Clinical Implementation of Simulation-Free Palliative Radiation Therapy: Consensus Recommendations From a Modified Delphi Study

Abstract

Purpose

Simulation-free radiation therapy (sim-free RT) uses diagnostic computed tomography scans in place of dedicated computed tomography simulations for treatment planning, potentially expediting palliative radiation therapy (PRT) while optimizing resource utilization. Broader adoption has been limited by the absence of standardized implementation guidance. This study aimed to establish expert consensus on key clinical, technical, and educational considerations for sim-free RT in PRT practice.

Methods and Materials

A modified Delphi process was conducted with international experts in sim-free RT, including radiation oncologists (n = 7), medical physicists (n = 2), a radiation therapist, and a PhD candidate specializing in PRT. The process included an open-ended first round followed by 3 Likert-scale survey rounds across 12 domains. Consensus was defined as ≥75% agreement; ≥90% agreement indicated strong support. “Trend toward agreement” was defined as 67% to 74%.

Results

Nine respondents, representing 11 individual experts from 8 institutions in 5 countries, completed all study rounds. Experts from the same institution responding as a group were considered as 1 participant. Consensus was achieved for 95% of statements. Strong support was observed for sim-free RT in palliative-intent treatments at low-risk sites (eg, thoracic/lumbar spine, pelvis) using conventional dose-fractionation. Recommended prerequisites included recent high-quality diagnostic scans (≤3 mm slice thickness, 100-140 kVp), robust image quality assurance, and multidisciplinary oversight. Areas of limited agreement included the use of older scans, application to cervical spine targets, and integration with intensity modulated radiotherapy/volumetric modulated arc therapy planning.

Conclusions

This Delphi study provides expert-driven recommendations to support safe implementation of sim-free RT in palliative practice. Findings highlight suitable indications, workflow safeguards, and training needs while underscoring the need for further dosimetric validation and broader evaluation across diverse practice settings.

Introduction

Palliative radiation therapy (PRT) is a cornerstone of palliative oncology,¹ widely used to alleviate pain and other symptoms of advanced cancer.² Expediency and convenience are key principles—treatments are ideally initiated quickly and delivered using single- or short-course fractionation schedules to optimize symptom relief, preserve quality of life, and minimize patient burden. Simulation-free radiation therapy (sim-free RT) workflows, which employ recent diagnostic computed tomography (CT) scans for treatment planning rather than requiring dedicated CT simulations, have attracted growing interest in radiation oncology. These workflows offer potential advantages, including shorter time to treatment, decreased logistical demands and patient burden, and improved resource utilization through innovative care models. Evidence supporting their dosimetric feasibility, clinical safety, stakeholder acceptability, and operational efficiency continues to accumulate across retrospective investigations, pilot projects,3, 4, 5, 6 prospective studies,7, 8, 9, 10 randomized trials,¹¹ and real-world reports. Notably, the sim-free RT-1000 study⁷ demonstrated meaningful reductions in consultation-to-treatment timelines with comparable plan quality and patient-reported outcomes. Despite these advances, widespread adoption remains limited by the lack of standardized implementation guidelines.

The Delphi technique is a structured consensus method frequently applied in health care to address emerging or complex clinical questions.^12,13 Within palliative care and radiation oncology, it provides a rigorous framework for integrating diverse expert perspectives to guide best practices where empirical data are scarce. This study used a modified Delphi process to develop expert consensus recommendations defining the clinical, technical, and educational requirements for implementing sim-free PRT in settings without online adaptive technologies.

Methods and Materials

Study design

A modified Delphi consensus process was employed to evaluate expert perspectives on sim-free RT. This included an open-ended first round of questions, followed by 3 rounds of Likert-scale surveys based on themes identified in round 1. Modifications from traditional Delphi methodology included the use of digital surveys for efficiency and a relatively small group of experts, reflective of the novelty of sim-free RT workflows. This approach blended qualitative input with quantitative consensus-building to effectively address the goals of this consensus process.

Participants

Experts in sim-free RT, including radiation oncologists (n = 7), medical physicists (n = 2), a radiation therapist, and a PhD candidate specializing in PRT, were invited to participate in this Delphi consensus process. An initial call for participants took place at a radiation oncology conference in October 2023, and additional experts were subsequently identified based on their clinical or academic contributions in the field of sim-free RT and invited to participate via email.

