Impact of technological advances in treatment planning, image guidance, and treatment delivery on target margin design for prostate cancer radiotherapy: an updated review

Abstract

Recent innovations in image guidance, treatment delivery, and adaptive radiotherapy (RT) have created a new paradigm for planning target volume (PTV) margin design for patients with prostate cancer. We performed a review of the recent literature on PTV margin selection and design for intact prostate RT, excluding post-operative RT, brachytherapy, and proton therapy. Our review describes the increased focus on prostate and seminal vesicles as heterogenous deforming structures with further emergence of intra-prostatic GTV boost and concurrent pelvic lymph node treatment. To capture recent innovations, we highlight the evolution in cone beam CT guidance, and increasing use of MRI for improved target delineation and image registration and supporting online adaptive RT. Moreover, we summarize new and evolving image-guidance treatment platforms as well as recent reports of novel immobilization strategies and motion tracking. Our report also captures recent implementations of artificial intelligence to support image guidance and adaptive RT. To characterize the clinical impact of PTV margin changes via model-based risk estimates and clinical trials, we highlight recent high impact reports. Our report focusses on topics in the context of PTV margins but also showcase studies attempting to move beyond the PTV margin recipes with robust optimization and probabilistic planning approaches. Although guidelines exist for target margins conventional using CT-based image guidance, further validation is required to understand the optimal margins for online adaptation either alone or combined with real-time motion compensation to minimize systematic and random uncertainties in the treatment of patients with prostate cancer.

Introduction

External beam radiation therapy is a cornerstone of treatment in localized prostate cancer and is associated with good biochemical outcomes and a favourable toxicity profile in most patients. Recently, there have been many technological developments focussed on precision delivery of prostate radiotherapy (RT).^1,2 The selection of appropriate target margins to ensure tumour coverage, while minimizing radiation dose to the organs at risk (OAR), relies on the technical factors used to plan and setup the patient.³ With recent advancements in treatment planning, image guidance and plan adaption, optimizing planning target volume (PTV) margin design is critical for minimizing irradiation of normal tissue, enabling dose escalation, and shifting to stereotactic body RT (SBRT) for a greater number of patients. In this article, we review the technical and clinical data to support evidence-based recommendations for margin design in photon RT for intact, localized prostate cancer.

Methods and materials

We started with a structured search of PubMed from January 2010 to March 2023 using the following search terms: (“margin” OR “planning target volume” OR “target volume”) AND “prostate” AND “radiotherapy” AND (“margin” OR “planning target volume” OR “target volume”) AND “prostate” AND “radiation.” The PubMed search resulted in more than 1900 articles after applying our inclusion criteria of English-language publications that involved external beam photon treatment of the intact prostate. We screened abstracts and articles to exclude studies focussed on post-operative RT, brachytherapy, and particle beam therapy. In addition to the 46 articles found from the structured search, we also added another 54 articles within the same time range to highlight other advances in prostate margin design. We categorized the articles based on the following topics and subtopics listed in Figure 1.

Figure 1.

Summary of key topics impacting PTV margins for localized prostate cancer treatment. Abbreviations: GTV = gross tumour volume, CBCT = cone beam CT, TCP = tumour control probability, NTCP = normal tissue complication probability, PTV = planning target volume.

