Adaptive Radiotherapy — Board Review Summary
PART I — DEFINITION, RATIONALE, AND HOW ONLINE ART WAS BUILT
Quick Definition: What Is Online ART?
Online adaptive radiotherapy is the process of adjusting the daily treatment plan in response to observed changes in target and/or organ-at-risk geometry while the patient remains on the treatment table. In practical terms, it condenses what would traditionally be 1-2 weeks of planning work into a single session lasting roughly 15-90 minutes.
Why ART Emerged from the Limitations of Standard IGRT
- Static RT plans cannot account for substantial day-to-day anatomic variation.
- Upper abdominal and pelvic sites are especially vulnerable because OAR position may change enough to invalidate an otherwise acceptable plan.
- ART therefore aims to preserve target coverage while avoiding unplanned OAR overdosing.
Three Core Pieces Required to Make ART Possible
| Requirement | Why it matters | Board-style takeaway |
| 1. On-board imaging of adequate quality | Must visualize target and OAR changes reliably enough to justify re-planning | No meaningful adaptation without anatomy you can trust |
| 2. Paired TPS for rapid re-contouring and re-optimization | Adaptation has to happen in real time while the patient remains on the table | Speed and contouring logic are part of the technology, not an afterthought |
| 3. Online pre-treatment QA | Every adapted plan still needs to be safe to deliver | New workflows still require old-fashioned rigor |
How Platforms Were Vetted: Stepwise Development Path
| Step | MR-guided example | Key result |
| Imaging study | Original low-field 0.35T MRgRT platform | No imaging dose, 4 or 8 f/s, automatic gating, TruFISP-based imaging; MRI visualized soft tissue better than standard C-arm CBCT |
| In silico workflow / TPS testing | Simulated delivery of 50 Gy / 5 fx SMART | Large daily anatomic shifts observed; predicted 63% daily ART need to protect OARs; adaptation predicted to reverse 100% of those OAR violations |
| Pilot / Phase I trial | Adaptive upper-GI SMART pilot | Demonstrated feasibility and safety before broader efficacy testing |
Board pearl: the slide deck explicitly argues that ART platforms should be developed in a repetitive, stepwise fashion: imaging validation, dosimetric/in silico testing, then feasibility and safety trials. That sequence is itself testable.
Current Clinical Platform Landscape
| Platform type | Clinical use timeline | Practical note |
| MR-guided ART | In clinical use since 2014 | Best soft-tissue visualization and integrated motion management |
| CT-guided ART | In clinical use since 2018 | Typically higher throughput and easier integration into conventional IGRT-heavy workflows |
| C-arm linac adaptive | FDA 510(k) cleared in January 2026 | Important because C-arm linacs are the most globally common treatment platform |
PART II — CLINICAL PILLAR #1: SAFE DOSE ESCALATION
Phase I Adaptive SMART: Upper GI / LAPC Feasibility
Phase I adaptive upper-GI SMART pilot (example LAPC, N = 10) established that online adaptive SBRT was safe and feasible. The key result was that the initial plan violated OAR constraints in 70 of 95 fractions, and online ART resolved all of those violations, with 0 grade 3+ acute GI toxicities.
SMART Pancreas Phase II Trial
| Feature | Details |
| Population | Borderline resectable / locally advanced pancreas |
| Platform / concept | MRI-guided online ART |
| Dose | 50 Gy / 5 fx (BED10 = 100 Gy), subject to strict OAR constraints |
| Primary endpoint | CTCAE grade 3+ definitely related GI toxicity within 90 days |
| Primary result | Zero definitely treatment-related grade 3+ toxicities; primary endpoint met |
| 2-year local control | Nearly 80% overall; about 70% in LAPC; >90% in BR disease that proceeded to resection |
| 2-year OS context | BR SMART about 60.1%; LAPC SMART about 34.3% |
Why SMART matters: it is the clearest example in the deck that ART is not just about convenience; it is about making an otherwise aggressive, ablative prescription deliverable without unacceptable GI toxicity.
Current and Ongoing GI / Hepatobiliary Adaptive Trials
| Trial | Disease / question | Adaptive takeaway |
| ARTIA | CT-guided ART for BR/LAPC | Single-arm phase II evaluating whether CTgART reduces acute and late toxicity compared with historical controls |
| NRG GI-011 "LAP-100" | Phase III dose-escalated RT in LAPC | Allows both CT- and MR-guided online ART |
| NRG GI-012 "HELIO-RT" | HCC with macrovascular invasion, IO +/- SBRT | Also allows both CTgART and MRgART |
HCC Example: Why Daily Adaptation Matters
In the HCC HELIO-RT example, a day-1 plan met stomach and bowel constraints, but by day 2 the large bowel constraint of < 33 Gy to 0.5 cc was violated, with parts of bowel exceeding 40 Gy. Post-adaptation, bowel was protected while target coverage was maintained. This is the board-style illustration of why ART matters for mobile upper-abdominal anatomy.
