- Email: [email protected]
93 PREDICTION OF RADIOSENSITIVITY USING PHOSPHORYLATION OF HISTONE H2AX
S35 handling of the beam lateral spreading in tissue, as well as to the roadmap for establishing a clinical workflow of in-vivo treatment verification based on a commercial offline PET/CT scanner (SIEMENS mCT) installed next to the treatment rooms. In particular, first examples of in-vivo PET/CT imaging after scanned proton and carbon ion irradiation of stationary tumour indications at HIT will be shown, highlighting the promises of this technique for both ion species. However, challenges of the current implementation will be discussed, together with an overview of on-going activities aiming to improve the methodology and mitigate the drawbacks of offline imaging. Additional investigations on the extension of the technique to time resolved 4D imaging will be addressed in a separate contribution to this conference [1]. References [1] C. Kurz et al., First Steps towards 4D Off-Beam PETBased Treatment Verification at the Heidelberg Ion Beam Therapy Center, submitted to this conference Acknowledgement Parts of this work have been supported by the BMBF Project DOTMOBI (grant agreement number 01IB08002F) and the European FP7 Project PARTNER (grant agreement number 215840-2). I. Rinaldi acknowledges the financial support by the Helmholtz Association of German Research Centres (HGF) in the Virtual Institut Project (contract number VH VI 303). The authors would like to thank the HIT colleagues for their invaluable help, in particular S. Brons, R. Panse and T. Haberer from the irradiation technique team, O. Jîkel and the medical physics team, A. Peters and the accelerator team, as well as J. Debus, S. Combs, T. Welzel, K. Herfarth and their medical / technical staff. Finally, support from the FLUKA collaboration is gratefully acknowledged. 92 DOSE CORRECTION STRATEGY FOR THE OPTIMIZATION OF VOLUMETRIC MODULATED ARC THERAPY G.S. Mageras, P. Zhang Memorial Sloan-Kettering Cancer Center, New York, NY Although a form of intensity-modulated radiation treatment (IMRT), volumetric modulated arc therapy (VMAT) poses additional challenges with regard to dose computation. This presentation discusses these challenges and surveys strategies that have been developed to address them, which are a result of the particular characteristics of VMAT delivery, namely, continuous gantry motion, multileaf collimator (MLC) motion and dose rate modulation. We examine two aspects: continuous arc delivery, and x-ray scatter. Continuous arc delivery: Because it is computationally expensive to perform ray tracing from a continuously moving source, most treatment planning systems approximate dose computation for VMAT (and other arc therapy modes) with a set of closely spaced static beams at discrete angles. The choice of angular spacing is a trade-off between computation time and dose accuracy. The appropriate spacing will depend on dose accuracy requirements of the planning target volume vis-a-vis more peripheral organs at risk, and on the particular treatment plan case. It has been shown that for coarsely spaced beams (every 10°), plans having
ICTR-PHE 2012 larger discrepancies between calculated and delivered dose distributions are with large leaf travel and large dose-rate modulation. This is the basis for approaches that use more finely interpolated dose calculation points for aperture and dose-rate but with fewer source positions. X-ray scatter: A major challenge is the dose computation time and memory requirements for the large number of beams used to model VMAT dose delivery. A common approach in fixed-angle IMRT optimization is to use dose deposition coefficients for computing dose from individual beamlets. The large number of beams imposes large memory requirements for beamlet storage. To address this, the lateral extent of the dose contribution of each beamlet is limited, thus the scatter contribution is less accurately computed, resulting in discrepancies between dose during optimization and dose computed with full scatter following optimization. To this end, strategies have evolved to apply corrections to the scatter dose as the optimization progresses, for example, by periodically computing dose with full scatter and applying corrections to the deposition coefficients. Alternatively, dose is calculated using a fast convolution algorithm at each iteration. 