- Email: [email protected]
29 disc Dose-volume effects in the rat spinal cord: mathematical NTCP models put to the test
Poster Workshop Discussion
lesions in the irradiated region, which revealed that, GM lesions may be attributed to blood spinal cord barrier (BSCB) disruption or vessel dilation, together with edema. In addition to BSCB disruption, WM lesions may reflect less perfused or fewer functional blood vessels. The blood volumes yielded were 6% in gray matter (GM) and 4% in white matter (WM) in control animals and 7% (GM) and 6% (WM) in irradiated animals, which is in the upper range of cerebral blood volume values reported in literature, 25%. An additional result is the imaging of large blood vessels which decrease in functionality after severe radiation damage, which resulted in a decreased accessibility of the large blood vessels to blood. In summary, vascular changes and tissue damage of the rat spinal cord were detected by USPIO enhanced MR imaging in a relatively early stage of late radiation damage. 26
Sunday, 1 June 2003
$9
ly implanted osmotic pump. IGF was injected subcutaneously. Animals were followed until development of myelopathy or for a maximum of 12 months. Cervical spinal cord was harvested for histology. Results: at the dose inducing 100% RM (ED100) for controls only 50% of the PDGF treated animals developed RM. Neither IGF nor VEGF prevented animals from developing RM at the ED100 level. The ED50 for controls was 33.8 Gy and for PDGF treatment 35.8 Gy (p=0.009). The dose modifying factor was 1.06 with a very steep slope in the dose response. Toxicity in all groups was comparable and related to radiation-induced mucositis. Conclusion: prevention of RM by i.t. PDGF infusion is possible and feasible. Further studies are needed to determine the magnitude of protection with more clinically relevant fractionation regimen, the optimal PDGF dose. scheduline, and the underlying mechanisms. Both IGF and VEGF treatment failed to show a protective effect.
disc
Partial volume proton irradiation of the rat cervical spinal cord indicates regional differences in radiosensitivity
28
H.P. Bijl 1, P. van Luijk 1,2, J.M. Schippers 2,4, R.P. Coppes 1,3, A.W. T. Konings 3, A.J. van der Kogel 5 1University Hospital Groningen, Department of Radiation Oncology, Groningen, The Netherlands 2Kemfysisch Versneller Instituut, Groningen, The Netherlands 3University of Groningen, Department of Radiation and Stress Cell Biology, Groningen, The Netherlands 4paul Scherrer Institut, Vllligen, Switzerland 5University Medical Center Nijmegen, Department of Radiation Oncology, Nijmegen, The Netherlands
A. Jiresova 1, H. Van Goor 2, A.W.T. Konings 1, H.H. Kampinga 1, R.P. Coppes 1/3 l Fac. of Medical Sciences, University of Groningen, Radiation & Stress Cell Biology, Groningen, The Netherlands 2University Hospital Groningen, Department of Pathology, Groningen, The Netherlands 3University Hospital Groningen, Department of Radiotherapy, Groningen, The Netherlands
Puroose: To estimate dose-response relationships for paralysis in rats after inhomogeneous irradiation of the cervical spinal cord. Methods and Materials: Two types of inhomogeneous dose-distributions were applied across the cervical spinal cord of male Wistar rats. In the first experiment (grazing beam) rats were irradiated on the left lateral half of the spinal cord (Dmax: 24-36 Gy). In the second experiment (central beam), the middle part of the spinal cord (Dmax: 20-80 Gy) was irradiated. The field length (C1-T2) was 20 mm in both experiments. The animals were irradiated with variable single doses of unmodulated protons (150 MeV) using the shoot through method, which employs the plateau of the depth-dose profile rather than the Bragg peak. The endpoint for estimating ED50 values was paralysis of fore and/or hind legs. The present data were compared with results of homogeneously irradiated 20 mm length of cervical cord. Results: High precision proton irradiation of the lateral as well as the central part of the spinal cord resulted in a significant shift of dose-response curves to higher dose values when compared with the homogeneously irradiated cervical cord to the same 20 mm length. The ED50 values were 33.4 Gy for the grazing beam, and even as high as 71.9 Gy for the central beam irradiations, compared with 20.4 Gy for the 20 mm homogeneous single field irradiation. Histological analysis of the spinal cords showed that the paralysis was due to white matter necrosis. The gray matter did not show major lesions like necrosis and hemorrhages. Conclusions: This study showed large regional differences in radiosensitivity in the spinal cord. The white matter of the lateral parts of the spinal cord seems more sensitive than the white matter of the middle part of the cord. The gray matter is highly resistant to radiation since no major lesions were observed even after a single dose of 80 Gy. These results may be important for the estimation of normal tissue complication probabilities for high precision radiation techniques such as IMRT or radiosurgery.