Delphi process

The process consisted of 4 rounds. Round 1 consisted of open-ended questions, and participants provided qualitative input on key clinical, technical, and educational considerations in sim-free RT. A complete list of round 1 questions is provided in the Supplemental Appendix. Participants were informed that these consensus statements apply to sim-free RT without online adaptation.

Subsequent survey questions were developed based on the themes from round 1, and organized into 11 domains for round 2, and a total of 12 domains for rounds 3 and 4 (see Supplementary Appendix and Table E1). Participants rated their level of agreement with each statement on a 5-point Likert scale (1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree, 5 = strongly agree). A free-text response option was available within each survey domain and at the end of the survey, providing participants with opportunities to elaborate on or explain their responses further, or to provide additional comments/feedback. Consensus was defined as ≥75% agreement (scores of 4 or 5).

In round 2, participants received a survey consisting of 65 questions across 11 domains. In rounds 3 and 4, responses from prior rounds were presented graphically (see Fig. 1) alongside deidentified free-text comments. Round 3 comprised 30 statements across 9 domains in which consensus was not yet achieved, with 1 domain added in response to participant feedback that consideration be given to implementation recommendations. The fourth round consisted of 5 questions across 5 remaining domains. A fifth round was not pursued as consensus had been achieved for nearly all questions, and because responses to statements without consensus were split, making consensus unlikely. The full list of Delphi statements from each round is provided in Table E2 (Supplementary Appendix).

Figure 1.

Examples of statements, responses, and revised statements presented to participants.

Abbreviation: BED = biological effective dose; CT = computed tomography; PTV = planning target volume; SABR = stereotactic ablative body radiotherapy.

For clarity, several commonly used terms were explicitly defined for the panel. “Clinically significant anatomic changes” were defined as changes in tumor or normal tissue anatomy expected to alter target coverage or organ-at-risk dose by ≥5%. “Generous margins” referred to target expansions of at least 10 mm beyond the gross tumor volume, with exact values varying by site and institutional practice. Although no universal value was agreed on, these ranges were consistently referenced in panel discussions and survey responses.

Round 1 was initiated in November 2023, and round 4 was completed in November 2024. All surveys were administered using Qualtrics XM, and survey links were distributed via email.

Analysis

Qualitative data from round 1 were analyzed thematically to identify key domains for further consideration. Likert-scale survey questions were grouped by domain and presented in subsequent rounds. Responses were analyzed, and descriptive statistics were used to describe the level of consensus for each statement. Consensus was defined as ≥75% agreement (ratings of 4 or 5). A “trend toward agreement” was defined as ≥67% agreement but <75%. Statements achieving ≥90% agreement were considered to have “strong support.” Minor differences in reported agreement percentages across rounds (eg, 76 % vs 78 %) reflect rounding and incremental participant reconsideration of items after exposure to group feedback, rather than statistical error.

Results

Participant demographics

Eighteen individual experts from 10 institutions were invited to participate in round 1 of this Delphi study, and 12 survey responses representing 16 individual experts from 8 institutions (89% individual and 80% institutional response rates) were received. In some cases, experts from the same institution formed groups and responded collectively. In these instances, the group was subsequently considered to be 1 collective participant. In round 2, 9 of 11 participants responded, representing 11 individual experts from 8 institutions (82% individual and 100% institutional response rates). Rounds 3 and 4 achieved full participation. In total, 9 panel representatives (reflecting 11 individual experts from 8 institutions across 5 countries) completed every study round. Most respondents were radiation oncologists with over 10 years of practice experience. Additional details on professional roles, geographic distribution, and years of practice are provided in the Supplementary Appendix.

Key recommendations

Table 1, Table 2, Table 3 to 4 present the key consensus recommendations in 4 main areas. Agreement rates (%) are displayed in brackets, and the results are discussed in further detail following the tables. Because agreement percentages reflect only the proportion of respondents selecting “agree” or “strongly agree,” the same percentage may represent differing qualitative strengths of endorsement depending on the survey round and question phrasing. All detailed Delphi statements, response distributions, and participant demographic data are provided in the Supplementary Appendix (Tables E1 and E2).

Table 1.