93 PREDICTION OF RADIOSENSITIVITY USING PHOSPHORYLATION OF HISTONE H2AX H. Kunogi, K. Sasai Juntendo University, Tokyo Japan Purpose: Predictive assays for radio-sensitivity are an important issue in radiation biology. Although many methods have been proposed, none have proven clinically useful yet. DNA double-strand breaks are the most important damage produced by radiation and relate directly to cell death. Phosphorylation of histone H2AX (γH2AX) occurs rapidly in response to the presence of DNA double-strand breaks and recruits repair enzymes to these sites. Therefore, some investigators demonstrated a good relationship between frequency of γH2AX and cell survival after radiation. However, others did not show such a correlation. Previously, some authors revealed a good relationship between the frequency of micronuclei and cell survival if they normalized the frequency using DNA content of each cell line. Therefore, we introduced DNA content of each cell line into the analysis of the relationship between radio-sensitivity and expression of γH2AX foci. Methods and Materials: Seven human tumor cell lines, including a fibrosarcoma (HT 1080), 4 esophagus carcinoma (TE-9, KYSE30, KYSE150, and KYSE220), and 2 breast carcinoma (HCC70, and ZR75-1) were used. Cellular radio-sensitivity was assessed by a standard colony-forming assay. To demonstrate the frequency of γH2AX foci in each cell, we counted the number of cells under fluorescence microscopy after immunofluorescence staining. DNA content of each cell line was determined by a flow cytometry assay. Results: The frequency of γH2AX foci after irradiation (4 Gy) in HT1080 cell showed a peak at 3-4 hours after irradiation and gradually decreased up to 24 hours. The 7 cell lines tested here showed dose (0-9 Gy) dependent increases of γH2AX foci per cell at 24 hours after irradiation. However, there was not a good correlation between the γH2AX foci frequency and surviving
S36 fraction after irradiation. When we normalized the γH2AX foci frequency by DNA content, there was a good correlation between the frequency and cell survival for 5 cell lines, but not 2 cell lines. Conclusions: It is difficult to predict cell survival using the γH2AX foci frequency, even after normalization of DNA content. We likely need to introduce other factors, such as an apoptotic cell frequency, in addition to DNA content of each cell line. 94 REGULATION OF PARACRINE SIGNALING BY MICROTUBULE STABILIZING AGENTS AND IONIZING RADIATION M. Pruschy University Hospital Zurich, Switzerland Ionizing radiation (IR) affects multiple cellular components, which induce a multilayered tumor stress response. Para- and autocrine factors are released into the tumor microenvironment in response to irradiation and treatment-activated intracellular stress-responses. During the time course of a fractionated radiation regimen they modulate thereby the tumor microenvironment and these processes co-determine the treatment sensitivity of the tumor and eventually treatment outcome. Using a panel of genetically defined tumor cells, derived from different tumor entities, we analyzed the expression and secretion of biologically active factors in response to IR. Such IRactivated modulators of tumor angiogenesis and tumor cell migration eventually represent interesting novel targets for combined treatment modalities. With a specific focus on microtubule stabilizing agents (MSA) we further investigated deregulation of important components of the tumor microenvironment by the combined treatment modality of IR with the clinically relevant MSA patupilone. In a lung carcinoma tumor model, IR alone induced hypoxia-dependent VEGFexpression. A strong dominant counteracting effect of patupilone was observed when IR was combined with patupilone, on the level of HIF-1a protein stability, VEGF-expression and VEGF-secretion, but only in patupilone-sensitive and not in patupilone-resistant tumor cell lines. Patupilone and IR thereby dysregulated hypoxia-induced tumor angiogenesis, contributing to the potency of this promising combined treatment modality. IR also enhanced the enzymatic activity of secreted matrix metalloproteinases and increased the invasive capacity of several tumor cells. Interestingly low dose patupilone-pretreatment also counteracted specifically IR-induced enhanced MMP activity and patupilone pretreatment completely abrogated IR-induced cell invasion. This effect is mainly regulated by the secretion of the corresponding tissue inhibitors of metalloproteinases (TIMPs). These results indicate that multiple proangiogenic and pro-invasive factors are specifically induced by ionizing radiation. Combined treatment modalities of IR with pharmacological agents, which prevent the secretion of these factors, represent a promising treatment approach. 95 AUTOMATED DETECTION OF ION BEAM MODIFICATIONS IN IN-BEAM PET IMAGES P. Kuess1, W. Birkfellner2, S. Helmbrecht3, F. Fiedler4,
