MR-guided stereotactic radiosurgery of an intracardiac tumoral thrombus using comprehensive motion management.

1
Introduction
Lung cancer is the leading cause of cancer-related death in industrialized countries. In Europe, it accounts for approximately 20% of all cancer deaths [1]. This type of cancer is frequently associated with hypercoagulability and vascular thrombus formation. These thrombi may be bland, tumoral, or mixed [2]. Intracardiac metastases are not uncommon and represent serious complications in many cancers, particularly lung cancer. Autopsy studies reveal cardiac involvement in 17–31% of patients with bronchogenic carcinoma. Non cardiac tumors may invade the heart by lymphatic or hematogenous spread, or by local extension [3]. Management on cardiac tumoral involvement requires a multidisciplinary approach. While surgical removal is an option, it is often limited by the tumor’s anatomical location [4], and systemic therapy alone is often insufficient. Retrospective data suggests that radiotherapy may improve survival in these patients [5], however, irradiating cardiac targets is technically challenging due to simultaneous motion of both heart and lungs.
The introduction of MR-Linac technology has provided improved soft tissue visualization, enabling more precise targeting. Specifically, Elekta’s 1.5 T Unity MR-Linac system (Elekta AB, Stockholm, Sweden) has recently incorporated comprehensive motion management (CMM), allowing for monitoring and accurate control of targets subject to respiratory motion during irradiation. CMM leverages real-time, high-quality 2D cine MRI to continuously track target anatomy, enabling automatic beam gating, where irradiation is delivered only while the target remains within a predefined boundary.
Here, we present the first ever described case of a patient with non-small cell lung cancer (NSCLC) and recurrent left atrial tumoral thrombus, successfully treated with stereotactic radiosurgery (SRS) using Elekta’s Unity MR-Linac with CMM.

