Machine-learning-based prediction of the effectiveness of the delivered dose by exhale-gated radiotherapy for locally advanced lung cancer: The additional value of geometric over dosimetric parameters alone.

Purpose: This study aimed to assess interfraction stability of the delivered dose distribution by exhale-gated volumetric modulated arc therapy (VMAT) or intensity-modulated arc therapy (IMAT) for lung cancer and to determine dominant prognostic dosimetric and geometric factors. Methods: Clinical target volume (CTVPlan) from the planning CT was deformed to the exhale-gated daily CBCT scans to determine CTVi, treated by the respective dose fraction. The equivalent uniform dose of the CTVi was determined by the power law (gEUDi) and cell survival model (EUDiSF) as effectiveness measure for the delivered dose distribution. The following prognostic factors were analyzed: (I) minimum dose within the CTVi (Dmin_i), (II) Hausdorff distance (HDDi) between CTVi and CTVPlan, (III) doses and deformations at the point in CTVPlan at which the global minimum dose over all fractions per patient occurs (PDmin_global_i), and (IV) deformations at the point over all CTVi margins per patient with the largest Hausdorff distance (HDPworst). Prognostic value and generalizability of the prognostic factors were examined using cross-validated random forest or multilayer perceptron neural network (MLP) classifiers. Dose accumulation was performed using back deformation of the dose distribution from CTVi to CTVPlan. Results: Altogether, 218 dose fractions (10 patients) were evaluated. There was a significant interpatient heterogeneity between the distributions of the normalized gEUDi values (p<0.0001, Kruskal-Wallis tests). Accumulated gEUD over all fractions per patient was 1.004-1.023 times of the prescribed dose. Accumulation led to tolerance of ~20% of fractions with gEUDi <93% of the prescribed dose. Normalized Dmin >60% was associated with predicted gEUD values above 95%. Dmin had the highest importance for predicting the gEUD over all analyzed prognostic parameters by out-of-bag loss reduction using the random forest procedure. Cross-validated random forest classifier based on Dmin as the sole input had the largest Pearson correlation coefficient (R=0.897) in comparison to classifiers using additional input variables. The neural network performed better than the random forest classifier, and the gEUD values predicted by the MLP classifier with Dmin as the sole input were correlated with the gEUD values characterized by R=0.933 (95% CI, 0.913-0.948). The performance of the full MLP model with all geometric input parameters was slightly better (R=0.952) than that based on Dmin (p=0.0034, Z-test). Conclusion: Accumulated dose distributions over the treatment series were robust against interfraction CTV deformations using exhale gating and online image guidance. Dmin was the most important parameter for gEUD prediction for a single fraction. All other parameters did not lead to a markedly improved generalizable prediction. Dosimetric information, especially location and value of Dmin within the CTV i , are vital information for image-guided radiation treatment.

Spatial and dosimetric evaluation of residual distortions of prostate and seminal vesicle bed after image-guided definitive and postoperative radiotherapy of prostate cancer with endorectal balloon.

PURPOSE: To quantify daily residual deviations from the planned geometry after image-guided prostate radiotherapy with endorectal balloon and to evaluate their effect on the delivered dose distribution. METHODS: Daily kV-CBCT imaging was used for online setup-correction in six degrees of freedom (6-dof) for 24 patients receiving definitive (12 RTdef patients) or postoperative (12 RTpostop patients) radiotherapy with endorectal balloon (overall 739 CBCTs). Residual deviations were evaluated using several spatial and dosimetric variables, including: (a) posterior Hausdorff distance HDpost (=maximum distance between planned and daily CTV contour), (b) point Pworst with largest HDpost over all fractions, (c) equivalent uniform dose using a cell survival model (EUDSF ) and the generalized EUD concept (gEUDa with parameter a = -7 and a = -20). EUD values were determined for planned ( EUD SF plan ), daily ( EUD SF ind ), and delivered dose distributions ( EUD SF accum ) for plans with 6 mm (=clinical plans) and 2 mm CTV-to-PTV margin. Time series analyses of interfractional spatial and dosimetric deviations were conducted. RESULTS: Large HDpost values ≥ 12.5 mm (≥15 mm) were observed in 20/739 (5/739) fractions distributed across 7 (3) patients. Points Pworst were predominantly located at the posterior CTV boundary in the seminal vesicle region (16/24 patients, 6/7 patients with HDpost  ≥ 12.5 mm). Time series analyses revealed a stationary white noise characteristic of HDpost and relative dose at Pworst . The EUDSF difference between planned and accumulated dose distributions was < 5.4% for all 6-mm plans. Evaluating 2-mm plans, EUDSF deteriorated by < 10% (<5%) in 75% (58.5%) of the patients. EUD SF accum was well described by the median value of the EUD SF ind distribution. PTV margin calculation at Pworst yielded 8.8 mm. CONCLUSIONS: Accumulated dose distributions in prostate radiotherapy with endorectal balloon are forgiving of considerable residual distortions after 6-dof patient setup if they are observed in a minority of fractions and the median value of EUD SF ind determined per fraction stays within 95% of prescribed dose. Common PTV margin calculations are overly conservative because after online correction of translational and rotational errors only residual deformations need to be included. These results provide guidelines regarding online navigation, margin optimization, and treatment adaptation strategies.

