MRI measurements of vessel calibre in tumour xenografts: comparison with vascular corrosion casting

Abstract

Vessel size index (R(v), μm) has been proposed as a quantitative magnetic resonance imaging (MRI) derived imaging biomarker in oncology, for the non-invasive assessment of tumour blood vessel architecture and vascular targeted therapies. Appropriate pre-clinical evaluation of R(v) in animal tumour models will improve the interpretation and guide the introduction of the biomarker into clinical studies. The objective of this study was to compare R(v) measured in vivo with vessel size measurements from high-resolution X-ray computed tomography (μCT) of vascular corrosion casts measured post mortem from the same tumours, with and without vascular targeted therapy. MRI measurements were first acquired from subcutaneous SW1222 colorectal xenografts in mice following treatment with 0 (n=6), 30 (n=6) or 200 mg/kg (n=3) of the vascular disrupting agent ZD6126. The mice were then immediately infused with a low viscosity resin and, following polymerisation and maceration of surrounding tissues, the resulting tumour vascular casts were dissected and subsequently imaged using an optimised μCT imaging approach. Vessel diameters were not measurable by μCT in the 200 mg/kg group as the high dose of ZD6126 precluded delivery of the resin to the tumour vascular bed. The mean R(v) for the three treatment groups was 24, 23 and 23.5 μm respectively; the corresponding μCT measurements from corrosion casts from the 0 and 30 mg/kg cohorts were 25 and 28 μm. The strong association between the in vivo MRI and post mortem μCT values supports the use of R(v) as an imaging biomarker in clinical trials of investigational vascular targeted therapies.

Fig. 1

a) Representative parametric R_v (μm) and fBV (%) MRI maps acquired from SW1222 colorectal xenografts 24 h after treatment with either vehicle, 30 or 200 mg/kg of the vascular disrupting agent ZD6126. b) μCT images of the corresponding vascular corrosion casts obtained from the same tumours from mice treated with vehicle or 30 mg/kg ZD6126. The images show the skeletonization of the CT data where the colour of the vessels is varied linearly with diameter, from blue ≤ 16.8 µm, to red ≥ 270 µm. The vessel thicknesses/diameters are rendered with reduced scale/size to aid visualization. No μCT data is presented from tumours treated with 200 mg/kg ZD6126, as treatment decreased the fBV so dramatically as to inhibit adequate resin perfusion.

Fig. 2

Normalised frequency histogram of R_v (μm) values, determined from SW1222 tumours 24 h after treatment with either vehicle (n = 6), 30 mg/kg (n = 6), or 200 mg/kg (n = 3) ZD6126.

Fig. 3

(a) Variation of mean vessel diameter with voxel size of the tomography data. The error bars show the standard deviations calculated from the vessel diameter distributions, and (b) comparison of the distribution of vessel diameters for a voxel size of 1.2 and 2.4 μm.