Introduction

Filter-exchange imaging (FEXI) is a promising non-invasive MRI technique to measure water exchange processes^{[1, 2]}. The technique has been used to measure vascular water exchange across the blood-brain barrier (BBB) in the human brain^[3], which could be important for improved understanding of neurological conditions such as neuroinflammation and neurodegeneration. The FEXI method has yet to be applied to study BBB exchange in the rodent brain, due to challenging implementation with the markedly smaller brain size, requiring thinner slices and thus higher magnitude crusher gradients^[4]. However, the application of FEXI in rodents would allow controlled studies of disease using transgenic rodent models to probe specific targets that underpin pathology relating to BBB function. Here, we establish a FEXI protocol optimised to measure water exchange across the BBB in the rat brain and provide estimates of apparent exchange rate (AXR).

Methods and Materials

To investigate the effects of crusher gradients and low b-value used to measure BBB exchange, synthetic signals relating to a diffusion encoding experiment were simulated in Matlab R2021a using a two compartment model as described previously^[4], with exchange rate, k = 2.5 s^-1, extravascular volume fraction, f_e = 0.95, intravascular and extravascular diffusivities, D₁ = 8 x 10^-3 mm²/s and D₂ = 0.8 x 10^-3 mm²/s ^[5], and all other acquisition parameters matching the in-vivo imaging protocol below.

Experimental data were acquired in Wistar rats (n = 10) using an Agilent 7T magnet interfaced to a Bruker Avance III console with a rat head receiver coil (Bruker BioSpin) using double diffusion-encoding FEXI sequence, developed in house, with a single shot EPI readout. Non-selective filter pulses with slab thickness = 30mm were implemented in first diffusion-encoding filter module, to minimise in-flow effects, and slice selective pulses were used for the second diffusion-encoding with slice thickness = 4mm. Other imaging parameters: filter b-value (b_f) = 0 s/mm2 and 250 s/mm2; b values = 0, 250 s/mm²; mixing times (t_m) = 25, 50, 100, 200, 300 ms; TE = 14 ms; TR = 3000 ms; single-slice; resolution = 0.5 x 0.5 x 4 mm; repetitions = 10. Total scan time was approximately 35 minutes. Scans were repeated in the same rats 9 ± 2 days later to assess the repeatability of the measurement.

Experimental data were evaluated using Matlab (R2021a), in a whole brain ROI, excluding the ventricular regions. Synthetic and experimental data were fit with the standard apparent exchange rate (AXR) model:
ADC’(t_m) = ADC^eq(1 – σ exp(-t_m AXR)) ^[2],
where ADC’ is the filtered apparent diffusion coefficient, ADC^eq is the non-filtered apparent diffusion coefficient, σ is the filter efficiency, using a non-linear least squares fitting with ADC^eq fixed at ADC^eq(t_m = 25ms). To reduce biases in ADC’ and ADC^eq, and hence AXR, due to crusher gradients, the experimental ADC’ was normalised to the ADC^eq at each t_m prior to fitting a normalised AXR model: ADC’(t_m)/ADC^eq(t_m) = 1 – σ exp(-t_m AXR). Repeatability was quantified using the coefficient of variation (CoV).

Results & Discussion

Figure 1 shows the simulated and experimental results of ADC’(t_m) and estimates of AXR using the standard model. The simulated ADC^eq data (Figure 1A) are biased due to the crusher gradients, leading to an underestimation of the AXR (from 2.5 s-1 to 1.22 s-1) ^[4]. The experimental ADC^eq (Figure 1B) exhibits a similar trend (seen across all subjects), with mean experimental AXR = 0.78 ± 0.59 s^-1 (scan), 0.51 ± 0.43 s^-1 (rescan), n = 10. Figure 2A shows normalised ADC’ map for each tm in a representative subject. Group mean normalised AXR plots are shown in Figure 2B, demonstrating the correspondence between the scan and rescan measurements. The mean normalised AXR values were determined to be 3.34 ± 1.14 s^-1 (scan) and 3.62 ± 0.96 s^-1 (rescan), with Bland-Altman plot of the average vs the difference presented in Figure 2C; CoV = 0.26. Overall, the experimental data show a recovery of the ADC with increasing t_m reflective of extravascular water entering the intravascular compartment. Though the mean AXR values obtained using the standard AXR model are slightly lower than expected, the normalised AXR measurements are in line with the average estimates of BBB water-exchange rate in the rodent brain^[6]. The difference in T1 and T2 relaxation effects between the intravascular an extravascular compartments may be contributing to the underestimation of the AXR, which will be explored in future work^[7]. This is a promising first step for using the non-invasive FEXI technique in the rat brain, and will be used to assess changes to BBB water permeability in disease conditions.

Conclusion

We have developed a FEXI imaging protocol to measure the apparent water exchange rate across the BBB in rat brain. This technique has the potential to be applied to study BBB water-exchange processes in a wide range of disease models, including neurodegeneration, stroke and neuroinflammation.

Acknowledgements

This work is funded by the EPSRC code EP/S031510/1 and the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 804746). SL is supported also by Random Walk Imaging. We would like to thank the staff at The Biological Service Facility at the University of Manchester for their help maintaining animal welfare and environmental enrichment during this study.