ICTR-PHE 2012 W. Enghardt3,4, D. Georg1 1 Department of Radiotherapy, Div. Medical Radiation Physics, Medical University Vienna / AKH Vienna, Austria 2 Department of Biomedical Engineering and Physics, MUW, Austria 3 OncoRay, TU Dresden, Germany 4 Helmholtz-Zentrum Dresden-Rossendorf, Germany Aim: Positron emission tomography (PET) is up to date the only clinically approved method to verify the range in ion beam therapy. Due to the different physical characteristics of delivered dose and measured activity Monte Carlo (MC) simulations has to be performed out of the treatment plan and the time structure of the irradiation to obtain a prediction of the β+-activity distributions. This simulation has to be compared to the measured PET images to evaluate beam-delivery. Until now the range assessment is performed by a group of experts via visual inspection. This procedure is very time consuming and requires well trained personnel. Hence, an automated objective comparison tool is highly desired for an efficient clinical workflow. Methods: 12 in-beam PET (ibPET) data sets were investigated during this study. Artificial range modifications of simulated β+-activity distributions were used, as described by Fiedler et al. (1). The simulations were altered to obtain range modifications of 4 mm, 6 mm and 10 mm water equivalent path length both in positive and negative beam direction as well as 0 mm to establish a discrimination level. To account for MC based fluctuations for each type of modification 10 simulations were performed. Different methods taken from image processing were applied on these data sets. The most promising results were obtained by calculating the Pearson’s correlation coefficient (PCC). Prior a median filter with a kernel of 11 pixels in all three spatial directions was applied on all images to reduce noise. For the calculation of the PCC both an unmodified simulation and an ibPET measurement were used as reference input. Based on a variable threshold, depending on the PCC, positive and negative values were classified for each patient. With these data sensitivity and specificity curves were calculated. Hence, Receiver Operating Characteristics (ROC) curves were used as a quality factor of the results. To compare ROC curves the area under the ROC curve (AUC) was taken as quality standard. Additionally statistical significance between PCC of modified and unmodified data was calculated (p < 0.05) using the Wilcoxon rank sum test. Results: Using ibPET measurements as reference images the PCC were significantly different regarding the unmodified and modified data sets for all modifications of +/-10 mm. Two third of +/-6 mm modified data sets showed a significant difference and half of the results of +/-4 mm modification. The resulting ROC curves are illustrated in figure 1 which shows a sensitivity of 90-95% for +/-10 mm modification and a specificity of 90-100%. For +/-6 mm alteration a sensitivity of about 75% and a specificity of 80- 90% was obtained. A range deviation of +/-4 mm defines the limitation of the method which is also close to the overall uncertainty of the ibPET method. The use of an unmodified simulation as reference showed similar ROC curves regarding 10 mm modification and about 10% improvement concerning 6 mm range
