Combined MRI and optical CT imaging of tumour vasculature in a preclinical model of neuroblastoma
Ciara M McErlean1, Yann Jamin1, Jessica KR Boult1, Alexander Koers1, Laura S Danielson1, David J Collins1, Martin O Leach1, Simon P Robinson1, and Simon J Doran1

1Institute of Cancer Research, London, United Kingdom


This study compares MRI functional measurements of the vasculature in a preclinical model of neuroblastoma with ex vivo optical CT high-resolution 3D imaging of the functional vasculature using India ink staining. MRI showed a heterogeneously perfused tumour with high fractional blood volume and vessel size index, characteristic of hypervascular neuroblastoma. The high resolution optical CT images allowed visualisation of individual vessels and corroborated the MRI findings. With improved registration, optical CT could help validate MRI functional biomarkers of the vasculature and accelerate both our understanding of vessel biology and the evaluation of vascular-targeted treatment in cancer and other vascular-related pathologies.


Tumour heterogeneity is a major obstacle to the delivery of precision anti-cancer medicine. Tumour vasculature itself presents with a high degree of heterogeneity both in its architecture and function, which affects drug delivery and response to vascular targeted therapies. Functional MRI measurement of tumour vasculature, such as fractional blood volume (fBV) and vessel size index (Rv, a weighted average measure of tumour blood vessel calibre), can non-invasively inform on drug delivery and provide predictive/prognostic biomarkers of response to treatment.

Before being deployed in clinical trials, these biomarkers need to be validated pre-clinically using histopathological correlates. However, conventional histopathology only affords a 2D representation of a 3D tumour structure. Optical computed tomography (CT) is a new ex vivo technique, which offers high resolution 3D imaging of the functional vasculature using the easily accessible stain, India ink.

This study compares optical CT and MR measurements of tumour vasculature using a preclinical allograft model of neuroblastoma, a childhood cancer of the sympathetic nervous system that presents with a characteristic hypervascular phenotype.


MRI was performed on a 129X1/SvJ mouse bearing bilateral subcutaneous allograft tumours derived from dissociated tumour cells taken from an abdominal tumour of a Th-MYCN mouse, the most established genetically modified mouse model of neuroblastoma.1

MRI was carried out on a 7T horizontal bore Bruker MicroImaging system using a 3cm birdcage coil. Tumours were localised on T2-weighted TurboRARE images (3×3cm field-of-view (FOV) and matrix size of 128×128) and field homogeneity was optimised on one of the bilateral tumour volumes using FASTMAP. Fractional blood volume (fBV, %) and vessel size index (Rv, µm) were measured as previously described2, using USPIO particles (P904, 150μmolFe/kg, Guerbet). Functional images were acquired from three contiguous 1mm slices through the largest cross-section of the tumour over a 3×3cm FOV and matrix size of 64×64.

Twenty four hours after MRI, to allow clearance of USPIO particles, India ink (15% in saline) was injected intravenously via the tail vein and allowed to circulate for 5 minutes before the mouse was sacrificed. The tumour was excised, fixed in 70% ethanol in PBS overnight at 4˚C, embedded in 0.75% agarose, dehydrated and optically cleared in a graded series of ethanol and 2:1 benzyl alcohol:benzyl benzoate (BABB) solution to reduce scattering attenuation in the tissue. India ink particles trapped in the vasculature absorb visible light even after clearing, giving optical contrast. Optical CT imaging was performed using a previously characterised in-house scanner.3 The optical CT image data were reconstructed to a 1024×1024×830 matrix with isotropic voxel size of (11.3 µm)3.


On T2-weighted and gradient echo (GE) images, the tumour appeared heterogeneous. Hypointense regions are consistent with the presence of the haemorrhage and large areas of aggregated erythrocytes (blood lakes) characteristic of this model (arrows, Fig.1).4 Calculated parametric maps of Rv (Fig.2a, median 74.1µm) and fBV (Fig.2b, median 22.8%) also demonstrate tumour regional heterogeneity and are consistent with a hypervascular tumour.

The optical CT data corroborate the MR findings, with haemorrhagic regions appearing bright and diffuse due to the presence of erythrocytes. India ink has been shown to remain within perfused blood vessels post mortem and thus the bright spots of Figs2c&d represent cross-sections within the imaging plane of individual perfused blood vessels. The optical CT images suggest a high degree of perfusion heterogeneity within the tumour, with some regions containing a high density of large vessels. Fig.3a shows the full high-resolution optical CT dataset from a different orientation, allowing visualisation of the entire perfused tumour vascular network and characterisation of the tumour heterogeneity in 3D. Fig.3b shows a magnified view of a large vessel with diameter 140±11µm.


This study demonstrates the potential of ex vivo optical CT for both complementing and validating in vivo MRI quantitative measurements of tumour vasculature development and response to therapy. India ink is inexpensive and is easily delivered to the capillary bed, unlike the resins used in vascular casting methods.5 Technical development of methodology for acquiring accurately co-registered MR and optical CT data, incorporating fiducial markers, is ongoing and promises to allow quantitative comparison between functional MR maps and measures of vessel calibre derived from optical CT, despite tissue shrinkage and deformation during sample preparation.


This feasibility study demonstrates the potential of optical CT for complementing MRI to study the role of vascular architecture and function in tumour biology, and for validating non-invasive predictive/prognostic MRI biomarkers of tumour response.


We acknowledge support from The Institute of Cancer Research Cancer Research UK and EPSRC Cancer Imaging Centre, in association with the MRC and Department of Health (England) grant C1060/A10334, NHS funding to the NIHR Biomedical Research Centre and a Paul O’Gorman Postdoctoral Fellowship funded by Children with Cancer UK.


1. Weiss, William A., et al. "Targeted expression of MYCN causes neuroblastoma in transgenic mice." The EMBO journal 16.11 (1997): 2985-2995.

2. Walker-Samuel, Simon, et al. "Non-invasive in vivo imaging of vessel calibre in orthotopic prostate tumour xenografts." International journal of cancer 130.6 (2012): 1284-1293.

3. Doran, Simon J., et al. "Establishing the suitability of quantitative optical CT microscopy of PRESAGE® radiochromic dosimeters for the verification of synchrotron microbeam therapy." Physics in medicine and biology 58.18 (2013): 6279.

4. Jamin, Yann, et al. "Intrinsic susceptibility MRI identifies tumors with ALK-F1174L mutation in genetically-engineered murine models of high-risk neuroblastoma." PloS one 9.3 (2014): e92886.

5. Burrell, Jake S., et al. "MRI measurements of vessel calibre in tumour xenografts: Comparison with vascular corrosion casting." Microvascular research 84.3 (2012): 323-329.


a. T2-weighted MRI of central tumour slice. Gradient echo (GE) images (TE = 9.5ms) b. prior to and c. post-injection of P904. Haemorrhagic and highly perfused regions marked (white and red arrows, respectively).

a. MR paramagnetic maps of vessel size index and b. fractional blood volume. c. optical CT image from a similar position (slice thickness 0.5mm), d. magnified view of the marked region (thickness 11.6µm). Haemorrhagic and highly perfused regions marked (white and red arrows, respectively).

a. Maximum intensity projection (MIP) from a different orientation of the optical CT dataset showing the heterogeneous distribution of blood vessels, stained with India ink, in the tumour. b. MIP of the marked region of the full image volume containing a branched vessel, diameter 140±11µm at the marked red line.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)