Introduction

Image-guided radiotherapy (IGRT) of pancreatic ductal adenocarcinoma (PDAC) based on implanted fiducial markers and daily orthogonal kV X-ray imaging or cone-beam computed tomography (CBCT) has been shown to significantly reduce the setup error as compared to bony alignment¹. In state-of-the-art IGRT solid gold markers are implanted into the pancreas using endoscopic ultrasonography (EUS), a procedure that is well established and generally well tolerated. However, solid gold markers not only deteriorate image quality in both CT and MRI, but additionally cause significant dose alterations in particle therapy, showing local dose perturbations up to 80% of the prescribed dose. Recently, a new biodegradable liquid marker has been developed, which forms a gel-like marker after injection into soft tissue. This marker may particularly benefit patients with PDAC who are scheduled for particle therapy, because it can be implanted using very thin (≤25 G) needles, its low Z-elemental (non-ferrous and non-magnetic) composition causes minimal proton dose perturbation in soft tissues, its size and visibility on X-ray images, CT and CBCT can be adjusted by controlling the injected volume and its soft-surface adhesiveness may decrease migration behaviour relative to solid markers. So far, the characteristics of the liquid marker on magnetic resonance imaging (MRI) have not been investigated. It is the aim of the present work to provide a quantitative, pulse sequence-independent assessment of the visibility and artefacts of the new liquid fiducial marker on MRI and compare them against those of two gold markers commonly applied in IGRT of PDAC.

Methods

To quantify the propensity of the different markers to generate signal voids and signal shifts on MRI, a spherical gel phantom mimicking the relaxation properties of healthy pancreatic tissue at 3 Tesla was constructed. Different volumes (10 µL, 25 µL, 50 µL and 100 µL) of the liquid marker (BioXmark®, Nanovi A/S) were casted into the gel as well as four Gold Anchor™ (Naslund Medical AB; 0.28 mm diameter, 10 mm and 20 mm length) and three VisiCoil™ (IBA Dosimetry; 0.35 mm diameter, 5 mm and 10 mm length) markers, implanted in different orientations. MR relaxometry was performed to quantify the size and magnitude of the decrease in the effective transversal relaxation time T2* and water proton density ρ(H) relative to pure water as a measure of potential visibility, and to quantify the size and magnitude of the increase in magnetic field inhomogeneity ΔB0 as a measure of potential signal artefacts. The phantom was scanned with a 3.0 T Philips Ingenuity TF PET/MR scanner using an 8-channel head coil.

Results

The solid fiducial markers showed a direct linear relationship between the potentially visible size and artefact size. The liquid fiducial marker showed a tendency towards a potentially visible size at smaller artefacts. Liquid markers from 25-100 μL generated visible volumes comparable to the visible size of the solid markers. The visible magnitude was the largest for the liquid fiducial marker with volumes of 25μL – 100μL showing no correlation with the magnitude of artefact. The solid markers showed a strong non-linear correlation between magnitude of visibility and artefact. The gold-iron alloy marker induced the strongest artefacts.

Discussion

The liquid fiducial marker causes signal voids on MRI due to its absence of water hydrogen atoms without strongly affecting the magnetic field in the surrounding tissue. The alteration of the static magnetic field was found to be the main effect leading to the visibility of the solid fiducial markers.

Conclusion

The liquid fiducial marker has beneficial MRI properties regarding the trade-off between potential visibility and artefacts compared to the tested solid gold markers that are currently being used for IGRT of PDAC. Contrary to the solid markers, an increase in visibility of the liquid marker was not directly coupled to an increase in artefact. Due to the proton density reduction effect, the liquid marker behaves comparably in all pulse sequences if acquired at similar resolution.

Acknowledgements

The authors thank G. Lymperopoulou (IBA Dosimetry GmbH) for providing the VisiCoil™ and Dr. I. Näslund (Naslund Medical AB) for providing the Gold Anchor™ markers. Furthermore we are grateful to T. Jepsen (Nanovi A/S) for providing BioXmark® markers and for the support in the construction of the phantom. Finally we express our gratitude to Dr. B. Mädler (Philips GmbH) for support and advice on technical aspects of MR image acquisition and processing.

References

1. Van der Horst A, Wognun S, Fajardo RD et al. Interfractional position variation of pancreatic tumors quantified using intratumoral fiducial markers and daily cone beam computed tomography. IJROBP 2013; 87, 202-8