Magnetic resonance-guided simultaneous multi-fo​cal adaptive radiotherapy for prostate, pElvis & metastases (MRgSMART-PEM) in prostate cancer: a prospective phase II study

Abstract

Purpose

To evaluate the safety, feasibility, and patient-reported outcomes of magnetic resonance-guided simultaneous multi-focal adaptive radiotherapy for the prostate, pelvis, and metastases (MRgSMART-PEM) in patients with high-risk, very-high-risk, or oligometastatic prostate cancer.

Methods and materials

In this prospective single-arm phase II study, 35 patients with pathologically confirmed prostate cancer (11 localized, 8 pelvic nodal, 16 oligometastatic) were treated with MRgSMART-PEM on a 1.5-T MR-Linac between June 2021 and March 2025. Radiotherapy was delivered in five alternate-day fractions using an Adapt-to-Shape workflow. The primary endpoint was clinician-reported acute grade ≥ 2 genitourinary (GU) and gastrointestinal (GI) toxicity within 12 weeks. Secondary endpoints included late toxicity, quality of life (QoL), and survival outcomes.

Results

Median follow-up was 17.2 months (range, 4.3–49.3). The median on-couch time was 59 min (range, 35–80). Acute grade ≥ 2 GU and GI toxicities occurred in 5.7% of patients each, including one grade 3 GI event; all resolved within 8 weeks. No late grade ≥ 2 toxicities were observed, and no grade 1 events persisted beyond 18 months. QoL scores declined at 2 weeks post-treatment but returned to baseline by 3 months. At 2 years, biochemical and clinical progression-free survival rates were both 87.4%, and in-field control was 100%.

Conclusions

MRgSMART-PEM is feasible, safe, and well tolerated in patients with high-risk, very-high-risk, and oligometastatic prostate cancer, achieving excellent local control with minimal toxicity and recovery of QoL. These results support further evaluation of MRgSMART-PEM in randomized trials.

Trial registration

NCT05183074 (Chinese Academy of Medical Sciences, First Posted 2022-01-10).

Supplementary Information

The online version contains supplementary material available at 10.1186/s13014-026-02799-9.

Introduction

In 2022, prostate cancer was the second most common malignancy among men worldwide and the fifth leading cause of cancer-related death in men [1, 2]. The National Cancer Center of China reported 134,200 newly diagnosed cases that year, ranking sixth in morbidity and eighth in mortality among male cancers [3]. With increasing life expectancy, the incidence of prostate cancer is expected to continue rising [4].

Prospective randomized controlled trials have shown that radiotherapy (RT) directed at either the primary tumor or all metastatic lesions improves overall survival (OS) in metastatic prostate cancer [5, 6]. The STAMPEDE trial demonstrated that, in patients with a low metastatic burden, adding RT to the primary tumor alongside systemic therapy significantly improved 3-year OS (81% vs. 73%) and failure-free survival (50% vs. 33%) [5]. A subsequent meta-analysis confirmed the benefit of primary RT in low-volume metastatic disease [7]. For metastatic lesions, the SABR-COMET trial showed that stereotactic body radiotherapy (SBRT) to all sites (≤ 5 metastases) improved 5-year OS compared with standard care alone (42% vs. 17%) [6]. These findings suggest that patients with low metastatic burden may benefit further from RT targeting both the primary tumor and metastases.

External-beam RT has long been the primary treatment for localized prostate cancer. Randomized trials confirm that moderate hypofractionation achieves similar efficacy and toxicity as conventional fractionation [8–11], and recent studies support condensing treatment into five to seven fractions [12, 13]. However, toxicity outcomes vary widely depending on image guidance. In HYPO-RT-PC, ultra-hypofractionation was associated with increased one-year urinary morbidity, as 80% of participants received three-dimensional conformal RT without intrafraction position verification (PV) [12]. In contrast, PACE-B used smaller margins, highly conformal volumetric-modulated arc therapy, and intrafraction monitoring, yet late grade ≥ 2 genitourinary (GU) toxicity at five years remained higher with ultra-hypofractionation than with conventional or moderate schedules (26.9% vs. 18.3%) [14, 15].

