ESTRO 2024 - Abstract Book

S5993

RTT - Treatment planning, OAR and target definitions

ESTRO 2024

The Gross Tumor Volume (GTV) was delineated by a dedicated radiation oncologist and the PTV was generated with 3mm around the GTV. For plan optimization and online plan review, PTVopt structure was generated consisting of the PTV but excluding any overlap with OARs. Figure 1 shows two axial views of online MRI for two pancreatic cancer patients with a different amount of OARs around the PTV. For each patient, a baseline treatment plan was generated, which was online adapted at each fraction to account for the inter-fractional anatomical changes. Prescribed dose to the PTVopt was V38Gy ≥ 95%. Dose constraints for the duodenum, bowel and stomach were V36Gy ≤ 0.1cc and V33Gy ≤1.0cc. Data from each fraction was extracted from the treatment planning system (TPS) and further analyzed to ascertain daily variations in GTV, PTVopt and target coverage. Linear regression analysis were performed with IBM SPSS (28.0.0). Variations are reported as 5th – 95th percentiles within brackets.

Figure 1: Axial view of two patients. Regarding the left patient OARs are present within 2 cm of the PTV. As for the right patient, there are no OARs within 2 cm. PTV is indicated in red, GTV in dark-blue, duodenum in cyan, stomach in pink and bowel in brown

Results:

Inter-fraction median online GTV and PTVopt variation was 0.3% (-3.1 – 5.8%) and 1.2% (-10.6% – 14.8%), respectively. Median overlap of OARs with the PTV was 1.2cc (0.0 – 4.5cc), resulting in a median difference between online PTV and PTVopt of (2.2%) (0.0 – 10.5%). The median volume of OARs within 2cm from the PTV was of 24.1cc (66.7 – 120.4cc). Stomach and duodenum were more prominent present within the first 2cm than bowel (average volume of 28.6 and 33.8cc vs. 16.1cc). Median PTVopt dose volume coverage was 93.6% (84.0 – 98.4%) at 38Gy. In 35% of the fractions was possible to achieve the prescribed dose-volume constraint V38Gy ≥ 95%. In the majority of fractions this was not possible due to the presence of radio-sensitive OARs next to the PTV. Only in two fractions the V33Gy constraint for the stomach was slightly exceeded (1.1cc), while the other dose constraints were achieved. Statistical analysis of online adapted plans revealed that the volume of OARs within 2cm from the PTV and their overlap with the PTV were statistically correlated (p < 0.001) with V38Gy for PTVopt. Linear regression analysis allowed to derive a model for upfront prediction of target coverage as a function of these two variables. The PTVopt V38Gy(%) could be predicted as: V38Gy(%)= 98.2 - (0.03 x OAR 2cm ) - (0.94 x overlap OAR-PTV ). Figure 2 shows a scatterplot visualize the correlation of the actual and the predicted target coverage. Thirteen outliers from the linear regression model