Consensus statements for treatment, anatomic site, and dose eligibility

Domain 1: Treatment eligibility
1.1	Diagnostic CT-based radiation planning is appropriate for palliative-intent treatments (100.0%) and may be used when delivering radiation concurrently with systemic treatment (88.9%). It may also be used in routine practice for palliative reirradiation if there is no risk of exceeding OAR tolerances (100.0%).
1.2	Diagnostic CT-based radiation planning is not appropriate for: curative intent treatment (88.9%); patients who have had significant clinical interventions since their diagnostic imaging was acquired (77.8%); highly mobile tumors (100.0%); stereotactic treatment (88.9%); treatment setups that require a thermoplastic mask for optimal immobilization (77.8%).
Domain 2: Anatomic site considerations
2.1	The following body sites may be routinely treated using diagnostic CT-based planning when all other eligibility requirements have been met: thoracic and lumbar spine, and pelvis (100.0%); mediastinum, lung, and abdomen (88.9%); brain, proximal and distal limbs (77.8%).
2.2	Diagnostic CT-based planning may be considered for neck or cervical spine radiation treatments at the discretion of the radiation oncologist but is not recommended when thermoplastic mask immobilization is indicated (66.7%).
Domain 3: Dose and fractionation
3.1	Diagnostic CT-based planning can be used in routine practice for low-dose, conventional palliative regimens, ie, 30 Gy in 10 fractions, 20 Gy in 5 fractions, and 8 Gy in 1 fraction (100%).
3.2	Diagnostic CT-based planning should not be used for high-dose conventional or hypofractionated regimens (88.9%), intermediate-dose regimens (88.9%) or for moderate- or high-dose stereotactic regimens (both 77.8%).

Abbreviations: CT = computed tomography; OAR = organ-at-risk.

Table 2.

Consensus statements for technical and logistical factors

Domain 4: workflow and imaging quality assurance
4.1	Identifying suitable patients for diagnostic CT-based planning should primarily be the role of the radiation oncologist (100.0%) although radiation therapists may also be involved when/when appropriately trained (practice environment dependent, may include advanced practice therapists) (66.7%).
4.2	Imaging quality assurance checks may be performed by radiation oncologists (88.9%), medical physicists (100.0%) and dosimetrists (88.9%), and/or by appropriately trained radiation therapists (77.8%).
Domain 5: diagnostic imaging eligibility
5.1	Diagnostic scans that are less than 72 h old are most preferable (100%), but scans that are 2 wk old or less are considered reasonable to use for diagnostic CT-based planning (77.8%).
5.2	Diagnostic scans that are older than 2 wk old may be considered for “slower” tumor histologies, where radiographic findings correlate well to patient symptoms, where generous target margins are employed to account for possible interval progression, and/or where the risk of an undeliverable treatment (and the need to schedule a CT simulation) is deemed acceptable. However, radiation oncologists should exercise caution, and the preference is for routine CT simulation (100.0%).
Domain 7: HU and respiratory motion considerations
7.1	Low-dose diagnostic scans are suitable for diagnostic CT-based planning (77.8%).
7.2	Technical variations between individual diagnostic scanners and CT simulators may result in differences in HU, geometry, etc. For diagnostic CT-based planning using direct beam, parallel-opposed pair or 3DCRT techniques, it is reasonable to accept these variations (and their effect on dosimetry) and individual diagnostic scanner-to-CT simulator calibrations are not required (100.0%).
7.3	It may be reasonable to treat intraparenchymal/abdominal targets suspected or known to have a significant degree of motion if this motion is accounted for through generous target margins/field borders, or by using planning techniques that are less sensitive to motion such as field-based parallel-opposed pair beam arrangements (100.0%).

Abbreviations: CT = computed tomography; HU = Hounsfield Units.

Table 3.