Results and discussion

Targets

Targets and planning risk volumes

Prostate and seminal vesicles treatment

Conventionally, the clinical target volume (CTV) encompasses the prostate with inclusion of the proximal seminal vesicles in patients with unfavourable risk features.⁴ Typical margin recipes rely on the assumption of a spherically symmetric target without deformations or rotations during RT.^5,6 Recent work indicates a standard 3-5 mm PTV margin sufficiently accounts for rotations,^7,8 whereas another report indicates accounting for CTV shape changes improves CTV coverage.⁹ Bladder and bowel preparation plays an important role in margin design, and a recent survey of American Society of Radiation Oncology radiation oncologists captured patterns of practice and determined the need for more evidence-based guidelines for preparation and PTV margin selection.¹⁰ With advances, such as MR-guided adaptive RT techniques, it is possible to better delineate and account for inter-fraction prostate volume changes¹¹ and non-rigid-body translations.¹² Ability to accurately characterize risk of microscopic extracapsular extension and delineate the appropriate CTV volume is a critical aspect for ensuring target coverage and as PTV margins decrease closer to 0 mm, the concept of CTV margins become critical. Moreover, unique and independent seminal vesicle motion between and within treatment fractions is increasingly being characterized and addressed in both margin design and target tracking.^13–15 PTV margin design that addresses intra-fraction seminal vesicle motion may require novel approaches, such as assessing dosimetric coverage for retrospectively generated plans with a range of margins.¹⁶

Concurrent target treatment

 

There is accruing evidence to support the treatment of pelvic lymph nodes (PLNs) in high-risk prostate cancer.¹⁷ Concurrent prostate and PLN irradiation create additional challenges in achieving target coverage due to the independent intra- and inter-fraction target motion. Initial work established the efficacy of using prostate-fiducial-based image-guided RT (IGRT) for treating the prostate and PLN.^18,19 Ferjani et al.²⁰ demonstrated the efficacy of a dual-registration technique, which assesses and limits translational shifts between an initial match to the pelvic bony anatomy and subsequent prostate alignment. The cited study demonstrated that clinically employed 8 mm prostate PTV (6 mm posterior) and 5 mm PLN PTV provided sufficient coverage, and interestingly the simulations showed no dosimetric benefits for adaptive RT or MLC-tracking. Although for sequential treatment, another report using dose reconstruction demonstrated that 5 mm PLN margins are sufficient for either prostate or bony anatomy registration, but prostate registration significantly reduced rectum D50.²¹ In contrast, another study showed that using a 5 mm PLN PTV margin only provided coverage for ∼75% of fractions, with a significant dosimetric impact on PLN coverage in 19% of fractions.²² In addition to large bony-to-prostate registration variations, large bladder height variations have been identified as a contributing factor in patients with sub-optimal nodal coverage.²³ With node-positive simultaneous integrated boost, prostate-fiducial registration led to significantly reduced coverage of gross node volumes, requiring 3-5 mm gross PTV margins for all directions except superior and posterior, which require 8-9 mm for 90% target coverage.²⁴ Multiple target volumes treated using a single isocentre creates opportunities for novel PTV margin validation strategies like the Lyons et al.²⁵ approach using cone beam CT (CBCT) anatomical data and incremental margin expansion to determine that a 5.2-mm prostate, 6.5-mm prostate and seminal vesicle, and 7.6-mm PLN margin were required to achieve 95% geometric coverage.

 

To date, the clinical literature has not clearly defined the optimal margin for gross tumour volume (GTV) boost. Isotoxic dose escalation to the primary GTV that targets the macroscopic tumour appears to offer increased biochemical disease-free survival without an impact on toxicity or quality of life, as reported in the phase III FLAME trial¹ in which investigators did not apply a CTV or PTV margin to the boost region. Isotoxic dosimetric criteria may create similar plans irrespective of PTV margins if the boost dose is limited by OAR constraints. Target coverage may be further improved using tracking and motion compensation approaches as demonstrated in a recent phantom study.²⁶ One study demonstrated sufficient coverage with a 3 mm PTV margin applied to the dose-escalated GTV volume without significant impact on the OARs.²⁷ With intra-prostatic microboost, urethral sparing becomes critical and Benhmida et al.²⁸ employed a 2 mm planning volume at risk margin on the urethra with a 3 mm PTV margin on the GTV. Studenski et al.²⁹ used a set of plans with a range of 0-5 mm margins to demonstrate that 3 mm is optimal for the intra-prostatic GTV based on institutional clinical goals. Location of the GTV is also important, as a previous dose accumulation study demonstrated significantly lower delivered dose for intra-prostatic GTVs in the posterior medial peripheral zone due to the proximity to the rectum.³⁰ Image guidance on CBCT typically relies on prostate gland or fiducials as surrogates, however, MR-guidance offers potential for direct GTV visualization and further opportunity for margin reduction.