To decrease the uncertainty in normal tissue complication probability predictions, dose-volume and dose-region animal experiments need to be performed to yield data for optimization of predictive models. Here we addressed the issue of lung radiosensitivity. Methods: Rats were irradiated by graded single doses to 100% (9-12 Gy), 50% (16-22 Gy) and 25% (27-36 Gy) of the total tung volume using a collimated X-ray beam. For the 50% volume, the irradiation was targeted to 6 different regions: right (R), left (L), apical (A), basal (B), mediastinal (M), lateral (LT). Two structural endpoints - histology and CT scans, and a functional endpoint - breathing frequency rates (BFR) were evaluated over a period of 38 weeks. Dose-effect curves for BFR were constructed for each volume and region. Results: Histological signs (H&E and ED-1 staining for macrophage activation) of pneumonitis (8 weeks) and fibrosis (26 weeks) were dose- but not region- dependent. They were limited to the irradiated parts of lungs. CT changes at 8 weeks pointed towards hypersensitivity of L and A regions. After the 50% irradiation, BFR increases displayed biphasic dynamics with a first peak between 4 and 10 weeks and a second increase starting at 16 weeks followed by stabilization at moderately elevated level. Regional differences were detected during the early peak with the BFR increases expressed in the following order of magnitude: A > L > B = LT > R = M. The differences were attenuated at the later time points (> week 20). After the 25% and 100% irradiation, only a mild elevation in BFR was observed between weeks 4-10. Discussion: The results indicate biphasic dynamics in expression of functional damage consistent with the onset of pneumonitis and fibrosis in temporal sequence. Regional differences in sensitivity were observed during the early phase. They could not be explained by structural damage but were likely linked to the varying distribution of sensitive gas-exchange structures throughout the lung as well as the involvement of the heart in the irradiation field. 29
27
disc
Reduction of radiation-induced myelopathy by plateletderived growth factor; a dose response study N. Andratschke 1, C. Niedet2, R. Price3, B. Rivera 3, S. Millet3, K.K. Ang 1 1Dept. of Radiation Oncology, MD Anderson Cancer Center, U.S.A. 2Dept. of Radiation Oncology Technical University of Munich, Germany 3Dept. of Diagnostic Imaging, MD Anderson Cancer Center, U.S.A. Purpose: to test the role of platelet-derived growth factor (PDGF), insulinlike growth factor (IGF) and vascular endothelial growth factor (VEGEF) in reducing radiation-induced spinal cord injury (radiation myelopathy, RM) and to establish a dose-response relationship in vivo. Methods: the cervical spinal cord of adult rats was irradiated with 16+14-24 Gy with concurrent growth factor (GF) treatment (n=41 for PDGF, n=32 for IGF, n=24 for VEGF) or saline as controls (n=50). PDGF and VEGF were administered intrathecally (i.t.) through a canula, inserted into the cisterna magna, that was connected via a polyethylene catheter to a subcutaneous-
disc
Dose-volume-region effects in partial irradiation of rat lung
disc
Dose-volume effects in the rat spinal cord: mathematical NTCP models put to the test P. van Luiik 1, H.P. Bijl 1, A.J. van der Koget2, A. W. T. Konings 3, J.M. Schippers 4 1University Hospital Groningen, Department of Radiation Oncology, Groningen, The Netherlands 2University Medical Center Nijmegen, Department of Radiation Oncology, Nijmegen, The Netherlands 3University of Groningen, Department of Radiation and Stress Cell Biology, Faculty of Medicine, Groningen, The Netherlands 4paul Scherrer Institute, Accelerator department, Proton Therapy Project, Vllllgen, Switzerland The spinal cord is a dose-limiting critical organ in head and neck tumour treatment. For volumes smaller than a few centimeter the tolerance dose of the spinal cord is known to depend on the irradiated volume. To predict the
$10 Sunday, 1 June 2003
