Bimodal magnetic resonance and optical imaging of extracellular matrix remodelling by orthotopic ovarian tumours

Abstract

Background: The extracellular matrix modulates the development of ovarian tumours. Currently, evaluation of the extracellular matrix in the ovary is limited to histological methods. Both magnetic resonance imaging (MRI) and two-photon microscopy (2PM) enable dynamic visualisation and quantification of fibrosis by endogenous contrast mechanisms: magnetisation transfer (MT) MRI and second-harmonic generation (SHG) 2PM, respectively.

Methods: Here, we applied the MT-MRI protocol for longitudinal imaging of the stroma in orthotopic human ovarian cancer ES-2 xenograft model in CD1 athymic nude mice, and for orthotopically implanted ovarian PDX using a MR-compatible imaging window chamber implanted into NSG mice.

Results: We observed differences between ECM deposition in ovarian and skin lesions, and heterogeneous collagen distribution in ES-2 lesions. An MR-compatible imaging window chamber enabled visual matching between T2 MRI maps of orthotopically implanted PDX grafts and anatomical images of their microenvironment acquired with a stereomicroscope and SHG-2PM intravital microscopy of the collagen. Bimodal MRI/2PM imaging allowed us to quantify the fibrosis within the same compartments, and demonstrated the consistent results across the modalities.

Conclusions: This work demonstrates a novel approach for measuring the stromal biomarkers in orthotopic ovarian tumours in mice, on both macroscopic and microscopic levels.

Fig. 1. Longitudinal MT-MRI of collagen remodelling in ES-2 xenograft model.

a T2-weighted MR images with MTR calculated within ROIs representing skin lesion (SL) and orthotopic ovarian lesion (OL). b Histogram of MTR values for skin lesion and orthotopic ovarian tumour normalised to the ROI size. c Mean MTR value calculated for each compartment and timepoint. Mixed model was fit to check for the influence of time on MTR, and the difference between the groups across all timepoints, both found significant with p < 0.01 (**) and p < 0.001 (***), respectively.

Fig. 2. Tumour microenvironment heterogeneity underlying the variations in the MTR ratio.

a MTR maps of the tumour area on days 6 and 13 after xenografting ES-2 cells (MRI in vivo). b Corresponding view of the same tumour within the anatomical context of the mouse abdomen post-mortem (stereomicroscopy ex vivo). Ovarian tumour with fluid-filled cyst encompassing the ovary, and part of the oviduct and uterine horn. c ES-2 cells imaged within de novo-forming collagen network in the mouse ovary (arrows) imaged throughout the imaging window with two-photon microscope 3 days post inoculation (Imaris rendering, MIP with 3D shading). The de novo-forming collagen fibres extend from the normal ovarian collagen. d The freshly excised tumour with reticuline network of collagen fibres, surrounding clusters of tumour cells. Cyan—tumour cells (eGFP), yellow—collagen (SHG).

Fig. 3. Histopathology of tumour stroma in ES-2 xenograft model 16 days post injection.

HE and Sirius Red staining of the stroma.

Fig. 4. Visual matching between images acquired with MRI, stereomicroscope, two-photon microscope and the corresponding histological sections stained with Sirius Red.

Optical access to the tissue is indispensable for finding the ROI for further quantification. The collagen distribution visible on in vivo two-photon images and on histological sections clearly represents the fibrotic character of the tumour lesion. FP fat pad, OV ovary, T tumour. Yellow—collagen (SHG).

Fig. 5. Quantification of MTR in PDOX and ovarian fat pad.

a ROIs for calculation of MTR were chosen based on anatomical cues derived from optical imaging. The MTR was calculated pixel-by-pixel [MTR = (MT_off–Mt_on)/MT_off], measured in percentage units. The colour bars were scaled to depict values in 0–50%. b Histogram of MTR values for tumour and fat pad, normalised to the ROI size, and calculation of the mean MTR value for both types of ROIs (n = 4 mice, respectively, two-tailed, unpaired Student's t test, p < 0.008).

Fig. 6. Two-photon imaging of PDOX and the native fat pad.

a Orthogonal views of the collagen matrix inside the PDOX lesion and native fat pad. b Quantification of volume and capsule thickness based on the binary mask in PDOX and fat pad (n = 4 mice, respectively, two-tailed unpaired Student's t test; p < 0.008, p < 6.5E⁻⁵, respectively).