. 2025;21(10):1648-1656.

doi: 10.1038/s41567-025-02993-8. Epub 2025 Aug 19.

Image-guided treatment of mouse tumours with radioactive ion beams

Daria Boscolo^#¹, Giulio Lovatti^#², Olga Sokol^#¹, Tamara Vitacchio¹, Martina Moglioni¹, Francesco Evangelista², Emma Haettner¹, Walter Tinganelli¹, Christian Graeff^{1

3}, Uli Weber^{1

4}, Christoph Schuy¹, Munetaka Nitta², Daria Kostyleva¹, Sivaji Purushothaman¹, Peter G Thirolf², Andreas Bückner¹, Jonathan Bortfeldt², Christoph Scheidenberger^{1

5

6}, Katia Parodi², Marco Durante^{1

7

8}

Affiliations

¹ GSI Helmholtzzentrum für Schwerionenforschung, Darmstadt, Germany.
² Department of Medical Physics, Ludwig-Maximilians-Universität München, Munich, Germany.
³ Department of Electrical Engineering and Information Technology, Technische Universität Darmstadt, Darmstadt, Germany.
⁴ Life Science Engineering Faculty, Technische Hochschule Mittelhessen, Gießen, Germany.
⁵ Institute of Physics, Justus-Liebig-Universität Gießen, Gießen, Germany.
⁶ Helmholtz Research Academy Hesse for FAIR, GSI Helmholtz Center for Heavy Ion Research, Campus Giessen, Giessen, Germany.
⁷ Institute of Condensed Matter Physics, Technische Universität Darmstadt, Darmstadt, Germany.
⁸ Department of Physics 'Ettore Pancini', University Federico II, Naples, Italy.

^# Contributed equally.

PMID: 41098546
PMCID: PMC12518140
DOI: 10.1038/s41567-025-02993-8

Image-guided treatment of mouse tumours with radioactive ion beams

Daria Boscolo et al. Nat Phys. 2025.

. 2025;21(10):1648-1656.

doi: 10.1038/s41567-025-02993-8. Epub 2025 Aug 19.

Authors

Affiliations

¹ GSI Helmholtzzentrum für Schwerionenforschung, Darmstadt, Germany.
² Department of Medical Physics, Ludwig-Maximilians-Universität München, Munich, Germany.
³ Department of Electrical Engineering and Information Technology, Technische Universität Darmstadt, Darmstadt, Germany.
⁴ Life Science Engineering Faculty, Technische Hochschule Mittelhessen, Gießen, Germany.
⁵ Institute of Physics, Justus-Liebig-Universität Gießen, Gießen, Germany.
⁶ Helmholtz Research Academy Hesse for FAIR, GSI Helmholtz Center for Heavy Ion Research, Campus Giessen, Giessen, Germany.
⁷ Institute of Condensed Matter Physics, Technische Universität Darmstadt, Darmstadt, Germany.
⁸ Department of Physics 'Ettore Pancini', University Federico II, Naples, Italy.

^# Contributed equally.

PMID: 41098546
PMCID: PMC12518140
DOI: 10.1038/s41567-025-02993-8

Abstract

Charged particle therapy with protons or heavier ions is one of the most effective radiotherapy techniques, but uncertainties in the beam range can limit its efficacy. Radioactive ion beams are ideal for image-guided particle therapy because isotopes that undergo β⁺ decay can be visualized with positron emission tomography. This allows spatial localization of the particle distribution in vivo, which can be correlated with the expected dose deposition for online beam range verification. Here we report the successful treatment of a mouse osteosarcoma using a radioactive ¹¹C-ion beam. The tumour was located in the neck, close to the spinal cord, where deviations of even a few millimetres in the beam range could lead to unintended dose deposition in the spine and radiation-induced myelopathy, an injury to the spinal cord. We achieved complete tumour control with the highest dose of 20 Gy while avoiding paralysis. Low-grade neurological side effects were correlated to the activity measured by positron emission tomography in the spine. The biological washout of the activity from the tumour volume was dependent on the dose, indicating a potential component of vascular damage at high doses. This experiment marks a step towards future clinical applications of radioactive ion beams.

