dose of IV-GBCA.^5,6 It has been reported that leakage from the cortical vein is not observed in young subjects below the age of 37 years.¹⁵ The degree of leakage of IV-GBCA into the CSF may be animaging biomarker for the evaluation of aging and neurodegeneration.¹⁵

Magnetic resonance fingerprinting (MRF) is a method that uses transient signal evolutions and data analysis to simultaneously estimate several quantitative tissue property parameters like T₁ and T₂ relaxation times.¹⁶ Spatial and temporal incoherence through varying acquisition parameters lead to very efficient signal encoding. The acquisition time for one slice of MRF takes 41 s. We now routinely use spiral MRF with fast imaging based on steady-state precession (FISP) to quantify the parameters of various tissues.¹⁷ MRF uses a pseudo-randomized acquisition that causes the signals from different tissues to have a unique evolution or “fingerprint”, which is a function of the multiple material properties under investigation. The processing after acquisition involves a pattern-matching algorithm to voxelwise match the fingerprints to a predefined dictionary of simulated signal evolutions featuring different T₁ and T₂ relaxation times. The pattern matching of all voxels leads to quantitative maps of the parameters of interest.¹⁸

The purpose of this study was to evaluate whether the leakage of IV-GBCA from the cortical veins into the CSF space that is observed by whole-brain 3D-real IR images can be detected by MRF in subjects receiving MR scanning for the evaluation of EH.

Materials and Methods

Patients

Fourteen patients (11 men, 3 women) with suspicion of endolymphatic hydrops were included (age range 39–79 years old, median 63 years). All patients received a single dose (0.1 mmol/kg) of intravenous administration of gadobutrol (Gadovist, Bayer Pharm, Osaka, Japan). They were in the range of older than 37 years, and therefore were expected to have GBCA leakage from the cortical veins.¹⁵

All patients had an estimated glomerular filtration rate of >60 mL/min/1.73 m². Patients received only a single dose (0.1 mmol/kg) of intravenously administered gadobutrol (Gadovist). The ethical committee of our institution approved this study with written consent from the patients.

MRI

The images were obtained on a 3T MR scanner (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany) using a 32-channel array head coil at 4 h after IV-GBCA.

Axial-slab 3D-real IR images covering the whole brain were obtained over 10 min. The axial slab was set parallel to the anterior commissure-posterior commissure (AC–PC) line. The slice thickness was 1 mm. Scan parameter details are shown in Table 1.

Table 1.

Pulse sequence parameters

Sequence name	Type	Repetition time (ms)	Echo time (ms)	Inversion time (ms)	Flip angle (°)	Section thickness/gap (mm)	Pixel size (mm)	Number of slices	Echo train length	Field of view (mm)	Matrix size	Number of excitations	Scan time (min)
3D-real IR	SPACE with inversion pulse	15130	549	2700	90/constant 145	1/0	0.5 × 0.5	256	256	165 × 196	324 × 384	1	10

SPACE, sampling perfection with application-optimized contrasts using different flip angle evolutions; 3D-real IR, 3D inversion recovery with phase sensitive reconstruction (real reconstruction).

This sequence utilizes a frequency selective fat suppression pre-pulse, non-slab selective excitation pulse, and non-slab selective inversion pulse.

3D slab is set in an axial orientation.

MR fingerprinting

The pulse sequence for 2D-MRF, based on FISP, uses a spiral readout trajectory, inversion recovery pulse, variable TR, and variable flip angles. The slice thickness for this acquisition was 5 mm and had a square FOV of 300 mm. The matrix size was 256 × 256. The scan time was 41 s with 3000 echoes for one slice.

The scan parameters included an echo time of 2.0 ms and an inversion time of 20.6 ms. The TR ranged from 12.1 to 15.0 ms, and the flip angle ranged from 0° to 74°. Each echo encoded an image by a single spiral readout with a variable-density k-space trajectory (24- to 48-fold undersampling), and the spiral trajectory was rotated by 82.5° for each repetition time to improve the temporal incoherence of the undersampling artifacts.^19,20

The axial slice location for the MRF scan was selected at 25 mm below the vertex of the cranium. The axial slice was set parallel to the 3D-real IR slab. At this parietal position, it has been reported that leakage from the cortical veins is frequently visible.¹⁵

