Tumor Hypoxia Assessment: In Vivo 3D Oxygen Imaging Through Electron Paramagnetic Resonance

Summary

Quantitative 3D oxygen maps of murine tumors were imaged noninvasively using Pulse Electron Paramagnetic Resonance. Ultrasound B-mode and Power Doppler were used for anatomy and vascular structure. Images from both modalities were superimposed enabling multi-parametric tumor analysis.

Abstract

The precise and real-time measurement of oxygen partial pressure (pO₂) brings valuable information in many pathologies, including cancer. Low tumor pO₂ (i.e., hypoxia) is connected to tumor aggressiveness and poor response to therapy. The quantification of tumor pO₂ allows the evaluation of treatment effectiveness. Electron Paramagnetic Resonance Imaging (EPRI), particularly Pulse EPRI, has emerged as an advanced three-dimensional (3D) method of assessing tissue oxygenation in vivo. This innovation was enabled by the technological developments in EPR (Electron Paramagnetic Resonance) and the application of the water-soluble oximetric spin probes from the triaryl family, offering fast and sensitive oxygenation data. The relaxation time of the spin probe (T1 and/or T2) provides accurate information about pO₂ in selected voxels.

Human glioblastoma LN229 tumors were grown in the interscapular fad pad of BALB/c nude mice. Ultrasound (US) imaging was used as a reference for tumor anatomical information. To image tissue pO₂, the animals were placed in a fixed position in the animal bed with fiducials, enabling registration between the imaging modalities. After the OX071 contrast agent was administered, EPRI was performed, followed by US B-mode. Because of the low spin probe toxicity, the procedure can be repeated during tumor growth or treatment. Following imaging, the registration process was carried out using software written in MATLAB. Ultimately, the hypoxic fraction can be calculated for a specific tumor, and the histogram of pO₂ tissue distribution can be compared over time. EPRI combined with ultrasound is an excellent tool for oxygen mapping of tumors in the preclinical setting.

Introduction

Comprehending the tumor microenvironment (TME), with its complex spatial and dynamic interactions, brings a fuller understanding of tumor biology. Hypoxia, or low oxygen levels, is the key component of TME and plays a critical role in the development of other life-threatening conditions, including cardiovascular diseases, metabolic disorders such as diabetes, and chronic kidney disease^1,2,3. Tissue oxygenation is a fundamental factor, particularly in the context of cancer, where partial tissue oxygen pressure (pO₂) is correlated with therapy resistance. A pO₂ level exceeding 10 mm Hg is associated with an increase in the effectiveness of low Linear Energy Transfer (LET) radiotherapy (oxygen enhancement effect).

Recent studies using Electron Paramagnetic Resonance Imaging (EPRI) have demonstrated that oxygen-guided radiation therapy can result in a twofold improvement in survival rates in different cancers in murine models^4,5. This is similar to human subjects whose tumor pO₂ was measured with multiple Eppendorf Electrode measurements and found to have median or mean pO₂ values below 10 torr⁶. Besides radiotherapy, tumor hypoxia has been directly correlated with tumor aggressiveness and the outcome of other therapies, such as immune therapy^7,8. This association underscores the importance of precise oxygen measurements in enhancing therapeutic outcomes and understanding the pathophysiology of diseases.

Optimal in vivo oximetry necessitates a direct measurement of partial tissue oxygen pressure independent of factors such as tissue perfusion and hemoglobin saturation. The procedure should be noninvasive, with a brief and precise imaging time to avoid potential impacts on the organism, such as prolonged anesthesia, alterations in tissue temperature, or significant changes in tissue pressure and pH. Tissue oximetry should exhibit high accuracy and reliability, ensuring consistent measurements regardless of variations in the tissue microenvironment, including differences in pH and redox state. For effective therapy planning, real-time image data reconstruction and straightforward interpretation are crucial. This entails not only achieving spatial resolution preferably less than 1 mm, but also enabling fast data collection to monitor dynamic changes in tissue oxygen status, such as cycling hypoxia.