Keywords: Applied physics; Biological physics; Computational biophysics; Experimental nuclear physics; Imaging techniques.

PubMed Disclaimer

Conflict of interest statement

Competing interestsThe authors declare no competing interests.

Figures

**Fig. 1. Mouse model and μCT.**
a, LM8 osteosarcoma as visible by eye or at the μCT at different times after cell inoculation. Green lines depict the GTV contours of the tumours. For the irradiation, a 2-week timepoint was chosen. b, Two slices of the CT depicting the contours of the individual tumour GTVs (green) near the OARs (spine and trachea with lungs, marked with yellow and blue, respectively). The generalized CTV contour (purple) was applied to all animals, covering all the possible GTV locations previously identified in the tumour induction study. Not all the GTVs are depicted here, as some of them were located on the different neighbouring CT slices. c, The CTV obtained from the contours of individual GTVs of tumours that grew in the animals used to establish and confirm the tumour model. To account for further biological variation, the resulting contour was smoothened and made symmetrical with respect to the spine. The CTV is depicted in purple; the mouse skeleton is shown in light yellow; the trachea and the lungs are shown in blue. The light-grey colour depicts the contours of the mouse body.

**Fig. 2. Experimental beamline.**
a, Drawing of the different elements along the experimental beamline. The mice were irradiated in a vertical position inside the SIRMIO PET scanner while a series of passive components shaped the beam to match the desired irradiation volume. In particular, calibrated large-plate ionization chambers were used as beam monitors for the pristine ¹¹C beam. A 2D range modulator shaped the beam energy, creating a 1.2-cm SOBP in water. A range shifter and aluminium degraders were then used to adjust the beam range to approximately match the tumour position. Two brass collimators were used to reduce the lateral irradiation field and block parts of the beam that did not contribute to the target dose. Finally, a plastic mouse collar acting as a compensator was fixed to the mouse bed. It was designed to partially absorb the beam outside the CTV and shape the distal edge of the SOBP to match the target contour. b, SIRMIO animal holder with the anaesthesia tubes and a mouse in position. c, Animal holder aligned in the beamline while the SIRMIO PET is raised. d, The SIRMIO PET scanner is then lowered to surround the animal. e, Lateral view of the full beamline. Panel a created with BioRender.com.

**Fig. 3. PET imaging in mouse.**
a, FLUKA simulation showing the expected ¹¹C-ion dose (in Gy) distribution in the µCT of the mouse in the sagittal view. Doses are normalized to the planned target dose. b, Corresponding Monte Carlo simulation of the PET activity. c, Online SIRMIO PET image of the positron activity distribution deposited during ¹¹C irradiation overlaid on the same pre-treatment µCT used for the simulations. All 2D distributions are overlaid on the same sagittal µCT slice of the same animal, which is shown in the background. The generalized CTV contour (Fig. 1) is highlighted with a black line, while the spine (OAR) contour is marked in red. All the images are integrated on the BEV aperture (±1 mm) in the x (axial) plane transversing the beam direction.

**Fig. 4. Activity profiles in mice.**
a, For the same mouse in Fig. 3, we show the z-axis depth profiles of the simulated dose (normalized to the target dose; blue), the simulated PET activity (dashed blue) and the measured (solid red) PET activity profiles, normalized to their maximum and laterally integrated on the BEV aperture (±1 mm) in the x–y plane orthogonal to the beam direction. The CTV and µCT spine regions are highlighted by pink and red bands, respectively, while the compensator is depicted in yellow and the dimple (Extended Data Fig. 3b) in light blue. b, Comparison between simulated dose and PET activity profiles for a mouse analysed at the CBCT in the SARRP in vertical position. The plot shows the z-axis depth profiles of the simulated dose (normalized to the target dose; solid black), the simulated PET activity (dashed black), and the measured (solid red) PET activity profiles, normalized to their maximum and laterally integrated on the BEV aperture (±1 mm) in the x–y plane orthogonal to the beam direction. Although the mouse is not the same as in a, we observed that switching to the vertical position consistently induces the same anatomical change in all animals. Therefore, we overlaid the measured PET activity profile of the mouse from Fig. 3 in red. As in a, The CTV and SARRP spine regions are highlighted by pink and red bands, respectively, while the compensator is depicted in yellow and the dimple (Extended Data Fig. 3b) in light blue.

