Prediction of Anti-cancer Nanotherapy Efficacy by Imaging

Abstract

Anticancer nanotherapeutics have shown mixed results in clinical trials, raising the questions of whether imaging should be used to i) identify patients with a higher likelihood of nanoparticle accumulation, ii) assess nanotherapeutic efficacy before traditional measures show response, and iii) guide adjuvant treatments to enhance therapeutic nanoparticle (TNP) delivery. Here we review the use of a clinically approved MRI nanoparticle (ferumoxytol, FMX) to predict TNP delivery and efficacy. It is becoming increasingly apparent that nanoparticles used for imaging, despite clearly distinct physicochemical properties, often co-localize with TNP in tumors. This evidence offers the possibility of using FMX as a generic "companion diagnostic" nanoparticle for multiple TNP formulations, thus potentially allowing many of the complex regulatory and cost challenges of other approaches to be avoided.

Keywords: dextran-coated iron oxide nanoparticle; enhanced permeability and retention effect; magnetic resonance imaging; nanomedicine; personalized medicine; tumor associated macrophage..

Figure 1

FMX imaging captures heterogeneous EPR effect in xenografts and patients. A) Schematic of solid tumors exhibiting varying degrees of EPR effect. B) Measurements of tumoral FMX accumulation were pooled from previously reported data across multiple xenograft mouse models (including orthotopic breast cancer) and tumor lesions in patients . Values were normalized to the average observed uptake in order to calculate coefficient of variation (CV = standard deviation / mean). Data are median ± interquartile range. C) Using quantitative FMX-MRI, concentrations of FMX in patient tumors were measured, in most cases across multiple metastatic lesions per patient. Patients were ordered according to median value for each tumor type (shown by bars). Figure is modified from .

Figure 2

FMX imaging correlates with vessel permeability. A) Early FMX-MRI of tumors in patients was fit to a compartmental pharmacokinetic model, and correlation was observed between computationally inferred permeability (y-axis) and FMX (x-axis). The thick line and shading denote the fit and confidence interval of a linear regression model. Figure modified from . B) Intravital microscopy in a xenograft mouse model was used to measure dynamic bursts of macromolecular extravasation, shown by rapid transport of dextran from vasculature into tumor tissue, as outlined by the orange dashed line. Scale bar = 20 μm. FMX was pre-injected 24 hr prior, and the timestamps denote time post-injection of the 60 kDa fluorescent dextran. C) Using data as in B, effective vessel permeability was measured for each image dataset and correlated with multiple other image features such as vessel width, cell density, and vessel branching. The enrichment of FMX+ tumor-associated phagocytes correlated most with permeability, suggesting these cells contribute to vessel leakiness. D) Bursts of dextran extravasation were identified by data as in B, and were found to occur more often in vessels with FMX+ phagocytes in close proximity. For B-D, figures are modified from .

Figure 3

Late FMX uptake is driven by extracellular tissue space and TAM uptake. A) FMX-MRI measurements in patients were fit to a compartmental pharmacokinetic model, and correlation was observed between computationally inferred extracellular volume fraction (y-axis) and FMX (x-axis). Thick line and shading denote the fit and confidence interval of a linear regression model. Figure modified from . B) T1-weighted and T2*-weighted FMX-MRI show positive and negative enhancement, respectively, upon FMX accumulation. In this example, T1-weighted images show extracellular FMX accumulation in the necrotic tumor core, while T2*-weighted images show intracellular uptake via phagocytes near the tumor periphery. Figure adapted from . C) 24 hr post-injection with fluorescently-labeled FMX, fibrosarcoma tumors from a xenograft mouse model were excised and stained for hematoxylin and an F4/80 antibody at x10 (top left; scale bar 100 μm) and x40 (top right; scale bar 50 μm). Adjacent sections were imaged by immunofluorescence (bottom; x40, scale bar 50 μm). Reproduced from .

Figure 4

Correlation of FMX with drug uptake in tumors. A) FMX-MRI of a pancreatic cancer patient shows uptake and perfusion of the liver, kidneys, spleen, and subtle changes at the pancreas-tumor interface, with an apparent enrichment in TAM near the margin. Modified from . B-C) FMX accumulation in a mouse model of breast cancer was used to stratify tumors into low, medium, and high FMX cohorts, which correlated more with accumulation of a co-administered fluorescent docetaxel TNP (B), compared to the un-encapsulated solvent formulation (C). Reproduced from .

Figure 5

FMX reports on enhanced TNP delivery following neo-adjuvant treatment. A) Fluorescently labeled versions of FMX and two FDA-approved TNP were co-injected into xenograft tumors that had been locally irradiated to enhance their EPR effects, and 24 hr later tumors were imaged by intravital microscopy (scale bar 20 μm). >90% of phagocytes accumulating one type of NP also accumulated the others. Reproduced from . B) Fluorescently labeled FMX and a model liposome TNP were co-adminstered in tumor-bearing mice, and 24 hr later tumors were imaged for FMX and TNP accumulation. Some tumors were primed with local irradiation to enhance TNP delivery, which was reported by FMX. Each dot represents a tumor, while thick and dashed lines denote the fit and 95% confidence interval of a linear regression .

Figure 6

FMX predicts response to nanomedicine. A-B) FMX-MRI in a xenograft mouse model stratified tumors into low, medium, and high FMX cohorts. Following subsequent treatment with paclitaxel-encapsulated TNP, tumor growth was monitored for several days, revealing that tumor growth was only halted in tumors that accumulated high FMX (A). Tumor cell DNA damage and cell-cycle defects were measured by γH2A.X staining and flow cytometry, also correlating with FMX uptake (B). Reproduced from . C) FMX-MRI measurements, performed across tumor lesions from 9 patients, were correlated with response to subsequently administered liposomal irinotecan TNP. Response was quantitated by changes in lesion size as imaged by CT at least 8 weeks after treatment initiation. Lesions with above-average FMX-MRI enhancement exhibited greater TNP response than lesions with below-average FMX accumulation. Modified from .

Figure 7

Mechanistic studies into how FMX predicts TNP efficacy. A-B) Polymeric micelle TNP were labeled with two fluorophores, such that the polymer vehicle and cytotoxic cisplatin-related payload could be simultaneously measured. A) Histological analysis of payload redistribution from TAM to tumor cells shows F4/80+ cells accumulate high levels of TNP, exhibit a local gradient of payload emanating from them, and tend to be neighboring tumor cells with high activity of the DNA damage response marker 53BP1. Scale bar = 50 μm. B) Using flow cytometry of excised xenograft tumors, cancer cells were observed to have more payload than TAM, relative to the corresponding amount of TNP vehicle in each cell. This data among others suggested payload had transferred from TAM to tumor cells. Reproduced from .