In this context, various techniques for measuring molecular oxygen or assessing hypoxia have been developed, each with unique applicability and advantages. The platinum electrode, considered the "gold standard" for cellular and live animal tissue oximetry, offers consistent measurements through precise insertion into tissues. Other approaches, such as optical methods using fluorescent probes, photoacoustics, monitoring of the effects of hypoxia through gene or protein expression, or comet assays, are easy to use but are indirect or limited by optical path in tissues. Promising alternatives to assess hypoxia and/or oxygenation appear to be magnetic resonance imaging (MRI)-OE-MRI¹⁰ -- or MOBILE¹¹, positron emission tomography (PET) with various hypoxia-sensitive probes¹², or electron paramagnetic resonance (EPR).

EPR has a long history in the field of biomedicine. The phenomenon itself was first reported in 1944 and was widely adopted as a tool for analyzing chemical structures and more recently, for biological systems and materials with unpaired electrons¹³. EPR spectroscopy has been used to study the dynamics and structure of biological systems such as photosynthesis, metalloproteins, radical enzymes, and phospholipid membranes^14,15,16. Electron Paramagnetic Resonance (EPR) spectroscopy and tomography have emerged as pivotal non-invasive methods for studying tumor oxygenation and microenvironment with a spatial resolution of ~1 mm, temporal resolution of 1-10 min, and pO₂ resolution of 1-3 torr^5,17,18.

Continuous Wave (CW) EPR methods remain widely used in most applications due to the simplicity of recording and interpreting spectra. The oxygen-spin probe interactions work by assessing alterations in EPR signal intensity or line shape, providing insights into oxygen levels within the sample. CW EPR has a notable advantage in sensitivity to a wider range of pO₂ compared to pulse methods. By applying various pulse sequences, information such as electron spin-spin relaxation times, spin-lattice relaxation times, and interactions with neighboring spins can be elucidated^18,19. Pulse EPR techniques, such as inversion recovery with electron spin echo (IRESE) readout, measure spin lattice relaxation rates, avoiding the artifact from relaxation caused by spin probe-spin probe relaxation at low oxygen concentrations^19,20. EPR can be used to monitor oxygen concentration changes with high temporal and spatial resolution; however, in oximetry at high oxygen concentrations, pulse EPR faces limitations due to the short relaxation times of transverse magnetization measured with electron spin echo (ESE). Ultimately, CW and pulse EPR are complementary, and a reliable understanding of the spin system requires the application of both methods.

EPR oximetry techniques rely upon the linear relationship between oxygen levels and the spin-lattice as well as spin-spin relaxation rates in solution. All oximetric probes are often divided into two types: soluble and particulate spin probes. Choosing the correct spin probe depends on the experimental setup and the information needed^21,22,23. Soluble spin probes, such as nitroxides or the trityl derivatives^24,25 such as OX063 and its deuterated form OX071, distributed throughout the tissue, provide information from the whole volume. Alternatively, for single-point measurement, and for prolonged and recurrent oxygen assessments, solid-state probes like LiPc, LiBuO or carbon derivatives may be used (see Table 1)^22,23,26.

Ultrasonography B-mode imaging is widely used in the clinic for soft tissue imaging. The resolution depends on the transducer frequency used, and for preclinical studies, 18 MHz and higher provide sufficient resolution in the plane and the depth of the image. An additional advantage of ultrasonography is the possibility of obtaining functional vasculature images using Power Doppler mode. Here, we present electron paramagnetic resonance oxygen imaging (EPROI) as a method for generating 3D oxygen maps of tumors in living mice. Corresponding ultrasonography enables the necessary anatomical reference for tumor definition within EPROI. Multiple imaging sessions are possible for every animal. The last step is the analysis, including image reconstruction and registration between the modalities to obtain a pO₂ histogram from the tumor volume.

Protocol

Mice were obtained from an approved animal breeding facility and all experiments were conducted in compliance with ethical guidelines (in our case - Permission No. 165/2023, First Local Ethics Committee, Kraków, Poland).

1. Animals and tumor line

NOTE: The mice were housed under standard laboratory conditions: Light/dark: 12 h/12 h, humidity: 60%, temperature: 23 °C. They were provided a standard chow diet with free access to drinking water in community cages.