Multi-scenario based robust intensity-modulated proton therapy (IMPT) plans can account for set-up errors more effectively in terms of normal tissue sparing than planning target volume (PTV) based intensity-modulated photon plans in the head and neck region.

BACKGROUND: In a previous report, we compared the conformity of robust intensity-modulated proton therapy (IMPT) plans with that of helical tomotherapy plans for re-irradiations of head and neck carcinomas using a fixed set-up error of 2 mm. Here, we varied the maximum set-up errors between 0 and 5 mm and compared the robust IMPT-plans with planning target volume (PTV) based intensity-modulated photon therapy (IMRT). FINDINGS: Seven patients were treated with a PTV-based tomotherapy plan. Set-up margins of 0, 2, and 5 mm were subtracted from the PTV to generate target volumes (TV) TV(0mm), TV(2mm), and TV(5mm), for which robust IMPT-plans were created assuming range uncertainties of ±3.5% and using worst case optimization assuming set-up errors of 0, 2, and 5 mm, respectively. Robust optimization makes use of the feature that set-up errors in beam direction alone do not affect the distal and proximal margin for that beam. With increasing set-up errors, the body volumes that were exposed to a selected minimum dose level between 20% and 95% of the prescribed dose decreased. In IMPT-plans with 0 mm set-up error, the exposed body volumes were on average 6.2% ± 0.9% larger than for IMPT-plans with 2 mm set-up error, independent of the considered dose level (p < 0.0001, F-test). In IMPT-plans accounting for 5 mm set-up error, the exposed body volumes were by 11.9% ± 0.8% smaller than for IMPT-plans with 2 mm set-up error at a fixed minimum dose (p < 0.0001, F-test). This set-up error dependence of the normal tissue exposure around the TV in robust IMPT-plans corresponding to the same IMRT-plan led to a decrease in the mean dose to the temporal lobes and the cerebellum, and in the D2% of the brain stem or spinal cord with increasing set-up errors considered during robust IMPT-planning. CONCLUSIONS: For recurrent head and neck cancer, robust IMPT-plan optimization led to a decrease in normal tissue exposure with increasing set-up error for target volumes corresponding to the same PTV.

Helical tomotherapy for whole-brain irradiation with integrated boost to multiple brain metastases: evaluation of dose distribution characteristics and comparison with alternative techniques.