Planning risk volumes

While the focus of the current review is PTVs, it is recognized that margins may also be necessary and applied to OAR. For example, the urethra is subject to uncertainty in both delineation and geometric motion, and there is emerging evidence to support dose-toxicity relationships with this OAR.^31,32 With regards to urethra delineation, the use of optimized MRI sequences, such as 3D Haste and 3D T2 SPACE, has been shown to reduce inter-observer variability and potentially improve the accuracy of the urethra contours.³³ The use of on-board MR-guidance has also highlighted the possibility of urethral interfractional motion and anatomic changes during treatment, potentially resulting in higher anticipated delivered doses to the urethra. The literature suggests that urethra-sparing protocols, which do not require on-treatment MR guidance/adaptive planning should use a 2-3 mm urethra PRV to account for these uncertainties.^34,35 It is possible to apply PRVs to other OARs, but use of PRVs has not been mandated in most large cooperative group study protocols as robust data does not exist to support routine use.

Imaging modalities

Cone beam CT

With increased availability, daily CBCT-guided radiation therapy has emerged as the conventional IGRT approach in clinical practice for localized prostate treatment. Compared with weekly CBCT imaging one study reported that daily CBCT increased target coverage in 90% of patients and reduced rectal dose in 80% of patients. Using contours from verification CBCT imaging rigidly mapped to the planning CT, the cited study suggested that a 5 mm PTV margin offered adequate coverage based on the CTV V95%.³⁶ Efficacy of daily vs weekly image guidance was demonstrated in a phase 3 randomized trial and found improvements in rectal toxicity and oncologic outcomes with daily image guidance.³⁷ Further work evaluating all sources of uncertainty, including intra-fraction prostate motion, suggest that 6 mm (3 mm posterior) PTV margins adequately account for 92% of intra-fraction motion but increasing the inferior margin and reducing the left-right margins would better reflect intra-fraction motion.³⁸ When considering margin reduction for soft-tissue registration, supportive evaluation with fiducial markers should be considered at least once to avoid underestimating the PTV margins.³⁹

Novel CBCT reconstruction approaches offer the potential to improve image quality in prostate adaptive RT applications to improve online contouring accuracy and soft-tissue alignment. One approach gaining traction is iterative reconstruction CBCT, which estimates and subtract scatter from the projections followed by an iterative reconstruction technique to reduce artefacts and noise.⁴⁰ Other advances in artificial intelligence (AI)-driven CBCT reconstruction have the potential to further improve image quality,⁴¹ but there is no clear data to show the image quality improvements may translate to PTV margin reductions.

MRI

One key technological advance is the use of MRI for prostate RT, with improved soft-tissue contrast (Figure 2), for delineation, image registration, and real-time motion monitoring.^42,43 With an MR-only simulation workflow, it is possible to reduce systematic uncertainties in target delineation without concomitant registration uncertainties associated with CT-to-MR fusion.⁴⁴ Recent work showed that inter-and intra-observer prostate delineation variability is dependent on training and experience, and a well-structured education programme has greatest impact on MRI contouring compared with CT.⁴⁵ With MR-guided RT (MRgRT), target margin reduction is possible based on multiple factors including improved target and normal tissue border delineation, improved soft-tissue contrast for image registration, and real-time target tracking via cinematic (CINE) MRI to drive gating and motion compensation strategies. Using this justification, a recent randomized clinical trial employed a 2 mm PTV margin for the MRgRT arm, whereas the CBCT-guided arm followed the standard 4 mm margins.⁴⁶ Interestingly, a secondary study end-point was the proportion of MRgRT fractions with online adaption, but in the end online adaptation was not used in the MRgRT arm.^2,46 Given the additional resources required for adaptation, a recent study suggested non-adaptive MRgRT SBRT with a 3 mm PTV margin was adequate, and online adaption should be reserved for applications with greater potential for clinical benefit.⁴⁷

Figure 2.