normal tissue complication probability NTCP as a function of the dose distribution, normally use is made of one of the many existing mathematical models. Puroose: To test a variety of NTCP models for their applicability to the endpoint of white matter necrosis in the cervical spinal cord. Materials/Methods: In the present study, many different mathematical models are fitted to data obtained after irradiating the cervical spinal cord of the rat with many different dose distributions using proton radiation. All dose distributions used in the current study are uniform in the transverse cross section of the spinal cord. Dose distributions included in the current study are: uniform (2, 4, 8 and 20 mm), non-contiguous (two 4 mm fields with different separations), non-uniform irradiations (high doses to either 2 or 4 mm in the center of a 20 mm low dose region or on the edge of a 12 mm tow dose region.) Results: The different models were fitted to the data and their goodness of fit was determined. The cell migration principle introduced by Shirato et al. applied to the critical element and critical volume model resulted in models that were capable of describing the NTCP-data obtained in uniform irradiations. However, none of the models Was able to describe the entire data set. Based on the experimental results it was suspected that a non-local repair process at the border between irradiated and non-irradiated tissue plays an important role. We were able to construct an empirical model for predicting the 50% tolerance dose (ED50) from the dose distribution. This model includes effects of possible non-local reapair effects at the boundaries of the dose distribution. Conclusions: One has to be extremely cautious when considering using NTCP models if there is insufficient data validating them for all types of dose distributions they will be used with. Using the concept of non-local repair on the border between irradiated and non-irradiated tissue, we could describe the variation of the ED50 with the dose distribution. For the spinal cord new models are needed that employ the spatial distribution of dose in addition to the dose volume histograms. 30
disc
NTCP modelling of radiation damage to the parotid glands of the rat assessed from stimulated flow measurements P. van Lu!ik 1,2, F. Cotteleet2, H. Fabet2, R.P. Coppes 1,2, H. Meertens 1, A. W. T. Konings2 1University Hospital Groningen, Department of Radiation Oncology, Groningen, The Netherlands 2University of Groningen, Department of Radiation and Stress Cell Biology, Faculty of Medicine, Groningen, The Netherlands The parotid gland is a radiation sensitive organ which is often damaged in radiotherapy of head and neck tumours. Damage to the parotid glands can lead to serious morbitity and a severe decrease in quality of life. Pureose: To obtain more insight into the mechanisms that turn the deposited dose into organ impairment, and subsequently better normal tissue complication probability (NTCP) predictions. This may lead to improved treatment strategies that better spare the parotid gland. Materials/Methods: In the current study, data obtained in a recent study [1,2], which was a part of the Groningen Proton Therapy Project, was analysed. The parotid glands of rats were irradiated using three different dose distributions. Three groups of rats were irradiated on the parotid gland using different dose distributions being a) complete 100%, b) the cranial 50% and c) the caudal 50% part. The stimulated flow as a function of dose was converted into two binary dose response data sets using the endpoints "less than 25% flow, 365 days after irradiation" and "less than 50% flow, 365 days after irradiation". Probit curves were fitted seperately to all three experiments in both data sets. From these curves the tolerance dose (ED50) was derived. Results: For the 25% flow end-point the tolerance doses were: 17.2 Gy (13.8-20.3 Gy 95% confidence interval), 21.4 Gy (16.8-25.6 Gy) and >40 Gy for the 100%, 50% cranial and 50% caudal irradiations respectively. For the 50% flow end-point the tolerance doses were: 11.7 Gy (5.5-15.0 Gy 95% confidence interval), 21.4 Gy (16.8-25.6 Gy) and >40 Gy for the 100%, 50% cranial and 50% caudal irradiations respectively. Conclusions: As expected, the tolerance dose of the parotid gland strongly depends on the irradiated volume, but more interestingly, is the observed regional dependence of the tolerance dose. As a result currently existing NTCP models that don't use any spatial information of the dose distribution need to be modified in order to be able to use them for the paretid gland. [1 ] AW.T. Konings, F. Cotteleer, H. Faber et al. Volume effects for early and late radiation damage in the parotid gland. Radiother Oncol 2002;64;$193 [2] F. Cotteleer, H. Faber, H. Meertens et al. Regional dependence of function loss after partial irradiation of the parotid gland. Radiother Oncol
Poster Workshop Discussion