2
Case presentation
A 54-year-old patient presented with dyspnea in October 2023. Thoracic computed tomography (CT) performed in December revealed a large right mediastino-pulmonary mass (12 x 10 x 8 cm) involving all three lobes, the main right bronchus, diaphragm and extending into both the left atrium and mediastinum (Fig. 1). Left atrial invasion was suspected on echocardiography. Broncoscopic biopsy confirmed adenocarcinoma histology with a PD-L1 status (TPS) of 20%. The next-generation sequencing panel showed no actionable mutations for first line therapy, but identified an ERBB2 mutation (p.Asp277Tyr, exon 7) and a TP53 mutation (p.Arg282Trp, exon 8).
Staging with Fluorine-18 fluorodeoxyglucose positron emission tomography/CT (18F-FDG PET/CT) showed a locally advanced tumor with mediastinal nodal involvement but no distant metastasis, corresponding to a cT4 cN2 cM0, stage IIIB (TNM 8th edition). Neurological examination revealed dizziness, an unstable tightrope walker test, and rightward deviation on the Unterberger test. Brain MRI showed multiple ischemic strokes, consistent with Trousseau’s syndrome.
On January 3, 2024, the patient began first-line chemo-immunotherapy with carboplatin, pemetrexed and pembrolizumab. After three cycles, restaging 18F-FDG PET/CT demonstrated a good partial response (Fig. 1) with significant regression of hilar mass (16 cm3 vs 225 cm3), mediastinal and pericardial involvement, and complete nodal response. Given the patient’s excellent response and ECOG performance status of 0, definitive radiotherapy was initiated concurrently with chemotherapy from the 4th cycle. Following completion of radiotherapy, systemic maintenance with pemetrexed and pembrolizumab was continued.
A planning CT was performed on a Philips Big Bore RT (Koninklijke Philips N. V., Amsterdam, Netherlands). Organs at risk (OARs) were defined using MVision AI (Helsinki, Finland). Clinical target volume (CTV) was delineated using post chemotherapy lung gross tumor volume (GTV) and pre-chemotherapy nodal involvement. 66 Gy was prescribed in 33 fractions. 98% of the planning target volume (PTV) was covered by at least 95% of the prescribed dose. Treatment plan was calculated using RayStation version 2024B (RaySearch Laboratories, Stockholm, Sweden) with 6MV FFF photons and delivered on our Elekta Versa HD clinical linear accelerator (Elekta AB, Sweden). Standard thoracic dose constraints were all respected and are summarized in Table 1A.
Treatment was well tolerated by the patient with little to no radiation induced side effects. Patient described grade 1 anorexia and diarrhea following chemotherapy. Radiation was completed on May 2, 2024 and followed by maintenance pembrolizumab and pemetrexed.
In October 2024, follow up 18F-FDG PET/CT revealed metabolic reactivation of the left atrial thrombus. Given the ERB2 mutation, the patient was switched to trastuzumab deruxtecan. However, after 4 cycles, a subsequent 18F-FDG PET/CT confirmed progressive disease. Cardiac MRI demonstrated a left atrial thrombus with both bland (22 x 11 mm) and tumoral (11 x11 mm) components. Cine MRI quantified intracardiac motion: 2 mm anteroposterior, 3 mm transverse, 5 mm craniocaudal, with an additional 4 mm respiratory craniocaudal motion.
The case was discussed in a multidisciplinary meeting, and reirradiation was proposed. Pulmonary scintigraphy revealed that the right lung contributed only 5% of overall function. T1VaneXD and b3VaneXD MRI sequences were acquired on our MR linac Unity (Elekta AB, Stockholm, Sweden), and a planning CT was acquired as previously described. Rigid and deformable registration of the 18F-FDG PET/CT was performed using the hybrid deformable image registration (DIR) algorithm of RayStation. This algorithm integrates image-based information, including voxel intensity values, with anatomical constraints derived from contoured structures [6], [7]. This combined approach enables improved preservation of anatomical structures, particularly in regions where image information is ambiguous or of reduced reliability. The registration error of the algorithm is around 1 mm [7]. The FDG-avid component of the thrombus was then delineated as the GTV.
Cardiac motion was accounted for by defining an internal target volume (ITV) using automatic margin expansion based on the aforementioned diagnostic cine cardiac MRI data: 2 mm anteroposterior, 3 mm transverse, 5 mm craniocaudal without the additional 4 mm respiratory craniocaudal motion. Indeed, respiratory motion was addressed by the CMM gating envelope. A 2 mm PTV margin was generated. The registration structure necessary for the CMM was created using the distal trachea as shown in Fig. 2. The registration structure was monitored in real-time to gate for respiratory motion, while the intracardiac target was managed via ITV margin described above. A single fraction of 18 Gy was prescribed to 95% of the PTV, with at least 16.2 Gy covering 99% of the PTV while limiting maximum in-target dose to ≤ 120% of the prescription.
Given the previous thoracic irradiation, a rigid and deformable registration was performed between the current and previous planning CTs using the same strategy as described above. Dose deformation and summation of previous and current treatments was performed in RayStation using the same vector of deformation. All previously delivered doses were converted to EQD2 using α/β = 3 Gy (except for the spinal canal, α/β = 2 Gy) and considered in the new treatment plan calculations.
No additional biological discount factor for the time interval was applied to the previous dose; instead, reirradiation dose constraints, adapted from Rulach et al., were used, as shown in Table 1B
[8]: bronchial tree Dmax < 90 Gy equivalent dose in 2 Gy fractions (EQD2) (α/β = 3 Gy), esophagus Dmax < 75–100 Gy EQD2 (α/β = 3 Gy), Dmax medullary canal < 50 Gy EQD2 (α/β = 2 Gy). Although no formal dose constraints exist, the sino-atrial node was limited to < 90 Gy EQD2 (α/β = 3 Gy) for the sum of both irradiations. Right lung dose constraints were disregarded based on the aforementioned scintigraphy, while left lung sparing was prioritized. The plan, generated in Monaco version 6.2.2.0 (Elekta AB, Stockholm), employed 7 MV FFF photons with a step-and-shoot technique. All constraints were respected (Table 1B). PTV was undercovered to respect OARs constraints, as shown in Fig. 3.
On May 5, 2025, treatment was delivered using CMM average gating on Unity MR-Linac using an Adapt to Position workflow. The gating tolerance was set to 2 mm isotropic margin around the reference position. The gating efficiency was 70%, with total beam-on time of 9 min out of a total treatment time of 12 min. The session lasted under 30 min and was well tolerated. To our knowledge, this represents the first reported case of cardiac SBRT with CMM.
Ten days later, the patient developed fever, tachycardia, and hypoxemia, requiring hospitalization. Workup revealed grade 3 bilateral pneumonitis attributed to a Pneumocystis jirovecii infection and drug-induced interstitial lung damage secondary to trastuzumab deruxtecan. He was discharged one month later. A restaging 18F-FDG PET/CT performed in July 2025 demonstrated a complete metabolic response of the left atrial thrombus, with, unfortunately, a suspicious hypermetabolism within the radiation-induced remodeling of the lower right lung. Although suspect in nature, an inflammatory or infectious origin was considered more likely. The patient was started on antibiotics and scheduled for reevaluation after 8 weeks. No distant lesions were detected on the 18F-FDG PET/CT. At the most recent follow up, on July 21, the patient presented with an ECOG of 0, without any complaint.