**Fig. 5. Tumour growth.**
a, Average tumour volumes calculated from 2D caliper measurements of the visible tumour (Methods) for the 0 Gy control group (n = 27 animals, purple line and circle symbols), 5 Gy group (n = 6 animals, light-blue line and square symbols) and 20 Gy group (n = 22 animals, orange line and triangle symbols). Over the course of observation period, 12 animals from the 0 Gy group had to be euthanized when the permitted tumour burden was reached. The vertical dashed line corresponds to the timepoint of irradiation. b, Measurements of the volumes using μCT. Data are more precise than caliper measurements, but they are less frequent than external measurements. c, Zoom of the data points for irradiated groups shows the recurrence of the tumour irradiated with 5 Gy. Bars are standard errors of the mean values of the different animals. For every type of tumour measurement data, a two-way analysis of variance was performed (GraphPad Prism version 10.5.0 (774)) to estimate the differences in the impact of different radiation doses on the tumour growth dynamics. All tests were two-sided, and effect sizes were not computed.

**Fig. 6. Radioactive washout.**
Top: individual activity data recorded after the end of irradiation in Supplementary Fig. 5 are grouped (left: 5 Gy (n = 7 animals), middle: 20 Gy (n = 8 animals), right: comparison of fit functions assuming the physics decay of the beam containing 96% ¹¹C, 3% ¹⁰C and 0.5% ¹⁵O ions). Filled circles represent the total measured activity while the crosses correspond to the activity normalized for a physical decay. As a fit function, a double-exponential decay function was chosen over a single-exponential decay following the results of the F test (ratio of the fit χ² with one or two parameters) with number of degrees of freedom d.f. >100. F(5 Gy) = 1126 (≫1), F(20 Gy) = 63 (≫1). The fit functions are depicted separately on the top left. Orange and blue lines are the fits of the 20 Gy and 5 Gy data, respectively. Dashed lines correspond to the fit of the full decay data, while the solid lines represent the fits of the biological washout data only. Bottom: double-exponential decay rates (k_s for the slow washout constant and k_f for the fast washout constant) and the weight of the slow component (W_s) from equation (1) for 5 Gy and 20 Gy. Box plots display the median (line), interquartile range (box spanning the 25th–75th percentiles) and whiskers extending to the furthest data points within 1.5× the interquartile range. Circles display the individual data points, and colours represent treatment groups (blue for 5 Gy, orange for 20 Gy). Significance of the differences was assessed by two-sided unpaired t-test assuming equal variances. Effect size and confidence intervals were not computed. Data visualization and analysis were performed in Python 3.10 using the Seaborn and SciPy libraries.

**Extended Data Fig. 1. Fragment separator at GSI.**
A schematic view of the FRS is shown. A primary beam of ¹²C-ions from the SIS-18 synchrotron was incident on a beryllium target to produce ¹¹C-ions, separated using the Bρ-ΔE-Bρ method. An achromatic degrader was placed at the mid-focal plane of the FRS. The FRS has three experimental branches for delivering the separated radioactive ion beam. The branch directed to Cave-M is indicated in the figure. The elements between the last dipole of the FRS and Cave-M belong to the high-energy beam transport line of GSI, designed for primary beams and therefore having smaller apertures than the FRS magnetic elements, leading to reduced transmission efficiency of secondary beams to the Cave-M. Figure courtesy of the DMU (Digital Mock Up) unit at GSI.