A conventional linear accelerator (Linac) cannot independently correct for inter- and intra-fraction prostate motion, particularly shifts caused by bladder or rectal filling, nor can it account for independent pelvic nodal movement [16, 17]. These displacements compromise target coverage and increase normal tissue exposure, making ultra-hypofractionation highly dependent on accuracy.

The integration of 1.5-T magnetic resonance imaging (MRI) with a Linac combines multiple advanced RT concepts [18]. MRI provides superior soft-tissue contrast for accurate prostate registration without fiducials or transponders. This platform supports online adaptive strategies, specifically Adapt-to-Position (ATP) and, more relevant in prostate cases, Adapt-to-Shape (ATS), with cine-MRI monitoring during treatment for continuous target verification. Since 2019, several prostate cancer fractionation schedules have been delivered using the 1.5-T magnetic resonance (MR)-Linac [19–21]. In a recent randomized trial, MR-guided RT significantly reduced acute grade ≥ 2 GU (24.4% vs. 43.4%, P = .01) and gastrointestinal (GI) toxicity (0% vs. 10.5%, P = .003) compared with Computed Tomography (CT)-guided RT, while improving patient-reported quality of life (QoL) [22].

To date, however, no study has reported clinical outcomes of ultra-hypofractionated RT that treats both primary and metastases. We hypothesized that MR-guided adaptive RT covering all sites would be safe, feasible, and might potentially confer further clinical benefit. Accordingly, we present the first prospective phase II evaluation of adaptive ultra-hypofractionated MR-guided RT for patients with high-risk, very-high-risk, or oligometastatic prostate cancer on a 1.5-T MR-Linac, with a focus on safety and efficacy(NCT05183074).

Methods and materials

Patient selection

We conducted a prospective study to evaluate the safety and efficacy of MR-guided simultaneous multi-focal adaptive RT for the prostate, pelvis, and metastases (MRgSMART-PEM), an approach designed to deliver simultaneously integrated, adaptive ultra-hypofractionation for patients with high- or very-high-risk prostate cancer requiring pelvic nodal prophylactic RT, as well as those with oligometastatic disease confined to the pelvis. Eligibility criteria included: (1) age ≥ 18 years; (2) pathologically confirmed prostate cancer; (3) localized high- or very-high-risk disease, pelvic lymph node metastases, or bone oligometastases (≤ 5 sites, with all safely treatable using ultra-hypofractionated RT) identified on multiparametric MRI and/or F18-prostate-specific membrane antigen (PSMA) PET; (4) prostate volume < 100 mL; (5) International Prostate Symptom Score (IPSS) ≤ 19; (6) no transurethral resection of the prostate within 6 months; (7) no prior pelvic RT; and (8) no contraindications to MRI (e.g., implanted devices or severe claustrophobia). Neoadjuvant or adjuvant androgen deprivation therapy (ADT) was administered according to National Comprehensive Cancer Network (NCCN) guidelines [23].

Target delineation and planning

Patient preparation procedures and simulation imaging for contouring followed previously published studies [24, 25]. Given the relatively large target volumes in MRgSMART-PEM and the longer on-couch time, patients were instructed to drink 500–800 mL of water over 20–30 min to achieve a comfortably full bladder. MR-guided RT was delivered in five alternate-day fractions.

Target contouring was performed as follows: The gross tumor volume (GTV) was defined as MRI-visible primary tumor. The clinical target volume (CTV) for the prostate (CTVprostate) encompassed the entire prostate, with a 3 mm expansion for extraprostatic extension (EPE) location. CTV4000 was defined as whole prostate with a 1 mm contraction. CTVsv included the proximal 1.5–2 cm of the seminal vesicles (SV) for ≤ cT3a disease, or the entire SV for cT3b. Lymph node metastases visible on baseline imaging were contoured as GTVnd. CTVp included pelvic lymph nodes region, from the L5-S1 junction to include the bilateral common, external, and internal iliac, presacral, and obturator nodes [26]. For bone metastasis, GTVbone was defined as MRI-visible bone disease, and CTVbone was created by applying a 1 cm isotropic margin to GTVbone, constrained to the bone cortex. Planning target volumes (PTVs) were generated by applying margins to the respective CTVs. A uniform 1 mm isotropic margin was applied to GTV and GTVbone to create planning gross tumor volume (PGTV) and PGTVbone. PGTVnd was generated by expanding GTVnd by 5 mm, limited to the confines of PTVp, and avoiding bowelbag. An isotropic 3 mm expansions were applied to CTVprostate, CTVp, and CTVbone to create PTVprostate, PTVp, and PTVbone, respectively. The expansion from CTVsv to PTVsv was anisotropic: 3 mm in the superior-inferior direction and 5 mm axially [24, 25]. The PTVprostate and PTVsv were subsequently combined to form PTV for treatment planning. Restricted by the maximum treatment field of MR-Linac, the cranio-caudal length of all targets was restricted to ≤ 18.5 cm.