Consensus statements for treatment technique and delivery

Domain 6: treatment techniques
6.1	Direct beam and parallel-opposed pair field-based treatment planning techniques are appropriate for diagnostic CT-based planning (88.9%); 3DCRT is also considered a reasonable technique (77.8%).
6.2	IMRT or VMAT treatment planning is not generally recommended because there is currently a lack of published evidence demonstrating dosimetric similarity between planned and delivered doses using diagnostic CT-enabled planning. These planning techniques may be employed at the radiation oncologist's discretion, but caution should be taken to reduce uncertainties (77.8%).
Domain 8: treatment delivery considerations
8.1	Cushions, headrests, and other passive immobilization supports (ie, arm boards, rice bags, etc.) are preferred for diagnostic CT-based planning patient setups (88.9%). Setups requiring thermoplastic immobilization are not generally recommended (66.7%).
8.2	Custom Vac-loks (created on the unit at the first fraction) may be used for patient immobilization but should only be considered where the clinical benefit is reasonably assumed to outweigh the disadvantage of additional patient time on the treatment couch (88.9%).
8.3	Surface guidance, where available, is appropriate for patient setup (77.8%). Alternately, anatomic landmarking is an appropriate technique to aid in the initial setup of patients before image guidance (100.0%). Note that this key recommendation refers to setup only, and neither surface guidance nor anatomic landmarking is a substitute for IGRT.
8.4	3D image guidance (ex. CBCT/MVCT) is required for accurate delivery of diagnostic CT-based treatment plans (100.0%). 2D portal imaging (ie, kV/MV) alone is not considered adequate image guidance (77.8%). However, after 3D imaging has been performed (ie, on fraction 1) and setup verified, 2D portal imaging may be substituted for 3D imaging at subsequent fractions given there are no setup concerns (77.8%).
8.5	In general, the dosimetric arising from the difference in couch curvature between diagnostic scanners and treatment units are clinically insignificant and reproduction of the diagnostic couch curvature is not required. However, in certain cases, such as posterior-lateral beams or lateral targets, correcting for this discrepancy would be reasonable (100.0%).
Domain 9: managing separation and/or volume change
9.1	It is preferred, but not required, that large separation or volume discrepancies between the diagnostic CT and the CBCT be accounted for either through dosimetric recalculation or by diverting the patient for routine CT simulation (77.8%).
Domain 10: nondeliverability
10.1	If a diagnostic CT-based radiation plan is deemed undeliverable because of anatomic differences, it is recommended to proceed with a routine CT simulation instead (100.0%).

Abbreviation: 3DCRT = Three-dimensional conformal radiotherapy; CT = computed tomography; CBCT = cone beam CT; IMRT = intensity modulated radiotherapy; MVCT = megavoltage CT; VMAT = volumetrically modulated arc therapy.

Table 4.

Consensus statements for training and implementation

Domain 11: training
11.1	In-house training (88.9%), step-by-step written protocols (including patient selection criteria, scan eligibility, and quality assurance, image guidance, etc.) (77.8%), senior expert availability (77.8%), and feedback processes (77.8%) are recommended educational and evaluation components of a diagnostic CT-based planning workflow.
Domain 12: implementation recommendations
12.1	Centers developing a diagnostic CT-based planning workflow should seek multidisciplinary input and collaboration when designing their workflows. Preimplementation strategy may include dosimetric assessments/studies, preferred IGRT approaches, QA procedures, and eligibility (patient, anatomic site, diagnostic imaging) criteria (88.9%).
12.2	Thoracic spine, lumbar spine, and pelvis are recommended as low-risk (ie, the least likely to have issues with setup or interval anatomic change) body sites, which are appropriate for early/initial implementation (100.0%).

Abbreviations: CT = computed tomography; IGRT = image guided radiotherapy; QA = quality assurance.

Treatment eligibility

The panel reached a strong consensus that diagnostic CT-based planning is appropriate for palliative-intent treatments, including reirradiation scenarios, provided there is no risk of exceeding organ-at-risk tolerance doses, and also may be used when RT is delivered concurrently with systemic therapy. Conversely, its use in curative intent treatment settings, to deliver stereotactic ablative RT, and in cases requiring thermoplastic immobilization, was discouraged because of concerns regarding precision and setup reproducibility. The panel unanimously agreed that the highly mobile tumors should be excluded, as should patients who have had significant clinical interventions (ie, thoracentesis, paracentesis, etc.) since diagnostic imaging.

Anatomic site considerations

Thoracic and lumbar spine and pelvis were unanimously deemed low-risk and suitable for diagnostic CT-based planning; they were also recommended as starting points when first implementing a sim-free workflow. There was also strong support for mediastinal and proximal limb targets, and for whole brain, lung, abdominal, and distal limb targets when eligibility requirements were otherwise satisfied, with strategies in place to account for target motion, and setup reproducibility could be verified through image guidance. Opinions on treatment of distal limbs were initially divided, but there was consensus in the third round that these targets may be appropriate where generous margins/image guidance are used to ensure reproducibility. Surface guidance technology was felt to be helpful in these cases as well. Consensus was not achieved regarding inclusion or exclusion of cervical spine/neck targets, but there was a trend toward agreement that these sites may be considered at the discretion of the radiation oncologist, but also that sim-free RT is generally not recommended if thermoplastic immobilization is indicated.