Representative patient (A) CT image and contours with the intra-prostatic GTV as the innermost contour, prostate CTV contour outlining the prostate, and 5 mm PTV in around the prostate CTV along with the corresponding CBCT image used for IGRT on a standard linear accelerator (B) and T2-weighted MR image collected on 1.5 T MR-linac (C) ∼1 h prior to the CBCT.

Technological advances

Adaptive RT

Initial work on adaptive RT in localized prostate PTV margin personalization dates back over two decades. With a prospective cohort of 30 patients the team at William Beaumont reported on the clinical implementation of adaptive RT using an offline adaptive strategy to create a confidence-limited PTV margin that removes systematic variation and helps to compensate for random variation in both the patient setup and target location across treatment⁴⁸ and further developed this approach for IMRT.⁴⁹ Increasingly, adaptive approaches are moving from an offline intervention to daily online adaptive RT, which removes systematic contouring uncertainty and better addresses target rotation and shape variation (Figure 3). An early report using CT-on-rails demonstrated the ability to automate online adaptation to compensate for target under coverage and limit dose to the OARs.⁵⁰ Implementation of daily adaptive prostate RT significantly benefits from integration of AI segmentation and planning into an efficient workflow.⁵¹ An initial report of CBCT-guided adaptive RT demonstrated improved coverage using the same margins as standard IGRT,⁵² but further investigation is required to realize potential margin reductions.

Figure 3.

Contrasting the target contouring and inter-fraction shape variation for both conventional and adaptive RT with the target (simulated prostate) in a transparent oval and outline-only oval representative of the target contour position on the reference and daily imaging. In conventional planning the reference plan contouring errors are systematic across all treatment fractions, whereas theoretically with online adaptive contouring uncertainties at each individual fraction represent a random error in the overall treatment delivery. Inter-fraction target differences, including rotations and target shape changes (a simple expansion in this example) cannot be accounted for with conventional RT, but with adaptive RT is able to properly account for these changes.

Online MR-guided adaptive RT offers the potential to adapt each fraction to the patient’s anatomy to reduce the impact of inter-fraction anatomical uncertainties for target coverage improvements and OAR-sparing. Several investigations have quantified the impact of margins on target coverage and OAR-sparing. Using automatic delivered dose reconstruction via CINE MRI target tracking and log files authors showed that intra-fraction anatomical changes had a small dosimetric impact using 5 mm PTV margins with 3 mm posterior expansion.⁵³ Evaluating the importance of adaptation, another study demonstrated that with daily adaptation the seminal vesicle target coverage was achievable using a 3 mm PTV margin but without adaptation coverage was insufficient in 75% of plans.⁵⁴ Another study looked at the dosimetric assessment of gating and demonstrated a negligible impact of gating on delivered dose and that 3 mm PTV margins are appropriate for the majority of patients.⁵⁵ As we drive towards smaller margins with MR-guidance subtle impact of local susceptibility distortions should be considered as some CTV voxels may have distortions >0.5 mm, particularly in presence of large rectal gas.⁵⁶ One key advantage of online adaptive MRgRT is the potential for neurovascular-sparing with the improved soft-tissue contrast for delineation.⁵⁷ Margins for online adaptive MRgRT for a neurovascular-sparing technique were 4 mm for the GTV microboost and 5 mm prostate margin with neurovascular optimization constraints providing a sparing effect.⁵⁸ For concurrent prostate and PLN treatment, MR-guided adaptive RT also offers potential to reduce margins compared with standard CBCT-guidance.⁵⁹