1. Murine glioblastoma multiforme LN229 cell line culture
   1. Culture LN229 cells in 25 cm² tissue culture flasks in DMEM medium supplemented with 10% heat-inactivated fetal bovine serum (FBS) and penicillin-streptomycin at concentrations of 10 U/mL and 0.1 mg/mL, respectively.
   2. Maintain cells in a humidified atmosphere containing 5% CO₂ at 37 °C, passaging every 48 h using 0.25% trypsin-EDTA and PBS without magnesium and calcium ions (pH 7.4).
2. Tumor inoculation
   1. One week before inoculation, handle mice daily to familiarize them with the investigator.
   2. On the day of inoculation, weigh mice.
   3. Suspend 200,000 LN229 cells in 50 µL of extracellular matrix without growth factors. Using a 29 G needle, inoculate the cell mix subcutaneously into the intrascapular fat pad of 16-week-old BALB/c nude mice (N = 5).

2. Doppler US imaging

The overall timeline of tumor imaging is shown in Figure 1. Ultrasound imaging is used both for vasculature imaging by Doppler US and Anatomy US as a reference just before EPROI (Figure 2). B-mode anatomical imaging is essential for tumor oxygenation analysis by EPR and is described in section 3. While Doppler ultrasound imaging (section 2) is not mandatory for successful registration performance, it nevertheless provides valuable information about the optimal time window for the EPR study and allows for the determination of active vasculature in the tumor area.

1. Mouse preparation
   1. Wait until tumors reach around 30 mm³. If necessary, shave the mice manually around the tumor site.
   2. Induce anesthesia using 2% isoflurane and maintain it with 1-1.5% isoflurane.
   3. Transfer the animal from the anesthesia chamber to a heating pad to maintain body temperature at 37 °C during preparation (based on the rectal temperature probe). Stabilize the animal temperature to obtain proper physiological parameter feedback.
2. Ultrasonography
   1. Use the ultrasound system with a 57-25 MHz transducer for preclinical imaging (Figure 2, step 1).
   2. Acquire 3D B-mode measurements for tumor morphology visualization in a sagittal direction. Use 0.1 mm step size.
      NOTE: Do not move the animal or the transducer to preserve the exact same settings.
   3. Employ Power Doppler mode to visualize functional tumor vasculature with the following parameters: Velocity: 1.5 kHz, Wall filter: Low, Priority: 75%, Persistence: Medium, Step size 0.10 mm.
   4. Rotate the transducer to repeat steps 2.2.2 and 2.2.3 in an axial direction.
   5. Perform data analysis with the software provided by the ultrasound manufacturer. Mark the tumor border, upload it into a 3D image, and calculate the tumor volume and the percentage of vasculature.