The prescribed doses were as follows: 36.25 Gy/5 fractions to PTV, 25 Gy/5 fractions to both PTVp and PTVbone, with a concomitant boost of 40 Gy/5 fractions to CTV 4000 and PGTV, as well as 35 Gy/5 fractions to PGTVnd and PGTVbone. Organs at risk (OARs), including bladder, bladder wall, rectum, rectum wall, urethra, bowel bag, pelvic bone, and bilateral femoral heads, were contoured. Although no catheter was inserted, the urethra was delineated on T2-weighted MRI using a 6-mm-diameter brush to minimize high-dose exposure. All target and organ-at-risk delineations were performed in accordance with the definitions and recommendations provided in the International Commission on Radiation Units and Measurements (ICRU) Report 83 [27].

Reference intensity-modulated RT plans were generated using the Monaco planning system (version 5.40, Elekta AB, Stockholm, Sweden). Plans typically used 9–12 beams and < 90 segments; <15 beams and < 120 segments were acceptable for complex cases, but subsequent experience led to a refined threshold of < 100 segments. Dose distributions were normalized to ensure 95% of PTVs received the prescription dose. Priority was given to meet normal tissue constraints.

Dose constraints were: bladder wall V37Gy < 10 cc and V18.1 Gy < 80%; rectum wall D_max < 40 Gy, V38Gy < 0.1 cc, V36Gy < 1 cc, V29Gy < 20%, and V18.1 Gy < 50%; bowel bag V30Gy < 0.1 cc and V25Gy < 50 cc; urethra D_50% <40 Gy; lumbosacral plexus D_0.1cc <32 Gy; and bilateral femoral heads V25Gy < 5%. The complete constraint set is summarized in Table S1. Figure 1 provides representative examples of plan dose prescriptions for patients treated with MRgSMART-PEM.

Fig. 1.

Examples of magnetic resonance-guided simultaneous multi-focal adaptive radiotherapy plans for an oligometastatic patient with metastases in the right ilium, sacrum, and left ischium, as well as involvement of iliac lymph nodes

Online Adapt-to-Shape procedure

During each treatment session, patients were instructed to prepare their bladder and rectum in the same manner as during simulation. A pre-MR was acquired using a T2-weighted three-dimensional sequence. For the first two patients, a standard sequence with a 6-minute acquisition time was used. For all subsequent patients, an optimized fast acquisition sequence, a T2-weighted three-dimensional 2-minute sequence, was employed, as it provided adequate image quality for contour refinement while shortening the imaging acquisition time. For the first fraction, the pre-MR was rigidly registered to the simulation CT; for subsequent fractions, it was registered to the pre-MR from the earlier fraction whose organ position and shape most closely matched the current anatomy. Contours were initially propagated to the pre-MRI via deformable registration and then refined by the physician.

A complete online re-optimization was performed using the Monaco treatment planning system (version 5.40, Elekta AB, Stockholm, Sweden), beginning with fluence re-calculation. Just prior to completion of optimization, a second PV MR-scan was obtained to confirm target and OAR positions. If all contoured targets remained entirely within their respective PTVs and the anterior rectal wall did not shift anteriorly, the ATS plan was approved, and treatment commenced under real-time cine-MR guidance; otherwise, the ATS workflow was repeated [20].

In the event of a machine crash before adaptive plan delivery, re‑execution of the ATS workflow was prioritized, unless the machine breakdown precluded its execution. If an interruption occurred during the beam‑on phase, an “interruption plan” containing only the remaining beams was generated. Following device restoration, a new pre-MRI was acquired. Using this interruption plan as the reference, the system performed online adaptive adjustment via the ATP workflow, ensuring the complete and accurate delivery of the prescribed dose. In the ATP workflow, the pre‑MR was rigidly registered to the pre‑MR from a prior fraction, and this registration quantified the spatial discrepancy between the current and reference anatomy. Based on the deviation, the isocenter was adjusted accordingly, and the reference plan was recomputed or re‑optimized using the updated coordinates.