PURPOSE: To quantitatively evaluate dose distribution characteristics achieved with helical tomotherapy (HT) for whole-brain irradiation (WBRT) with integrated boost (IB) to multiple brain metastases in comparison with alternative techniques. METHODS AND MATERIALS: Dose distributions for 23 patients with 81 metastases treated with WBRT (30 Gy/10 fractions) and IB (50 Gy) were analyzed. The median number of metastases per patient (N(mets)) was 3 (range, 2-8). Mean values of the composite planning target volume of all metastases per patient (PTV(mets)) and of the individual metastasis planning target volume (PTV(ind met)) were 8.7 ± 8.9 cm(3) (range, 1.3-35.5 cm(3)) and 2.5 ± 4.5 cm(3) (range, 0.19-24.7 cm(3)), respectively. Dose distributions in PTV(mets) and PTV(ind met) were evaluated with respect to dose conformity (conformation number [CN], RTOG conformity index [PITV]), target coverage (TC), and homogeneity (homogeneity index [HI], ratio of maximum dose to prescription dose [MDPD]). The dependence of dose conformity on target size and N(mets) was investigated. The dose distribution characteristics were benchmarked against alternative irradiation techniques identified in a systematic literature review. RESULTS: Mean ± standard deviation of dose distribution characteristics derived for PTV(mets) amounted to CN = 0.790 ± 0.101, PITV = 1.161 ± 0.154, TC = 0.95 ± 0.01, HI = 0.142 ± 0.022, and MDPD = 1.147 ± 0.029, respectively, demonstrating high dose conformity with acceptable homogeneity. Corresponding numbers for PTV(ind met) were CN = 0.708 ± 0.128, PITV = 1.174 ± 0.237, TC = 0.90 ± 0.10, HI = 0.140 ± 0.027, and MDPD = 1.129 ± 0.030, respectively. The target size had a statistically significant influence on dose conformity to PTV(mets) (CN = 0.737 for PTV(mets) ≤4.32 cm(3) vs CN = 0.848 for PTV(mets) >4.32 cm(3), P=.006), in contrast to N(mets). The achieved dose conformity to PTV(mets), assessed by both CN and PITV, was in all investigated volume strata well within the best quartile of the values reported for alternative irradiation techniques. CONCLUSIONS: HT is a well-suited technique to deliver WBRT with IB to multiple brain metastases, yielding high-quality dose distributions. A multi-institutional prospective randomized phase 2 clinical trial to exploit efficacy and safety of the treatment concept is currently under way.

Re-irradiation of recurrent head and neck carcinomas: comparison of robust intensity modulated proton therapy treatment plans with helical tomotherapy.

BACKGROUND: To test the hypothesis that the therapeutic ratio of intensity-modulated photon therapy using helical tomotherapy (HT) for retreatment of head and neck carcinomas can be improved by robust intensity-modulated proton therapy (IMPT). METHODS: Comparative dose planning with robust IMPT was performed for 7 patients retreated with HT. RESULTS: On average, HT yielded dose gradients steeper in a distance ≤ 7.5 mm outside the target (p<0.0001, F-test) and more conformal high dose regions down to the 50% isodose than IMPT. Both methods proved comparably robust against set-up errors of up to 2 mm, and normal tissue exposure was satisfactory. The mean body dose was smaller with IMPT. CONCLUSIONS: IMPT was found not to be uniformly superior to HT and the steeper average dose fall-off around the target volume is an argument pro HT under the methodological implementations used. However, looking at single organs at risk, the normal tissue sparing of IMPT can surpass tomotherapy for an individual patient. Therefore, comparative dose planning is recommended, if both methods are available.

Megavoltage computed tomography image guidance with helical tomotherapy in patients with vertebral tumors: analysis of factors influencing interobserver variability.

PURPOSE: To evaluate megavoltage computed tomography (MVCT)-based image guidance with helical tomotherapy in patients with vertebral tumors by analyzing factors influencing interobserver variability, considered as quality criterion of image guidance. METHODS AND MATERIALS: Five radiation oncologists retrospectively registered 103 MVCTs in 10 patients to planning kilovoltage CTs by rigid transformations in 4 df. Interobserver variabilities were quantified using the standard deviations (SDs) of the distributions of the correction vector components about the observers' fraction mean. To assess intraobserver variabilities, registrations were repeated after ≥4 weeks. Residual deviations after setup correction due to uncorrectable rotational errors and elastic deformations were determined at 3 craniocaudal target positions. To differentiate observer-related variations in minimizing these residual deviations across the 3-dimensional MVCT from image resolution effects, 2-dimensional registrations were performed in 30 single transverse and sagittal MVCT slices. Axial and longitudinal MVCT image resolutions were quantified. For comparison, image resolution of kilovoltage cone-beam CTs (CBCTs) and interobserver variability in registrations of 43 CBCTs were determined. RESULTS: Axial MVCT image resolution is 3.9 lp/cm. Longitudinal MVCT resolution amounts to 6.3 mm, assessed as full-width at half-maximum of thin objects in MVCTs with finest pitch. Longitudinal CBCT resolution is better (full-width at half-maximum, 2.5 mm for CBCTs with 1-mm slices). In MVCT registrations, interobserver variability in the craniocaudal direction (SD 1.23 mm) is significantly larger than in the lateral and ventrodorsal directions (SD 0.84 and 0.91 mm, respectively) and significantly larger compared with CBCT alignments (SD 1.04 mm). Intraobserver variabilities are significantly smaller than corresponding interobserver variabilities (variance ratio [VR] 1.8-3.1). Compared with 3-dimensional registrations, 2-dimensional registrations have significantly smaller interobserver variability in the lateral and ventrodorsal directions (VR 3.8 and 2.8, respectively) but not in the craniocaudal direction (VR 0.75). CONCLUSION: Tomotherapy image guidance precision is affected by image resolution and residual deviations after setup correction. Eliminating the effect of residual deviations yields small interobserver variabilities with submillimeter precision in the axial plane. In contrast, interobserver variability in the craniocaudal direction is dominated by the poorer longitudinal MVCT image resolution. Residual deviations after image guidance exist and need to be considered when dose gradients ultimately achievable with image guided radiation therapy techniques are analyzed.

