Sensitivity of Multiphase Pseudocontinuous Arterial Spin Labelling (MP pCASL) Magnetic Resonance Imaging for Measuring Brain and Tumour Blood Flow in Mice

Abstract

Brain and tumour blood flow can be measured noninvasively using arterial spin labelling (ASL) magnetic resonance imaging (MRI), but reliable quantification in mouse models remains difficult. Pseudocontinuous ASL (pCASL) is recommended as the clinical standard for ASL and can be improved using multiphase labelling (MP pCASL). The aim of this study was to optimise and validate MP pCASL MRI for cerebral blood flow (CBF) measurement in mice and to assess its sensitivity to tumour perfusion. Following optimization of the MP pCASL sequence, CBF data were compared with gold-standard autoradiography, showing close agreement. Subsequently, MP pCASL data were acquired at weekly intervals in models of primary and secondary brain tumours, and tumour microvessel density was determined histologically. MP pCASL measurements in a secondary brain tumour model revealed a significant reduction in blood flow at day 35 after induction, despite a higher density of blood vessels. Tumour core regions also showed reduced blood flow compared with the tumour rim. Similarly, significant reductions in CBF were found in a model of glioma 28 days after tumour induction, together with an increased density of blood vessels. These findings indicate that MP pCASL MRI provides accurate and robust measurements of cerebral blood flow in naïve mice and is sensitive to changes in tumour perfusion.

Figure 1

Optimisation of EPI readout using respiratory triggering to acquire the imaging readout during the plateau phase of respiration, reducing signal fluctuation at the base of the brain. Black trace = raw respiration trace; red trace = digitised respiration trace which acts as the trigger signal; and blue trace = readout of the post-trigger delay period, followed by the imaging slice acquisition. A variable encoding the post-trigger delay was implemented in the sequence to allow adjustment for changing respiration rates.

Figure 2

Optimisation of MP pCASL bolus duration. (a) A−10° labelling plane (blue) relative to the axial imaging plane (yellow) was optimal, as this was perpendicular to the carotid and vertebral arteries visible by angiography (red). Label plane placement posterior to the medulla oblongata, visible on a sagittal T ₂ weighted anatomical image (green), enabled consistent label plane location between animals, while not interfering with imaging slices (yellow). (b) No significant differences were found in calculated CBF across the range of label duration times tested (n=3).

Figure 3

Optimisation of postlabel delay (a–b). Multi-PLD scans were acquired to measure arterial transit time at the front and back of the brain. Example maps of bolus arrival time in (a) anterior and (b) posterior slices. (c) Arrival maps showed bolus arrival in over 95% of voxels in the anterior slice and 99% of voxels in the posterior slice within 0.4 s (n=4).

Figure 4

Cerebral blood flow maps and regional values. (a) Representative CBF maps produced using gold-standard autoradiography (top row), and the optimised MP pCASL sequence (bottom row). (b) Regional CBF values as measured by MP pCASL and autoradiography (AR). Average CBF values were 96 ± 18 mL/100 g/min across the whole brain using MP pCASL, and 101 ± 32 mL/100 g/min using autoradiography. No significant differences were found between MP pCASL and autoradiography CBF measurements either for whole brain, or by region. CC–corpus callosum.

Figure 5

Application of the MP pCASL sequence in mice with intracerebral 4T1-GFP metastatic tumours showing gadolinium enhancement. (a) T ₁-weighted image showing enhancement (dotted outline) of metastatic foci. (b) CBF maps with the same ROI indicated. (c) Graph showing CBF values in tumour and contralateral regions; no significant decrease in perfusion is evident (d-e). Histological comparison of microvessel area fraction in tumour (d), and contralateral (e) regions. Scale bars = 20 µm. (f) Graph showing quantitation of microvessel area fraction in tumour and contralateral striatum regions; a significant increase in microvessel area fraction is evident in tumour compared with normal tissue in the contralateral striatum (n=7). ^∗∗∗ p < 0.01.

Figure 6

Application of the MP pCASL sequence in mice with intracerebral 4T1-GFP metastatic tumours showing a gadolinium enhancing rim and a nonenhancing core region. (a) T ₁-weighted image showing core (nongadolinium-enhancing central regions; dotted outline) and rim (gadolinium-enhancing regions; solid outline) of metastatic foci. (b) CBF maps with the same ROIs indicated. (c) Graph showing CBF values in core, rim, and contralateral regions; a significant decrease (n=6) in perfusion is evident between the core and rim regions of tumours (d-e). Histological comparison of microvessel area fraction in core (d) and rim (e) regions. Scale bars = 20 µm. (f) Graph showing quantitation of microvessel area fraction in core, rim, and contralateral striatum regions; a significant increase in vessel area fraction is evident in core regions compared with both the rim regions (n=6) and normal tissue in the contralateral striatum. ^∗ p < 0.05; ^∗∗∗ p < 0.0001.

Figure 7

Application of the MP pCASL sequence in mice with intracerebral U87 glioma at the day 28 time point. (a) T ₁-weighted image showing no evidence of gadolinium enhancement in the tumour injected (left) hemisphere. (b) Immunohistochemical image showing an overlay of tumour area (red) combined from tissue sections spanning 500 µm corresponding to the MRI slice in (a). (c) CBF maps showing reduction in local CBF (arrowheads) (d-e). Immunohistochemical image showing CD31 staining of vessels (brown) in tumour-injected striatum (d) and contralateral striatum (e). (f) Graph showing CBF values in the tumour-bearing (black) and contralateral (white) striatum. CBF is reduced in the tumour-bearing compared with the contralateral striatum at day 28 time point (p=0.02). Scale bars = 20 µm.