To assess the effectiveness of different on-treatment correction strategies on set-up
accuracy in patients with head and neck cancer (HNC) treated on a TomoTherapy HiArt
system. To assess the adequacy of clinical target volume (CTV) to planning target
volume (PTV) treatment planning margins when treating with intensity-modulated radiotherapy
without daily image guidance.
The set-up accuracy measured by daily online volumetric imaging was retrospectively
reviewed for the first 15 patients with HNC treated on the TomoTherapy unit at Addenbrooke's
Hospital. For each fraction, megavoltage computed tomography was carried out, any
discrepancy from the planning scan was noted, and corrected, before treatment. These
data were used to evaluate imaging correction protocols using three different action
levels. The first three fractions were imaged and used to correct for systematic error,
using a 5 mm action level (5 mmAL), a 3 mm action level (3 mmAL), and no action level
(NAL). All imaging strategies were applied, to assess the number of fractions that
would potentially have exceeded a 5 and 3 mm margin. Systematic and random errors
were calculated for the population, assuming the NAL protocol had been applied, and
minimum CTV-PTV margins, required to allow for errors attributable only to set-up,
were calculated using van Herk's formula.
In total, 490 fractions were analysed. Using a 5 mmAL imaging protocol, potentially
198/490 fractions (40%) were outside a 5 mm CTV-PTV margin and 400/490 (82%) were
outside a 3 mm margin. Using a 3 mmAL imaging protocol, potentially 67/490 fractions
(14%) were outside a 5 mm CTV-PTV margin and 253/490 (52%) were outside a 3 mm margin.
A small systematic error was identified in the system; once corrected this would improve
these results. Using the NAL imaging protocol, potentially 31/490 fractions (6%) were
outside a 5 mm CTV-PTV margin and 143/490 fractions (29%) were outside a 3 mm margin.
Estimated minimum CTV-PTV margins to account only for set-up errors, with three-fraction
image-guided radiotherapy and a NAL protocol, were 2.8, 3.1 and 4.1 mm in the mediolateral,
superior-inferior and anterior-posterior directions, respectively.
Reducing the action level at which the systematic error is corrected improves the
probability of treatment delivery accuracy. Using the NAL correction protocol reduces
the number of fractions that have set-up displacements outside a 5 mm CTV-PTV margin.
Although a 5 mm margin is probably sufficient for standard HNC radiotherapy, change
to a 3 mm margin is not favoured at our centre without access to daily image-guided
radiotherapy.