Abstract Book

S295

ESTRO 37

SP-0560 Against the motion M. Schwarz 1 1 Centro di Protonterapia, Protontherapy, Trento, Italy Abstract text Patient specific quality assurance(PSQA) may in principle take several different forms, but the introduction of Intensity Modulated Radiation Therapy(IMRT) did promote and made popular quite a specific interpretation of what PSQA means. It is this interpretation, and not the general concept of PSQA, that deserves plenty of criticism. Current PSQA typically consists in a)recalculating the planned dose on a simplified geometry; b)generating 2D dose map(s) by measuring the dose on that simplified geometry and c)deciding on the acceptability of the treatment by comparing calculations vs. measurements with a dedicated metric (e.g. the gamma analysis). Is this process “patient specific”? Very little. The only specific component which is verified is whether the treatment head model in the TPS does not fail for the irradiation pattern of that patient. Is this process ”quality assurance” of the treatment workflow? Very little. From the process perspective, we are only verifying the consistency of information between treatment planning and delivery system. Is the metric used to assess the plan quality -correlated to clinically relevant indices? There are quite some indications that this is not the case. -sensitive enough to point out possible causes of disagreement? Questionable at least. -sensitive enough to detect systematic equipment errors that can be easily corrected (e.g. in MLC positioning)? Most likely not. Are there guidelines, or commonly agreed criteria, for the conditions under which individual patient specific QA can be discontinued? No. Were recent major accidents due to lack of PSQA? Probably not. Could they have been prevented by much simpler checks? Yes. Can PSQA as currently performed be a hurdle to improvements of radiotherapy techniques? Yes. In 2018, the chance that the best way to detect a clinically significant error is via a measurement in a water equivalent phantom and gamma analysis is small at best. Then why are we doing all this? Because the software and hardware of early days IMRT was not entirely up to the challenge, a proper commissioning was impossible, so we accepted the idea of doing continuous commissioning and called it “Patient specific QA”. The process then became ingrained in our way of working, and we’re struggling to stop it or to change it into something better. Is there a way out? Yes, there are a few options.The most pragmatic approach is to: Get back to accurate TPS commissioning, thus ending “patient specific” measurements much earlier than now; Rely on accurate equipment QA to detect (small) systematic errors that can be corrected; Use treatment delivery information (e.g. via logfiles) or add-on tools (e.g. transmission chambers) to perform workflow QA and avoid large errors; Last but not least, use the time we saved to work and improve overall treatment quality.

How many attendees at ESTRO (or any RT conference) will wander through the exhibition hall and think “if only we could buy and run that! We just don’t have the money, staff or skills” When every RT department in the world has the best treatment equipment available with highly skilled workers to optimally treat every kind of disease that comes their way, we won’t have to “QA” every single plan. Everything will work as expected. No humans will be involved, there won’t be room for error. Radiotherapy in 2018 is not there yet. Good Patient Specific QA is an efficient, robust, sensitive and specific check by a qualified person that the patient’s tumour, or target, receives the prescribed dose. It compares an intended dose with a result, from as early to as late in the treatment process as possible. The term generally implies it is, at least in part, performed by a human. Of course any system that wastes time and doesn’t work should be made obsolete. No one would argue that there is still a role for problem-ridden, bad Patient Specfic QA . The technology to implement a good version exists today, it is just not (yet) a priority for major vendors. We should be striving for good patient specific QA. There are many forms of non-patient specific QA in place, for the beam model, small fields, treatment planning system, mechanics of the delivery systems, image formation, image registration and positioning systems. If everything works as it should, why do you also need patient QA? There are two increases in Radiotherapy that make this necessary 1) personalisation and 2) complexity. First, an impersonal square field or two used to be sufficient to treat most tumours. In the past, the resulting dose distribution depended on the anatomy of the patient. It’s hard to get that wrong. Today the objective is to achieve a uniform, maximum dose distribution in the tumour and very little everywhere else. The result is that you have 2 identical plans about as often as you have have 2 identical patients. Imaging biomarkers in PET, MRI and now CT are being used customise plans to not only diagnosis, but also prognosis, moving away from one-size-fits-all. It is difficult for protocols to keep up with so many exceptions, which can increase the chance of mistakes. Secondly, we have increasing precision, automation, dose and accuracy in Radiotherapy. The result is complicated protocols for even simple treatments. In addition, the level of training for staff and the time allowed to carry out additional tasks does not always keep up. Consequently, both the risk of human error and the impact of accidents is also increasing. While complicated plans get all the attention, major accidents that get reported are often standard, simple treatments that went wrong. When 1 accident in a million good treatments makes headline news, doubt is cast over all Radiotherapy. Due the personalised nature of plans and increasingly complicated protocols, patient specific QA still has an important role to play. Finally patient specific QA provides reassurance that the patient was treated as intended. When a patient or a lawyer asks for evidence that the treatment went according to plan, the department should have something to show. They are unlikely to be satisfied with evidence that the previous 100 patients were checked and found to be fine. It is estimated 1 in 20 patients will suffer injuries from radiation treatment. We don’t know how many are due to mistakes unless we have plan-specific verification data. Treatment prescriptions and randomised trials also rely on information about accuracy and trend changes. You need data for this, and the best kind is not sampled. On the other side, worker satisfaction improves when people know they’ve not made a mistake. It’s effective and good practise to provide feedback to your department: “well done, we’re within x%”. For all these reasons and more, we need the significant, useful data that patient specific QA provides.