To compensate for the inhomogeneity of the RF-field (${\text{B}}_1^{+}$), an automated field correction was applied. Before the actual MRF scan was started, an RF-field-mapping sequence was triggered.²¹

After acquisition of the MRF data, the scanner automatically performed pattern recognition between the calculated dictionary and the measured time course. The dictionary was calculated for a range of discrete T₁, T₂ and (${\text{B}}_1^{+}$ values by performing a Bloch simulation over a voxel of 200 spins taking the slice profile into account.²² A subset was extracted from the dictionary according to the ${\text{B}}_1^{+}$ value from the separately measured RF-field-mapping sequence for every voxel. The actual voxelwise matching process then took place in this subset in T₁ and T₂ dimensions. The inner products between the normalized measured time course of each pixel and all entries of the normalized dictionary were calculated, and the dictionary entry corresponding to the maximum value of the inner product was taken to represent the closest signal evolution to the one acquired. The T₁ and T₂ values were then derived from this entry to generate the T₁ and T₂ relaxation maps.¹⁷

The range and step size of the calculated dictionary for the MRF utilized in this study is shown below:

• T₁: range 10–4500 ms.

  Step-size: 10 ms (10–100 ms), 20 ms (100–1000 ms), 40 ms (1000–2000 ms) and 100 ms (2050–2950, 3100–4500 ms).

• T₂: range 2–3000 ms.

  Step-size: 2 ms (2–100 ms), 5 ms (100–150 ms), 10 ms (160–300 ms), 50 ms (300–800 ms), 100 ms (800–1600 ms) and 200 ms (1600–3000 ms).

• (${\text{B}}_1^{+}$: range 0.60–1.40.

  Step-size: 0.01.

Image evaluation

In the 3D-real IR image at 25 mm below the convex, an experienced neuroradiologist (S.N.) placed a circular ROI with a diameter of 3 mm in the CSF area containing GBCA leakage as well as in the CSF area without discernable GBCA leakage. The ROIs were set at least 5 mm away from the inner surface of the skull to reduce partial-volume effects. If there were multiple CSF areas with GBCA leakage, the observer placed the ROI in the most uniform high signal CSF area where the brain parenchyma and vessels could be avoided. To select the position of the CSF area without GBCA leakage, the observer placed the ROI in the most uniformly low-signal CSF area where the brain parenchyma and vessels could be avoided. The images were evaluated on a PACS viewer (RapideyeCore, Canon Medical Systems, Tokyo, Japan). The display condition was set as constant with a window width of 80 and a window level of 10. Then the ROIs were copied and pasted onto the MRF maps.

The T₁, T₂ values and the signal intensity values on 3D-real IR, were compared between the ROIs in the CSF area with GBCA leakage and those without GBCA leakage. An example of the ROI location is shown in Fig. 1. We used 5% as a threshold to determine statistical significance. The software R (version 3.3.2, R Foundation for Statistical Computing, Vienna, Austria, https://www.R-project.org/) was used for statistical analyses.

Fig. 1.

Exemplary images of (a) 3D-real inversion recovery (IR), (b) T₁ map, (c) T₂ map. Circles indicate ROIs with gadolinium-based contrast agents (GBCA) leakage (solid arrows) and without GBCA leakage (dotted arrows), respectively. On the T₁ map (d) and T₂ map (e) with narrow window setting, cerebrospinal fluid area with GBCA leakage is slightly discernible (open arrowheads).

A paired Student’s t-test was used for the comparison between the ROIs in the area of GBCA leakage and the non-leakage area in the CSF.

Results

The mean T₁ value of the ROI in the area of GBCA leakage was 2422 ± 261 ms and that in the non-leakage area was 3851 ± 235 ms (P < 0.01). There was no overlap between the T₁ values in the area of GBCA leakage and those in the non-leakage area (Fig. 2a).

Fig. 2.

