Radiation dose escalation has been highly effective in achieving high rates of durable local control for several tumors, which can further cause improved overall survival (1–3). However, the delivery of ablative doses is often compromised by the proximity of certain organs at risk (OARs), and this has been a major limiting factor, especially when intrafraction organ motion adds a layer of dosimetric uncertainty, as is frequently observed in the thorax and abdomen. Given the lack of correlation between surface changes and internal organ position, location, and movement, patient surface anatomy cannot be used as a reliable surrogate for the internal motion of the thorax and upper abdominal organs (4).

Real-time tracking and automatic gating of radiation delivery is an effective solution to spare the OARs and mitigate treatment-related toxicity. This enables the safe delivery of ablative doses to the gastrointestinal (GI)/pelvic structures, which was previously impractical due to limited visualization and intrafraction motion. Magnetic resonance-guided radiation therapy (MRgRT) enables online adaptation to customize the radiation to the anatomy of the day, addressing both intra- and interfraction anatomic variation while providing excellent soft tissue visualization. The ability to continuously track in real time with MR imaging addresses the uncertainty associated with intrafraction motion (5), while eliminating the need for invasive procedures such as internal fiducials (required for x-ray-based real-time tracking) or electromagnetic transponders (6). It also eliminates the need to expand the true target to account for motion using the concept of an internal target volume (ITV) and thereby reduces the required setup margin, therefore exposing less normal tissue to unnecessary radiation doses (7).

In contrast to conventional radiation therapy, where homogeneous target coverage is the primary objective, isotoxic dose escalation increases the dose to the target volume until the pre-selected adjacent OAR dose constraint is reached. An isotoxic approach is generally applied when certain OARs are in proximity to the target volume, and these OARs are constrained to a lower dose level than the target volume. As such, this isotoxicity approach is characterized by heterogeneous target coverage, with a higher dose covering the core of the target [usually the gross tumor volume (GTV)] (8). Furthermore, this approach necessitates ensuring a rapid dose falloff of the ablative doses from the core to the periphery of the target, so as to further minimize OAR doses in close proximity. Because of the anatomically constrained dosimetric gradients between the OARs and target volumes in such cases, it is essential to ensure that the motion of the OAR-target geometry is minimized and/or accounted for in isotoxicity planning and delivery.

Liu et al. implemented a multitarget MLC-based motion tracking system for MRgRT that can simultaneously track two independently moving structures and gate the radiation beam in real time, compensating for motion (9). This approach can be used to track the tumor and adjacent dose-limiting OARs concurrently. Such an approach makes isotoxic dose escalation in real time feasible, thus broadening the therapeutic window. MRgRT provides continuous real-time cine MR imaging of the radiation field and is deployed utilizing a simple target margin, i.e., a boundary, for standard anatomic tracking and gating. The addition of isodose line-based tracking can further enhance the intrafraction motion accuracy beyond the use of simple anatomic margin expansion, as has been used historically in MRgRT.

This is the first report to describe the feasibility and workflow of this innovative approach to dose-guided tracking of GI OAR using MRgRT. Additionally, we report the institutional dose constraints for abdominal and pelvic targets to ablative doses (50 Gy in five fractions) and time analysis for the current workflow.

A 63-year-old woman was initially diagnosed with stage IB carcinosarcoma of the endometrium in September 2020. She was treated with robotic total laparoscopic hysterectomy with bilateral salpingo-oophorectomy and bilateral pelvic sentinel lymph node dissection, followed by high-dose-rate intracavitary brachytherapy to the upper vagina (21 Gy in three fractions of 7 Gy prescribed to 0.5 cm depth), and completed six cycles of adjuvant carboplatin/paclitaxel chemotherapy in February 2021. A follow-up positron emission tomography–computed tomography (PET‐CT) scan in June 2022 revealed an oligometastatic disease in the left para-aortic (PA) node (1.1 cm), which has further enlarged to 1.3 cm by October 2022, with an increase in maximum standardized uptake value (SUV_max) from 5.75 to 10.9.