PART III — CLINICAL PILLAR #2: MARGIN REDUCTION
DARTBOARD Phase II Randomized Trial
| Feature | Details |
| Core strategy | 1 mm PTV margin with daily ART instead of 5 mm standard IGRT margin |
| Dosimetric effect | Reduced parotid and submandibular gland dose, smaller targets |
| Primary endpoint | Xerostomia Questionnaire (XQ) at 1 year |
| Workflow timing | CBCT to end of RT: 33 min; door-in to door-out: 39.4 min; MD contour time: 12.6 min; MD at console: 22 min |
DARTBOARD Toxicity and Oncologic Outcomes
| Outcome | IGRT | DART | Comment |
| Dermatitis grade 2+ | 31% | 8% | Statistically lower with adaptive treatment |
| Mucositis grade 2+ | 92% | 75% | Trend favoring ART |
| Dysphagia grade 2+ | 81% | 75% | No clear difference |
| Dysphagia grade 3+ | 19% | 8% | Numerically lower with ART |
Safety signal: median follow-up was about 18 months, there were no marginal recurrences, and the single local recurrence was in-field and salvaged. That is exactly the kind of disease-control reassurance required for margin-reduction strategies.
PART IV — CLINICAL PILLAR #3: EXTREME HYPOFRACTIONATION, SAFELY
SMART ONE Phase II Trial
| Feature | Details |
| Patients / lesions | 30 patients, 32 sites treated |
| Platform | 0.35T MR-Linac |
| Design goal | Single-fraction SBRT with grade 3+ toxicity <15% |
| Adaptive use | Online ART used in 17/32 sites (53%) |
| Primary safety result | Acute grade 3+ toxicity 10%; late grade 3+ toxicity 0%; zero definitely or probably related to SBRT |
| Volumes | Median GTV 3.7 cc; median PTV 15.8 cc |
SMART ONE Prescription Table
| Target location | Prescription | BED10 |
| Lung | 30-34 Gy x 1 | 120-149.6 Gy |
| Liver | 35-40 Gy x 1 | 157.5-200 Gy |
| Adrenal | 25 Gy x 1 | 87.5 Gy |
| Abdominal / pelvic LN | 25 Gy x 1 | 87.5 Gy |
| Kidney | 25 Gy x 1 | 87.5 Gy |
| Pancreas | 25 Gy x 1 | 87.5 Gy |
MOMENTUM Registry and the Prostate Model
The prospective MOMENTUM registry across 1.5T MRgART sites now contains over 8,000 patients, about 98,000 MRI scans, and 33,800 dose plans. One registry-generated hypothesis is that MRgART may lower toxicity in 5-fraction prostate regimens relative to historical standards, whereas that advantage is less obvious with conventional fractionation.
HERMES Randomized Phase II Trial
| Feature | Arm 1 | Arm 2 |
| Population | Intermediate-risk / lower high-risk prostate cancer |
| Adaptive strategy | Daily ART for all patients, 3 mm PTV for all |
| Dose | 36.25 Gy / 5 fx | 24 Gy / 2 fx |
| Boost | CTV to 40 Gy | GTV SIB to 27 Gy / 2 fx |
| Primary endpoint | Acute grade 2+ GU toxicity, benchmarked against a 62% historical rate (PACE-B based) |
Board pearl: HERMES is not just a prostate trial; it is a proof-of-concept that ART may help push SBRT toward fewer fractions while maintaining acceptable toxicity. The slide explicitly frames the next step as the larger SABR-DUAL phase II/III effort.
ARCHER (NRG GU-015) — Phase III Bladder Trial
| Feature | Details |
| Question | Can 5-fraction adaptive SBRT be non-inferior to 20-fraction hypofractionated RT for bladder-intact event-free survival? |
| ART requirement | Mandatory ART, CT- or MR-guided |
| Systemic therapy | Concurrent chemo allowed: weekly cisplatin, gemcitabine, or 5-FU + mitomycin-C |
| Primary endpoint | 3-year BI-EFS with a 10% non-inferiority margin (HR < 1.32) |
| Sample size | N = 486 |
| PTV margins shown on slide | Adaptive arm approximately 7 mm vs conventional arm 1 to 1.5 cm |
PART V — IMPLEMENTATION: HOW TO BUILD AN ART PROGRAM
ART Has Reached the Mainstream
The deck emphasizes that online ART is no longer a niche physics project. There are now published safety white paper guidelines addressing quality, personnel, and workflow, and ART has already entered phase III cooperative-group trials.