To mitigate acute GU toxicities and ensure adequate bladder filling, prophylactic administration of solifenacin (Vesicare; Astellas Pharma Inc., Tokyo, Japan) was recommended after the third or fourth fraction until completion of RT. In addition, prophylactic rectal lavage with recombinant human superoxide dismutase (rhSOD; Yuanda Pharmaceutical Co., Ltd., Changchun, China) was encouraged to protect rectal mucosa from radiation-induced crypt loss [28].

Patient follow-up and endpoints

The primary endpoint was the incidence of clinician-reported acute grade ≥ 2 GU (urinary frequency, urgency, obstruction, incontinence, hematuria, radiation cystitis) or GI (diarrhea, incontinence, proctitis, rectal pain, bleeding) toxicity occurring within 12 weeks of MRgSMART-PEM. Toxicities were graded according to the Common Terminology Criteria for Adverse Events (CTCAE), version 5.0. Assessments were performed at baseline, at the end of RT, every two weeks for 1 month, and monthly for 3 months thereafter.

Secondary endpoints included: (1) late toxicities assessed using Radiation Therapy Oncology Group (RTOG) criteria, with follow-up scheduled every 3 months in year 1, every 6 months in years 2–3, and annually thereafter; and (2) patient-reported quality of QoL. QoL was measured using the IPSS [29], the Expanded Prostate Cancer Index Composite-26 (EPIC-26) [30], the Functional Assessment of Cancer Therapy-Prostate (FACT-P) [31], the European Organisation for Research and Treatment of Cancer Quality-of-Life Questionnaire Core-30 (EORTC QLQ-C30) [32], and the International Index of Erectile Function-5 (IIEF-5) [33]. QoL assessments were conducted concurrently with toxicity evaluations, with an additional EPIC-26 assessment administered at the 6-week follow-up. Exploratory endpoints included failure-free survival and in-field control.

Statistical analyses

Continuous variables were summarized as mean, median (range), or frequency with percentage, depending on distribution. Changes in QoL over time were evaluated using linear mixed models. A two-sided P value < 0.05 was considered statistically significant. Failure-free survival was defined as the interval from enrollment to either biochemical or clinical failure. Prostate-specific antigen (PSA) progression was defined according to the RTOG and American Society for Radiation Oncology (ASTRO) Phoenix definition as nadir + 2.0 ng/mL. Clinical failure was classified as either local or distant progression. In-field control was defined as time from enrollment to in-field recurrence. Recurrence was categorized as in-field if > 80% of the tumor volume was covered by the 95% isodose line, marginal if 20–80% coverage was achieved, and out-of-field if < 20% coverage was achieved. All statistical analyses were performed using SPSS software (version 26.0; IBM, Armonk, NY, USA). Survival outcomes were estimated using the Kaplan-Meier method.

Results

Between June 2021 and March 2025, thirty-five patients were enrolled, with a median age at treatment initiation of 70 years (range, 53–84). Baseline characteristics are summarized in Table 1. Of these patients, 11 (31.4%) had clinically localized disease, 8 (22.9%) had pelvic nodal involvement, and 16 (45.7%) presented with oligometastatic disease. Nearly half (45.7%) were classified as Group 5 based on histologic grade, and 74% had a baseline PSA level ≥ 20 ng/ml. The ATS workflow was used for all fractions except one, which required ATP due to machine breakdown. The median on-couch time was 59 min (range, 35–80), including patient set-up (3 min), pre-MR scan (6 min), image registration (4 min), contour adaptation (12 min), plan re-optimization (13 min), PV dosecheck and plan transfer (6 min), and treatment delivery (15 min), as shown in Figure S1.

Table 1.

