Purpose

Arterial spin labelling (ASL) MRI has numerous potential applications in the clinic. In 2015, a white paper with guidelines for standardising ASL-MRI in the clinic was published¹. Similar standardisation, however, is still lacking in pre-clinical MRI, with many laboratories performing ASL on rodents and publishing a range quantitative cerebral blood flow values. Optimised parameters for ASL in rats were previously determined by using a multiphase pseudo-continuous ASL (pCASL) in three strains of naïve rats. Here, our aim was to determine how these optimised pCASL approach could be used to measure perfusion changes occurring in tumours in a rat model of brain metastasis.

Methods

Female Berlin-Druckrey IX (BD-IX) rats were injected in the left striatum with 1000 ENU1564 (N-ethyl-N-nitrosourea-induced rat mammary adenocarcinoma) cells in 1 µL PBS, or with PBS alone as control. MRI experiments were performed on isoflurane anaesthetised rats at weeks 2, 3, and 4 post-tumour cell injection using a 9.4 T MRI spectrometer (Agilent) with a 72 mm volume transmit coil and a 4-channel surface receive array (Rapid Biomedical). Pre- and post-gadolinium T₁-weighted along with T₂-weighted anatomical (FSEMS) multi-slice images were acquired for each animal (FOV=32x32 mm, slice thickness=1 mm, 256x256 matrix). Additionally, T₁ and T₂ maps were acquired, together with multiphase² pCASL data in the same slice pattern (single-shot spin echo EPI, FOV=32x32 mm, 64x64 matrix, slice thickness=1 mm). For pCASL, a 6.2 mm wide tagging plane was placed perpendicular to the carotid arteries with a labelling bolus duration of 1.4 s (TR=4 s, TE=28.7 ms; Figure 1). Post label delays used for quantitative pCASL were either 0.55 s or 0.65 s at later time points in rats injected with ENU cells to account for tortuosity of tumour vessels and possible delayed blood arrival times. Multiphase pCASL used radiofrequency (RF) pulses with eight phase increments between 0˚ and 315˚, in steps of 45˚ (4.4 µT, FA=40˚). pCASL data from eight phases were fit to a modified Fermi function with α=66 and β=21 using a modified version of BASIL³ to get cerebral blood flow (CBF) maps. Immunohistochemistry targeting the endothelial cell marker CD31 (vessels) and pimonidazole (hypoxia) was performed on 10 µm brain cryo-sections of metastasis-bearing rats for tumour characterisation.

Results

Metastatic tumours were barely visible on T₂-weighted images of metastasis-bearing rats 2 weeks post-injection, but were quite evident from week 3 (Figure 2B). Similarly, blood-brain barrier breakdown occurred between weeks 2 and 3 post-tumour cell injection, as demonstrated by hyperintense regions on post-gadolinium T₁-weighted images (Figure 2C). At this time, a significant decrease in CBF (Figures 2A & 3) was evident within the tumour with respect to both PBS injected controls and the contralateral striatum in ENU-injected rats. The tumour core showed the greatest decrease in CBF and histologically was found to be hypoxic in regions distant to blood vessels (Figure 4).

Discussion

The significant decrease in tumour perfusion from week 3 is indicative of dysregulated blood flow within the tumour. The elevated presence of pimonidazole in areas further from CD31 stained vessels within brain metastases provides evidence of hypoxia within these tumours, reflecting the decreased CBF. The low tumour CBF values present in the metastases at a later stage are similar to the findings of Silva and colleagues in a rat primary brain tumour (glioma) model using continuous ASL⁴.

Conclusion

With our optimised pCASL methodology, we obtained reproducible CBF values in tumour cell and PBS injected rats. The metastatic tumours proved to be hypoxic and this correlated with reduced CBF observed using multiphase pCASL. CBF values in normal brain are in line with gold-standard autoradiography measurements of CBF.

Acknowledgements

This work was funded by the CRUK/EPSRC Cancer Imaging Centre (grant number C5255/A16466) in Oxford, Cancer Research UK (grant number C5255/A15935), a Medical Research Council studentship (MC_ST_U13080) and MRC supplementary award (MR/K501256/1). The authors thank James A. Meakin and Peter Jezzard (FMRIB, Oxford) for their pCASL MRI sequence, which was modified for use on the Agilent system.