Volume 34, Issue 6, June 2007
Index of content:
- Therapy Scientific Session: Room M100J
Radiobiology: Fundamental and Outcome Modeling
WE‐D‐M100J‐01: Enhancement of Cell Killing by X‐Ray Irradiation in the Presence of Gold Nanoparticles34(2007); http://dx.doi.org/10.1118/1.2761562View Description Hide Description
Purpose: This investigation was initiated to determine the radiobiologic enhancement for x‐ray irradiation of cells in the presence of citrate‐stabilized goldnanoparticles.Method and materials: Citrate‐stabilized goldnanoparticles were added to a strain of lymphoma cells in suspension at a concentration of 0.017 mg of gold per ml. Cell suspensions prepared included a two sham controls, and two samples with goldnanoparticles. One sham and one goldnanoparticle based sample were irradiated with x‐ray photons from a clinical x‐ray device (120 kVp, 6.67 mm Al equivalent beam). A total dose of 0.91 Gy was delivered in two approximately equal parts (24 hour separation). The non‐irradiated cell suspensions were treated identically. Cell survival for non‐irradiated cell suspensions and irradiated cell suspensions were determined over 72 hours using methylene blue dye exclusion. Results: The ratio of live:dead cells for all cell suspensions was 7.5:1 at the start of the experiment. After 72 hours, the irradiated cells exhibited a ratio approximately 50% lower than the non‐irradiated cells. The irradiated cell suspension with goldnanoparticles showed enhanced cell killing. Conclusions: Even at the low gold concentration and low x‐ray energy, goldnanoparticles exhibit some enhancement of cell killing. Further studies are ongoing using increasing gold concentrations and varying radiationdoses. It is likely that the presence of higher Z materials contributes to secondary electrondose in a way that may be described only through microdosimetry methods.
Goldnanoparticle‐enhancedradiation therapy is a possible outcome of this work. Additional effort is directed toward functionalized goldnanoparticles using targeting moieties for specific cancers. This work contributes toward an understanding of the therapeutic enhancement that may be expected with specific concentrations of goldnanoparticles.
Supported by: NIH R01 CA119412‐01.
WE‐D‐M100J‐02: A Closed Form of Linear‐Quadratic Model with Reciprocal‐Time for Radiation Damage Repair34(2007); http://dx.doi.org/10.1118/1.2761563View Description Hide Description
Purpose: Repair of sublethal damage plays an important role in radiation therapy. Conventional radiobiology concepts are based on the assumption that the repair rate remains constant during the entire radiation course. However, increasing evidence from animal experiments report that the repair process may slow down with time and the data does not fit an exponential pattern. To address this dilemma, we developed a new closed form of Linear‐Quadratic (LQ) model based on the repair pattern with a reciprocal time. The new formulas were tested with published experimental data. Methods and Materials: The LQ model has been widely used in radiation therapy,. The parameter G represents the repair process of sublethal damage with Tr as the repair half‐time: in the reciprocal pattern. A closed form of G was derived analytically for arbitrary radiation schemes: dose rate . A set of published animal data was adopted to test the reciprocal formulas, in which rat foot skin was irradiated in split‐doses with increasing time intervals (0–22h). The complication data of moist desquamation were analyzed using the generalized LQ model. Results: A closed form of the LQ model to describe the repair process in a reciprocal pattern was obtained: , where function . Formulas for special cases were derived from this general form. The reciprocal model showed a better fit to the animal data than the exponential model, especially for the ED50 data (reduced of 2.0 vs. 4.3, p=0.11 vs. 0.006), with the following LQ parameters: α/β = 2.6–4.8Gy, Tr = 3.2–3.9h. Conclusions: The repair process following a reciprocal‐time has been investigated in this study and a closed form of the generalized LQ model was presented and validated. These formulas can be used to analyze the experimental and clinical data, where a slowing‐down repair process appears during the radiation course.
34(2007); http://dx.doi.org/10.1118/1.2761564View Description Hide Description
Purpose: To build and test a feed‐forward neural networkmodel to predict the occurrence of lung radiation‐induced Grade 2+ pneumonitis. Method and Materials: The database comprised 235 patients with lungcancertreated using radiotherapy (34 diagnosed with pneumonitis). The neural network was constructed using a unique algorithm that alternately grew and pruned it, starting from the smallest possible network, until a satisfactory solution was found. The weights and biases of the network were computed using the error back‐propagation approach. The network was tested using ten‐fold cross‐validation, wherein 1/10th of the data were tested, in turn, using the model built with the remaining 9/10th of the data. Results: The network was constructed with input features selected from dose and non‐dose variables. The selected input features were: lung volume receiving > 16 Gy (V16), mean lungdose, generalized equivalent uniform dose (gEUD) for the exponent a=3.5, free expiratory volume in 1s (FEV1), diffusion capacity of Carbon Monoxide (DLCO%), and whether or not the patient underwent chemotherapy prior to radiotherapy. With the exception of FEV1, all input features were found to be individually significant (p < 0.05). The area under the Receiver Operating Characteristics (ROC) curve for cross‐validated testing was 0.76 (sensitivity: 68%, specificity: 69%). To gauge the impact of non‐dose variables on model predictive capability, a second network was constructed with input features selected only from lungdose‐volume histogram variables. The area under the ROC curve for cross‐validation was 0.67 (sensitivity: 53%, specificity: 69%). The network constructed from dose and non‐dose variables was statistically superior (p=0.020), indicating that the addition of non‐dose features significantly improves the generalization capability of the network. Conclusions: The neural network constructed from dose and non‐dose variables can be used to prospectively predict radiotherapy‐induced pneumonitis and, thereby, appropriately alter radiotherapy plans to reduce this possibility.
