A prospective analysis of inter- and intrafractional errors to calculate CTV to PTV margins in head and neck patients

To evaluate an institute-specific CTV-PTV margin for head and neck (HN) patients according to a 3-mm action level protocol. Twenty-three HN patients were prospectively analysed. Patients were immobilized with a thermoplastic mask. Inter- and intrafractional set-up errors (in the three dimensions) were assessed from portal images (PI) registration. Digitally reconstructed radiographs (DRRs) were compared with two orthogonal PI by matching bone anatomy landmarks. The isocenter was verified during the first five consecutive days of treatment: if the mean error detected was greater than 2 mm the isocenter position was corrected for the rest of the treatment. Isocenter was checked weekly thereafter. Set-up images were obtained before and after treatment administration on 10, 20 and 30 fractions to quantify the intrafractional displacement. For the set-up errors, systematic (I ) pound, random (sigma), overall standard deviations, and the overall mean displacement (M), were determined. CTV to PTV margin was calculated considering both inter- and intrafractional errors. A total of 396 portal images was analysed in 23 patients. Systematic interfractional (I (inter)) pound set-up errors ranged between 0.77 and 1.42 mm in the three directions, whereas the random (sigma (inter)) errors were around 1-1.31 mm. Systematic intrafractional (I (intra)) pound errors ranged between 0.65 and 1.11 mm, whereas the random (sigma (intra)) errors were around 1.13-1.16 mm. A verification protocol (3-mm action level) provided by EPIDs improves the set-up accuracy. Intrafractional error is not negligible and contributes to create a larger CTV-PTV margin. The appropriate CTV-PTV margin for our institute is between 3 and 4.5 mm considering both inter- and intrafractional errors.