Multicentre dosimetry audit in the European randomized phase III study of neoadjuvant chemoradiotherapy using proton versus photon radiotherapy in locally advanced oesophageal cancer

Graphical abstract

Highlights

• •Agreement within 3% on dosimetry audits for proton and photon therapy centres.
• •Consistent dosimetry practices are essential for patient safety in multi-centre trials.
• •External on-site dosimetry audits confirm accuracy of dose delivery.

Abstract

This study presents the results of a multicentre external dosimetry audit conducted as part of the PROTECT trial, involving nine European radiotherapy centres delivering proton and photon treatments. A dosimetry equipment vendor performed beam output audits at participating sites, verifying the absorbed dose-to-water measurements. Measurements used reference conditions based on the IAEA TRS-398, with ionization chambers traceable to primary standards. The audit compared local centre measurements to independent assessments, finding all within the acceptable 3% agreement threshold. The study is part of the successful implementation of a comprehensive, multicentre QA programme in the PROTECT trial.

1. Introduction

Oesophageal cancer treatment with chemoradiotherapy followed by surgery carries substantial risk of radiation-induced toxicity to nearby organs at risk. Proton therapy (PT) can reduce dose to normal tissues compared with conventional photon radiotherapy (XT) while maintaining target coverage [1] and may therefore reduce treatment-related morbidity and potentially improve clinical outcomes [1], [2], [3], [4], [5], [6], [7], [8], [9]. In a former phase IIB randomized trial, the total toxicity burden was 2.3 times higher for XT compared to PT, highlighting the potential benefits of proton therapy [10]. Robust evidence on clinical benefits is still emerging, and differences in dose delivery across centres could confound multi-centre comparisons.

Reliable, consistent dosimetry across participating centres is essential for the scientific validity and patient safety of multi-centre radiotherapy studies [11], [12]. External beam output audits and standardized radiotherapy quality assurance (RTQA) procedures reduce inter-site dose variability, lower the rate of protocol deviations, and are associated with improved trial data quality and patient outcomes. While well-established audit frameworks exist for high-energy photon and electron beams [13], [14], [15], [16], [17], [18], [19], comprehensive, commissioned audit systems for proton therapy are still being developed and implemented.

This study evaluates the accuracy and consistency of beam output measurements across multiple European radiotherapy centres participating in a multicentre comparative trial of proton versus photon radiotherapy. We report results from external on-site reference dosimetry audits and summarize outcomes from independently conducted audits to assess whether participating centres meet accepted dose-agreement thresholds.

2. Materials and methods

The centres participating in this study were sites enrolled in the multicentre randomized phase III trial comparing proton versus photon radiotherapy for oesophageal cancer (PROTECT; PROton versus photon Therapy for Oesophageal Cancer – a Trimodality strategy; NCT05055648) [4]. To ensure consistency of treatment delivery across sites and preserve the scientific validity of the trial endpoints, a comprehensive radiotherapy quality assurance (RTQA) programme was implemented, which is crucial in reducing major protocol deviations and ensuring high-quality clinical trial results [5], [6], [7], [8], [9]. Protocol deviations are associated with increased risk of treatment failure and mortality [20] and are more frequent at sites with fewer patient recruitments [7], [21].

Key elements of the RTQA programme included mandatory pre-trial activities (benchmark delineation of target volumes and organs at risk [22], and treatment planning benchmark cases [23]), prospective individual case reviews for the first two patients and every fifth patient thereafter at each site, facility questionnaires to document local treatment workflows, site visits, and beam output audits (BOAs), and patient-specific QA at all participating sites [24]. These measures were intended to reduce protocol deviations and harmonize practice across participating centres.