3
Discussion
This case demonstrates the first successful use of CMM for cardiac SBRT on a 1.5 T MR-Linac, achieving metabolic response two months after delivering a single 18 Gy fractions to a left atrial tumoral thrombus secondary to a NSCLC.
CMM has been established for managing motion in the abdomen, pelvis and even thoracic locations, its application to a highly mobile target, such as the heart, represents an important advance in its potential clinical use. [9], [10], [11]. In fact, the only other published case of cardiac SBRT using Unity MR-Linac is the report by Batumalai et al. [4], who treated a recurrent cardiac sarcoma. While their work demonstrates the feasibility of delivering SBRT to cardiac target on Unity, the treatment did not incorporate CMM nor cine MRI for motion characterization. In contrast, our case represents the first use of CMM in combination with cine MRI ITV definition, compensating for respiratory displacement and improving accuracy.
Intracardiac extension of lung cancer is life threatening and associated with poor prognosis, refractory to systemic therapy and unsuitable for surgery. While radiotherapy is a non-invasive alternative, delivering an ablative dose to a mobile intracardiac structure is technically demanding. The dual motion of the heart and lungs necessitate larger treatment margins, with the risk of increased toxicity. The application of MR-Linac guided radiotherapy, specifically with Elekta’s Unity, allowed for superior soft-tissue contrast. CMM implementation was crucial to improve target motion management. CMM average gating was selected for its simplicity and efficiency, which prevented prolonged treatment time. Although respiratory motion was effectively accounted for in the gating envelope, cardiac motion was not and required an ITV. We also evaluated tracking the target based on cardiac motion, but the thrombus’s movement was too erratic on live 2D sagittal and coronal cine images, with temporal resolution (balanced fast field echo cine scan resolution of 5 frames per second) insufficient for the system to establish a reliable tracking lock for gating. Accordingly, the distal trachea was selected as a surrogate tracking structure for respiratory motion, as its craniocaudal displacement serves as a reliable proxy for the respiratory-induced motion of the base of the heart while not being affected by erratic cardiac motion [12]. Alternative CMM strategies such as Breath-hold gating would have reduced the gating envelope volume, allowing for tighter margins and would have been the preferred method if real-time visual feedback during treatment had been available.
Furthermore, our case underscores the importance of combining multi-modal imaging and functional studies. The combination of 18F-FDG PET/CT and cardiac MRI cine sequences allowed for precise target definition, while pulmonary scintigraphy was essential in optimizing treatment delivery and avoiding functional pulmonary tissue, particularly relevant in the re-irradiation setting. This comprehensive approach was key to deliver highly accurate and conformal treatment in a challenging location.
A significant challenge was determining a safe and effective re-irradiation dose, as validated constraints are lacking. To be conservative we estimated DIR dose uncertainties to be less than 2% [7]. As shown in Table 1B, cumulative dose delivered to OARs, were well below the tolerated reirradiation threshold, even a worse-case deviation of 2% would not result in a threshold violation. Our protocol was largely based on Rulach et al.’s guidelines [8]. The choice of 18 Gy SRS was a matter of careful consideration. Data from STOP STORM trial for ventricular tachycardia suggest doses up to 25–30 Gy, but considering the reirradiation context and lack of data on the tolerance of cardiac substructures like the sinoatrial node, we deemed it safer to reduce the dose [13]. Conversely, lung SBRT typically delivers > 100 Gy biological effective dose (α/β = 10 Gy). Safety of such a dose to the heart is unknown, but this comparison raised the possibility that the dose delivered in our case may have been insufficient for optimal tumor control.
The patient’s post-treatment course was complicated by trastuzumab deruxtecan-related interstitial pneumonitis and Pneumocystic jirovecii pneumonia. In our opinion, this is unlikely to be linked to the cardiac SBRT: drug-induced pneumonitis is a common complication of trastuzumab deruxtecan [14], the timing is not coherent with radiation-induced pneumonitis, and the bilateral nature of the inflammation argues against a post-radiation cause. Similarly, the suspicious hypermetabolism observed in the remodeled, irradiated right lung in the follow up 18F-FDG PET/CT warranted a cautious interpretation given the timing, reirradiation context and recent history of pneumonitis and pneumonia, all of which could suggest a non-tumoral origin.
While this case demonstrates a promising outcome, the limitations of a single report must be acknowledged. Long-term follow-up is necessary to assess durability of local control and monitor for late cardiac toxicities. Nonetheless, this experience establishes that MR-guided SRS with CMM is feasible, safe, and an effective non-invasive treatment for mobile intracardiac tumors, thereby expanding the therapeutic options available to our patients.

4
Conclusion
In conclusion, we present the first successful case of SRS delivered to a recurrent left atrial tumoral thrombus using a 1.5 T MR-Linac with CMM. This advanced approach allowed for the precise delivery of an ablative radiation dose, resulting in a complete metabolic response two months post treatment, despite the significant technical challenges posed by simultaneous cardiac and respiratory motion in a re-irradiation setting. Our case highlights the potential of MR-guided cardiac SRS to address lesions previously considered untreatable with radiotherapy. The integration of visual feedback within the CMM system will further increase treatment possibilities, enhance precision, and improve the practical application of cardiac radiosurgery.

CRediT authorship contribution statement
Shaïma El Chammah: Writing – original draft, Writing – review & editing. Mickael Bendahman: Writing – review & editing. Sarah Ghandour: Writing – review & editing. Olivier Pisaturo: Writing – review & editing. Marc Pachoud: Writing – review & editing. Asteria Nikolopoulou: Writing – review & editing. Mahmut Ozsahin: Writing – review & editing, Supervision.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.