Box plots for (a) T₁ values (ms) and (b) T₂ values (ms) obtained by MR fingerprinting between high signal (with gadolinium-based contrast agents [GBCA] leakage) and low signal CSF (without GBCA leakage). Box plots for (c) signal intensity values on 3D-real inversion recovery (IR) between high signal (with GBCA leakage) and low signal CSF (without GBCA leakage). There was significant difference between the mean values of the ROI in high signal CSF and that in low signal CSF for T₁ value, T₂ value and signal intensity value of 3D-real IR (P < 0.01). Upper edge of the box indicates the first quartile and lower edge indicates the third quartile. The line in the box shows the median. The whiskers extend up from the top of the box to the largest data element that is ≤1.5 times the interquartile range (IQR) and down from the bottom of the box to the smallest data element that is >1.5 times the IQR. Values outside this range are considered to be outliers and are represented by small circles. CSF, cerebrospinal fluid.

The mean T₂ value in the area of GBCA leakage was 319 ± 90 ms and that in the non-leakage area was 670 ± 166 ms (P < 0.01). There was overlap between the T₂ values in the area of GBCA leakage and those in the non-leakage area (Fig. 2b).

The mean signal intensity value of 3D-real IR in the area of GBCA leakage was 64.4 ± 27.5 and that in the non-leakage area was −46.4 ± 5.3 (P < 0.01). There was no overlap between the signal intensity values of 3D-real IR in the area of GBCA leakage and those in the non-leakage area (Fig. 2c).

Discussion

MRF is a rapid quantification method to assess multiple parameters simultaneously. Multiple kinds of quantitative data can be obtained without pixel misregistration.^16,17,23 The stability and accuracy of MRF has been reported in phantoms²³ and in vivo.²⁴

IV-GBCA leaks into the CSF in older subjects.^25–28 This finding may be a part of the normal aging process; however, it also might be a biomarker of neurodegeneration.^15,26,28 To better understand the importance of GBCA leakage into the CSF from the cortical veins, we need more research. MRF can facilitate studies to further understand this phenomenon due to its brief scan time when we scan the limited number of slices of MRF. With the data from the larger number of subjects, we might be able to correlate the degree of GBCA leakage into the CSF with aging, the status of various diseases, cognitive function and so on. The short acquisition time, quantitative nature and reproducibility allow MR fingerprinting to serve as the valuable tool for the longitudinal studies, especially focused to evaluate the function of glymphatic system.

Heavily T₂-weighted 3D-FLAIR and 3D-real IR images are quite sensitive to low concentrations of GBCA in fluid. However, they are not sensitive to T₁ changes in brain parenchyma.^4–6 The longitudinal magnetization of the brain parenchyma fully recovers relative to CSF after a long inversion time as conventionally used, therefore the presence of GBCA in the brain parenchyma cannot be measured, if these sequences focus on the signal changes inside CSF. In other words, these sequences have a narrow dynamic range to evaluate the differences in T₁. MRF has a wide dynamic range, which is applicable to both brain parenchyma as well as CSF.

This study had some limitations. The number of subjects was small. The evaluation was done with manual placement of small ROIs. We evaluated only post-contrast-enhanced images. We did not evaluate the sensitivity of MRF to very slight CSF enhancement, which can be barely detected by 3D-real IR. Partial-volume effects may affect the T₁ and T₂ values due to the use of a relatively thick slice (5 mm) for MRF. However, we placed the ROI so that it was centered on the area of GBCA leakage into the CSF according to the 3D-real IR image (1 mm). In addition, there could be some CSF flow. While flow effects on MRF have not been clarified, recent studies have indicated that T₂ values are more sensitive than T₁ values to in-plane and especially through-plane motions in 2D FISP–MRF.^29,30 The wider range of T₂ values compared with T₁ values might be caused by motion within the CSF. Therefore, the use of T₁ values to detect GBCA leakage into the CSF rather than T₂ values might be a reasonable strategy. Furthermore, it has been reported, that when using time spatial spin labeling inversion pulse imaging, there is no CSF pulsation or flow in the subarachnoid space over the convexity in the normal brain, nor in the hydrocephalic brain.³¹ Therefore, we believe that the CSF movement in the convexity area is too small to cause a significant error in the T₁ value measurement. The other limitation is that the maximum T₁ values of the MRF dictionary (4500 ms) could underestimate the CSF T₁ values over 4500 ms. From the above, further research is needed to assess whether MRF is useful for CSF; however, we could clearly detect GBCA leakage into the CSF. Our result has the potential to extend the clinical applications for MRF.

Conclusion

Leaked GBCA from the cortical veins into the surrounding CSF can be detected using MRF obtained in <1 min.