All participating centres were required to perform an external BOA once for each modality (XT or PT) used for trial patients. For XT, only BOAs accepted by the European Organization for Research and Treatment of Cancer (EORTC) [25] were acceptable for enrolment; for PT, local dosimetry audits will be accepted. The resulting report is to be submitted to the EORTC. This process is essential to ensure that the correct dose is consistently administered to all patients participating in a clinical trial [26], [27]. A BOA is an external dosimetry audit in which the dose delivered by a radiation beam is independently verified under reference conditions (e.g. 10 × 10 cm², 5/10 cm in water). Dosimetry audits can be performed by different providers (e.g., vendor on-site audits using calibrated ionization chambers, TLD/alanine postal audits, or national laboratory comparisons). While well-established formal audit systems exist for high-energy photon and electron beams, equivalent structured systems for clinical proton beams are in development; therefore, the RTQA group accepted a range of established independent audits for proton centres.

The IBA Secondary Standard Dosimetry Laboratory (SSDL, Schwarzenbruck, Germany) offered to perform on-site BOAs for centres enrolled in the trial by May 2023. Four PT facilities and five XT centres participated in these vendor-organized on-site measurements, while nine sites conducted BOAs independently of the trial collaboration vendor.

Measurements were initially performed by the local team at each centre using their own dosimetry equipment and were subsequently repeated by the auditing team from the dosimetry equipment vendor using ionization chambers traceable to the German PTB Primary Standards Dosimetry Laboratory (PSDL). To minimize the influence of potential beam instabilities, results are reported as the average dose per monitor unit based on a minimum of three repetitions.

For protons, six beams were used: three monoenergetic (100, 170, and 220 MeV) 10x10 cm² fields with a measurement depth of 2 cm, and three 10 × 10 × 10 cm³ modulated spread-out Bragg peak (SOBP) beams measured at the SOBP midpoint at depths of 10, 17 and 25 cm. Vender-organized measurements were performed at gantry 90° (or 270°) with a Blue Phantom PT water phantom (IBA, Dosimetry) and a plane-parallel ionization chamber PPC05 (IBA) with a calibrated Dose-1 or Dose-X electrometer (IBA). The PPC05 (SN475) was used at −500 V or −300 V for pulsed or quasi-continuous proton beams, respectively. The beam quality correction factors (kQ) were 1.003 for all energies; values from Palmans et al. [28] were used by local medical physicists.

For XT, the vendor-organized BOA followed the IAEA TRS-398 (2000) [29] Code of Practice (CoP), using a 10 × 10 cm² field at 100 cm source-to-surface distance. Vendor measurements were performed in a vertical beam setup using an WP1D water phantom (IBA Dosimetry) with a FC65-G (SN1224) Farmer ionization chamber coupled with either a Dose-1 or Dose-X electrometer (IBA Dosimetry). Local teams used the same conditions but varied in equipment: proton sites used plane parallel chamber, PPC05 (IBA), Semiflex Type 31,013 (PTW) Roos plane-parallel, type 34,001 (PTW) chambers, with electrometers including Dose-1 (IBA) and UNIDOS (PTW); photon sites employed Farmer FC65-G, Semiflex type 31003, Farmer 30013 and Farmer NE2571 (Phoenix Dosimetry Ltd., Sandhurst, UK) chambers, with Dose-1, UNIDOS and SNC PC (Sun Nuclear, USA) electrometers.

Nine sites did not participate in the dosimetry equipment vendor-organized audit and conducted their own BOAs independently: two XT and one PT sites used IROC with TLDs [30]; one PT site used Alanine pellets with NPL [31]; one PT site employed Farmer chamber measurements certified by their national centre [32]; one XT and one PT site used TLDs with ESTRO-Equal [16]; two PT sites compared their dosimetry with another PT site using ESTRO-Equal. Both TLDs and pellets are received from and returned to the audit organization. For centres that conducted BOAs independently of the vendor, a range of established external audit programs was accepted; these programs differ in methodology (e.g., TLD, alanine, chamber-based, national-laboratory comparisons) and thus have measurement characteristics that are not strictly identical to the vendor on-site audits. Nonetheless, all audits were traceable to national/primary standards and reviewed by the RTQA group for EORTC compliance.

The ratio of the absorbed dose-to-water reported by each centre to the corresponding dose measured by the vendor (or reference audit) was used as the primary metric for inter-centre comparison.