3. EPROI

1. Probe preparation
   1. Dissolve OX071 powder stored at -80 °C in injectable H₂O to a final concentration of 1 g/10 mL (~70 mM).
2. Mouse preparation
   1. Anesthetize the mice with 1%-3% isoflurane mixed with room air and position them on an animal holder. 
   2. Administer 1 mL of physiological saline subcutaneously to maintain the proper hydration level.
   3. Monitor the animal's breathing rate (80 ± 20 BPM) and temperature (37 ± 1 °C) using a respiratory pillow sensor and a surface thermometer attached to the mouse skin. Monitor the rectal temperature as a reference point (37 °C ± 1 °C).
   4. Insert polytetrafluoroethylene (PTFE) tubing (0.7 mm external diameter) intraperitoneally for spin probe administration. Secure the cannula with the use of vinyl polysiloxane (VPS) to prevent it from falling out of the abdominal cavity.
   5. Insert a urine catheter (24 G) to collect the excreted spin probe.
   6. Secure the mouse in an animal bed using VPS dental clay for immobilization (Figure 2A).
3. Anatomical US imaging for registration
   1. Turn on the 3D-controlled table and level bed. Set the platform temperature to 60 °C so that the temperature of the animal isolated by the plastic bed holder is maintained at 37 °C.
   2. Place the animal bed with the immobilized animal in the bed holder and transfer it to a 3D-controlled table for anatomical imaging before EPR measurement (Figure 2B). Make sure that the animal is secured within the bed holder and not touching the heating platform.
   3. Fix the position of the bed holder in three dimensions to prevent rotation, particularly XZ (sagittal plane) rotation, which is crucial for accurate registration.
   4. Place a position marker (fishing wire on the tape, 0.35 mm diameter) on the bed holder, marking the beginning of the resonator.
   5. Conduct B-mode imaging manually with a 1 mm step in both axial and sagittal directions towards the Y axis.
   6. Image the tumor structure with the following parameters dependent on the specific transducer frequency; for an 18 MHz transducer, depth 20 mm, dynamic range 84 dB, power 8 dB, gain 80%; for 40 MHz transducer, depth 20 mm, dynamic range 32 dB, gain 100%; for 57 MHz transducer, depth 13 mm, gain 6 dB, power 100%. Adjust the setting on the transducer focus according to the particular tumor location.
   7. At the end of anatomical imaging, wipe off the excess gel from the mouse and remove the immobilized mouse in the animal bed from the bed holder.
4. EPR oxygen imaging
   1. Utilize the oxygen imager for Pulse EPR, operating within radio frequencies ranging from 685 MHz to 735 MHz. For setup, employ offset coils for 3D imaging and analysis of relaxation times. Use a horizontal resonator sized 32 mm x 35 m.
   2. Transfer the immobilized mouse in the animal bed promptly to the EPR imager after anatomical ultrasound (Figure 2C). Maintain the position of the animal bed carefully to minimize rotations within the resonator, similar to the procedure outlined in section 3.3.
   3. Reconnect the temperature probe to ensure continuous monitoring and maintenance of the animal's temperature, correlating it with the temperature inside the resonator.
   4. In the EPR spectrometer software, perform resonator tuning by using the tuning wheel to center the frequency around 725 MHz (Figure 3A).
      NOTE: A well matched resonator should have a dip of 25 dB or better.
   5. Optimize the microwave power to obtain a peak at 60 ns (Figure 3B) by adjusting the attenuation value from 20 dB to 3 dB.
   6. Tune the instrument so that the free induction decay (FID) of fiducials positioned within the animal bed is centered in the magnetic field and phased (Figure 3C).
   7. Administer 100 µL of OXO71 intraperitoneally through the previously inserted cannula, followed by a flush with 50-100 µL of saline solution.
   8. Using a programmed queue sequence, acquire measurements with set parameters; a typical sequence contains T1, T2 relaxation times, 3D Electron Spin Echo (ESE) for fiducial images, and 4D Inversion Recovery Electron Spin Echo (IRESE) for the animal image. Collect the IRESE imaging with a 1.5 G/cm gradient, repetition time of 55 µs, pretriggering -250 ns, t90 is 60 ns, and tau 400 ns; total acquisition time ~12 min. Repeat IRESE imaging at least for 30-40 min (3-4x) after the probe injection.
   9. After the imaging, transfer the mouse from the animal bed to the heating pad. Administer 1 mL of physiological saline subcutaneously to maintain the proper hydration level. Monitor until the mouse recovers upright locomotion.