2002;64;S193 31
disc
Normal tissue damage models with particular attention on the volume dependence M. Adamus-G6rka, B.K. Lind, A. Brahme Karolinska Institutet & Stockholm University, Department of Medical Radiation Physics, Stockholm, Sweden Backaround: Studying radiation response in radiotherapy is motivated by necessary improvement of knowledge about basic mechanisms involved in the response of living tissue to radiation and also by the need for understanding the nature of response in human when irradiating a patient for therapeutic reasons. In order to estimate the expected clinical outcome radiobiological models are applied more and more commonly in the radiotherapy treatment planning and optimization processes. Method: Seven models for normal tissue complication probability (NTCP) have been tested in the present study. Following four of them are based on the cell survival expression: relative seriality, inverse tumor, critical element and critical volume models, while the three following: Lyman and Kutcher, parallel architecture and Weibull distribution models are phenomenological. The experimental data for paralysis after irradiation of cervical spinal cord of rats (white matter necrosis) has been used in this work. The values for chi-square have been obtained to judge the shape of dose-response curves for respective models. In order to co check which model handles best the volume effect values of chi-square for fits, both including as well as excluding volume effect, have been calculated and compared. The F-test method has been used to compare both kinds of chi-square values for each respective model and the chi-square value for fit including the volume effect has been taken as the reference one. Results: On the basis of the present study the following conclusions can be drawn: - the normal distribution based Lyman and Kutcher model seems to give a slightly better curve shape in this particular case, - the parallel architecture model deals best with the volume effect, - the relative seriality, the critical element and the inverse tumor models describe well the shape of dose-response curve, but the two latter are not recommended for describing the volume effect. Conclusions: This work reviews the existing models as regards doseresponse curve shape and the way volume effect is handled and shows the advantages of using specific models in a specific respect. It also indicates suggestions for the future developing a new improved model to be used in clinically optimized radiation therapy. Perhaps a combination of the above mentioned best NTCP models could be an option.
lesions in the irradiated region, which revealed that, GM lesions may be attributed to blood spinal cord barrier (BSCB) disruption or vessel dilation, together with edema. In addition to BSCB disruption, WM lesions may reflect less perfused or fewer functional blood vessels. The blood volumes yielded were 6% in gray matter (GM) and 4% in white matter (WM) in control animals and 7% (GM) and 6% (WM) in irradiated animals, which is in the upper range of cerebral blood volume values reported in literature, 25%. An additional result is the imaging of large blood vessels which decrease in functionality after severe radiation damage, which resulted in a decreased accessibility of the large blood vessels to blood. In summary, vascular changes and tissue damage of the rat spinal cord were detected by USPIO enhanced MR imaging in a relatively early stage of late radiation damage. 26
Sunday, 1 June 2003
$9
ly implanted osmotic pump. IGF was injected subcutaneously. Animals were followed until development of myelopathy or for a maximum of 12 months. Cervical spinal cord was harvested for histology. Results: at the dose inducing 100% RM (ED100) for controls only 50% of the PDGF treated animals developed RM. Neither IGF nor VEGF prevented animals from developing RM at the ED100 level. The ED50 for controls was 33.8 Gy and for PDGF treatment 35.8 Gy (p=0.009). The dose modifying factor was 1.06 with a very steep slope in the dose response. Toxicity in all groups was comparable and related to radiation-induced mucositis. Conclusion: prevention of RM by i.t. PDGF infusion is possible and feasible. Further studies are needed to determine the magnitude of protection with more clinically relevant fractionation regimen, the optimal PDGF dose. scheduline, and the underlying mechanisms. Both IGF and VEGF treatment failed to show a protective effect.