Baseline characteristics for 35 patients

	No.	% (range)
Median age, years	70(53–84)
T-category
T2	4	11.4
T3	21	60.0
T4	10	28.6
N-category
N0	17	48.6
N1	18	51.4
M-category
M0	19	54.3
M1a	3	8.6
M1b	13	37.1
Gleason score
7	4	11.4
8	14	40.0
9	12	34.3
10	4	11.4
Unknown	1	2.9
Baseline PSA (ng/ml)
<10	2	5.7
10 ≤ PSA<20	7	20.0
≥20	26	74.3
Risk group
High	3	8.6
Very high	8	22.85
N1M0	8	22.85
N0-1M1	16	45.7
Androgen deprivation therapy
Yes	35	100
No	0	0
α-blocker therapy
Before RT	1	2.9
During RT	0	0
Solifenacin (Vesicare)
Before RT	3	8.6
During RT	17	48.6

Abbreviations: PSA prostate specific antigen

Clinician-reported outcome measurements

All patients completed a minimum follow-up of 12 weeks after MRgSMART-PEM. At baseline, grade 1 GU toxicities were present in 25.7% of patients, including one who required α-blocker therapy. Acute GU and GI toxicities peaked at 2 weeks following treatment (Fig. 2A and B, and Table 2). The maximum cumulative incidence of grade ≥ 2 acute events was 5.7% (2/35) for GU and 5.7% (2/35) for GI toxicity. Two patients developed grade 2 acute GU toxicity, presenting with urinary frequency and urgency, which resolved to grade 1 at 4 and 8 weeks after treatment, respectively. No grade 3 or higher GU toxicities were observed at any time point. Acute GI toxicities included one grade 2 and one grade 3 event, both manifesting as diarrhea; all cases improved to grade 1 by the 4-week assessment. During treatment, 17 patients received solifenacin (Vesicare) for urinary symptom management, and 33 underwent prophylactic rectal lavage with rhSOD. Both interventions were discontinued immediately after RT. None of the patients received a rectal spacer.

Fig. 2.

Physician-recorded genitourinary (GU, A) and gastrointestinal (GI, B) toxicities were measured according to the CTCAE v5.0 criteria. Patient-reported outcomes included International Prostate Symptom Score (IPSS, C) symptom scores according to the American Urological Association classification and International Index of Erectile Function-5 (IIEF-5, D), only one patient demonstrating mild erectile dysfunction due to irregular androgen deprivation therapy after radiotherapy

Table 2.

The cumulative incidence of clinician-reported genitourinary (GU) and Gastrointestinal (GI) acute and late toxicities according to Common Terminology Criteria for Adverse Events (CTCAE) 5.0 and Radiation Therapy Oncology Group (RTOG)

Incidence	Grade 0 (n)	Grade 1 (n)	Grade 2 (n)	Grade 3 (n)	Grade 4
GU Adverse Effects
Baseline	74.3% (26)	25.7% (9)	0	0	0
Acute	43.9% (15)	51.4% (18)	5.7% (2)	0	0
Late	88.6% (31)	11.4% (4)	0	0	0
GI Adverse Effects
Baseline	100% (35)	0	0	0	0
Acute	77.1% (27)	17.1% (6)	2.9% (1)	2.9% (1)	0
Late	97.1% (34)	2.9% (1)	0	0	0

Late toxicity, assessed by RTOG criteria, was evaluated at 3, 6, 9, 12, 18, 24, 30, and 36 months post-treatment, with data available from 35 (100%), 32 (91.4%), 26 (74.3%), 25 (71.4%), 17 (48.6%), 12 (34.3%), 9 (25.7%), and 6 (17.1%) patients, respectively. The median follow-up was 17.2 months (range, 4.3–49.3), with 32 patients followed for more than 6 months. Among the 35 patients, 31 patients (88.6%) reported no late GU symptoms. The remaining 4 patients (11.4%) experienced only Grade 1 late GU toxicity, presenting with urinary frequency. Late GI toxicity was even less frequent. Thirty-four patients (97.1%) had no GI symptoms. A single patient (2.9%) experienced Grade 1 late diarrhea. No patient experienced grade ≥ 2 late toxicity, and no grade 1 toxicities persisted beyond 18 months (Table 2).

Patient-reported outcome measurements

Peak IPSS symptom burden occurred at 2 weeks post-treatment (Fig. 2C), with 20% (7/35) and 5.7% (2/35) of patients experiencing moderate and severe symptoms, respectively (excluding six patients who already had moderate symptoms at baseline). Because the majority of patients were receiving ADT, nearly all exhibited severe erectile dysfunction (ED), with only one patient demonstrating mild ED due to irregular ADT use (Fig. 2D).