gamma-H2AX foci formation in peripheral blood lymphocytes of tumor patients after local radiotherapy to different sites of the body: dependence on the dose-distribution, irradiated site and time from start of treatment.

PURPOSE: To evaluate the relationship between an estimated integral total body radiation dose delivered and phosphorylated histone H2AX protein (gamma-H2AX) foci formation in peripheral blood lymphocytes of cancer patients. MATERIAL AND METHODS: gamma-H2AX formation was quantified as the mean number of foci per lymphocyte (N(meanH2AX)) and the percentage of lymphocytes with > or =n foci. The integrated total body radiation dose was estimated from the dose volume histogram of patient's body corrected for the proportion of the body scanned by computed tomography for 3D treatment planning. RESULTS: There was a strong linear correlation between the mean number of gamma-H2AX foci per lymphocyte in the peripheral blood sample and integrated total body radiation dose (r = 0.83, p < 0.0001). The slope of the relationship was dependent on the site of body irradiated. In comparison to chest irradiation with a slope of 8.7 +/- 0.8 foci Gy(-1), the slopes for brain, upper leg and pelvic sites were significantly shallower by -4.7, -4.3, and -3.8 Gy(-1), respectively (p < 0.0001), while the slope for upper abdomen irradiation was significantly larger by 9.1 +/- 2.6 Gy(-1) (p = 0.0007). There was a slight time effect since the start of radiotherapy on the slopes of the in vivo dose responses leading to shallower slopes (-1.5 +/- 0.7 Gy(-1), p = 0.03) later (> or =10 day) during radiotherapy. After in vitro irradiation, lymphocytes showed 10.41 +/- 0.12 foci per Gy with no evidence of inter-individual heterogeneity. CONCLUSIONS: gamma-H2AX measurements in peripheral lymphocytes after local radiotherapy allow the estimation of the applied integral body dose. The site and time dependence have to be considered.

Value of 18F-fluoro-2-deoxy-D-glucose-positron emission tomography/computed tomography in non-small-cell lung cancer for prediction of pathologic response and times to relapse after neoadjuvant chemoradiotherapy.

PURPOSE: To determine the value of combined positron emission tomography/computed tomography (PET/CT) during induction chemotherapy (CTx) followed by chemoradiotherapy (CTx/RTx) for non-small-cell lung cancer to predict histopathologic response in primary tumor and mediastinum and prognosis of the patient. EXPERIMENTAL DESIGN: Fifty consecutive patients with locally advanced non-small-cell lung cancer received induction therapy and, if considered resectable, proceeded to surgery (37 of 50 patients). Patients had at least two repeated 18F-2-fluoro-2-deoxy-D-glucose (FDG)-PET/CT scans either before treatment (t0) or after induction CTx (t1) or CTx/RTx (t2). Variables from the PET/CT studies [e.g., lesion volume and corrected maximum standardized glucose uptake values (SUV(max,corr))] were correlated with histopathologic response (graded as 3, 2b, or 2a: 0%, >0-10%, or >10% residual tumor cells) and times to failure. RESULTS: Primary tumors showed a percentage decrease in SUV(max,corr) during induction significantly larger in grade 2b/3 than in grade 2a responding tumors (67% versus 34% at t(1), 73% versus 49% at t(2); both P < 0.005). SUV(max,corr) at t(2) was significantly correlated with histopathologic response in tumors smaller than the median volume (7.5 cm(3); r = -0.54, P = 0.02). In the mediastinal lymph nodes, SUV(max,corr) values at t2 predicted an ypN0 status with a sensitivity and specificity of 73% and 89%, respectively (SUV(max,corr) threshold of 4.1, r = -0.54, P = 0.0005). Freedom from extracerebral relapse was significantly better in grade 2b/3 patients (86% at 16 months versus 20% in 2a responders; P = 0.003) and in patients with a greater percentage decrease in SUV(max,corr) in the primary tumor at t2 in relation to t0 than in patients with lesser response (83% at 16 months versus 43%; P = 0.03 for cutoff points between 0.45 and 0.55). CONCLUSIONS: SUV(max,corr) values from two serial PET/CT scans, before and after three chemotherapy cycles or later, allow prediction of histopathologic response in the primary tumor and mediastinal lymph nodes and have prognostic value.

