Synopsis

Keywords: Molecular Imaging, Cancer

Here we performed a molecular MR imaging study on patient-derived tumor xenografts (PDX) mouse model. In vivo molecular MR imaging demonstrated that proliferative vs. non-proliferative regions exist in PDX model and highly specific contrast agent such as PEGMnCap (pH-responsive nanoparticle based on Mn) can delineate active and hypoxic regions while small-molecular size contrast agent such as Gd-DOTA (gadoterate meglumine) can delineate non-active tumor regions. These MR image separations were not clearly obtained with a cancer cell line, suggesting more complex microenvironment of PDX model and may provide valid translational feedbacks for clinical practice.

Introduction

There has been increasing interests in the development and characterization of patient-derived tumor xenograft (PDX) models for cancer research. PDX models mostly retain the basic histological and genetic characteristics of their donor cancer patients¹. These models have been shown to predict clinical outcomes and are used for preclinical drug screening, biomarker identification, and personalized medicine strategies². To our knowledge, no preclinical MR imaging study on PDX models has been reported in the literature, although these models may provide direct translational imaging evidence that may not be possible with cultured cells³. In this study, we obtained a commercially available PDX colon cancer mouse model and performed a molecular MR imaging study using MR contrast agents that have been developed in our institute for cancer research^4,5.

Methods

The PDX mouse model was purchased from the Central Institute for Experimental Animals, Kawasaki, Japan. The ID was COL-01-JCK that was established from male colon cancer patients without any drug treatment history. The original PDX mouse model was expanded to other athymic nude mice (BALB/c-nu/nu; SLC, Hamamatsu, Japan) using a tissue transplantation kit (Ez-Plant, Kyushu Organ Needle, Kumamoto, Japan) and a total of 10 PDX mouse models were used. MRI experiments were performed 4-8 weeks after transplantation once the tumor size was over 10-mm in diameter (Fig 1A). Mice were anesthetized with Isoflurane and MRI scans were acquired using either 7-tesla Biospec 70/40 or 70/20 with a cryogenically cooled RF coil (Bruker Biospin, Ettlingen, Germany). T1-weighted images were acquired with 3D-MPRAGE sequence using the following parameters: TR, 4000-ms; TE, 2.2-ms; inversion time, 1050-ms; resolution, 0.05-mm isotropic; NEX, 10; and scanning time, 30-minutes. T1-weighted scans were repeated before and after administration of contrast agents through the mouse tail vein. Gd-DOTA (gadoterate meglumine) and Gadolisome⁵ were purchased and PEGMnCap (Mn²⁺-doped calcium phosphate nanoparticles with a PEG shell) were prepared according to the previous paper⁴. Colon 26 mouse model was also prepared according to the previous paper for model comparison. Histopathological analysis such as hematoxylin and eosin (H&E) and Azan staining as well as immuno-stainings (Pimonidazole, Ki-67, and TUNEL) were performed from the same animals after MR imaging.

Results

To characterize the tissue microenvironment of the PDX model, we adopted PEGMnCap that is pH-responsive MR contrast agent (Fig 1C). T1-weighted MRI before and after PEGMnCap administration (0.225 mmol/kg based on Mn) revealed that peripheral regions were selectively enhanced (Fig 1D) and histopathological analysis revealed that peripheral regions were indeed hypoxic and enriched with proliferation positive tumor cells (Fig 1E). TUNEL positive regions were not enhanced with PEGMnCap. To compare with conventional contrast agent, we adopted Gd-DOTA, where molecular size is around 1-nm, which were expected to leak from the vascular wall into the tumor. T1-weighted MRI before and after Gd-DOTA administration (0.4 mmol/kg based on Gd) revealed that central regions were selectively enhanced after 60-minutes of injection. Histopathological analysis revealed that central regions were enriched with TUNEL positive and without proliferation. Angiography imaging after Gadolisome administration (0.023 mmol/kg base on Gd) confirmed that there was no leakage from vessel wall (Fig 2C) as the average molecular size of Gadolisome was around 110-nm. When compared with classical Colon 26 model no such region-specific enhancement was observed (Fig 2D). High magnification image of histopathological analysis revealed that tumor vasculature was surrounded with collagen fiver, and TUNEL positive regions were found in the distal area from the vasculature (Fig 3A). Diffusion-based parametric image such as apparent diffusion coefficient (ADC) revealed that Gd-DOTA positive areas were not entirely overlapped with ADC image (Fig 3B), suggesting that Gd-DOTA positive regions were not fully accounted for water diffusion.

Discussion

In vivo molecular MR imaging demonstrated that proliferative vs. non-proliferative regions exist in PDX model and highly specific contrast agent such as PEGMnCap can delineate active and hypoxic regions while small-molecular size contrast agent such as Gd-DOTA can delineate non-active tumor regions. These MR image separations were not clearly obtained with a cancer cell line, suggesting more complex microenvironment of PDX model and may provide valid translational feedbacks for clinical practice. Further preclinical MR studies such as extensions to other types of PDX model, adaptations of other types of contrast agent, and investigations on drug treatment response were expected.

Acknowledgements

This work was supported by JSPS KAKENHI Grant Numbers JP21H04966, JP21H04966, JP21KK0201 and MRI devices were partly supported by JST/MEXT project for promoting public utilization of advanced research infrastructure, Grant Number JPMXS0450400422.