3. Results

Fig. 1 (top) displays results from the mono-energetic and modulated proton beams audits. Across the four centres, consistent agreement was observed between the reported ratios and measurements, maintained across various beam qualities. Table 1 provides the ratios per centre with standard deviations.

Fig. 1.

Ratio between absorbed dose-to-water reported by the institution (medical physicist undertaking the measurements) and dose measured by the dosimetry equipment vendor in monoenergetic (top) and modulated (middle) proton beams, and photon beams (bottom).

Table 1.

Ratio reported/measured obtained for each centre: C1–4 for protons and C1–5 for photons. In the bottom are stated BOAs by other enrolled centres. NPL: National Physical Laboratory, IROC: Imaging and Radiation Oncology Core – MD Anderson.

	Monoenergetic proton beam	Modulated proton beam
C1	1.009 ± 0.001	0.993 ± 0.007
C2	1.022 ± 0.006	1.021 ± 0.001
C3	1.009 ± 0.004	1.007 ± 0.003
C4	0.998 ± 0.002	0.994 ± 0.002
	Photon beam
C1	0.980 ± 0.005
C2	0.994 ± 0.000
C3	0.987 ± 0.002
C4	1.014 ± 0.002
C5	0.992 ± 0.002
Other Proton BOAs
	Institution	Ratio [range]
C1	NPL	1.019 [1.01–1.03]
C2	Equal-ESTRO	1.013 [1.01–1.02]
C3	Equal-ESTRO	1.018 [1.02–1.02]
C4	Equal-ESTRO	1.004 [1.00–1.01]
C5	IROC	0.990 [0.98–1.00]
C6	IROC	1.010 [1.00–1.02]
Other Photon BOAs
	Institution	Ratio [range]
C1	Equal-ESTRO	0.999 [0.98–1.02]
C2	IROC	1.008 [0.95–1.03]
C3	IROC	0.992 [0.97–1.02]

Except for centre C2, the agreement remained within ±1%. The discrepancy noted for C2 (2.2%) was attributed to a measurement depth of 2.3 cm instead of the prescribed 2 cm. When accounting for this, agreement improved to ratios of 1.006 (±0.004) and 1.011 (±0.001), as illustrated by the open symbols in Fig. 1.

Fig. 1 (bottom) presents results from the photon beam audit. All five centres showed consistent agreement between local measurements and vendor measurements across beam qualities (Table 1), with all centres within 2%. Notably, while centre C4 did not adhere to the IAEA TRS-398 (2000) [29] recommendations and relied on a different CoP for reference dosimetry, a closer agreement was observed when considering the differences in the CoP, as shown by the open symbols in Fig. 1 (bottom).

All centres met EORTC requirements for a beam output audit (“All measurements preferably within 3% agreement, 3%–5% is acceptable” [25]). The relative standard uncertainty (k = 2) associated with vendor results was 2.3% and 3.1% for photon and proton measurements, respectively. Overall, for both proton and photon beams, agreement was well within the 3% EORTC acceptance threshold.

All nine sites performing independent BOAs showed agreement within 3% as recommended by EORTC [8] (Table 1, bottom). Centres 2 and 3 used ESTRO-Equal, comparing results with centre 1. In total, eighteen sites conducted BOAs — either vendor-organized or independently — and all were approved for trial participation.

4. Discussion

This paper describes the results of a multicentre external on-site reference dosimetry audit in nine European centres participating in the PROTECT trial, covering both proton and photon radiotherapy. The study demonstrates a high level of agreement in absorbed dose-to-water measurements between the centres and the auditing vendor for both beam types, well within the 3% EORTC threshold [8]. The consistency observed across multiple centres provides confidence in safety and consistency in multi-centre clinical trials.

The increasing complexity of radiotherapy technologies necessitates robust QA measures across treatment centres to ensure safety and effectiveness [11], [15], [18], [26], [33]. Ferreira et al. [16] established the ESTRO‐QUALity assurance network (EQUAL), which uses TLDs to conduct stringent audits of radiotherapy centres across Europe. Approximately 3% of centres exceeded established tolerance levels in beam output measurements, primarily in larger fields and wedged beams, underlining the necessity for continued QA evolution to adapt to new treatments.