QoL data were available for all patients at baseline and at the end of treatment, for 91.4% (32/35) at 3 months, 81.3% (26/32) at 6 months, and 72.0% (18/25) at 12 months. Figures 3 and 4 illustrate mean changes in EPIC-26, EORTC QLQ-C30, and FACT-P scores over time. Most QoL scales reached their lowest levels 2 weeks after treatment and returned to baseline by approximately 3 months. Significant declines in QoL related to GU and GI domains were observed compared with baseline. Changes from baseline at 2 weeks, 1 month, 2 months, and 3 months were as follows: − 2.2, − 1.0, − 1.0, and 0.4 for the FACT-P prostate cancer subscale (P < .023); − 2.2, − 0.7, − 0.8, and 0.5 for EPIC-26 urinary incontinence (P < .001); − 7.1, − 4.6, − 2.6, and − 1.4 for EPIC-26 urinary irritative symptoms (P < .001); − 14.9, − 4.0, − 0.8, and − 0.4 for EPIC-26 bowel symptoms (P < .001); and − 14.3, − 7.6, − 5.4, and − 1.0 for EORTC QLQ-C30 global health status (P < .001), respectively.

Fig. 3.

Longitudinal changes in quality of life (QoL) across Expanded Prostate Cancer Index Composite subscales: (A) Urinary Irritative/Obstructive, (B) Bowel, (C) Urinary Incontinence, (D) Sexual Function, and (E) Hormonal Function, along with global health status assessed by EORTC quality-of-life questionnaire Core-30

Fig. 4.

Longitudinal changes in quality of life (QoL) across Functional Assessment of Cancer Therapy: (A) Physical, (B) Social, (C) Emotional, (D) Functional, (E) Prostate cancer subscale, and (F) FACT-P total score

Survivals

At 2 years, both biochemical progression-free survival (bPFS) and clinical progression-free survival (cPFS) were 87.4%. Four patients experienced both biochemical and clinical progression: one developed lung metastases, two developed extrapelvic bone metastases, and one developed recurrence involving the ramus of the ischium and sacrum. The 2-year in-field control rate was 100%, with no marginal recurrences observed.

Discussion

In this prospective trial, MRgSMART-PEM was employed for the first time to deliver comprehensive irradiation of the prostate, SV, entire pelvic nodal chain, and all visible metastatic sites. The regimen was both safe and well tolerated, with low toxicity documented by clinicians and patients alike, and demonstrated encouraging efficacy.

Daily online adaptive contouring and plan re-optimization provided by MR-Linac allowed real-time correction for target or organ-at-risk deformation and motion, enabling precise SBRT delivery. Alongi et al. reported grade ≥ 2 acute GU toxicity in 12% and GI toxicity in 4% of patients treated with 1.5-T MR-Linac [21]. Similarly, Poon et al. analyzed 51 patients and found maximum cumulative rates of 11.8% for GU and 2.0% for GI grade ≥ 2 acute events [34]. Tetar et al. observed peak acute rates of 19.8% (GU) and 3.0% (GI) at treatment completion, decreasing to 7.9% and 1.0% at 6 weeks, and further to 4.0% and 1.0% at 3 months [35]. Multiple studies have confirmed that MR-guided adaptive RT for localized prostate cancer, which targets only the prostate with or without SV, delivers low toxicity and high efficacy [21, 35]. More recently, the randomized MIRAGE trial further established that MR-guided RT significantly reduces adverse effects compared with CT-guided techniques [22].