Sensorineural hearing loss in combined modality treatment of nasopharyngeal carcinoma.

BACKGROUND: Combined modality therapy has become the standard of care for nasopharyngeal carcinoma, yet the combined ototoxic effects of radiation and cisplatin are poorly understood. The incidence and severity of sensorineural hearing loss (SNHL) with combined modality therapy was evaluated and the dose-response relation between radiation and hearing loss was investigated. METHODS: Patients with newly diagnosed AJCC Stage II-IVB nasopharynx carcinoma treated from 1994-2003 were identified. The records of 44 ears in 22 patients who received a preirradiation pure tone audiogram and followup audiograms 12+ months postirradiation were included in the analysis. All patients were treated with conformal radiotherapy to 70 Gy and received platinum-based chemotherapy similar to the Intergroup 0099 trial. Composite cochlear dose distributions were calculated. Ototoxicity was measured using intrasubject audiogram comparisons and SNHL was defined as per the American Speech and Hearing Association guidelines, with standard range of speech between 2000-4000 Hz. SNHL was analyzed using Fisher exact test and linear and logistic regression models. PATIENT CHARACTERISTICS: median age, 45; 27% Asian; 68% male; 64% WHO III. Median audiologic followup was 29 months (range, 12-76 mos). Mean cochlear dose (Dmean) ranged from 28.4-70.0 Gy (median, 48.5 Gy). SNHL was detected in 25 of the 44 ears (57%) studied. There was an increased risk of SNHL for ears receiving Dmean > 48 Gy compared with those receiving < or = 48 Gy at all frequencies within the range of speech (P = 0.04). Using univariate logistic regression analysis, Dmean to the cochlea, cycles of cisplatin, and time postradiotherapy were independently significant factors in determining the incidence of SNHL (P = 0.02, P = 0.03, and P = 0.04, respectively). In univariate and multivariate linear regression analysis, Dmean was statistically significant at all frequencies in affecting degree of SNHL, whereas the significance of cisplatin and time was variable. CONCLUSIONS: There was a significant increase in risk of SNHL among patients receiving > 48 Gy, suggesting a threshold in cochlear radiation dose-response in the setting of combined modality therapy. This dose should serve as a Dmean constraint maximum for intensity-modulated radiotherapy treatment of nasopharynx carcinoma.

Radiation-induced changes of brain tissue after radiosurgery in patients with arteriovenous malformations: dose/volume-response relations.