The role of BOAs and their impact on reducing patient safety issues was further examined by Clark et al. [11], who demonstrated that BOAs effectively reduce dose variability across treatment centres while ensuring patient safety during implementation of advanced radiotherapy techniques. Taylor et al. [13] analysed anthropomorphic phantoms used for QA across PT centres, finding lower pass rates in more complex phantoms and emphasizing the need for improved motion management strategies and institutional training. A failure modes and effects analysis comparing XT and PT found significant failure risks for both modalities, underscoring that rigorous QA is essential to prevent errors in clinical trials [34].

The challenges of credentialing and ensuring quality across institutions participating in multi-institutional clinical trials are discussed by Ibbott et al. [17], which advocate for rigorous standards for treatment delivery. The accreditation processes ensure improved adherence to clinical protocols and thereby improve the reliability of clinical trial output.

Pettersen et al. [26] have quantified how improved dosimetry can reduce required patient samples in clinical trials, consequently improving their efficiency. The correlation established between dosimetric uncertainties and patient sample sizes illustrates the necessity for robust QA practices that not only protect patient safety but also streamline clinical research processes. Bentzen et al. [27] have assessed the significant clinical implications of beam output variations. The study modelled the effects of small deviations on clinical outcomes, emphasizing that even minor inaccuracies in dose delivery led to substantial variations in treatment efficacy and patient morbidity. This supports the assertion that continuous QA practices, like the EORTC TLD program, are essential for ensuring optimal clinical results, which highlights the need for strict QA programmes and BOAs in clinical trials.

A high RTQA standard is aimed at in the PROTECT trial. However, the RTQA programme must be pragmatic to ensure that centers have the necessary capacity and resources to realistically enter the trial in due time — as many trials have suffered slow accrual due to logistical burden. Permitting flexibility in output audit timing was a pragmatic compromise balancing feasibility with resource availability. Similarly, mandating a BOA under reference conditions rather than a full audit using an anthropomorphic phantom was a pragmatic choice [35], [36].

In summary, all centres’ measurements agreed within 3%, including the nine sites performing independent audits, and all 18 evaluated centres were accepted for trial participation. This study demonstrates the successful implementation of a comprehensive, multicentre QA process in the PROTECT trial. Identifying and correcting discrepancies, while adapting audit practices to new treatment technologies, is vital for improving patient safety and clinical outcomes in both photon and proton radiotherapy.

CRediT authorship contribution statement

Mai L. Ehmsen: Writing – original draft, Visualization, Methodology, Conceptualization. Camilla S. Byskov: Methodology, Conceptualization. Hanna R. Mortensen: Methodology, Conceptualization. Rebecca Bütof: Methodology, Conceptualization. Richard Canters: Methodology, Conceptualization. Gilles Defraene: Methodology, Conceptualization. Karin Haustermans: Methodology, Conceptualization. Maria Fuglsang Jensen: Methodology, Conceptualization. Arturs Meijers: Conceptualization, Methodology. Christina T. Muijs: Methodology, Conceptualization. Ditte S. Møller: Methodology, Conceptualization. Marianne Nordsmark: Methodology, Conceptualization. Pieter Populaire: Methodology, Conceptualization. Gloria Vilches-Freixas: Methodology, Conceptualization. Lars Nyvang: Investigation. Liliana Stolarczyk: Investigation. Sebastian Makocki: Investigation. Sara Broggi: Investigation. Francesco Fracchiolla: Investigation. Andrea Martignano: Investigation. Kenneth Poels: Investigation. Séverine Rossomme: Visualization, Investigation, Formal analysis. Stefan Vasiliniuc: Visualization, Investigation, Formal analysis. Lone Hoffmann: Writing – review & editing, Methodology, Conceptualization.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Authors SR and SV are currently employed at IBA and collected and analyzed all data from the dosimetry vendor audits.

Data availability statement

The data that support the findings of this study are available from the corresponding author, MLE, upon reasonable request.