disc
Partial volume proton irradiation of the rat cervical spinal cord indicates regional differences in radiosensitivity
28
H.P. Bijl 1, P. van Luijk 1,2, J.M. Schippers 2,4, R.P. Coppes 1,3, A.W. T. Konings 3, A.J. van der Kogel 5 1University Hospital Groningen, Department of Radiation Oncology, Groningen, The Netherlands 2Kemfysisch Versneller Instituut, Groningen, The Netherlands 3University of Groningen, Department of Radiation and Stress Cell Biology, Groningen, The Netherlands 4paul Scherrer Institut, Vllligen, Switzerland 5University Medical Center Nijmegen, Department of Radiation Oncology, Nijmegen, The Netherlands
A. Jiresova 1, H. Van Goor 2, A.W.T. Konings 1, H.H. Kampinga 1, R.P. Coppes 1/3 l Fac. of Medical Sciences, University of Groningen, Radiation & Stress Cell Biology, Groningen, The Netherlands 2University Hospital Groningen, Department of Pathology, Groningen, The Netherlands 3University Hospital Groningen, Department of Radiotherapy, Groningen, The Netherlands
Puroose: To estimate dose-response relationships for paralysis in rats after inhomogeneous irradiation of the cervical spinal cord. Methods and Materials: Two types of inhomogeneous dose-distributions were applied across the cervical spinal cord of male Wistar rats. In the first experiment (grazing beam) rats were irradiated on the left lateral half of the spinal cord (Dmax: 24-36 Gy). In the second experiment (central beam), the middle part of the spinal cord (Dmax: 20-80 Gy) was irradiated. The field length (C1-T2) was 20 mm in both experiments. The animals were irradiated with variable single doses of unmodulated protons (150 MeV) using the shoot through method, which employs the plateau of the depth-dose profile rather than the Bragg peak. The endpoint for estimating ED50 values was paralysis of fore and/or hind legs. The present data were compared with results of homogeneously irradiated 20 mm length of cervical cord. Results: High precision proton irradiation of the lateral as well as the central part of the spinal cord resulted in a significant shift of dose-response curves to higher dose values when compared with the homogeneously irradiated cervical cord to the same 20 mm length. The ED50 values were 33.4 Gy for the grazing beam, and even as high as 71.9 Gy for the central beam irradiations, compared with 20.4 Gy for the 20 mm homogeneous single field irradiation. Histological analysis of the spinal cords showed that the paralysis was due to white matter necrosis. The gray matter did not show major lesions like necrosis and hemorrhages. Conclusions: This study showed large regional differences in radiosensitivity in the spinal cord. The white matter of the lateral parts of the spinal cord seems more sensitive than the white matter of the middle part of the cord. The gray matter is highly resistant to radiation since no major lesions were observed even after a single dose of 80 Gy. These results may be important for the estimation of normal tissue complication probabilities for high precision radiation techniques such as IMRT or radiosurgery.
To decrease the uncertainty in normal tissue complication probability predictions, dose-volume and dose-region animal experiments need to be performed to yield data for optimization of predictive models. Here we addressed the issue of lung radiosensitivity. Methods: Rats were irradiated by graded single doses to 100% (9-12 Gy), 50% (16-22 Gy) and 25% (27-36 Gy) of the total tung volume using a collimated X-ray beam. For the 50% volume, the irradiation was targeted to 6 different regions: right (R), left (L), apical (A), basal (B), mediastinal (M), lateral (LT). Two structural endpoints - histology and CT scans, and a functional endpoint - breathing frequency rates (BFR) were evaluated over a period of 38 weeks. Dose-effect curves for BFR were constructed for each volume and region. Results: Histological signs (H&E and ED-1 staining for macrophage activation) of pneumonitis (8 weeks) and fibrosis (26 weeks) were dose- but not region- dependent. They were limited to the irradiated parts of lungs. CT changes at 8 weeks pointed towards hypersensitivity of L and A regions. After the 50% irradiation, BFR increases displayed biphasic dynamics with a first peak between 4 and 10 weeks and a second increase starting at 16 weeks followed by stabilization at moderately elevated level. Regional differences were detected during the early peak with the BFR increases expressed in the following order of magnitude: A > L > B = LT > R = M. The differences were attenuated at the later time points (> week 20). After the 25% and 100% irradiation, only a mild elevation in BFR was observed between weeks 4-10. Discussion: The results indicate biphasic dynamics in expression of functional damage consistent with the onset of pneumonitis and fibrosis in temporal sequence. Regional differences in sensitivity were observed during the early phase. They could not be explained by structural damage but were likely linked to the varying distribution of sensitive gas-exchange structures throughout the lung as well as the involvement of the heart in the irradiation field. 29
