Session Item

The Imaging and Radiation Oncology Core provides end-to-end phantom audits for institutional credentialing for clinical trials. These indicate how well an institution can deliver their intended treatment. However, assessment of institutional performance, and pass/fail criteria, are based on dosimetric metrics rather than clinical acceptability. It is unclear what the performance on this phantom, or the phantom criteria, correspond to in terms of expected patient outcome. Therefore, we used TCP modelling to estimate clinical outcomes associated with actual phantom deliveries in the community. 

A total of 322 irradiations of the IROC H&N phantom were conducted between 2017 and 2019 by different radiotherapy centres. For each irradiation absolute doses to 6 TLD, centrally located within the CTV, were measured. These 6 values were used to generate CTV dDVH. Variation of outcome as a result of dose variation was determined with the Marsden TCP model (1)(BioSuiteVn12). The “ideal” outcome was obtained by assigning 100% of the prescribed dose to the CTV. Model parameters were derived from the literature and the value of s_a was adjusted to fit the model with published clinical data in terms of local control. Because this phantom is used to credential for all anatomical sites where IMRT is the focus, TCP modelling for three sites was done: prostate, H&N and lung.  For lung, because there was substantial difference between the standard and adjusted models, both were evaluated. Parameters were: Intact prostate:a=0.185(1/Gy) s_a=0.053, a/b=3 Gy; H&N:a=0.3(1/Gy) s_a=0.09, a/b=10 Gy; lung:a=0.307(1/Gy) s_a=0.037 standard-0.080 adjusted, a/b=10 Gy, r_c=1.0e7(per cc) for all sites. TCP results were sorted based on phantom result: good(95%-105% dose agreement between predicted and delivered in the CTV), poor(93-95% and 105-107%) and fail(< 93%, >107%).

Of the 322 phantom results, 13 failed, 38 were poor and 271 were good (Fig1). For prostate and H&N, median TCPs were 78.4% and 76.5% respectively. TCP decreased by 3% for poor phantom results, and by a further 5% for failing phantoms (with TCP values as low as 46.1%(prostate) and 40%(H&N)); differences were all significant (One-way ANOVA p<0.001). For the lung, median TCPs were 74.8% with the published and 62.9% for the adjusted parameters. TCP dropped substantially when the delivered plan did not match the predicted plan; poor results were 8%-4% lower and failing results were a further 20%-10% lower, with TCP values as low as 1.3% and 14.4% for standard and adjusted respectively.

The variability in phantom results indicate a substantial impact on patient outcome. In particular, failing phantom results predicted a dramatically poorer expected tumour control. As these results are based on clinical audit results, this work highlights a dramatic need for continual improvement in radiation oncology.

(1)    Nahum AE, Uzan J (Radio)biological optimization of external-beam radiotherapy. Comput Math Methods Med 2012;2012:329214