One concern for online MR-guided adaptive RT is the extended treatment times for image acquisition and online adaptive processes.⁶⁰ Verification imaging and time-efficient virtual couch shifts can correct for this motion but it creates a new paradigm for margin generation dependent on the threshold for performing a correction.⁶¹ With new auto-planning techniques it is possible to achieve improved PTV coverage and reduce the time required for contouring and generating the online adapted plan.⁶²

New and evolving IGRT platforms

The evolution and introduction of non-C-arm IGRT systems offers opportunities to further optimize prostate RT margin design. For example, Cyberknife technology (Accuray Incorporated, Sunnyvale, CA, United States) has evolved to include a multi-leaf collimator capable of delivering intensity modulated RT as well as more recently demonstrated arc therapy, capable of halving prostate SBRT treatment times.⁶³ Using a 2 mm PTV margin the room-mounted orthogonal x-ray imaging system is capable of tracking and correcting for intra-fraction prostate motion with <0.1 Gy dosimetric impact over the course of treatment.⁶⁴ Comparing 0 and 3 mm margins, van de Water et al.⁶⁵ used dose reconstruction simulations to demonstrate the 0 mm margin required rotational corrections up to 10°, but only 5° for 3 mm margins but the study did not account for beam delivery and target delineation errors.

Another evolving platform is the O-ring kV-x-ray-guided linear accelerator system. With the Radixact Tomotherapy delivery system (Accuray Incorporated, Sunnyvale, CA, United States), recent work indicates 2 mm PTV margins with motion compensation had delivered dose distributions that matched the intended treatment plan.⁶⁶ Another O-ring technology available for prostate RT is the Halcyon (Varian Medical Systems, Palo Alto, CA, United States) bore-linac, which offers an efficient <10 min prostate SBRT treatment workflow⁶⁷ but impact of this technology on PTV margin design is unclear.

Real-time tracking

Use of implant fiducial markers and electromagnetic beacons allow for the implementation of real-time tracking combined with exception gating or MLC-tracking to minimize the intra-fraction component of motion. Use of 2D kV imaging for exception gating is a mature and validated technology that improves PTV dose coverage for 5 mm PTV margins (3 mm posterior)⁶⁸ and may allow for PTV margin reduction as low as 3 mm.⁶⁹ Clinical implementation of electromagnetic-guided MLC-tracking for localized prostate RT shows improvement in beam-target alignment from a root mean-square error of 1.4-0.9 mm, improved target coverage and reduced uncertainty in OAR doses.⁷⁰ Most recently, independent MLC-tracking for prostate and lymph node displacements demonstrated reduced targeting errors compared with single-target tracking.⁷¹

Hydrogels and immobilization

Various immobilization devices may facilitate margin reduction by limiting the potential for the intra-fraction motion component of the margin design. Previously, the most commonly used devices were endorectal balloons, which are inflated with air or water and used to displace the posterior and lateral walls away from the prostate.⁷² Several studies have consistently shown that endorectal balloons can reduce intra-fraction motion.⁷²

Recent reports also have described the use of rigid immobilization devices to minimize motion and reduce PTV margins.^73,74 For example, the GU-Lok (Sunnybrook Health Sciences Centre, Toronto, Canada) is a rigid immobilization device that is inserted into the patient’s rectum, when the patient lies in a lateral decubitus position with the legs flexed to 90°.⁷³ Using the GU-Lok along with fiducial markers, PTV margins could be reduced to 2 mm (2.5 mm superior/inferior) along with the ability to move small bowel away from the target in patients with unfavourable anatomy. Similarly, the Rectafix device (Scanflex Medical AB, Tumstocksvägen, Sweden) is another rigid immobilization device, which has demonstrated the potential to reduce margins to 5 mm margin everywhere except posteriorly (3 mm).⁷⁴ In recent years, the use of hydrogel rectal spacer has become more prevalent.⁷⁵ Originally designed to provide physical space between the rectum and prostate to improve rectal dosimetry, various studies have also investigated whether the placement of a rectal placer affects prostate displacement.⁷⁶ While the results are mixed, most studies suggest that prostate motion is not consistently reduced with this approach.⁷²