PURPOSE: . To evaluate late radiation effects in the brain after radiosurgery of patients with cerebral arteriovenous malformations (AVMs) and to quantify dose/volume-response relations for radiation-induced changes of brain tissue identified on follow-up neuroimaging. PATIENTS AND METHODS: . Data from 73 AVM patients who had stereotactic Linac radiosurgery at DKFZ (German Cancer Research Center), Heidelberg, Germany, were retrospectively analyzed. The endpoint of radiation-induced changes of brain tissue on follow-up magnetic resonance (MR) neuroimaging (i.e., edema and blood-brain-barrier breakdown [BBBB]) was evaluated. Each endpoint was further differentiated into three levels with respect to the extent of the image change (small, intermediate, and large). A previous analysis of the data found correlation of the endpoints with several dose/volume variables (DV) derived from each patient's dose distribution in the brain, including the mean dose in a volume of 20 cm(3) (Dmean20) and the absolute brain volume (including the AVM target) receiving a dose of at least 12 Gy (V12). To quantify dose/volume-response relations, patients were ranked according to DV (i.e., Dmean20 and V12) and classified into four groups of equal size. For each group, the actuarial rates of developing the considered endpoints within 2.5 years after radiosurgery were determined from Kaplan-Meier estimates. The dose/volume-response curves were fitted with a sigmoid-shape logistic function and characterized by DV(50), the dose for a 50% incidence, and the slope parameter k. RESULTS: . Dose/volume-response relations, based on two alternative, but correlated, dose distribution variables that are a function of both dose and volume, were observed for radiation-induced changes of brain tissue. DV(50) values of fitted dose/volume-response curves for tissue changes of large extent (e.g., V12(50) = 22.0 +/- 2.6 cm(3) and Dmean20(50) = 17.8 +/- 2.0 Gy for the combined endpoint of edema and/or BBBB) were significantly higher than those for small tissue changes (V12(50) = 4.0 +/- 0.3 cm(3) and Dmean20(50) = 7.6 +/- 0.3 Gy). CONCLUSION: . The derived dose/volume-response relations allow to quantitatively assess the risk of radiation-induced changes of brain tissue after radiosurgery in AVM patients. However, further understanding of the mechanism leading to brain tissue changes and their correlation with the desired obliteration is required. This knowledge will eventually help to optimize radiosurgical treatments in AVM patients.

Radiation-induced changes of brain tissue after radiosurgery in patients with arteriovenous malformations: correlation with dose distribution parameters.

PURPOSE: To investigate the correlation of radiation-induced changes of brain tissue after radiosurgery in patients with cerebral arteriovenous malformations (AVMs) with treatment planning and dose distribution parameters. METHODS AND MATERIALS: The data from 73 AVM patients with complete follow-up information who underwent stereotactic linear accelerator radiosurgery at our institution between 1993 and 1998 were analyzed. Patients were treated with 11-14 noncoplanar fields shaped by a micromultileaf collimator. A median dose of 19 Gy (range, 13.3-22 Gy) was prescribed to the 80% isodose, which completely encompassed the target. Patients were followed at 3-month intervals the first year and then every 6 months with MRI and neurologic examinations. No patient developed radiation necrosis. The end point of radiation-induced tissue changes on follow-up neuroimaging (i.e., edema, blood-brain barrier breakdown [BBBB], and edema and/or BBBB combined) was evaluated. Each end point was further differentiated into four levels with respect to the extent of the image change (i.e., small, intermediate, large, and very large). The correlation of each end point was investigated for several treatment planning parameters, including prescribed dose and the absolute size of the AVM target volume. In addition, a number of dose-volume variables were calculated from each patient's dose distribution in the brain, including the mean dose to a specified volume of 16 and 20 cm(3) that was given the highest dose (Dmean16 and Dmean20, respectively), and the absolute and percentage of brain volume (including the AVM target) receiving a dose of at least 8, 10, and 12 Gy (V8-V12, and V8(rel)-V12(rel), respectively). These parameters were also determined excluding the AVM target volume from the considered volume (subscript "excl"). The correlation of all treatment planning and dose-volume parameters with outcome was assessed in univariate Cox proportional hazards models. The results were assessed by p values (statistical significance for p < or =0.05), residual deviance (ResDev) of the fits, and odds ratios. RESULTS: The prescribed dose was not predictive of outcome (p >0.05 for all end points). The AVM target volume correlated significantly with large edema, as well as large edema and/or BBBB. V12 and Dmean20 were significantly associated with all end points, except very large edema and large BBBB. Patients with V12 of 27.6 cm(3) (Dmean20 of 18.9 Gy) had a 2.8-fold (fourfold) higher risk of developing edema and/or BBBB with large extent than those with V12 of 4.2 cm(3) (Dmean20 of 8.4 Gy). For all end points, V12(rel) correlated worse with outcome compared with V12 (e.g., end point of large edema and/or BBBB: ResDev = 85.8 and 86.5 for V12 and V12(rel), respectively). Excluding the AVM target volume from the considered irradiated volume led to only small changes in the resulting correlations (e.g., end point of small edema and/or BBBB: ResDev = 99.0 and 98.7 for V12 and V12(excl), respectively, and ResDev = 96.1 and 96.1 for Dmean20 and Dmean20(excl), respectively). Throughout the analysis, V8-V12, Dmean20, and Dmean16 yielded similar results and none of these parameters could be favored over the others. CONCLUSION: Radiation-induced changes of brain tissue after AVM radiosurgery can be well predicted by single dose distribution parameters that are a function of both dose and volume. These can be used to quantify dose-volume response relations. Studies of this nature will eventually help to improve our current understanding of the mechanisms leading to radiation-induced tissue changes after AVM radiosurgery and to optimize radiosurgery treatment planning.