4. Data analysis

1. 4D reconstruction analysis in "ProcessGUI"
   1. Prior to reconstruction, apply baseline correction on the projections by selecting scenario "PulseRecon.scn" and loading the parameters file "IRESE_64pts_mouse_STANDARD_CHIRALITY.par" to analyze IRESE raw data.
   2. Filter each projection with a Gaussian filter of a width of 4 points, a default setting for selected scenario.
   3. Sub-sample the filtered projections to 64 points (Matrix size), a default setting for selected scenario.
   4. Further filter the projections with a Ram-Lak filter with apodization at 0.6 times the Nyquist frequency (click Reconstruction parameter | FilterCutOff).
   5. Back-project the filtered projections to produce a 4D spectral-spatial image.
   6. Utilize a fitting algorithm to extract the spin packet linewidth from the spectrum in each voxel. Section Fitting Parameters setup: Last point extender - 3, fitting method - default.
   7. Use the default settings for parameters in the fitting procedure, including the amplitude of the spectrum, the phase, spectral center, and the spectral spin packet linewidth.
2. Registration between anatomical US and EPROI in "ArbuzGUI" (Figure 4)
   NOTE: The "ArbuzGUI" MATLAB procedure developed by Boris Epel from the University of Chicago was used for performing the registration. The software is available at EPR-IT https://github.com/o2mdev/eprit. The registration toolbox was explained previously elsewhere²⁷. See the user manual for detailed step-by step instructions²⁸.
   1. Load 2D US images as a stack with 1 mm step (step is strictly related to frame acquisition on B-mode imaging, point 3.3.7) to create 3D US images.
   2. Add collected pO2 images (reconstructed at step 4.1) as 3D image type set as "PO2_pEPRI" data. Store data in the project.
   3. Create a registration sequence and transformation by selecting Special | MRI-EPRI registration.
   4. Select the image that needs to be adjusted into EPRI data (e.g., US axial 3D) by selecting Sequence | add action. Use Stage 5:T2 for the best performance. In the outcome, a black plus symbol will appear in front of the name of the selected image.
   5. Open the US image in SliceMaskEditPLG. Adjust the scale of the US image based on the ruler presented on the images. Mark US position marker, animal outline, and tumor position based on selected frames.
   6. In the SliceMaskEditPLG, mark the pO₂ map outline and fiducial in reconstructed 4D pO₂ images.
   7. Using the figure viewer and RotateImagePLG toolbox, rotate the US image according to the US position marker versus fiducial positions to register pO₂ map with US outline.
   8. Transform the tumor mask from the US image to the pO₂ map.
   9. Visualize the tumor mask in the pO₂ map of the mouse and export pO₂ values for each voxel.

Results

A representative cross section from the ultrasound image of an LN229 tumor growing in the intrascapular fat pad, together with vasculature is shown in Figure 5. Some vasculature is seen outside the tumor border. Unexpectedly, the percentage of tumor vasculature volume did not decrease and remained stable with tumor growth.

As outlined in Figure 2, step 2 involves immobilization of the mouse on the bed with all the accessories to ensur...

Discussion

There are a few critical steps in the described imaging protocol. First, to register the anatomy images with the oxygen maps, MRI might be a better choice than ultrasound due to better resolution and the ability to provide detailed 3D data¹⁹. Ultrasound with a high-frequency transducer provides excellent resolution and sufficient imaging depth for preclinical studies. Both MRI and US, however, require the use of fiducials to secure the tumor position across different imaging modalities.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
aqua pro injectione	Polpharma	1280610	-
ArbuzGUI	O2M Technologies	-	accesible in the github repository
disodium phosphate	POCH S.A.	799280115	-
Dulbecco′s Modified Eagle′s Medium - high glucose	Merck Life Science	D5648	4500 mg/L glucose and L-glutamine
fetal bovine serum	Gibco, Thermo Fisher Scientific	10500064	-
fishing wire	Good Fish	A-55A-035	US position marker - 0.35 mm
Geltrex	Gibco, Thermo Fisher Scientific	A1413302	reduced growth factor basement membrane matrix
ibGUI	O2M Technologies	-	accesible in the github repository
injectio natrii chlorati isotonica	Polpharma	multipe items were used	9 mg/mL
insulin needles 29 G	Becton, Dickinson and Company	multipe items were used	-
Jiva 25	O2M Technologies	-	EPROI
MATLAB	MathWorks	-	version R2021b
penicillin-streptomycin	Merck Life Science	P4333	with 10,000 units penicillin and 10 mg streptomycin/mL
potassium chloride	POCH S.A.	739740114	-
potassium dihydrogen phosphate	POCH S.A.	742020112	-
ProcessGUI	O2M Technologies	-	accesible in the github repository
PTFE tubing	Cole Palmer Instrument Co	06412-11	-
sodium chloride	POCH S.A.	794121116	-
SpecMan4EPR	FEMI Instruments	-	version 3.4 CS 64bit
Surflash I.V. Catheter	Terumo	SR*FF2419	size: 24G x ¾"
tape	3M	multipe items were used	micropore
Trypsin-EDTA	Gibco, Thermo Fisher Scientific	25200072	-
Ultrasonography	Telemed	-	Anatomical US
US gel	KONIX	NUG-0019	-
Vetflurane	Virbac	137317	1000 mg/g
Vevo F2	FujiFilms, Visual Sonics	-	B-mode and Doppler
vinyl polysiloxane dental clay	3M ESPE	multiple items were used	-