The Roadmap to Launch
| Phase | Core tasks |
| Pre-work | Choose platform and supporting technologies; identify champions; align with department strategy |
| Team preparation | Train the whole team; build workflows and procedures by disease site / indication |
| Go-live and beyond | Peer support, quality assurance, continuing education, and program growth |
Platform Selection: MR-Linac vs CT-Guided ART
| Platform | Strengths | Tradeoffs |
| MR-Linac | Real-time MR guidance; cine MRI gating; continuous motion management; excellent soft-tissue visualization | Slower throughput and setup limitations for standard IGRT-style workflow |
| CT-guided ART | Built on high-throughput CBCT platforms; broader day-to-day clinical versatility; lower capital cost | Motion management is less intrinsically robust and may require paired technologies (ABC, surface guidance, etc.) |
Where to Start: Choose High-Yield Champion Sites
The speaker's own center started with pancreas SBRT and prostate SBRT with nodal SIB, explicitly because they had the volume and the physician/physics champions to support them. The board-style implementation lesson is simple: volume does not appear from thin air; ART programs start where there is real clinical pull and committed local expertise.
Operational Pieces That Need to Exist Before Go-Live
- Coverage model: choose it deliberately and expect it to evolve.
- Team-wide training: ASTRO task-force guidance recommends broad multidisciplinary training, not just physician and physics education.
- Adaptive school curriculum: system overview, clinical rationale, planning templates, imaging, case-based ART, documentation, billing, and even retention quizzes.
- Handoff design: communicate goals of ART, what to re-contour, what constraints matter most, key anatomy issues, setup needs, and whether the patient may need analgesics or anxiolytics.
- Documentation: the ASTRO white paper includes practical templates.
- Ongoing quality culture: regular peer review and continuing education remain essential after launch.
PART VI — FUTURE DIRECTIONS: SIM-FREE WORKFLOWS AND NEW PLATFORMS
FAST METS: Direct-to-Unit / Sim-Free Palliative ART
FAST METS treated 47 patients using a simulation-free ETHOS-based adaptive workflow for palliative treatments. Diagnostic CT was used to pre-plan, symptom review was performed before arrival, and final plan recalculation / optimization occurred on the treatment unit. Average total treatment time, including consult, was 85 minutes, with average on-couch adaptive time of 30 minutes.
ONE-STOP Phase II Trial
| Feature | Details |
| Concept | Simulation-free, one-fraction CT-guided stereotactic adaptive treatment for early-stage NSCLC or oligometastatic lung lesions |
| Primary endpoint | Feasibility of the sim-free workflow in at least 70% of selected patients |
| Prescription target | Low-risk peripheral lung lesions suitable for 34 Gy x 1 |
| Inclusion highlights | Target <5 cm, >2 cm from proximal bronchial tree / mediastinum, estimated SI motion ≤1 cm |
| Status shown on slide | 10 / 10 patients successfully accrued and treated |
Board pearl: for the purpose of ONE-STOP, a backup conventional sim CT was still obtained for dosimetric comparison and as a rescue plan if direct-to-unit treatment could not be completed. "Sim-free" does not mean "no safety net" in early clinical development.
Adaptive on Additional Systems
- C-arm linac adaptive RT: first commercial platform announced in 2024, CE mark granted 9/16/24, FDA 510(k) cleared 1/19/2026.
- Proton online ART: the deck highlights one of the first online ART proton cases at WashU in June 2024, within the PARTy clinical trial framework.
- Functional MR / biologic guidance: the lecture flags MRgART and BgART as the next layer of adaptation beyond purely geometric replanning.
Where ART Is Likely to Land Long-Term
The speaker's long-term view is that ART becomes a versatile, common solution rather than a niche technique: not used in every case, but applicable in most modalities and treatment styles, much like IMRT, VMAT, and SBRT eventually became.
CROSS-CUTTING HIGH-YIELD POINTS
- Definition: online ART = same-day re-contouring / re-optimization on the treatment table.
- Three prerequisites: adequate on-board imaging, fast paired TPS, and online pretreatment QA.
- Development logic matters: imaging study → in silico workflow testing → pilot / phase I → larger prospective trials.
- MRgRT low-field platform: 0.35T, no imaging dose, 4 or 8 f/s, automatic gating, better soft-tissue visualization than standard C-arm CBCT.
- In silico pancreas SMART: daily ART predicted to be needed in 63% of fractions to protect OARs.
- Phase I adaptive upper-GI SMART: OAR violations in 70/95 fractions, all resolved by ART, with 0 grade 3+ acute GI toxicity.
- SMART pancreas phase II: 50 Gy / 5 fx, BED10 100 Gy, with zero definitely related grade 3+ GI toxicity.
- HELIO-RT liver example: large bowel constraint on the slide was < 33 Gy to 0.5 cc; adaptation corrected a day-to-day violation.