27
disc
Reduction of radiation-induced myelopathy by plateletderived growth factor; a dose response study N. Andratschke 1, C. Niedet2, R. Price3, B. Rivera 3, S. Millet3, K.K. Ang 1 1Dept. of Radiation Oncology, MD Anderson Cancer Center, U.S.A. 2Dept. of Radiation Oncology Technical University of Munich, Germany 3Dept. of Diagnostic Imaging, MD Anderson Cancer Center, U.S.A. Purpose: to test the role of platelet-derived growth factor (PDGF), insulinlike growth factor (IGF) and vascular endothelial growth factor (VEGEF) in reducing radiation-induced spinal cord injury (radiation myelopathy, RM) and to establish a dose-response relationship in vivo. Methods: the cervical spinal cord of adult rats was irradiated with 16+14-24 Gy with concurrent growth factor (GF) treatment (n=41 for PDGF, n=32 for IGF, n=24 for VEGF) or saline as controls (n=50). PDGF and VEGF were administered intrathecally (i.t.) through a canula, inserted into the cisterna magna, that was connected via a polyethylene catheter to a subcutaneous-
disc
Dose-volume-region effects in partial irradiation of rat lung
disc
Dose-volume effects in the rat spinal cord: mathematical NTCP models put to the test P. van Luiik 1, H.P. Bijl 1, A.J. van der Koget2, A. W. T. Konings 3, J.M. Schippers 4 1University Hospital Groningen, Department of Radiation Oncology, Groningen, The Netherlands 2University Medical Center Nijmegen, Department of Radiation Oncology, Nijmegen, The Netherlands 3University of Groningen, Department of Radiation and Stress Cell Biology, Faculty of Medicine, Groningen, The Netherlands 4paul Scherrer Institute, Accelerator department, Proton Therapy Project, Vllllgen, Switzerland The spinal cord is a dose-limiting critical organ in head and neck tumour treatment. For volumes smaller than a few centimeter the tolerance dose of the spinal cord is known to depend on the irradiated volume. To predict the
$10 Sunday, 1 June 2003
normal tissue complication probability NTCP as a function of the dose distribution, normally use is made of one of the many existing mathematical models. Puroose: To test a variety of NTCP models for their applicability to the endpoint of white matter necrosis in the cervical spinal cord. Materials/Methods: In the present study, many different mathematical models are fitted to data obtained after irradiating the cervical spinal cord of the rat with many different dose distributions using proton radiation. All dose distributions used in the current study are uniform in the transverse cross section of the spinal cord. Dose distributions included in the current study are: uniform (2, 4, 8 and 20 mm), non-contiguous (two 4 mm fields with different separations), non-uniform irradiations (high doses to either 2 or 4 mm in the center of a 20 mm low dose region or on the edge of a 12 mm tow dose region.) Results: The different models were fitted to the data and their goodness of fit was determined. The cell migration principle introduced by Shirato et al. applied to the critical element and critical volume model resulted in models that were capable of describing the NTCP-data obtained in uniform irradiations. However, none of the models Was able to describe the entire data set. Based on the experimental results it was suspected that a non-local repair process at the border between irradiated and non-irradiated tissue plays an important role. We were able to construct an empirical model for predicting the 50% tolerance dose (ED50) from the dose distribution. This model includes effects of possible non-local reapair effects at the boundaries of the dose distribution. Conclusions: One has to be extremely cautious when considering using NTCP models if there is insufficient data validating them for all types of dose distributions they will be used with. Using the concept of non-local repair on the border between irradiated and non-irradiated tissue, we could describe the variation of the ED50 with the dose distribution. For the spinal cord new models are needed that employ the spatial distribution of dose in addition to the dose volume histograms. 30
disc
