Optimization and control in deep hyperthermia: Clinical implementation of hyperthermia treatment planning in cervical cancer treatment to obtain a higher treatment quality

Deep hyperthermia is a treatment used in concurrence with radiation therapy or chemotherapy in the treatment of deep seated tumors. In hyperthermia, tumor temperatures are elevated 3 to 7oC above normal body temperature, up to a temperature of 44oC. In a randomized trial, the 3 year overall survival of cervical cancer patients was almost doubled by adding hyperthermia to radiotherapy. There is a clear dose-effect relation in hyperthermia, and therefore increasing the temperature in the tumor is an important factor to further increase survival rates in cervical cancer. Until recently, hyperthermia treatments in Rotterdam were performed by aiming a focus point that was calculated using a cylindrical representation of the patient. Because of the inhomogeneous nature of a patient, this representation is far from accurate. For the 4 antenna Sigma 60, the calculated focus point may still be close to the optimum, but for applicators with more antennas, and a high number of degrees of freedom, this approach will certainly be inadequate. Originating in the 1970’s, electromagnetic numerical and thermal modeling of 3D structures is currently possible with a precision and speed that is sufficient for routine use. When the electromagnetic and thermal properties of a patient are known, the energy and thermal distributions can be calculated for each antenna of the applicator. With this information, the interference pattern can be determined, dependent on phase and amplitude of the emitted signals by the antennas, and thus can be optimized. When performing these patient specific calculations, i.e. treatment planning, and optimizations, the resulting settings can be applied on-line in the clinic. This thesis covers the clinical introduction of hyperthermia treatment planning, the assessment of the various uncertainties that should be taken into account, and the results of clinical implementation. Optimization The successful application of hyperthermia treatment planning requires optimization routines that optimize the SAR distribution in such manner that the eventual dose in the tumor is maximized. In chapter 2, various SAR based goal functions were assessed. This assessment showed that a goal function taking into account hotspot minimization as well as maximization of the SAR in the tumor has the highest probability to lead to high tumor temperatures. Eventually, two goal functions were chosen for clinical assessment: average tumor SAR normalized on whole body average SAR (Opt1), and hotspot tumor quotient (HTQ), the ratio between SAR in the 0.1th percentile and the tumor SAR (Opt2). Further, the concept of complaint adaptive steering is tested, i.e. local reduction of SAR in case of patient discomfort by adapting the goal function. The phantom test and a sensitivity study in 10 patient models, show that complaint adaptive steering is most effective in peripheral complaint regions. Clinical evaluation in two groups of five patients shows that time between complaints is longer using Opt2 (p=0.007), i.e. a higher comfort of the patient during treatment. It does, however, not yet lead to significantly different temperatures [T50’s of 40.3(Opt1) vs. 40.1oC (Opt2) (p=0.898)]. From this study we concluded that complaint adaptive steering is feasible in terms of SAR-reduction in complaint regions and in time consumption. Moreover, complaint adaptive HTP guided steering has the potential for further improvement and thus higher temperatures, when the degrees of freedom are increased, i.e. in more advanced applicators. Opt2 ( i.e. HTQ) is used in further clinical application, because of better complaint reduction and control. Uncertainties The clinical use of hypethermia treatment planning can be influenced considerably by various uncertainties. These uncertainties are either related to the reproduction of the model setup in the clinic (e.g. positioning, water bolus shape, antenna signals), or deviations of patient tissue properties from literature values, and cause differences in heating between model predictions and the actual patient. In chapter 4, we investigated the influence of positioning uncertainties on power deposition in the Sigma 60 applicator. Position inaccuracies of less than 1 cm appear not to affect SAR patterns relevantly. Current positioning precision is sufficient in the X (right-left)-direction but precision measurements are needed to reach the desired accuracy in the Y (anterior posterior)-direction. In chapter 5, a closer look was taken at the influence of tissue parameter uncertainties on the tumor SAR and temperature levels for 20 patient models. A Monte Carlo analysis, simulating many uncertainty scenarios, shows a variation of HTQ of approximately 25% (interquartile range) and a variation of 0.7 to 1oC (interquartile range) for temperatures, due to the uncertainties in tissue parameters. Difference between the Sigma 60 and Sigma Eye applicators however, still remain significant (p<0.001 for SAR and temperature distributions). The additional benefit that could be expected from temperature modeling is canceled out by the uncertainties. This means that SAR modeling is sufficient, as long as uncertainties persist. Moreover, these results show that with uncertainty reduction, the potential of HTP guided steering can be increased considerably.

• View at Worldcat

Free Full Text ( Final Version , 3mb )