In vivo and predictive interplay evaluation methodology for lung and esophageal cancer patients treated in free breathing with IMPT
Giorgio Cartechini, Esther Kneepkens, Gloria Vilches-Freixas, Indra Lubken, Marije Velders, Sebastiaan Nijsten, Mirko Unipan, Ilaria Rinaldi · 2025
Read In vivo and predictive interplay evaluation methodology for lung and esophageal cancer patients treated in free breathing with IMPT by Giorgio Cartechini, Esther Kneepkens, Gloria Vilches-Freixas, Indra Lubken, Marije Velders, Sebastiaan Nijsten, Mirko Unipan, Ilaria Rinaldi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-17432.txt.
Introduction
Giorgio Cartechini and seven co-authors tackle a dosing risk in proton therapy in a 2025 arXiv paper. Pencil beam scanning delivers protons in fine spots, but a breathing patient's motion can smear the dose. The team built a way to predict that interplay effect for lung and esophageal cancer patients who breathe freely during treatment.
Their predictive model, built on synthetic breathing, runs inside the RayStation planning system. An in vivo check uses each patient's real breathing traces and machine log files. Both routes were benchmarked against the FRED Monte Carlo engine, and they agreed closely. Two clinical cases treated without rescanning showed the dose settling after five fractions, under breathing, range, and setup errors. The authors pitch the frame as an automated, clinic-ready check that can run before treatment starts.
Catalog record
| CREATOR | Giorgio Cartechini, Esther Kneepkens, Gloria Vilches-Freixas, Indra Lubken, Marije Velders, Sebastiaan Nijsten, Mirko Unipan, Ilaria Rinaldi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-17432
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
In vivo and predictive interplay evaluation methodology for lung and esophageal cancer patients treated in free breathing with IMPT Giorgio Cartechini, Esther Kneepkens, Gloria Vilches-Freixas, Indra Lubken, Marije Velders, Sebastiaan Nijsten, Mirko Unipan, Ilaria Rinaldi Abstract Purpose: Pencil beam scanning proton therapy is sensitive to respiratory motion, leading to potential dose inhomogeneities due to interplay effects. We developed and validated a predictive framework to assess interplay and motion robustness in lung and esophageal cancer patients treated under free-breathing conditions. Methods: A synthetic-breathing-based predictive model was implemented in the RayStation treatment planning system (TPS) and validated against an in vivo approach using patient-specific respiratory traces and machine log files. Both were benchmarked against the Monte Carlo engine FRED. To demonstrate the methodology, the framework was applied to two clinical cases treated on the Mevion S250i system without rescanning. Dose accumulation incorporated respiratory phase, range ( ± \pm 3%), and setup ( ± \pm 5 mm) uncertainties. Results: Excellent agreement was observed between TPS and FRED ( < < 1% mean dose difference) and between predictive and in vivo models ( < 2 <2 % across DVH metrics). Cumulative dose distributions for the primary CTV converged after five fractions, confirming robust delivery. Conclusion: This automated, clinically integrated framework enables pre-treatment prediction and in vivo validation of interplay and motion robustness in PBS plans. Preliminary results support its clinical utility, especially for hypofractionation and targets with large motion ( > > 2 cm). A larger cohort study is ongoing and will be reported separately. roton therapy, interplay, motion robustness, in vivo dosimetry, Monte Carlo, lung cancer, esophageal cancer. 1 Introduction Pencil beam scanning (PBS) proton therapy, particularly in the form of intensity-modulated proton therapy (IMPT), is the most advanced proton delivery technique, offering excellent dose conformity and enhanced sparing of surrounding healthy tissues [ Chang et al., 2017 ] . However, its layer-by-layer, spot-scanning delivery is inherently sensitive to respiratory motion, particularly in thoracic and upper abdominal tumors [ Chang et al., 2014 ] . This motion sensitivity can lead to the so-called interplay effect, a temporal mismatch between tumor motion and beam delivery, resulting in dose inhomogeneities such as cold and hot spots within the target volume [ Bert and Durante, 2011 ] . Several studies have investigated the magnitude and clinical implications of this effect. Li et al. [ Li et al., 2014 ] quantified interplay-induced deviations in stage III lung cancer patients, showing that although fractionation mitigates some heterogeneity, significant variations can still occur in individual fractions. Complementary, Monte Carlo simulations by Grassberger et al. [ Grassberger et al., 2015 ] further demonstrated that small spot sizes and rapid energy switching can exacerbate dose non-uniformity, particularly in single-fraction