WE‐D‐M100J‐04: Prediction of Lung Radiation‐Induced Pneumonitis Using the Support Vector Machine Algorithm34(2007); http://dx.doi.org/10.1118/1.2761565View Description Hide Description
Purpose: To build and test a Support Vector Machine (SVM) model to predict the occurrence of lung radiation‐induced Grade 2+ pneumonitis. SVM is a sophisticated statistical technique that is capable of using complex hypersurfaces to separate the cases with and without pneumonitis. Method and Materials: Two SVM models were built using data from 235 patients with lungcancertreated using radiotherapy (34 diagnosed with pneumonitis). One model (SVMall) selected input features from all dose‐volume and non‐dose factors. For comparison, the other model (SVMdose) selected input features only from lungdose‐volume factors. The models were built with in‐house developed software that employed a unique strategy to sequentially add/remove/substitute features. The SVM models were tested using ten‐fold cross‐validation, wherein 1/10th of the data were tested, in turn, using the model built with the remaining 9/10th of the data. Results: The input features selected to build SVMall were the lung generalized equivalent uniform dose (EUD) with exponents a=1.2, 1.3, 1.4, chemotherapy prior to radiotherapy (yes/no), tumor location (central/peripheral), gender, and histology (adenocarcinoma/other; small cell/other). The input features for SVMdose were EUD a = 1.1, 1.3, 1.4, lung volume receiving > 48 Gy (V48), and V50. Both models selected EUD a ≈ 1 (EUD a=1 is the mean lungdose, which frequently appears as a strong predictor of radiation pneumonitis in literature). The area under the cross‐validated SVMall Receiver Operating Characteristics curve was 0.76 (sensitivity/specificity = 74%/75%), compared to the corresponding SVMdose area of 0.71 (sensitivity/specificity = 68%/68%). SVMall was statistically superior (p=0.01), indicating that non‐dose features significantly contribute to separating patients with and without pneumonitis. Conclusions: The SVM model constructed from dose and non‐dose input factors is a valuable prospective tool for predicting the occurrence of radiation‐induced lung pneumonitis.
WE‐D‐M100J‐05: Estimation of Tumor Control Probability Model Parameters for Early Stage Non‐Small‐Cell Lung Cancer (NSCLC)34(2007); http://dx.doi.org/10.1118/1.2761566View Description Hide Description
Purpose: To estimate parameters for the tumorcontrol probability (TCP) model based on the linear‐quadratic (LQ) cell survival formalism from published clinical data for early stage NSCLC and to investigate potential outcomes of alternate dose fractionation schemes. Method and Materials: A comprehensive literature search was performed to identify studies reporting local control after conventionally fractionated 3DCRT and hypofractionated stereotactic body radiation therapy(SBRT) for T1‐2N0M0 NSCLC. Parameters of the LQ‐based TCP model were estimated by fitting the selected clinical data sets using the maximum likelihood method. Fits were obtained and compared for three versions of the TCP model: without the time effect, with an exponential growth of clonogens, and with delayed onset of clonogen proliferation. Obtained parameter estimates were used to compare different fractionation schemes that take advantage of increase in tumorcontrol as the overall treatment time is reduced. Results: A fit to the exponential growth model was significantly better (p < 0.0001) than a fit to the model with no time effect. Also, the model with delayed onset provided a significantly better fit (p = 0.0142) to the data than the exponential growth model assuming no delay in clonogen repopulation. Parameter estimates for the model with delayed onset were: α = 0.066 Gy−1, α/β = 21.1 Gy, γ = 1.85 Gy/day, T k = 32.2 days, and k (the number of clonogens) = 6.3. Our analysis of different hypofractionated regiments confirms observations by others and suggests that a moderate reduction in the number of fractions (with the increase in the dose per fraction) may result in better tumorcontrol for the same level of late complications.
Conclusion: The derived TCP model parameter estimates may prove useful for biologically based treatment planning of Stage I NSCLC, especially when unconventional fractionation schemes, such as in SBRT, are employed.