A multi-disciplinary discussion at our institutional gynecology tumor board recommended treating isolated nodal recurrence with ablative radiation therapy. Clinically, she reported intermittent rectal pain and sporadic rectal bleeding, with the last occurrence 1 month before. She was also diagnosed with co-existing acute inflammatory bowel disease (IBD), which raised the concern of severe acute GI toxicity if treated with conventional abdominal and pelvic radiation fields. Therefore, online adaptive stereotactic body radiation therapy (SBRT) was recommended at a dose of 50 Gy in five fractions on the MRIdian system (ViewRay, Cleveland, OH, USA). The MRIdian system combines a 0.35-T MR scanner with a 6-MV flattening filter-free Linac and a double-stacked, double-focused MLC offset by half a leaf width, enabling a leaf resolution of 4.15 mm at the isocenter (10). The patient was included in an institutional review board (IRB)-approved protocol.

She completed the planned treatment course, with Common Terminology Criteria for Adverse Events (CTCAE) grade 1 nausea being the only acute toxicity. The follow-up PET-CT 3 at months and 6 months post-treatment confirmed complete metabolic and anatomic resolution of the node. With a follow-up at 12 months post-radiation, the patient had durable local control of the treated lesion and the entire PA chain. She has not developed any further late GI toxicity but had one out-of-field failure (left external iliac lymph node). This was also treated with SBRT using MRIdian; however, owing to its quite posterior position and the absence of any abutting GI OAR with respiratory-induced motion, no GI OAR management was needed during cine tracking, and only multi-planar anatomical tumor tracking was used. She continues to have no late toxicity and no evidence of disease on the latest scans in November 2023.

Although multi-object tracking has been technically described, to our knowledge, this is the first report that describes the clinical workflow, successful implementation, and treatment timeline of an isotoxicity approach with dose-guided tracking of a GI OAR (9, 13). With the MRIdian A3i 510K approved for clinical use, several features now facilitate modern SBRT, including multi-planar tracking (up to three planes simultaneously at 4 FPS), dose-guided and anatomical tracking capabilities, and a parallel adaptive workflow that significantly reduces the on-table treatment times. This novel technology was applied to this patient’s case, in which a high-risk OAR was in close proximity to the target in a patient who had a higher predisposition for GI toxicity due to active IBD. The challenge of delivering an ablative dose in this case was that the intrafraction motion of the GI luminal OAR was not synchronous with the target motion. This approach was deemed successful because the patient did not have any significant GI toxicity (only grade 1 nausea), despite being at higher risk and she had complete resolution of the tumor after treatment.

Radiotherapy in the setting of IBD may be at increased risk of severe toxicity, with estimates upwards of 40%–50% in some series (14). In this cohort, 21% of patients required cessation of treatment due to toxicity, which, in the context of curative intent therapy, may result in suboptimal overall survival. There are other cohorts, including a systematic review, suggesting lower rates of toxicity, similar to the baseline risk following pelvic radiation, and some showed a similar risk of toxicity in the setting of IMRT vs. 3D-CRT techniques (15, 16). There is a paucity of data reporting dose constraints for the SBRT technique in the setting of IBD, and for consistency within our institution, similar OAR constraints were used as compared to other indications for adaptive abdominal radiotherapy.

As previously detailed, real-time tracking was performed for the target on the coronal plane and the duodenum, with dose-guided tracking on the sagittal plane. The institutional target coverage and dose constraints for abdominal targets at ablative doses (50 Gy in five fractions) are presented in Table 1 . A prophylactic dose of 30 Gy was selected in line with the SPARTACUS protocol for elective lymph node coverage (17). With the duodenum adjacent to the target in this case and considering its independent intrafraction motion with respiration, an isotoxicity approach was implemented with the 35-Gy isodose line selected for the dose-guided tracking.

An in-room patient monitor was installed to guide patients regarding the adequacy of their breath hold ( Figure 4 ). The monitor displays the real-time motion of the target volume on the right and the OAR on the left. Note that the user has the option to display all planes, a subset of planes, or none. A smiling emoji is displayed on the monitor as real-time feedback to the patient when the breath hold is adequate, and all gating parameters are satisfied.