In summary, the use of endorectal balloons and rigid immobilization devices can reduce PTV margins through reducing prostate displacement and increase the reproducibility of the rectal wall. However, these devices need to be inserted for every fraction and may be associated with increased treatment time and patient discomfort.^77,78

Artificial intelligence

Clinical integration of AI into the prostate RT workflow has the potential to influence margin design. With autonomous un-supervised plan generation and segmentation, a recent report demonstrated the clinical acceptability for generating a basic plan for online adaptation with a future aim of real-time ART delivery.⁷⁹ Autonomous adaptive planning may reduce the overall time for the adaptive process and potential for patient motion, allowing for margin reduction, but more data are required.

Setup uncertainty related to inter-observer differences aligning the planning CT with the CBCT is a key determinant of the PTV margin. AI has been employed to help automate the soft-tissue registration of the CTV by training on reference CTV volume shifts, with residual errors in the validation test set of 1.01 ± 1.09 mm with a single feature.⁸⁰ Another study demonstrated the feasibility of training a deep-learning model to localize the prostate target on CBCT without the need for fiducial markers via automatic detection of inherent landmarks within the prostate PTV and found 91% of translations and 95% of rotations were within the clinically accepted tolerance (±3 mm/±2°).⁸¹ With deep learning it is possible to track marker seeds in 98% of projection kV images from various gantry angles within ±1 mm in real time.⁸² Markerless prostate localization on projection kV x-ray imaging can also be achieved with a deep neural network with potential applications for patient setup and real-time target tracking with results showing average predicted minus actual differences <2 mm.⁸³

Beyond PTV recipes

Ultra-hypofractionation

As prostate cancer is thought to be characterized by a low α/β ratio, ultra-hypofractionation is increasingly being utilized to optimize the therapeutic ratio and improve patient convenience.⁸⁴ PTV margin formulas assume systematic and random errors are independently normally distributed,⁶ as such shortened treatment schedules require alternative methods to validate PTV margins. For example, Studenski et al.⁸⁵ employed dose accumulation with iterative CBCT to retrospectively validate 5 mm (3 mm posterior) PTV margins for 16 SBRT patients with prostate cancer.

Accounting for the effects of random (intra-fraction) errors has increased importance with ultra-hypofractionation. Yang et al.⁸⁶ retrospectively evaluated PTV margin requirements for 10 prostate patients receiving MR-guided SBRT using the van Herk formula. The required PTV was calculated based on the intra-fraction MR image sets and demonstrated anisotropic margins of 2.8, 5.3, and 3.9 mm were required in the left-right, superior-inferior, and anterior-posterior directions, with increasing margin requirements over elapsed treatment time. Levin-Epstein et al.⁸⁷ reported PTV requirements based on inter- and intra-fraction motion on 205 SBRT patients derived from rigid registration to implanted fiducials through kV orthogonal imaging. Using the van Herk formula, a 3 mm PTV margin was recommended for all directions except for the posterior direction, where 4 mm was deemed more appropriate to account for intra-fraction motion.

Robust optimization

Driven by a need to minimize range uncertainties, proton therapy techniques have employed robust optimization as an alternative to the conventional CTV-to-PTV margin approach to ensure target coverage.⁸⁸ Principles of robust optimization have recently been incorporated into photon planning for prostate RT. In a cohort of 13 intermediate-to-high-risk patients Wada et al.⁸⁹ used multiple setup error scenarios within the bounds of the PTV margins of 10 mm superior-inferior, 8 mm anterior-posterior, and 6 mm left-right to generate a robust plan with the CTV-only optimization function and a hybrid robust plan, which added a PTV minus rectum dose constraint. The cited study found that the dose robustness to the CTV for the hybrid robust plan was greater than the PTV-based plan with no appreciable difference in OAR dose. Moreover, robust optimization demonstrated benefits in OAR-sparing and target coverage in helical planning related to systematic setup errors, but no coverage improvements were observed with anatomical changes.⁹⁰

