Interplay effect modeling in stereotactic body radiotherapy treatment of liver cancer using volumetric modulated arc therapy.


Thaper D(1)(2), Oinam AS(3), Kamal R(1)(2), Singh G(1), Handa B(1), Kumar V(1), Yadav HP(2).
Author information:
(1)Centre for Medical Physics, Panjab University, Chandigarh, Punjab, India.
(2)Radiation Oncology Department, Institute of Liver and Biliary Sciences, New Delhi, India.
(3)Radiotherapy Department, PGIMER, Regional Cancer Centre, Chandigarh, Punjab, India. [Email]


To model the interplay effect and minimize it by a selection of optimum parameters value using a predictive model for SBRT of liver cancers. Ten cases of liver tumors treated with the VMAT technique were selected retrospectively. The dosimetric error due to the interplay effect was measured with a micro ionization chamber (0.015cm3) in a Quasar phantom simulating the moving tumor. The interplay effect dependent parameter's viz. patient breaths per minute, the amplitude of respiration, fractional dose (FD), plan complexity due to different energies (Relative degree of modulation), degree of modulation due to a different level of dose optimization constraints, and dose rate (DR) were measured. For the predictive model, mathematical equations were modeled in python from 300 combinations of proposed parameters using multivariate regression analysis. It was observed that the dose variation reduced from -8.44% to -5.16% for change in the BPM values from 7 to 31 and similarly for amplitude, the dose variation reduced from -9.44% to -4.93% for change in amplitude value from 16 mm to 2 mm. The DR and FD have a prominent effect with R2 values of 0.990 and 0.880 respectively. The calculated mean square errors of equations excluding amplitude for the predictive model were 0.90 and 0.82 whereas those for equations excluding BPM were 1.31 and 1.41 for 6 MV and 10 MV beams respectively. The values of the parameters can be prospectively optimized by the use of the predictive model according to clinical situations, so dose variation can be minimized.