ESTRO 2024 - Abstract Book

S5991

RTT - Treatment planning, OAR and target definitions

ESTRO 2024

Purpose/Objective:

Skull base chordoma, is a rare primary bone tumor arising from residues of the notochord; in 32% of cases it is located in the clivus. Treatment options include surgery, radiotherapy and especially proton therapy (PT). PT is a type of hadron therapy that uses proton beams often employed to treat tumors that are inoperable or resistant to traditional radiotherapy treatments. Protons, unlike photons, take advantage of the Bragg Peak, sparing dose to surrounding healthy tissues. Aim of the study is to make a comparison between contours performed directly on MRI images and those instead obtained downstream of contour propagation originally performed on CT images, calculating accuracy and precision of the contour propagation procedure from CT images.

Material/Methods:

Twenty patients with skull base chordoma, all treated with PT at the CNAO center, were enrolled in the study. 50% of patients had involvement of adjacent organs at risk (OARs), including brainstem, optic chiasm and optic pathways (right optic nerve and left optic nerve), while the remaining 50% showed no OARS involvement. All patients were treated in the supine position, with immobilization performed using custom thermoplastic masks. For each patient, OAR contouring was performed on CT images and T2 TSE sequences of baseline and first follow-up MRI. Subsequently, a subtraction of the contouring volumes obtained from CT and MRI images was performed. Volume differences obtained from the subtraction were then compared: comparisons, in terms of volume differences between two contouring segments, were calculated based on parameters such as maximum and avarage distance expressed in mm and the coefficient of similarity (DSC).

Results:

The mean values obtained show some variability between the individual OARs of the two different groups, higher in case of no involvement, standard deviation (SD) = 1.4, lower in case of involvement, SD = 0.5 For the first group, the maximum value was highest for the brainstem and lowest for the optic chiasm. Thus, it can be inferred that the contouring volumes performed on CT and MR images have a wider range of discordance for the brainstem. For the optic chiasm, the same parameter indicates that the contours performed on CT and MRI images are more similar and accurate. In the second group, maximum values between CT and MRI have less discrepancy. The right optic nerve was found to be the most precisely contoured OAR: both in the case of involvement and in the absence of involvement, for one out of 10 patients the value of DSC was 1 which means that there was perfect agreement between the contoured volumes on CT and MRI images.

Conclusion:

The contoured volumes performed and automatically recorded on CT scans alone show differences from the same ones derived from MRI images. Only in 10% of cases and only in the right optic nerve, a perfect overlap of the derived volumes was recorded. In the case of involvement of OARs, there was a greater margin of error between the volumes contoured on CT and MRI: for 20% of patients, the contours of the left optic nerve and optic chiasm were completely discordant. In conclusion, integration of MRI images with CT images can reduce the margin of error in the contouring phase.