Probabilistic planning

Probabilistic planning is another alternative to margin recipes. One recent prostate RT approach reported is a novel adaptive beam-dependent margins that provided greater bladder and rectum sparing while maintaining CTV dose coverage probability.⁹¹ Coverage-based planning techniques also offer the ability to produce better plans than the standard fixed margin approach with patient-specific relative advantages.⁹² Another recent prostate RT report described a novel approach combining probabilistic target definition and planning for both uniform and integrated boost that generated plans that are comparable to conventional optimization but with greater flexibility and versatility.⁹³

Clinical impact of margin reduction and IGRT

A well-accepted hypothesis is that reducing PTV margins can lead to toxicity profile improvements, which is well supported by Normal Tissue Complication Probability (NTCP) modelling. For example, Maund et al.⁹⁴ showed that CBCT-enabled PTV margin reduction for 74 Gy/37 fraction IMRT plans resulted in significantly improved rectal NTCP. The relative change in rectal NTCP was −5% with 5-4 mm reduction and −35% with 5-3 mm reduction. More recently, Christiansen et al.⁵⁹ investigated the impact of MR-guided adaptive RT and consequential PTV margin reduction on VMAT plans of 78 Gy to the prostate and 56 Gy to elective PLNs. The standard margins for prostate and seminal vesicles were 7 mm uniform, and for elective PLNs were 5 mm R-L, 7 mm A-P, and 12 mm S-I. The total mean PTV volume was reduced from 1060 to 569 cm³, translating to reduced median risks of urine incontinence (ΔNTCP 2.8%), urine voiding pain (ΔNTCP 2.8%), and acute gastrointestinal toxicity (ΔNTCP 17.4%).

Reports from recent clinical trials have directly interrogated the impact of margin reduction and image guidance on patient outcomes. The MIRAGE trial was the first prospective, randomized study to evaluate the clinical impact of margin reduction.² The study compared MR-guided vs CBCT-guided SBRT in 156 patients with localized prostate cancer. MR-guided RT was delivered using the MRIdian LINAC (ViewRay, Inc., Oakwood, GA, United States), whereas CT-guided SBRT was delivered using TrueBeam (Varian Medical Systems, Palo Alto, California, USA). PTV expansion was 2 mm isotropic for MRgRT and 4 mm for CBCT, and fiducials were only used in the CBCT cohort. The primary end-point of the study was acute Grade2+ GU toxic effects. The trial was closed early after futility analysis and results from 154 patients were analysed. The study showed that acute Grade2+ GU toxic effects were significantly lower with MRI vs CT guidance (24.4% vs 43.4%); P = .01. Also seen were statistically significant improvements in Grade2+ gastrointestinal toxic effects, and percentage of patients who experienced large changes in urinary and bowel-related symptoms at 1-month post RT.

Several other studies have assessed the impact of image guidance on biochemical outcomes and toxicity. In a recent meta-analysis of 18 studies involving 6521 men, the use of IGRT was associated with reduced acute and late GI and acute GU toxicity.⁹⁵ IGRT was also associated with improved prostate-specific antigen relapse free survival and biochemical failure-free survival. Recent emerging data have further demonstrated a potential benefit for intra-fraction monitoring during IGRT. The TROG 15.01 SPARK trial, which investigated the role of real-time kilovoltage intra-fraction monitoring, showed that the CTV D98% dose was within 5% of planned for all patients treated with intra-fraction monitoring, whereas 5.5% of treatment had CTV D98% <5% of planned without intra-fraction monitoring.⁹⁶ Patient outcomes in the whole cohort were excellent with no change in the 12-month patient-reported outcomes compared with baseline and no grade ≥3 genitourinary or gastrointestinal toxicities.

