Abstract Book

S324

ESTRO 37

Purpose or Objective Previous studies have focused on developing plan complexity metrics to correlate with dose delivery accuracy for IMRT and VMAT in an effort to understand, and thereby minimize, radiation dose uncertainties. The aim of this study is to evaluate whether several plan complexity metrics predicted dosimetric performance in IROC Houston’s anthropomorphic head and neck (H&N) phantom. This study is of particular interest because these plans were designed by hundreds of radiotherapy institutions for phantom credentialing and have been evaluated on a single geometry. Material and Methods 343 H&N phantom irradiations (September 2011 – December 2016) from 312 institutions were reviewed for dosimetric plan errors. Plan error was defined as the average magnitude percent difference between the TPS calculated dose and the corresponding measured TLD doses (6 TLDs per phantom). This error was compared against plan complexity using the following metrics: total MU delivered, modulation complexity score (McNiven et al., 2010), edge metric (Younge, et al., 2012), plan irregularity (average aperture irregularity from Du, et al. 2014), plan modulation (average beam modulation from Du et al., 2014), leaf travel normalized by arc length (Masi et al., 2013), and modulation indices for speed, acceleration, and total modulation (Park, et al., 2014). Correlations between complexity metrics and phantom plan error were determined using Spearman’s rank-order correlation coefficients (with Bonferroni corrections applied for multiple comparisons). Correlations were also evaluated according to TPS (Pinnacle and Eclipse) and delivery technique: step-and-shoot (N=43), dynamic MLC (N=93), and VMAT (N=207). Results For this wide-ranging cohort of plans, treatment plan complexity and average dose error showed no clear relationship and were not correlated statistically (p>0.02) for any of the metrics tested. Additionally, correlation coefficients were minimal and did not exceed ±0.119 for all the plans evaluated or ±0.240 for any of the delivery methods evaluated. Segregating plans by TPS yielded little additional information: correlations did not exceed ±0.233 and were insignificant. These negligible correlations suggest that plan complexity is not related to the dose accuracy of these phantom plans.

tolerance audit result and ± 3.3 % for pass (action level). A single point out of tolerance is considered sufficient evidence for an out of tolerance in the relevant modality (ie IMRT or VMAT). The Level II audit scores against a 2D 3%/3mm 20% threshold global gamma criteria. A 90% passing rate is used as the threshold for an out of Tolerance result, and a 97.5% passing rate indicates a Pass (Action Level) outcome. A summary of the results from the Level III audit are shown in Figure 1. The results are shown as both the raw audit results, and corrected for if the facility uses a Linac measured k Q or if they used a tabulated k Q from TRS-398. The average dose difference for the corrected results was + 0.5 % with standard deviation 1.7 % for IMRT and – 0.3 % with standard deviation 1.9 % for VMAT from 630 and 624 in volume point dose measurements respectively.

Figure 1 Summary of dose differences from all in target points measured in the Level III audit. Both audits measure in the low dose tissue sparing region in the centre of the wrap around c-shape. Figure 2 shows the relationship between lower dose in the tissue sparing region and increased discrepancy between the planned and measured dose. This may be due to increasingly modulated plans to produce the reduced dose.

Figure 2 Results from the low dose region in the Level II audit. Conclusion A two level approach to auditing IMRT/VMAT is giving the ACDS a detailed insight to doses delivered to patients across Australia and indications of limitations in different commercial algorithms. OC-0614 Complexity metrics do not predict plan performance in IROC Houston head and neck phantom irradiations M. Glenn 1 , V. Hernández 2 , J. Sáez 3 , D. Followill 1 , S. Zhou 4 , S. Kry 1 1 The University of Texas MD Anderson Cancer Center, Radiation Physics, Houston, USA 2 Hospital Universitari Sant Joan de Reus, Medical Physics, Reus, Spain 3 Hospital Clínic de Barcelona, Medical Physics, Barcelona, Spain 4 The University of Texas MD Anderson Cancer Center, Biostatistics, Houston, USA

Conclusion Surprisingly, existing complexity metrics were not predictive of dosimetric errors in the IROC H&N phantom irradiations. This is interesting, because unlike other experiments, the irradiated geometry of the IROC phantom is constant. These results indicate that other factors affecting treatment delivery, such as beam modeling inaccuracies, dictate the accuracy of phantom treatment plans.