Dose and fractionation

Consensus supported low-dose, conventional palliative regimens. Intermediate- and high-dose regimens (including moderate- and high-dose stereotactic ablative RT) were not endorsed for reasons of safety, accuracy, and the lack of evidence demonstrating dosimetric equivalence to CT simulation-based workflows.

Workflow, imaging quality assurance, and Hounsfield Units considerations

The panel unanimously supported radiation oncologists as the primary clinical decision-makers in identifying suitable candidates for sim-free workflows. Radiation therapists and dosimetrists were recognized for their important contributing roles. Although there was general support that adequately trained radiation therapists may be involved in assessing patient eligibility, consensus was not achieved. There was strong support for medical physicists and dosimetrists performing quality checks on diagnostic imaging, and a third round endorsement that radiation oncologists and appropriately trained radiation therapists could also perform this task.

Technical calibration of Hounsfield Units between diagnostic scanners and CT simulators was not considered mandatory in the setting of conventional PRT. A clinically insignificant degree of dosimetric variation was acknowledged as an inherent but acceptable limitation of diagnostic CT-based planning. Low-dose diagnostic scans were deemed suitable for use in planning.

Conventional CT simulation was strongly advised if there was any concern for clinically significant dosimetric differences, which may affect plan quality or accuracy.

Diagnostic imaging eligibility

Agreement on the acceptable age of diagnostic CT scans was reached only after multiple statement revisions. Scans conducted within 48 to 72 hours of treatment were unanimously endorsed, and respondents also supported the use of scans up to 2 weeks old, acknowledging general acceptability for their use in the palliative setting.

Opinions were more divided for images that were greater than 2 weeks old. Consensus was achieved in round 4 that such scans may be appropriate in cases of slower tumor histologies (generally associated with decreased risk of rapid interval radiographic change), or when sufficient margins are applied to account for potential anatomic change. The panel recommended that the sim-free expediency advantage be weighed against the risk of an undeliverable plan when using older scans and discouraged scans older than 2 weeks unless radiographic findings correlated closely with clinical symptoms, and target volume uncertainties could be mitigated.

Regarding image quality, panelists agreed that diagnostic scans should meet minimum technical standards for planning: axial slice thickness ≤3 mm, absence of significant motion artifact, and acquisition using a diagnostic kVp range (typically 100-140 kVp) suitable for accurate Hounsfield unit–based dose calculation.

Treatment techniques

There was consensus that direct beam and parallel-opposed field-based planning techniques were appropriate for use in sim-free workflows. The panel also endorsed 3D conformal RT planning as a reasonable approach. Intensity modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) were not generally recommended and are discussed later as areas for future consideration.

Motion management considerations

The need to account for tumor and organ motion variation arising from differences in diagnostic scan acquisition parameters (ie, breath hold) and treatment delivery parameters (ie, free breathing) emerged as a critical consideration. Consensus was achieved that motion-related uncertainties could be mitigated through generous target margins and the use of simpler, field-based beam arrangements, which are less sensitive to target motion.

Treatment delivery considerations

There was consensus that immobilization devices such as cushions and head rests were appropriate for patient setup, and agreement that custom Vac-lok fabrication should only be considered when the clinical benefit is assumed to outweigh the additional time required for treatment delivery for staff and patients. Anatomic landmarking and surface guidance were considered reasonable methods to assist in initial patient setup, provided setup accuracy was verified by imaging. The panel unanimously agreed that 3D imaging (ie, cone beam CT [CBCT] or megavoltage CT) was mandatory at the first fraction, but that 2D portal imaging may be performed at subsequent fractions, given that there are no setup concerns. Exclusive reliance on 2D imaging was strongly discouraged. The apparent difference in percent agreement regarding the use of thermoplastic masks when considering eligibility (domain 1) versus treatment delivery (domain 8) reflects that participants evaluated different aspects of immobilization. In domain 1, respondents were considering whether cases requiring a thermoplastic mask should be included in sim-free workflows (generally not recommended), whereas domain 8 addressed the broader question of immobilization adequacy once a patient was already deemed eligible. Thus, the same numeric agreement value represents distinct clinical contexts rather than discordant opinions.