Despite these advances, a substantial proportion of patients require elective pelvic irradiation, and increasing evidence suggests that comprehensive RT encompassing the primary tumor and all metastatic deposits can extend survival in those with limited metastatic burden [5, 6]. Current NCCN guidelines recommend ultra-hypofractionation for high- and very-high-risk disease and permit its selective use in N1 or low-volume M1 settings [23]. However, ultra-hypofractionation encompassing the pelvis has only been explored in small, fragmented studies. For example, the FASTR [36] and SATURN [37] phase I/II trials prescribed 25 Gy to the pelvis and 40 Gy to the prostate in five weekly fractions; acute grade ≥ 2 GU/GI toxicities reached 24%/0% in FASTR and 46.7%/3.3% in SATURN, while late grade ≥ 2 events were reported in up to 40–53%. Similarly, Telkhade et al. treated 60 patients with 35–37.5 Gy to the prostate/nodal disease and 25 Gy to the pelvis, reporting acute grade 2 GU and GI rates of 8.3% and 11.7%, and late toxicities of 3.3% and 8.3%, respectively [38]. These heterogeneous outcomes underscore the uncertainties inherent to conventional image guidance, as interfraction motion and organ deformation can displace dose from target volumes to adjacent structures. Peng et al. [16] compared 486 CT scans from CT-on-rail of 20 prostate cancer patients (treated with conventionally fractionated RT) with their planning CT scans, and found that the actual dose received by rectum was higher than the planned dose, with increase of V45Gy by > 15% in 5.6% of fractions and increase of V70Gy by > 5% in 11.6% fractions. If intrafraction motions are considered, the dose variations of the rectum might be larger. Moreover, Gunnlaugsson et al. [39] further demonstrated that significant prostate swelling during extreme hypofractionation could only be adequately managed by adaptive RT with tight margins.

Data on ultra-hypofractionated whole-pelvic RT remain sparse. Poon et al. [40] reported outcomes in 42 high-risk patients treated on a 1.5-T MR-Linac with 40 Gy to the prostate and 25 Gy to the pelvic nodes, showing acute grade 2 GI and GU toxicities of 2.4% and 7.1%, respectively, and no grade 3 events. Their workflow combined 5-mm isotropic margins with daily adaptation using ATS in 10% of fractions and ATP in 90%. However, 25% of patients received a rectal spacer, and SBRT was delivered twice weekly.

Our study advances this paradigm by employing daily MRgSMART-PEM to irradiate the prostate, SV, entire pelvic nodal chain, involved lymph nodes, and every visible bone metastasis within a single, simultaneous-integrated-boost plan. Treatment was delivered three times per week, a more condensed schedule than that of Poon et al. [40]. In our study, at most a 3-mm margin was applied to most targets, except for PTVsv, which received a 5-mm margin in the axial direction. Additionally, the adaptation was performed exclusively with ATS to maximize geometric fidelity. Notably, no rectal spacers were used. Despite the larger target volumes and higher dose painting, the rates of grade ≥ 2 acute GI and GU toxicities were only 5.7% each, with all events resolving within 8 weeks. These findings suggest that MRgSMART-PEM can safely encompass both regional and metastatic disease, offering a potent yet tolerable therapeutic approach for high-risk and oligometastatic prostate cancer. It should be noted, however, that nearly all patients underwent prophylactic rectal lavage with rhSOD. Along with the use of solifenacin for urinary symptom management, these proactive supportive care measures may have contributed to the observed low rates of acute GU and GI toxicity. While the primary aim of this study was to evaluate the feasibility and initial safety profile of the MRIgSMART adaptive procedure, the potential mitigating effect of these adjunctive interventions should be considered when interpreting the toxicity outcomes.

QoL trajectories in this study were consistent with prior reports. Most scales reached their nadir at 2 weeks post-treatment, paralleling physician-reported acute GU and GI toxicities, and returned to baseline within 3 months. Poon et al. similarly observed declines in EPIC urinary and bowel scores at 1 month with recovery by 4 months [34]. Although we noted modest declines at 6–9 months in a smaller subset, this pattern mirrors late-effect domains documented in PACE-B [15], suggesting that mild late effects may affect QoL months later.

At a median follow-up of 17.2 months, the 2-year in-field control rate was 100% with no marginal failures, underscoring the accuracy of target delineation and delivery. Although randomized trials have already shown that RT to primary or metastatic sites improves survival in metastatic prostate cancer [5, 6], there is no evidence that treating both areas simultaneously provides a further survival advantage. By treating the prostate, SV, pelvic nodes, and bone metastases with MRgSMART-PEM, our study achieved excellent local control; however, it is still exploratory. Larger, longer-term studies are warranted to determine whether these early outcomes translate into durable survival benefit.