ICTR-PHE 2012 larger discrepancies between calculated and delivered dose distributions are with large leaf travel and large dose-rate modulation. This is the basis for approaches that use more finely interpolated dose calculation points for aperture and dose-rate but with fewer source positions. X-ray scatter: A major challenge is the dose computation time and memory requirements for the large number of beams used to model VMAT dose delivery. A common approach in fixed-angle IMRT optimization is to use dose deposition coefficients for computing dose from individual beamlets. The large number of beams imposes large memory requirements for beamlet storage. To address this, the lateral extent of the dose contribution of each beamlet is limited, thus the scatter contribution is less accurately computed, resulting in discrepancies between dose during optimization and dose computed with full scatter following optimization. To this end, strategies have evolved to apply corrections to the scatter dose as the optimization progresses, for example, by periodically computing dose with full scatter and applying corrections to the deposition coefficients. Alternatively, dose is calculated using a fast convolution algorithm at each iteration. 93 PREDICTION OF RADIOSENSITIVITY USING PHOSPHORYLATION OF HISTONE H2AX H. Kunogi, K. Sasai Juntendo University, Tokyo Japan Purpose: Predictive assays for radio-sensitivity are an important issue in radiation biology. Although many methods have been proposed, none have proven clinically useful yet. DNA double-strand breaks are the most important damage produced by radiation and relate directly to cell death. Phosphorylation of histone H2AX (γH2AX) occurs rapidly in response to the presence of DNA double-strand breaks and recruits repair enzymes to these sites. Therefore, some investigators demonstrated a good relationship between frequency of γH2AX and cell survival after radiation. However, others did not show such a correlation. Previously, some authors revealed a good relationship between the frequency of micronuclei and cell survival if they normalized the frequency using DNA content of each cell line. Therefore, we introduced DNA content of each cell line into the analysis of the relationship between radio-sensitivity and expression of γH2AX foci. Methods and Materials: Seven human tumor cell lines, including a fibrosarcoma (HT 1080), 4 esophagus carcinoma (TE-9, KYSE30, KYSE150, and KYSE220), and 2 breast carcinoma (HCC70, and ZR75-1) were used. Cellular radio-sensitivity was assessed by a standard colony-forming assay. To demonstrate the frequency of γH2AX foci in each cell, we counted the number of cells under fluorescence microscopy after immunofluorescence staining. DNA content of each cell line was determined by a flow cytometry assay. Results: The frequency of γH2AX foci after irradiation (4 Gy) in HT1080 cell showed a peak at 3-4 hours after irradiation and gradually decreased up to 24 hours. The 7 cell lines tested here showed dose (0-9 Gy) dependent increases of γH2AX foci per cell at 24 hours after irradiation. However, there was not a good correlation between the γH2AX foci frequency and surviving
S36 fraction after irradiation. When we normalized the γH2AX foci frequency by DNA content, there was a good correlation between the frequency and cell survival for 5 cell lines, but not 2 cell lines. Conclusions: It is difficult to predict cell survival using the γH2AX foci frequency, even after normalization of DNA content. We likely need to introduce other factors, such as an apoptotic cell frequency, in addition to DNA content of each cell line. 94 REGULATION OF PARACRINE SIGNALING BY MICROTUBULE STABILIZING AGENTS AND IONIZING RADIATION M. Pruschy University Hospital Zurich, Switzerland Ionizing radiation (IR) affects multiple cellular components, which induce a multilayered tumor stress response. Para- and autocrine factors are released into the tumor microenvironment in response to irradiation and treatment-activated intracellular stress-responses. During the time course of a fractionated radiation regimen they modulate thereby the tumor microenvironment and these processes co-determine the treatment sensitivity of the tumor and eventually treatment outcome. Using a panel of genetically defined tumor cells, derived from different tumor entities, we analyzed the expression and secretion of biologically active factors in response to IR. Such IRactivated modulators of tumor angiogenesis and tumor cell migration eventually represent interesting novel targets for combined treatment modalities. With a specific focus on microtubule stabilizing agents (MSA) we further investigated deregulation of important components of the tumor microenvironment by the combined treatment modality of IR with the clinically relevant MSA patupilone. In a lung carcinoma tumor model, IR alone induced hypoxia-dependent VEGFexpression. A strong dominant counteracting effect of patupilone was observed when IR was combined with patupilone, on the level of HIF-1a