In our case, we chose gating settings that required both planes to be satisfied for beam-on delivery. As the target and OAR motion are independent of each other with respiration, at times there can be good target coverage but not a separation with the OAR, and vice versa. This non-synchronous motion is also demonstrated in Figure 4 , emphasizing the need for independent, simultaneous tracking of the target and OAR. The treatment beam is automatically gated off at such moments and delivered only when the target is inside and the OAR is outside of the defined boundary.

An increase in the gating events is generally expected, but note that in this case, the target was relatively static, improving the treatment delivery duty cycle. The in-room monitor that provides the visual position of the tumor in real time and feedback to the patient greatly enhanced the adequacy of the breath hold. Moreover, the patients felt actively involved in the treatment process, resulting in improved patient satisfaction, which might have compensated for the treatment time.

AAPM TG 76 recommends the use of appropriate respiratory motion management whenever the motion is >5 mm (18). In clinical practice, image-guided tracking and motion management are restricted to the target, while the OAR is considered secondary; there is no ideal solution to track two ROIs simultaneously in real time. However, SBRT for ablative doses relies on geometric accuracy even when the OAR and target are observed to have independent motion. The technique of multitarget tracking was demonstrated by Liu et al. to simultaneously track targets such as multiple primary lung tumors, prostate, and pelvic lymph nodes (9).

We have adapted this technique to track a target and the adjacent OAR, which have independent respiratory motion to enhance the accuracy of treatment delivery. Conventionally, abdominal and thoracic sites are treated using ITV and PRV, which can lead to inferior target coverage and/or a higher dose to the OAR. While our technique used a GI PRV to define the gradient of the ablative dose of 50 Gy, we performed plan normalization based on the isotoxicity to the nominal GI OAR wall rather than the GI PRV. Without adaptive planning and/or dose-guided tracking, performing isotoxicity to the GI PRV is recommended over the GI OAR wall. Therefore, our approach would rather allow tighter margins for both target and OAR without compromising on target coverage or OAR constraints. The median treatment time of 15 min was equivalent to other MRIdian A3i deliveries of tumor tracking only (19). MRIdian A3i parallel workflow efficiency gains of 30% reduction in online adaptive replanning times have been previously reported (19). The workflow presented here can be adopted for the definitive treatment of any thoracic, abdominal, or pelvic target with dose-limiting OAR in close proximity to ablative doses with an isotoxicity approach.

Beyond this case report, we have implemented dose-guided tracking for all MRgRT patients who have asynchronous motion between the proximal OAR and the tumor. In practice, we have found that the majority of applicable clinical sites are pelvic nodes, in which the target remains static with respiratory motion but the surrounding GI OARs have respiratory-induced motion. Criteria for the section are performed at the time of MRgRT simulation, upon which the geometry of the OAR and target motion is assessed in both the sagittal and coronal cine MR planes. Our future aim is to assess the toxicity profile of patients treated with tumor-based track alone versus dual tumor and isodose-based OAR tracking to quantify the potential reduced toxicity of this method.

Dose-guided tracking of a GI OAR for isotoxicity delivery is feasible on the MRIdian system for SBRT delivery of a target with an adjacent OAR having non-synchronous respiratory motion to ablative doses. This approach takes advantage of the soft tissue visualization of MR and the real-time multi-planar tracking capabilities of an MR Linac and aids in reducing toxicity while maintaining target coverage.

The datasets presented in this article are not readily available because this is a case report. Requests to access the datasets should be directed to the corresponding author.

This research was conducted as part of an institutional review board (IRB)-approved retrospective chart review of MR-guided radiotherapy that does not require prospective informed consent. There were no potentially identifiable images or data included in the article.

SY: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing. YW: Conceptualization, Investigation, Methodology, Supervision, Writing – review & editing. MC: Investigation, Methodology, Supervision, Validation, Writing – review & editing. NB: Conceptualization, Supervision, Validation, Writing – review & editing. AG: Conceptualization, Supervision, Writing – review & editing. RK: Writing – review & editing. MM: Supervision, Writing – review & editing. KM: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Supervision, Validation, Writing – original draft, Writing – review & editing.