**Extended Data Fig. 2. Dosimetry of the ¹¹C beam.**
PEAKFINDER^TM measurements of the laterally integrated monoenergetic (pristine) beam depth dose profile in water. b. Water phantom measurement of the beam spot 2D dose distribution at the minimal water equivalent depth (z = 21 mm) inside the phantom. c, d. Water phantom measurements of the horizontal and vertical 2D dose distributions in the central plane along the beam direction respectively. Figures are normalized to their maximum values and contours represent iso-dose lines.

**Extended Data Fig. 3. Cave M beamline components.**
The scheme of the beamline is shown in Fig. 2a. a. Range modulator used to generate the SOBP from a monoenergetic pencil beam scan. b. Plastic compensator to shape the lateral and distal edges of the irradiation field to the CTV. The length is 25 mm to shield the base of the skull and the rest of the animal spine. Top: inner side with an anatomical cut for fixing the animal neck area Bottom: outer side with a dimple corresponding to the distal edge of the CTV calculated in water-equivalent thickness. For a precise description of the compensator material see the section “Dosimetry” in the Supplementary Materials.

**Extended Data Fig. 4. Dosimetry of the SOBP.**
a. PEAKFINDER^TM measurements of the laterally integrated SOBP depth dose profile in water. b. Water phantom measurement of the beam spot 2D SOBP dose distribution at the minimal water equivalent depth inside the phantom. c, d. Water phantom measurement of the horizontal and vertical SOBP 2D dose distributions in the central plane along the beam direction respectively. Figures are normalized to their maximum values and contours represent iso-dose lines.

**Extended Data Fig. 5. RIB imaging in phantoms.**
a. FLUKA simulation showing the expected ¹¹C-ion dose (in Gy) distribution in the µCT of the plastic phantom in the axial view. Doses are normalized to one in the target region. b. Corresponding Monte Carlo simulation of the PET activity. c. Online SIRMIO PET image of the activity distribution deposited during ¹¹C-irradiation overlaid on the same pre-treatment µCT used for the simulations. All the 2D images are integrated over the BEV aperture (±1 mm) in the plane transversing the beam direction and overlayed in the same axial µCT slice of the plastic phantom, which is shown in the background. d–f. The same as before, but in the sagittal view, with all the 2D images integrated on the same BEV aperture and overlayed in the same sagittal µCT slice of the plastic phantom, which is shown in the background. g. Simulated dose normalized to the target dose (blue), simulated PET activity (dashed blue), and measured (solid red) PET activity depth profiles, normalized to their maximum, along the z-axis direction, integrated over the BEV aperture (±1 mm) in the x-y plane transversing the beam direction.

**Extended Data Fig. 6. Mouse CT.**
Comparison of a scans with the μCT (left) and the SARRP CBCT (middle) for the same mouse. Red color depicts the contour of the mouse skeleton segmented from the μCT, and the yellow one from SARRP. The two scans are superimposed in the image on the left to highlight the small differences in the spine position.

**Extended Data Fig. 7. Simulation of PET imaging of a RIB damaging the spinal cord.**
a. FLUKA simulation showing a hypothetical case of a¹¹C-ion SOBP dose (in Gy) distribution damaging the spinal cord for a vertical CBCT SARRP scan of a mouse in the axial view (shown in the background). Doses are normalized to the planned target dose. b. Corresponding Monte Carlo simulation of the PET activity as should have been seen with the SIRMIO PET scanner, overlaid on the same axial SARRP slice for the same mouse (displayed in the background). c. The activity simulation shown in b is now overlaid on the µCT scan (in the axial view, shown in the background) of the same mouse to illustrate what we would have seen online in case of RIB range exceeding the target distal position. The generalized CTV contour (Fig. 1) is highlighted with a black line, while the OAR contour is marked in red. All the images are integrated over the BEV aperture in the y-plane transversing the beam direction. In panels d, e, the same sagittal SARRP slice of the mouse is shown in the background of the 2D distributions. In panel f, the sagittal view of the µCT scan is displayed in the background. g. The corresponding simulated dose normalized to the target dose (solid black) and the simulated PET activity (dashed black) depth profiles, normalized to their maximum, along the z-axis direction, integrated over the BEV aperture in the x-y plane transversing the beam direction. The CTV and spine regions are highlighted by pink and red bands, respectively, while the compensator is depicted in yellow and the dimple in light blue. The plot shows that an online image as depicted in panels c and f, never actually observed in our experiments, would have required the insertion of a range shifter to reduce the range during the irradiation.