NTCP modelling of radiation damage to the parotid glands of the rat assessed from stimulated flow measurements P. van Lu!ik 1,2, F. Cotteleet2, H. Fabet2, R.P. Coppes 1,2, H. Meertens 1, A. W. T. Konings2 1University Hospital Groningen, Department of Radiation Oncology, Groningen, The Netherlands 2University of Groningen, Department of Radiation and Stress Cell Biology, Faculty of Medicine, Groningen, The Netherlands The parotid gland is a radiation sensitive organ which is often damaged in radiotherapy of head and neck tumours. Damage to the parotid glands can lead to serious morbitity and a severe decrease in quality of life. Pureose: To obtain more insight into the mechanisms that turn the deposited dose into organ impairment, and subsequently better normal tissue complication probability (NTCP) predictions. This may lead to improved treatment strategies that better spare the parotid gland. Materials/Methods: In the current study, data obtained in a recent study [1,2], which was a part of the Groningen Proton Therapy Project, was analysed. The parotid glands of rats were irradiated using three different dose distributions. Three groups of rats were irradiated on the parotid gland using different dose distributions being a) complete 100%, b) the cranial 50% and c) the caudal 50% part. The stimulated flow as a function of dose was converted into two binary dose response data sets using the endpoints "less than 25% flow, 365 days after irradiation" and "less than 50% flow, 365 days after irradiation". Probit curves were fitted seperately to all three experiments in both data sets. From these curves the tolerance dose (ED50) was derived. Results: For the 25% flow end-point the tolerance doses were: 17.2 Gy (13.8-20.3 Gy 95% confidence interval), 21.4 Gy (16.8-25.6 Gy) and >40 Gy for the 100%, 50% cranial and 50% caudal irradiations respectively. For the 50% flow end-point the tolerance doses were: 11.7 Gy (5.5-15.0 Gy 95% confidence interval), 21.4 Gy (16.8-25.6 Gy) and >40 Gy for the 100%, 50% cranial and 50% caudal irradiations respectively. Conclusions: As expected, the tolerance dose of the parotid gland strongly depends on the irradiated volume, but more interestingly, is the observed regional dependence of the tolerance dose. As a result currently existing NTCP models that don't use any spatial information of the dose distribution need to be modified in order to be able to use them for the paretid gland. [1 ] AW.T. Konings, F. Cotteleer, H. Faber et al. Volume effects for early and late radiation damage in the parotid gland. Radiother Oncol 2002;64;$193 [2] F. Cotteleer, H. Faber, H. Meertens et al. Regional dependence of function loss after partial irradiation of the parotid gland. Radiother Oncol
Poster Workshop Discussion
2002;64;S193 31
disc
Normal tissue damage models with particular attention on the volume dependence M. Adamus-G6rka, B.K. Lind, A. Brahme Karolinska Institutet & Stockholm University, Department of Medical Radiation Physics, Stockholm, Sweden Backaround: Studying radiation response in radiotherapy is motivated by necessary improvement of knowledge about basic mechanisms involved in the response of living tissue to radiation and also by the need for understanding the nature of response in human when irradiating a patient for therapeutic reasons. In order to estimate the expected clinical outcome radiobiological models are applied more and more commonly in the radiotherapy treatment planning and optimization processes. Method: Seven models for normal tissue complication probability (NTCP) have been tested in the present study. Following four of them are based on the cell survival expression: relative seriality, inverse tumor, critical element and critical volume models, while the three following: Lyman and Kutcher, parallel architecture and Weibull distribution models are phenomenological. The experimental data for paralysis after irradiation of cervical spinal cord of rats (white matter necrosis) has been used in this work. The values for chi-square have been obtained to judge the shape of dose-response curves for respective models. In order to co check which model handles best the volume effect values of chi-square for fits, both including as well as excluding volume effect, have been calculated and compared. The F-test method has been used to compare both kinds of chi-square values for each respective model and the chi-square value for fit including the volume effect has been taken as the reference one. Results: On the basis of the present study the following conclusions can be drawn: - the normal distribution based Lyman and Kutcher model seems to give a slightly better curve shape in this particular case, - the parallel architecture model deals best with the volume effect, - the relative seriality, the critical element and the inverse tumor models describe well the shape of dose-response curve, but the two latter are not recommended for describing the volume effect. Conclusions: This work reviews the existing models as regards doseresponse curve shape and the way volume effect is handled and shows the advantages of using specific models in a specific respect. It also indicates suggestions for the future developing a new improved model to be used in clinically optimized radiation therapy. Perhaps a combination of the above mentioned best NTCP models could be an option.