Variations between diagnostic scanner and treatment table curvatures were generally considered clinically insignificant in this setting, and couch curvature correction was not viewed as a requirement for treatment delivery. However, curvature discrepancy corrections may be considered for posterior-lateral beams or lateral targets where variations in soft tissue could have more of an effect on external body contour and dose distribution. A preshaped Vac-lok bag replicating the diagnostic couch curvature is appropriate in this setting and described in the literature.⁸

Undeliverable treatment

Panelists emphasized that contingency plans are essential. If a sim-free RT plan is considered clinically suboptimal or undeliverable for any reason, for instance, because of interval radiographic change or anatomically/setup incongruence, there was unanimous agreement that diagnostic CT-enabled treatment delivery be abandoned and a standard CT simulation performed.

Managing separation and/or volume change

Panelists agreed that large differences in separation/volume between the diagnostic scan and the CBCT, especially those that may impact dosimetric accuracy and/or treatment deliverability, should be accounted for, when feasible, through dosimetric recalculation or by performing a conventional CT simulation. Mandatory recalculation or CT simulation for large separation differences did not achieve consensus because of practical considerations and variability in clinical scenarios. Clear guidelines outlining acceptable and unacceptable deviation are essential, as is education for radiation therapists regarding anticipated setup variabilities, that is, differences arising from anatomic change, disease progression, poorer image quality with free-breathing CBCTs, etc. Clinical judgment should be exercised when managing anatomic changes, ensuring that the risk of dosimetric variation is weighed against treatment intent and workflow efficiency.

Implementation and training

Panelists emphasized comprehensive training and robust preimplementation planning, including step-by-step protocols, multidisciplinary collaboration, senior expert oversight, and regular audits or feedback reviews. Furthermore, dosimetric assessments and pilot studies were encouraged to ensure safe and effective adoption of sim-free RT workflows. The thoracic spine, lumbar spine, and pelvis were recommended as low-risk body sites appropriate for the initial phases of workflow implementation.

Discussion

This study employed a rigorous modified Delphi process to generate expert-driven recommendations for the implementation of sim-free RT workflows in PRT. Across 12 domains, the panel reached a strong consensus on clinical indications, technical requirements, workflow logistics, and training considerations, providing a foundation for integrating diagnostic CT-enabled planning into routine practice. However, these findings reflect expert consensus rather than empirical validation. Although this process yielded a high level of agreement across domains, it reflects collective expert opinion rather than empirical evidence. These recommendations should therefore be interpreted as guidance for safe implementation, not as definitive proof of safety or efficacy.

Findings support sim-free RT for palliative-intent treatments, particularly for low-risk anatomic sites such as the thoracic/lumbar spine and pelvis, and the importance of multidisciplinary collaboration, with radiation oncologists playing a primary role in patient selection, and medical physicists/dosimetrists contributing to image quality assurance (QA). In many United States practices, image QA is typically overseen by the treating radiation oncologist, who verifies target visualization and anatomic correspondence between diagnostic and planning data sets before approval. However, this task may also be delegated to medical physicists or dosimetrists, depending on institutional workflow and staffing. Our findings reflect this variability and the importance of clear role delineation in sim-free RT implementation. Although the panel supported the use of field-based and 3DCRT planning, the application of IMRT and VMAT within sim-free RT workflows remains controversial because of dosimetric uncertainties. Prospective studies refining scan-age eligibility and evaluating IMRT/VMAT dosimetric accuracy represent important future directions for sim-free RT research.

Areas of nonconsensus and debate

Despite substantial agreement on many aspects of the diagnostic CT-based planning workflow, some areas required several rounds to reach consensus, and others remained unresolved after round 4. For example, the acceptable age of diagnostic imaging scans raised significant debate. The panel eventually agreed that it may be reasonable to use diagnostic images older than 2 weeks, but only with several caveats, including acknowledging the risk of an undeliverable plan. The use of custom Vac-lok immobilization was also contentious, with some experts in favor and others arguing that cases requiring custom immobilization should undergo routine CT simulation by default. Consensus was ultimately reached that custom Vac-lok fabrication should only be considered when the clinical benefit is assumed to outweigh the additional time required of the treatment delivery staff and patient.