Modeling and computer simulations of tumor growth and tumor response to radiotherapy.

A model of tumor growth and tumor response to radiation is introduced in which each tumor cell is taken into account individually. Each cell is assigned a set of radiobiological parameters, and the status of each cell is checked in discrete intervals. Tumor proliferation is governed by the cell cycle times of tumor cells, the growth fraction, the apoptotic capacity of the tumor, and the degree of tumor angiogenesis. The response of tumor cells to radiation is determined by the radiosensitivities and the oxygenation status. Computer simulation is performed on a 3D rigid cubic lattice, starting out from a single tumor cell. Random processes are simulated by Monte Carlo methods. Short cell cycle time, high growth fraction, and tumor angiogenesis all increase tumor proliferation rates. Accelerated time-dose patterns result in lower total doses needed for tumor control, but the extent of dose reduction depends on the kinetics and the radiosensitivities of tumor cells. Tumor angiogenesis alters fully oxygenated and hypoxic fractions within the tumor and subsequently affects the radiation response. It is demonstrated for selected radiobiological parameters that the simulation tools are suitable to quantitatively assess the total doses needed for tumor control. Using the simulation tools, it is feasible to simulate time-dependent effects during fractionated radiotherapy and to compare different time-dose patterns in terms of their tumor control.

Risk group dependence of dose-response for biopsy outcome after three-dimensional conformal radiation therapy of prostate cancer.

BACKGROUND AND PURPOSE: We fit phenomenological tumor control probability (TCP) models to biopsy outcome after three-dimensional conformal radiation therapy (3D-CRT) of prostate cancer patients to quantify the local dose-response of prostate cancer. MATERIALS AND METHODS: We analyzed the outcome after photon beam 3D-CRT of 103 patients with stage T1c-T3 prostate cancer treated at Memorial Sloan-Kettering Cancer Center (MSKCC) (prescribed target doses between 64.8 and 81Gy) who had a prostate biopsy performed >or=2.5 years after end of treatment. A univariate logistic regression model based on D(mean) (mean dose in the planning target volume of each patient) was fit to the whole data set and separately to subgroups characterized by low and high values of tumor-related prognostic factors T-stage (or=T2c), Gleason score (6), and pre-treatment prostate-specific antigen (PSA) (10 ng/ml). In addition, we evaluated five different classifications of the patients into three risk groups, based on all possible combinations of two or three prognostic factors, and fit bivariate logistic regression models with D(mean) and the risk group category to all patients. Dose-response curves were characterized by TCD(50), the dose to control 50% of the tumors, and gamma(50), the normalized slope of the dose-response curve at TCD(50). RESULTS: D(mean) correlates significantly with biopsy outcome in all patient subgroups and larger values of TCD(50) are observed for patients with unfavorable compared to favorable prognostic factors. For example, TCD(50) for high T-stage patients is 7Gy higher than for low T-stage patients. For all evaluated risk group definitions, D(mean) and the risk group category are independent predictors of biopsy outcome in bivariate analysis. The fit values of TCD(50) show a clear separation of 9-10.6Gy between low and high risk patients. The corresponding dose-response curves are steeper (gamma(50)=3.4-5.2) than those obtained when all patients are analyzed together (gamma(50)=2.9). CONCLUSIONS: Dose-response of prostate cancer, quantified by TCD(50) and gamma(50), varies by prognostic subgroup. Our observations are consistent with the hypothesis that the shallow nature of clinically observed dose-response curves for local control result from a patient population that is a heterogeneous mixture of sub-populations with steeper dose-response curves and varying values of TCD(50). Such results may eventually help to identify patients, based on their individual pre-treatment prognostic factors, that would benefit most from dose-escalation, and to guide dose prescription.