ESTRO 36 Abstract Book

S818

ESTRO 36 2017

_______________________________________________________________________________________________

CT volumes at arbitrary positions within the respiratory

cycle. Daily CBCT images are used to calculate a baseline

shift between the planning CT and the actual anatomy

using deformable image registration. Treatment planning

was done by the TPS Pinnacle³ v9.8 (Phillips Radiation

Oncology Systems, Fitchburg, WI, USA). The motion model

was used to estimate CT volumes in intervals of 100 ms

that were imported into the TPS to calculate a partial

delivered dose. The Vero system uses rotations of the

LINAC head for DT, so the resulting images have to be

rotated according to the pan/tilt motion since the TPS is

not able to model a non-axial beam line. The partial doses

were back-rotated and superposed resulting in the

actually delivered 4D dose distribution. In addition, 4D

dose-volume-histograms (

DVH

) were calculated by

warping the reference contours onto every respiratory

phase using the motion model. The resulting 4D dose

distributions were evaluated and compared to phantom

measurements using the ArcCheck (Sun Nuclear,

Melbourne, FL, USA) in combination with a motion

platform jointly developed with QRM, Möhrendorf,

Germany.

Results

The accuracy of the motion model was validated for 20

patients with a geometrical uncertainty of < 3 mm. Dose

comparison of the estimated volumes to clinical ground-

truth CTs resulted in a pass rate above 98.9% for a γ-

criterion of 2% / 2 mm. 4D dose distributions could be

reconstructed and were in good agreement to phantom

measurements. The actually delivered dose for two liver

patients was recalculated and additional patients are

currently ongoing.

Conclusion

The presented approach is feasible to reconstruct 4D dose

distributions for DT patients using a common TPS together

with the presented motion model. External surrogates to

calculate the breathing state are essential as well as daily

CBCT or kV images to correct for baseline shifts.

EP-1542 Can proton therapy for head and neck cancer

reduce side effects while maintaining target

robustness?

D. Scandurra

, R.G.J. Kierkels

, M. Gelderman

, H.M.

Credoe

, H.P. Bijl

, R.J.H.M. Steenbakkers

, J.G.M.

Vemer - van de Hoek

, J.A. Langendijk

University Medical Center Groningen, Department of

Radiation Oncology, Groningen, The Netherlands

Purpose or Objective

Head and neck cancer generally consists of targets

requiring high doses of radiation in close proximity to vital

healthy organs. Proton therapy leads to a reduction of

integral dose and therefore can reduce normal tissue

complication probabilities (NTCP). The proton range,

however, depends on the material along its path and is

therefore susceptible to uncertainties, potentially

degrading target coverage. These uncertainties can be

partially accounted for by conventional target margins

(PTV) or robust-optimisation approaches. In this in-silico

study, four variations of pencil beam scanning (PBS)

proton therapy plans were designed for each patient and

compared to the clinical photon plan in terms of

robustness and NTCP reduction.

Material and Methods

Four PBS plans were made for each patient using

RayStation (v4.99 RaySearch Laboratories AB, Sweden):

1) PTV based, multi-field optimisation (PTV-MFO),

2) PTV based, single field optimisation (PTV-SFO),

3) CTV robustly optimised MFO (CTV-MFO), and

4) CTV robustly optimised SFO (CTV-SFO).

Each plan was designed to a similar target homogeneity

index and normalised to CTV70 D98%. Four beams were

used in an ‘x’ arrangement and a 4 cm range shifter was

used where appropriate.The PTV was defined as 5 mm

geometric expansion to CTV, identical to current photon

plans in our clinic (VMAT). Robust optimisation settings

were 5 mm isotropically and ±3% HU uncertainty. For each

plan, robust evaluation was performed for a combination

of ±5 mm translational, ±2

⁰

rotational and ±3% HU errors

(>50 scenarios per plan). The error scenario with the

lowest target coverage (V95% to CTV70) is deemed the

‘worst-case’

scenario.

Organ at risk (OAR) doses and NTCP values of the nominal

proton plans and the clinical VMAT plan were compared.

Results

Robust evaluation showed CTV70 target coverage

degraded significantly for PTV based plans, particularly

using MFO (-15%) but also for SFO (-5.4%). Robustly

optimised plans performed much better, particularly CTV-

MFO (-1.7%) which comes very close to matching the

robustness of the clinical VMAT plan (-1.2%) (figure 1).

CTV-SFO under-performed due to the beam arrangement

and the high weighting on OAR dose reduction.

OAR doses were significantly better in MFO plans, as this

gave the optimiser the most freedom in per–beam dose

distribution. In particular, CTV-MFO plans had the lowest

OAR doses of all proton and photon plans. NTCP model

calculations show that, compared to our clinical VMAT

plans, reductions across all patients for xerostomia (-

5.9%), dysphagia (-7.0%) and tube feeding (-4.7%) can be

expected, but these values varied widely among different

tumour sites, stages and individual patients (table 1).

Conclusion

Compared to photon VMAT plans, robustly optimised PBS

proton therapy plans reduce OAR doses in head and neck

cancer while maintaining a high degree of target coverage

robustness. NTCP calculations show a clinical benefit can

be expected for a significant proportion of patients.

EP-1543 Dominant intraprostatic lesions boosting:

comparison of tomotherapy, VMAT and IMPT

P. Andrzejewski

, A. Jodda

, P. Kuess

, D. Georg

, J.

Malicki

, T. Piotrowski

Medical University Vienna, Dept. of Radiation Oncology-

Christian Doppler Laboratory for Medical Radiation

Research for Radiation Oncology, Vienna, Austria

Greater Poland Cancer Centre, Department of Medical

Physics, Poznan, Poland