deliveries. To mitigate the impact of respiratory motion, several strategies have been investigated, including 4D robust optimization [ Engwall et al., 2018 , Spautz et al., 2023 , Bengtsson et al., 2025 ] and various techniques, including rescanning [ Rana and Rosenfeld, 2021 , Knopf et al., 2011 ] . Despite their effectiveness, these methods can be complex and time-consuming, limiting their routine clinical implementation. As a result, more streamlined and time-efficient approaches are often preferred. At our institution, patients with lung [ Taasti et al., 2025b ] and esophageal cancers [ Canters et al., 2024 ] are treated in free-breathing using 3D robust optimization based on an Internal Target Volume (ITV), defined as the union of Clinical Target Volume (CTV) contours across all respiratory phases. This ITV-based strategy, combined with planning on an average 4DCT and 3D and 4D robust evaluations [ Taasti et al., 2021 ] , incorporates motion into the optimization while preserving clinical feasibility. Treatments are delivered using the Mevion Hyperscan S250i system (Mevion Medical Systems, Littleton, MA, USA) [ Vilches-Freixas et al., 2020 ] , which accelerates a fixed energy beam of around 230 MeV. The beam energy is modulated via eighteen range shifter plates within the nozzle. This configuration results in relatively large spot sizes, advantageous characteristics for mitigating the impact of motion in PBS delivery [ Rana and Rosenfeld, 2022 ] . The so-called ‘4D evaluation’ framework implemented at our institution to assess plan robustness [ Taasti et al., 2021 ] accounts for anatomical variations across the breathing cycle but does not incorporate the temporal relationship between beam delivery and respiratory motion. As a result, it does not fully capture the dosimetric consequences of the interplay effect. To address this limitation, we propose a comprehensive framework for evaluating both interplay effects and motion robustness in PBS plans for lung and esophageal cancer patients treated under free-breathing conditions. We developed a predictive method to simulate interplay effects over the entire treatment course using synthetic respiratory scenarios, enabling evaluation to be performed prior to the start of treatment. The model is validated against an in vivo approach that integrates fraction-specific respiratory traces and machine log files recorded during treatment. The evaluation framework is fully automated, implemented within the clinical treatment planning system, and further validated using the independent Monte Carlo dose engine FRED [ Schiavi et al., 2017 ] , clinically validated in our proton facility [ Gajewski et al., 2020 ] . This article focuses on presenting the methodology for both predictive and in vivo evaluations; extension of this work to a larger patient cohort is currently underway and will be the subject of a separate publication. 2 Materials and Methods 2.1 Interplay effect evaluation: Predictive and in vivo methods The interplay evaluation was conducted using two approaches: predictive and in vivo . The predictive approach is designed to be performed before the start of the treatment, without prior knowledge of the patient’s breathing trace. In contrast, the in vivo approach retrospectively evaluates plan robustness to organ motion by incorporating the patient-specific breathing traces per fraction acquired during treatment and the corresponding machine log files recorded after each fraction. In vivo and predictive interplay evaluations were performed using two Monte Carlo dose engines: FRED v3.76.4 [ Schiavi et al., 2017 ] , and the clinical treatment planning system (TPS) RayStation (RaySearch Laboratories, Stockholm, Sweden) version 12B (Monte Carlo v5.4). The interplay workflow is illustrated in Figure 1 . The 4DCT, breathing traces, and a machine log file are used as input for both Monte Carlo systems. The 4DCT represents the anatomical model of the patient’s geometry. It consists of eight 3D CTs corresponding to different respiratory phases: 0%, 25%, 50%, 75%, and 100% inhale, and 25%, 50%, 75% exhale. The ANZAI belt system (Anzai Medical Co. Ltd, Tokyo, Japan) was used to measure the patient’s respiratory amplitude during the CT scan. The 3D volumetric images were sorted based on the acquired breathing signals to assign them to the appropriate respiratory phases [ Lu et al., 2006 , Li et al., 2006 ] . The in vivo approach exploits the fraction-specific breathing traces acquired by the C-RAD system [ Walter et al., 2016 ] installed in the treatment room. Since the predictive approach is intended to be performed before treatment is delivered and patient-specific breathing traces are not available, we generated 24 synthetic breathing curves based on a sinusoidal function model, varying the breathing period (2 s, 3.5 s, and 5 s) and the starting phase according