However, while the results of the MIRAGE trial are promising and support the potential clinical benefits for aggressive margin reduction, results from longer follow-up are awaited to ensure that this approach does not compromise oncological outcomes. One example study shows the dangers of aggressive margin reduction based on IGRT changes, with patients treated using 2D kV imaging with PTV margins of 6 mm LR, 10 mm AP/CC for patients without fiducials and 3 mm LR, 5 mm AP/CC with implanted fiducial markers.⁹⁷ Multi-variate analysis of freedom from biochemical failure (FFBF) showed that rectal distension on planning CT and unexpectedly the use of markers was significantly related to worse FFBF.

Conclusion

In this article, we review the impact of recent innovations in image guidance, treatment delivery, and adaptive RT on PTV margin design in patients with localized prostate cancer. With advances in pre- and on-treatment imaging, we recognize that prostate cancer targets are a heterogenous cohort and can exhibit both patient-specific and session-specific dynamic non-rigid changes.^1,98 Improved identification of intra-prostatic and PLN sub-volumes, facilitated by molecular imaging and multi-parametric MRI, requires novel approaches to margin design to mitigate the impact of non-rigid uncertainties. The available evidence suggests that a 3-4 mm margin is sufficient for GTV boost, whereas larger margins ≥5 mm (particularly superiorly/posteriorly) are required for PLN targets. Although complex interactions exist, Figure 4 provides a simplified illustration of techniques that can be deployed for different degrees of margin reduction.

Figure 4.

Simplified illustration summarizing the factors and technology typically deployed to support different magnitudes of margin reduction.

Over the last decade, CT-based IGRT has been established as the standard for image guidance at most institutions with access to modern LINAC technology. Guidelines exist for appropriate margins for fiducial and soft-tissue CT-based IGRT in localized prostate cancer, including support for prostate margins of 2-4 mm with online IGRT combined with tracking.⁹⁹ Novel CBCT reconstruction approaches have demonstrated significant improvements in image quality and may facilitate the use of smaller PTV margins without fiducial markers, although this concept requires validation.

Another promising alternative to CBCT IGRT is the use of MRI-guidance, with commercial products becoming available in the last few years including ViewRay MRIdian and Elekta Unity (Stockholm, Sweden). With the use of MR-guidance and real-time imaging for gating, 2 mm PTV margins have been proposed without the need for fiducial markers or adaptive RT.⁴⁶ With online MR-guided adaptive RT it is possible to capture and account for non-rigid inter-fraction changes combined with real-time imaging, and gating can address the intra-fraction uncertainties allowing for 2 and 3 mm PTV margins.¹⁰⁰ The use of smaller PTV margins, facilitated by MRI-guidance, was recently evaluated in the randomized MRIAGE trial.² The study demonstrated a reduction in acute genitourinary side effects using this approach, and conclusively established a relationship between margin reduction and toxicities, underscoring the importance and clinical impact of optimizing PTV margin design.

In addition to image-guidance advances, the use of adaptive RT and AI may provide further opportunities to personalize margin design and treatment delivery. For example, investigators are currently building tools and evidence for future implementation of deformable MLC-tracking, such as the recent report from Lombardo et al.¹⁰¹ using AI to predict the future position of the target centroid on CINE MR to reduce residual tracking errors. However, there is a paucity of data evaluating the impact of uncertainties in the online ART activities themselves on the overall treatment uncertainties especially as we shift to AI-driven adaptive processes. With a future of greater personalization and real-time motion compensated delivery, new approaches beyond margins should be explored to apply concepts, such as robust optimization and probabilistic planning within the online adaptive and motion compensation processes. Importantly, any updates to target margins requires clinical validation to ensure that toxicity and tumour control outcomes are not adversely impacted.

Funding

None declared.

Conflicts of interest

None declared.