WE‐D‐M100J‐06: Hypofractionation Results in a Decrease in Tumor Cell Killing Compared to Standard Fractionation as a Result of Tumor Hypoxia34(2007); http://dx.doi.org/10.1118/1.2761567View Description Hide Description
Purpose:Tumor hypoxia has been observed in many human cancers and has been shown to correlate with treatment failure in radiotherapy. The purpose of this study is to quantify the effect of radiation fractionation on tumor cell killing assuming a realistic distribution of tumor oxygenation and full reoxygenation between fractions. The sensitivity of the results to variations in the radiobiologically hypoxic fraction, the dose per fraction, and tumor cell intrinsic radiosensitivity is evaluated. Method and Materials: A probability density function for the partial pressure of oxygen in a tumor cell population is quantified as a function of radial distance from the capillary wall. Estimates of the oxygen partial pressures for subpopulations of tumor cells are used to determine the corresponding oxygen enhancement ratios (OERs) for cell killing. The overall surviving fraction of a tumor cell population consisting of maximally resistant cells, cells at intermediate levels of hypoxia, and well‐oxygenated cells is calculated as a function of dose per fraction for an equivalent tumor biological effective dose (BED). Results: Our model predicts that tumor cell killing decreases by a factor of 105 over a radial distance of 130 μm assuming a partial oxygen pressure of 60 mmHg at the capillary wall. For head and neck (α/β = 10 Gy) and prostate (α/β = 3.0 Gy) cancer, the surviving fraction of cells over a full treatment course increases by a factor of 103 as the dose per fraction is increased from 1–24 Gy and 1–18 Gy, respectively. The total dose delivered for each dose per fraction is calculated to achieve equivalent tumor BED values for reference head and neck and prostate treatments.Conclusion: Hypofractionation of a radiotherapy regimen results in reduced tumor cell killing compared to conventional fractionation for tumors with regions of hypoxia.
34(2007); http://dx.doi.org/10.1118/1.2761568View Description Hide Description
Purpose: Hemoglobin (Hgb) has been considered as an important factor on radiation therapy for cervical cancer. However, recent studies reported that anemia is only a secondary symptom associated with disease stage and tumor volume, but not an independent predictor for outcome. Our study is to investigate whether low Hgb actively influences the effectiveness of radiation therapy or is merely a symptom of tumor burden. Materials/Methods: The study was based on 67 cervical cancer patients (stage IB2‐IVA). Serial Hgb measurements were performed before and during therapy. Tumor volume was obtained from 3D MRI volumetry. Median follow‐up was 2.9 years (range 0.09∼6.4 years). Treatment outcome was evaluated with local tumor‐control and disease‐free survival. Correlations between Hgb parameters and clinical variables were evaluated with Spearman‐rank method. Outcome prediction was based on univariate and multivariate analyses with Cox regression model. Survival analysis was assessed using Kaplan‐Meier method. Results: Hgb levels were not correlated with tumor stage (correlation coefficient −0.1∼ −0.2) and weakly correlated with tumor volume (correlation coefficient −0.4∼ −0.5). In univariate analysis, local tumor‐control was predicted by mean Hgb (p=0.009) and nadir Hgb (p=0.004); disease‐free survival was only predicted by nadir Hgb (p=0.031). In multivariate analysis, Hgb parameters were the best predictors of outcome: local tumor‐control was predicted by mean Hgb (p=0.013) and nadir Hgb (p=0.007); disease‐free survival was only marginally predicted by nadir Hgb (p=0.053). Kaplan‐Meier analysis demonstrated that Hgb (>12g/dl vs <12g/dl) could potentially differentiate patients in terms of local tumor‐control and disease‐free survival. Conclusion: Preliminary results suggest that Hgb may not be associated with tumor burden (volume and stage). This early work suggests that chronically low Hgb during RT course negatively affects outcome. The independent role of Hgb in combination with other factors may be useful in outcome prediction.
34(2007); http://dx.doi.org/10.1118/1.2761569View Description Hide Description
Purpose: To investigate cell survival following exposure to spatially modulated beams, as created by intensity modulated radiotherapy(IMRT), using in vitro experiments. Method and Materials: We compared cell survival in modulated fields with cell survival in a uniform control field using malignant melanoma cells (MM576) exposed to a therapeutic megavoltage photon beam. Three different spatial modulations of the field were used: a control “Open” field in which all cells in a flask were uniformly exposed; a “Quarter” field in which 25% of cells at one end of the flask were exposed; and a “Striped” field in which 25% of cells were exposed in three parallel stripes. The cell survival in both the shielded and unshielded regions of the modulated fields, as determined by a clonogenic assay, were compared to the cell survival in the Open field. Results: In the unshielded regions of the irradiated flasks, the cell survival was seen to differ between the three fields for the same delivered dose. For the modulated fields, the regions which received only scattered and leakage dose displayed decreased survival at lower doses, relative to the “Open” field, and increased survival at higher doses.Conclusion: We have found three distinct ways in which the cell survival is influenced by the fate of neighboring cells. The first of these (Type I effect) is the classical Bystander Effect whereby cell survival is reduced when nearby cells receive a high radiationdose but some survive. The Type II effect is an increase in cell survival when nearby cells receive a lethal dose. The Type III effect is an increase in survival for cells receiving a high dose of radiation when nearby cells receive a low dose of radiation. Our observations of the Bystander Effects emphasize the need for improved radiobiological models, which include communicated effects.