The main limitation of this study is its non-randomized, single-arm design, which carries inherent risks of selection bias and unmeasured confounding. Additionally, the heterogeneity of the cohort, which includes patients with localized, pelvic nodal, and oligometastatic disease, may further affect the generalizability of the findings. Furthermore, the use of protective co-interventions, such as radioprotectants, was not standardized, potentially influencing the interpretation of safety outcomes and the broader applicability of this approach. Finally, the relatively small sample size and limited follow-up duration constrain the statistical power and maturity of the data. These constraints impact the assessment of late toxicities, long-term efficacy, and particularly the interpretation of survival outcomes, which remain exploratory and immature at this stage and require validation in future studies with longer follow-up.

Conclusions

This study presents the first clinical experience with MRgSMART-PEM, a novel approach that delivers simultaneous, adaptive ultra-hypofractionated RT to the prostate, SV, pelvic nodes, and metastatic lesions. The regimen was well tolerated, with minimal acute toxicity, no clinically significant late toxicity, rapid recovery of QoL measures, and excellent local control. These findings demonstrate that MRgSMART-PEM can safely and effectively encompass both regional and metastatic disease in patients with very-high-risk or oligometastatic prostate cancer. The favorable toxicity profile, coupled with promising efficacy, supports further evaluation of this strategy in larger, randomized studies to validate long-term outcomes and establish its role in routine clinical practice.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

MRgSMART-PEM

Multi-fo​cal Adaptive Radiotherapy for Prostate, pElvis & Metastases

GU

Genitourinary

GI

Gastrointestinal

QoL

Quality of life

RT

Radiotherapy

OS

Overall survival

SBRT

Stereotactic body radiotherapy

PV

Position verification

Linac

Linear accelerator

MRI

Magnetic resonance imaging

ATP

Adapt-to-Position

ATS

Adapt-to-Shape

MR

Magnetic resonance

CT

Computed Tomography

PSMA

Prostate-specific membrane antigen

IPSS

International Prostate Symptom Score

ADT

Androgen deprivation therapy

NCCN

National Comprehensive Cancer Network

GTV

Primary tumor volume

CTV

Clinical target volume

EPE

Extraprostatic extension

SV

Seminal vesicles

PTV

Planning target volume

PGTV

Planning gross tumor volume

OAR

Organs at risk

ICRU

International Commission on Radiation Units and Measurements

rhSOD

Recombinant human superoxide dismutase

CTCAE

Common Terminology Criteria for Adverse Events

RTOG

Radiation Therapy Oncology Group

EPIC

Expanded Prostate Cancer Index Composite

FACT-P

Functional Assessment of Cancer Therapy-Prostate

EORTC QLQ-C30

European Organisation for Research and Treatment of Cancer Quality-of-Life Questionnaire Core-30

IIEF-5

International Index of Erectile Function-5

PSA

Prostate-specific antigen

ASRTO

American Society for Radiation Oncology

ED

Erectile dysfunction

bPFS

biochemical progression-free survival

cPFS

clinical progression-free survival

Author contributions

N.-N.L., N.-Z.X., and M.S. designed the study; M.S., S.-R.Q., R.W., and N.-N.L. analyzed the data and wrote the manuscript; N.-N.L. and N.-Z.X. contributed to the study concept and coordination. M.S. and N.-N.L. performed the statistical analysis. All authors contributed to patient enrollment, data collection, and the interpretation of the results, and approved the final version of the manuscript.

Funding

This work was supported by the National High Level Hospital Clinical Research Funding, LC2024A19, the “Beneficial Ear Program-Public Welfare Project on Hearing and Related Cancerous Diseases-Clinical Research Project on Radiation Therapy for Tumors and Hearing Protection” of the Audiology Development Foundation of China, ADFC-XM-202200707, and the Non-profit Central Research Institute Fund of Chinese Academy of Medical Sciences, Longevity and Health Project, 2021-JKCS-003. The funder of this study had no role in the study design, data collection, data analysis, data interpretation, or writing of the manuscript.

Data availability

Research data are stored in an institutional repository and will be shared upon request to the corresponding author.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). This clinical research was approved by the Ethics Committee of National Cancer Center/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College (Approval No.: 20/121–2317). All participants signed informed consent forms.

Consent for publication

Consent to be enrolled in this trial and data for publication had been collected for all patients.

Competing interests

The authors declare no competing interests.