Vienna, Austria

ESTRO 2023

Session Item

May 14
10:30 - 11:30
Translational outputs from clinical trials
Chloe Brooks, United Kingdom;
Jens Overgaard, Denmark
Proffered Papers
10:30 - 10:40
Are we so different? An international risk comparison of photon vs. proton clinical trial QA
Paige Taylor, USA


Are we so different? An international risk comparison of photon vs. proton clinical trial QA

Paige Taylor1, Catharine Clark2, Joerg Lehmann3, Mitsuhiro Nakamura4, Hugo Palmans5, Lone Hoffman6, Sarah Kelly7, Elizabeth Miles8, Ditte Moller6, Coreen Corning7, Stephen Kry1, Kamal Akbarov9, Markus Stock5, Marianne Aznar10, Rachel Effeney11, Enrico Clementel7, Samir Patel12, Pavel Kazanstev9, Mauro Carrara9, Brendan Healy13

1UT MD Anderson Cancer Center, Imaging and Radiation Oncology Core, Houston, USA; 2NCRI, Radiotherapy Trials QA Group, London, United Kingdom; 3Trans-Tasman Radiation Oncology Group, Radiation Oncology, Waratah, Australia; 4Japan Clinical Oncology Group, Radiation Oncology, Kyoto, Japan; 5MedAustron, Radiation Oncology, Wiener Neustadt, Austria; 6Aarhus University Hospital, Radiation Oncology, Aarhus, Denmark; 7European Organisation for Research and Treatment of Cancer, Radiation Oncology, Brussels, Belgium; 8NCRI, Radiotherapy Trials QA Group , London, United Kingdom; 9International Atomic Energy Agency, Radiation Oncology, Vienna, Austria; 10University of Manchester, Radiation Oncology, Manchester, United Kingdom; 11Queensland Health, Radiation Oncology, Queensland, Australia; 12University of Alberta, Radiation Oncology, Edmonton, Canada; 13Australian Radiation Protection and Nuclear Safety Agency, Australian Clinical Dosimetry Service , Melbourne, Australia

Show Affiliations
Purpose or Objective

The extent and components of quality assurance (QA) in clinical trials may impact the outcome of the trials. The Global QA of Radiation Therapy Clinical Trials Harmonization Group  (GHG), an international consortium of clinical trial QA groups, compared the clinical trial QA for photon versus proton therapy to ascertain where differences might lie and where priorities should be.  

Material and Methods

A list of 25 clinical trial QA tests including phantom tests, dosimetry and image guidance was compiled (Table 1) . Seven participating GHG QA groups provided details on how they performed each test for photon and for proton therapy. Each group then individually performed a QA Failure Effects Analysis (FEA), modeled after Failure Modes and Effects Analysis, for the 25 tests. Each test was given a score for Occurrence, Severity (S), and Detectability. A Risk Priority Number (RPN) was defined as the product of these scores. The scores were combined and analyzed for trends. High-risk failures were defined as those with a median RPN  score in the top 20th percentile or those with a median S score ≥8, as recommended by AAPM Task Group Report 100 and noted in Table 1. A sub-analysis was performed to include an up-scaling of the S score, depending on the number of patients potentially affected by the error. A score of +0 was added to S if the failure would only affect one or a few patients; a score of +2 was added if the failure affected all patients of a certain disease site; a score of +4 was added if the failure affected all patients treated on the machine or at the facility. A scaled RPN was calculated based on the modified S score.


The median RPN for the unscaled  and scaled scores are shown in Figure 1, with top scorers noted in Table 1. Median and mean RPN showed the same highest RPNs, with deviations between groups largely driven by differences in system detectability. Proton and photon clinical trial QA had four out of five same top median RPNs for the unscaled scores, but proton RPNs were higher than photon RPNs for 18 of the 25 tests. The higher proton scores were driven by higher severity for the various failures. While only one photon test had a median S score of 8, five of the proton tests had median S scores of 8. The severity scaling did not substantially alter the order of RPNs, but it did increase the RPNs by an average of 38% for photons and 51% for protons.


Photon and proton clinical trial QA have many similarities, but proton failures tend to have higher severity. Severity scaling is novel for this type of analysis and helps capture the increased impact of some errors on the accuracy of clinical trial data, something not previously accounted for in standard FMEA. The RPNs from this review of international clinical trials QA have identified QA tests that check high-risk failures that should receive focused QA efforts for clinical trial design, prospective peer-review, and focused QA efforts in routine clinical practice.