Abstract Book

S1090

ESTRO 37

Conclusion Bayesian networks provide a framework for combining patient characteristics with population based patterns of lymphatic progression, and thereby provide a method to individualize CTV definition. References: [1] G. Sanguineti et al, IJROBP, V. 74(5), p.1356-64, 2009 [2] V. Grégoire et al, Radiother. Oncol., V. 110(1), p.172- 81, 2014 EP-2002 Ethnicity Information in Toxicity-Related Radiogenomic Studies: A Systematic Review N. Yahya 1 , S. Zuber 1 , H. Manan 2 1 The National University of Malaysia, Faculty of Health Sciences, Kuala Lumpur, Malaysia 2 The National University of Malaysia, Department of Radiology, Kuala Lumpur, Malaysia Purpose or Objective Associations of genes and radiotherapy-related toxicity obtained from one ethnicity or racial group are not necessarily applicable to individuals whose ancestry is primarily from a different ethnic or racial group. This study aims to map the statement of ethnicity in radiogenomic studies involving radiotherapy-related toxicity and determine the level of inclusion of this information in the analysis. Material and Methods Studies were identified by searching Scopus, PubMed and Medline until November 2016. References from the retrieved articles were also searched for additional publications. Studies satisfying the following criteria were eligible for inclusion: related to the effects of radiation dose from radiotherapy to normal tissues and involving human subjects. Studies on animals, cell cultures, case reports, meta- analysis and systematic review articles were excluded. The completeness of the ethnicity information was determined by mining the statement in the method/results. The consideration of ethnicity as a covariate and its significance was based on the statement mentioned in the statistical analysis/results of the study. Results 177 studies were found to satisfy the inclusion criteria. 75/177 stated the ethnicity of which 21/75 specifically stated the numbers for each ethnic and another 54/75 either reported a homogenous cohort or mentioned “mixed” without further specification. Of studies with ethnicity details, 10/21 has substantial mixed (second most common ethnicity > 20% of the most common) and 14/21 considered ethnicity as a covariate. From 14 studies, 2 found significant associations between ethnicity and toxicity. Conclusion Large proportion of radiogenomic studies are from homogenous cohorts which can make comparisons across studies with differing ancestry difficult. Studies combining cohorts from different ethnicity need to be encouraged. EP-2003 On the impact of dose variations in contrast to mean dose on EQDx and α/β estimation N. Niebuhr 1,2,3 , A. Pfaffenberger 1,3 1 German Cancer Research Center DKFZ, Medical Physics in Radiation Oncology, Heidelberg, Germany 2 Heidelberg University, Department of Physics and Astronomy, Heidelberg, Germany 3 National Center for Radiation Research in Oncology, Heidelberg Institute for Radiooncology, Heidelberg, Germany Purpose or Objective Multiple factors in radiotherapy cause local deviations between delivered and planned doses. Daily imaging

enables the estimation of the delivered dose, yet spatial and temporal accumulation and averaging of the dose leads to a loss of the newly gained information of the impact of (fractional) dose variations on clinical parameters. Here, we investigate the impact of dose variations on the delivered EQDx as well as on caused uncertainties in the estimation of α/β values. Material and Methods The analyses theoretically investigate the impact of dose variations of different amplitudes and in different fractionation schemes (conventional vs. hypofractio- nation) on the EQDx and the α/β estimation. The EQD2 formalism was adapted to compare a constant (mean) x Gy per fraction schedule to one that accounts for each fraction dose. This enables the comparison of treatment with constant dose x with that of varying doses around a mean of x (Figure 1). Furthermore, assuming a constant α-value, the ratio of the accumulated mean dose to the fractional delivered dose was studied as an estimate of the impact of dose variation on the prediction accuracy of α/β (Figure 2). Results Figure 1 shows that dose variations induce deviations in EQDx that increase with higher variation amplitude and mean dose, and with lower α/β as well as for hypofractionation scenarios. The difference from EQDx from constant mean dose to that from the same mean dose but with a standard deviation of 50% exceeds 15%. For lower mean doses, the effect is smaller, yet it is in the same order of magnitude (Figure 1, right).

Assuming a constant mean dose in contrast to accounting for dose variations systematically underestimates the resulting α/β (Figure 2). The estimated impact on the α/β estimation is independent of the mean fraction dose and the number of fractions.The underestimation is by 2%, 5% and 20% for 12.5%, 25% and 50% dose standard deviation, respectively, and represents an additional factor in the uncertainty of α/β estimation to the uncertainty from deviations of delivered to planned total dose (Figure 2, right).