Compatibility of diagnostic CT images with IMRT or VMAT planning (in the absence of online adaptive platforms) was a frequent topic of debate. Ultimately, panelists agreed that neither approach is generally recommended because of a lack of published evidence demonstrating dosimetric equivalency in planned versus delivered dose with diagnostic CT-enabled planning, underscoring the importance of limiting sim-free RT to less complex clinical scenarios until further data become available. There might also be an argument that the extra work of patient-specific QA involved in planning IMRT/VMAT could insert an undesirable delay before treatment administration for a patient for whom symptoms and overall clinical status call for expediency. If IMRT/VMAT planning techniques are used at the discretion of the radiation oncologist, robust safeguards to minimize uncertainties should be employed (ie, limiting beam angles/arcs where significant setup variation could exist or where diagnostic scans have incomplete fields of view, and by avoiding complex modulation).

Two statements did not reach consensus by round 4. Concerning the appropriateness of sim-free RT for neck or cervical spine anatomic targets, 67% of the panel agreed that these sites could be considered at the discretion of the radiation oncologist, but are not recommended when a thermoplastic mask would be required to ensure optimal immobilization. Differences in agreement percentages related to thermoplastic mask usage across domains primarily reflected differences in survey context—eligibility versus delivery—rather than true disagreement among participants. There was also 67% agreement that radiation therapists, in addition to radiation oncologists and only when adequately trained, could be involved in identifying patients suitable for the sim-free workflow. Based on participant feedback, the lack of consensus in this area is likely reflective of diversity in both practice environments and in radiation therapists skillsets/scopes of practice. For example, experts from institutions or countries in which advanced or enhanced practice RT roles are common were more likely to endorse that radiation therapists with the necessary training may be involved in performing eligibility assessment and/or identifying appropriate candidates.

Strengths and limitations

This Delphi process combined qualitative and quantitative methods—open-ended questions in round 1 and Likert-scale surveys in later rounds—allowing effective capture of both expert opinion and consensus. However, several limitations remain. Our targeted sampling approach of identifying experts with relevant expertise is an established method for ensuring that participants possess the necessary expertise to contribute valuable insights to novel areas of research. This sampling method, however, is susceptible to bias, and the selection process may have inadvertently excluded unpublished experts or non-English speakers, limiting the diversity of the panel. Although this Delphi process incorporated participants from multiple countries and professional disciplines, the overall panel size was small, and the majority represented academic or tertiary centers. Such limitations are common in early Delphi studies of emerging workflows, but they may restrict generalizability to community-based or resource-limited settings. A summary of information on participant roles, geographic distribution, and experience levels is provided in the Supplementary Appendix to enhance transparency.

Conclusion

In conclusion, this Delphi consensus study established expert-driven recommendations for integrating sim-free RT into PRT practice. Strong agreement was reached on treatment eligibility, appropriate anatomic sites, workflow logistics, and training considerations. Sim-free RT was deemed most suitable for palliative treatments in low-risk sites, such as the thoracic/lumbar spine and pelvis, using conventional dose-fractionation schedules. Radiation oncologists were identified as the primary decision-makers, with medical physicists and dosimetrists ensuring imaging quality.

Although consensus was achieved for most aspects of implementation, key areas of debate included the suitability of older diagnostic CT scans, the appropriateness of sim-free RT for neck or cervical spine anatomic targets, and the compatibility of sim-free workflows with IMRT/VMAT techniques. Further research is needed to refine scan eligibility criteria, validate dosimetric accuracy, and assess workflow feasibility in diverse clinical settings. Obtaining the informed consent required for clinical trial enrollment can be challenging in urgent situations where rapid treatment is needed for symptom relief because of the time required to communicate and ensure understanding of trial information. However, reviewing emerging clinical evidence and, when possible, conducting formal prospective studies will be valuable for guiding broader implementation and expanding access to efficient, patient-centered RT.

Disclosures

David A. Palma receives research funding from the Ontario Institute for Cancer Research, royalties from UptoDate.com and a consultant role with equity with Need Inc, unrelated to the current study. Vivian S. Tan holds a leadership/fiduciary role on the Canadian Radiation Oncology Foundation Board, unrelated to the current study. Simon Boeke has received speaker’s honoraria and travel expenses from Merck AG and Elekta AG, unrelated to the current study. Ashwin Shinde has received payment as the Disease Site Editor for RadOnc Questions, unrelated to the current study. Eva Versteijne receives research funding from Varian, a Siemes Healthineers Company (Palo Alto, CA, USA), unrelated to the current study.

All other authors declare no conflicts of interest.