- DARTBOARD: margin reduction from 5 mm to 1 mm lowered dermatitis and produced no marginal recurrences.
- SMART ONE: 1-fraction SBRT on 0.35T MR-Linac; ART used in 53% of sites; 10% acute grade 3+, 0% late grade 3+, and zero definitely/probably related events.
- SMART ONE key doses: lung 30-34 Gy x 1, liver 35-40 Gy x 1, adrenal / kidney / pancreas / abdominal-pelvic LN 25 Gy x 1.
- HERMES: prostate daily ART with 36.25 Gy / 5 or 24 Gy / 2 regimens, all with 3 mm PTV.
- ARCHER: phase III bladder trial; mandatory ART; 5-fx SBRT vs 20-fx hypofractionated RT; N = 486.
- ART is now mainstream: safety white paper exists, and cooperative-group phase III trials are open.
- Platform tradeoff: MR offers superior visualization and motion management; CT-guided adaptive offers higher throughput and versatility.
- Implementation lesson: start with high-volume disease sites and true local champions.
- Sim-free future: FAST METS and ONE-STOP show that ART can compress consult + planning + treatment into a single visit for selected indications.
- C-arm adaptive and proton adaptive mark the expansion of ART beyond ring-gantry MR/CT adaptive systems.
CONSOLIDATED DOSE AND WORKFLOW TABLE
| Use case / trial | Dose / margin / workflow | Why it matters |
| Upper-GI / pancreas SMART development | 50 Gy / 5 fx | Canonical ablative adaptive GI example |
| SMART pancreas phase II | 50 Gy / 5 fx | Ablation with strict OAR governance and favorable early safety |
| HCC adaptive example | Large bowel constraint < 33 Gy to 0.5 cc | Illustrates day-to-day OAR violation rescue |
| DARTBOARD | 1 mm PTV vs 5 mm | Margin reduction as a therapeutic-index tool |
| SMART ONE lung | 30-34 Gy x 1 | Extreme adaptive hypofractionation |
| SMART ONE liver | 35-40 Gy x 1 | Single-fraction adaptive liver SBRT |
| SMART ONE adrenal / pancreas / kidney / A/P LN | 25 Gy x 1 | Single-fraction adaptive multi-site use |
| HERMES arm 1 | 36.25 Gy / 5 fx | Adaptive prostate SBRT reference arm |
| HERMES arm 2 | 24 Gy / 2 fx | More extreme adaptive prostate hypofractionation |
| ONE-STOP | 34 Gy x 1 | Simulation-free direct-to-unit lung workflow |
| FAST METS | Average on-couch adapt time 30 min | Demonstrates palliative sim-free feasibility |
KEY LANDMARK TRIALS (memorize)
| Trial / study | Topic | One-line takeaway |
| Noel et al. 2015 | MRgRT imaging quality | Low-field MR provided superior soft-tissue visualization vs standard C-arm CBCT |
| Henke et al. 2016 | In silico SMART | Predicted daily ART need in 63% of fractions and reversal of all modeled OAR violations |
| Henke et al. 2018 | Phase I adaptive upper-GI SBRT | Feasible and safe; 70/95 fractions needed adaptation to fix OAR problems |
| Parikh et al. 2023 | SMART pancreas phase II design | 50 Gy / 5 fx ablative adaptive pancreas platform trial |
| Chuong et al. 2024 | SMART pancreas outcomes | About 80% 2y LC overall with favorable BR/LAPC survival context |
| DARTBOARD (Scher et al. 2025) | Margin reduction | 1 mm PTV adaptive strategy reduced dermatitis and produced no marginal failures |
| SMART ONE (Chuong et al. 2025) | Extreme adaptive hypofractionation | Single-fraction MR-guided SBRT was feasible with acceptable toxicity |
| MOMENTUM registry | 1.5T MRgART | Large prospective registry now generating site-specific hypotheses and trial ideas |
| HERMES (Cooper et al. 2025) | Adaptive prostate SBRT | Randomized phase II platform for 5-fx vs 2-fx daily adaptive prostate treatment |
| NRG GU-015 ARCHER | Adaptive bladder preservation | Phase III non-inferiority test of 5-fx adaptive SBRT vs 20-fx hypoRT |
| FAST METS (Nellison et al. 2023) | Sim-free palliative ART | Showed direct-to-unit adaptive workflows can be practical in selected palliative cases |
| ONE-STOP | Sim-free lung SBRT | Prospective test of direct-to-unit 34 Gy x 1 CT-guided adaptive lung treatment |
| ARTIA | CT-guided pancreas ART | Extends adaptive evaluation beyond MR-guidance for BR/LAPC |
| NRG GI-011 / GI-012 | Phase III GI / liver adaptive integration | Marks ART's transition into cooperative-group phase III trial design |