protein stability, VEGF-expression and VEGF-secretion, but only in patupilone-sensitive and not in patupilone-resistant tumor cell lines. Patupilone and IR thereby dysregulated hypoxia-induced tumor angiogenesis, contributing to the potency of this promising combined treatment modality. IR also enhanced the enzymatic activity of secreted matrix metalloproteinases and increased the invasive capacity of several tumor cells. Interestingly low dose patupilone-pretreatment also counteracted specifically IR-induced enhanced MMP activity and patupilone pretreatment completely abrogated IR-induced cell invasion. This effect is mainly regulated by the secretion of the corresponding tissue inhibitors of metalloproteinases (TIMPs). These results indicate that multiple proangiogenic and pro-invasive factors are specifically induced by ionizing radiation. Combined treatment modalities of IR with pharmacological agents, which prevent the secretion of these factors, represent a promising treatment approach. 95 AUTOMATED DETECTION OF ION BEAM MODIFICATIONS IN IN-BEAM PET IMAGES P. Kuess1, W. Birkfellner2, S. Helmbrecht3, F. Fiedler4,
ICTR-PHE 2012 W. Enghardt3,4, D. Georg1 1 Department of Radiotherapy, Div. Medical Radiation Physics, Medical University Vienna / AKH Vienna, Austria 2 Department of Biomedical Engineering and Physics, MUW, Austria 3 OncoRay, TU Dresden, Germany 4 Helmholtz-Zentrum Dresden-Rossendorf, Germany Aim: Positron emission tomography (PET) is up to date the only clinically approved method to verify the range in ion beam therapy. Due to the different physical characteristics of delivered dose and measured activity Monte Carlo (MC) simulations has to be performed out of the treatment plan and the time structure of the irradiation to obtain a prediction of the β+-activity distributions. This simulation has to be compared to the measured PET images to evaluate beam-delivery. Until now the range assessment is performed by a group of experts via visual inspection. This procedure is very time consuming and requires well trained personnel. Hence, an automated objective comparison tool is highly desired for an efficient clinical workflow. Methods: 12 in-beam PET (ibPET) data sets were investigated during this study. Artificial range modifications of simulated β+-activity distributions were used, as described by Fiedler et al. (1). The simulations were altered to obtain range modifications of 4 mm, 6 mm and 10 mm water equivalent path length both in positive and negative beam direction as well as 0 mm to establish a discrimination level. To account for MC based fluctuations for each type of modification 10 simulations were performed. Different methods taken from image processing were applied on these data sets. The most promising results were obtained by calculating the Pearson’s correlation coefficient (PCC). Prior a median filter with a kernel of 11 pixels in all three spatial directions was applied on all images to reduce noise. For the calculation of the PCC both an unmodified simulation and an ibPET measurement were used as reference input. Based on a variable threshold, depending on the PCC, positive and negative values were classified for each patient. With these data sensitivity and specificity curves were calculated. Hence, Receiver Operating Characteristics (ROC) curves were used as a quality factor of the results. To compare ROC curves the area under the ROC curve (AUC) was taken as quality standard. Additionally statistical significance between PCC of modified and unmodified data was calculated (p < 0.05) using the Wilcoxon rank sum test. Results: Using ibPET measurements as reference images the PCC were significantly different regarding the unmodified and modified data sets for all modifications of +/-10 mm. Two third of +/-6 mm modified data sets showed a significant difference and half of the results of +/-4 mm modification. The resulting ROC curves are illustrated in figure 1 which shows a sensitivity of 90-95% for +/-10 mm modification and a specificity of 90-100%. For +/-6 mm alteration a sensitivity of about 75% and a specificity of 80- 90% was obtained. A range deviation of +/-4 mm defines the limitation of the method which is also close to the overall uncertainty of the ibPET method. The use of an unmodified simulation as reference showed similar ROC curves regarding 10 mm modification and about 10% improvement concerning 6 mm range