**Extended Data Fig. 8. Grip strength test used to estimate the cervical myelopathy.**
a. In this quantitative test the animal is pulled by the tail and the strength of the grip on the bar is measured in newton (N). b. Control mouse during the test (lateral view). c. 20 Gy irradiated mouse during the test. The area behind the head is shaved prior to the tumor inoculation, and after the irradiation only some white fur grows back. Panel a created with BioRender.com.

**Extended Data Fig. 9. Toxicity vs. activity in the spinal cord.**
Animals exposed to ¹¹C-ions shows a lower forelimb strength in the grip test (Extended Data Fig. 8) compared to controls. Individual data are shown in Supplementary Fig. 3. Data for irradiated animals were only considered from week 6, assuming no radiation effect in the first month post-irradiation. a. Median grip strength values in the control (8 animals) and irradiated (13 animals) groups. Bars are standard errors of the median values ( $= 1.2533 ∙ σ / \sqrt{n}$ ). Median strength in the control group is significantly higher than in the irradiated group (Mood’s median test, p = 0.0152). b. Fraction of time points (Supplementary Fig. 3) where the measured peak force F was lower than 100 N. Bars are standard errors of the mean. The fraction is significantly higher in irradiated animals (Mann-Whitney test, p = 0.00024). c. Correlation between median grip strength values in single irradiated animals and total PET counts in the spinal cord. d. Correlation between fraction of tests with F < 100 N in individual irradiated animals and total PET counts in the spinal cord. Each point represents a single animal. The counts in the spinal cord are normalized to the total number of counts observed in the whole image reconstructed with the OSEM method. The grey area is the 95% CI around the regression line, r is the correlation coefficient and significance of the correlation was evaluated by Pearson’s test both in c and d panels. Plots produced with R v.4.2.1 ggplot package.

See this image and copyright information in PMC

References

1. Durante, M. & Paganetti, H. Nuclear physics in particle therapy: a review. Rep. Prog. Phys.79, 096702 (2016). - DOI - PubMed
1. Particle therapy facilities in operation. PTCOGhttps://www.ptcog.online/ (2025).
1. Durante, M., Debus, J. & Loeffler, J. S. Physics and biomedical challenges of cancer therapy with accelerated heavy ions. Nat. Rev. Phys.3, 777–790 (2021). - DOI - PMC - PubMed
1. Lomax, A. J. Myths and realities of range uncertainty. Br. J. Radiol.93, 20190582 (2020). - DOI - PMC - PubMed
1. Paganetti, H. Range uncertainties in proton therapy and the role of Monte Carlo simulations. Phys. Med. Biol.57, R99–R117 (2012). - DOI - PMC - PubMed

Grants and funding

883425/ERC_/European Research Council/International

LinkOut - more resources

Full Text Sources
- Nature Publishing Group
- PubMed Central
Research Materials
- National BioResource Project

Save citation to file

Email citation

Add to Collections

Add to My Bibliography

Your saved search

Create a file for external citation management software

Your RSS Feed

Image-guided treatment of mouse tumours with radioactive ion beams

Affiliations

Image-guided treatment of mouse tumours with radioactive ion beams

Authors

Affiliations

Abstract

Conflict of interest statement

Figures

References

Grants and funding

LinkOut - more resources

Full Text Sources

Research Materials