Which Lung Volumes to Use for Radiotherapy Planning of Lung Cancer: Inspiration, Expiration, Averaged, or Free-breathing?

Purpose/Objective(s)With 4DCT imaging gradually moving into routine practice in the management of thoracic cancers, questions are being raised as to how best to standardize dose-volume reporting for various lung volumes defined using different CT data sets. The goals of this study were (1) to investigate various lung volume definitions and their inter-relationships; (2) to study the impact on lung dose volume histograms (DVHs); and (3) to propose a population-based model for converting one lung volume definition to another for prospective or retrospective dose reporting.Materials/Methods40 retrospectively selected stage III/IV non-small-cell lung cancer patients were included. Each patient had a 4DCT and a fast free-breathing (FB) helical CT scan. All lungs were contoured in the FBCT and in the 4DCT data sets at the end of expiration (EXP) and the end of inspiration (INSP). A special time-averaged CT was created by averaging CT numbers in the same spatial voxel over all reconstructed phases of the 4DCT. This represents the time-averaged lung density (AVE). We compared various lung volumes using EXP CT as the reference volume. In order to compare the dose variations and dose-volume effects for lung tissue, we calculated the dose distribution using the same treatment plan in EXP, INSP, AVE, and FB CT data sets in 15 of 40 patients. We compared the percentages of lung exposed to more than 5 (V5), 20 (V20), and 50 (V50) Gy. In addition, we considered an alternative DVH calculation method that used the same dose distribution calculated on a single CT but normalized to the lung volume on a different CT data set. Finally, we used the population-based lung volumetric relationships in various CT data sets to convert DVHs and compared the predictive accuracy.ResultsOn average, the total lung volumes and masses defined on INSP CT sets were 13.9% and 4.6% (1SD: 5.1% and 1.8%) higher than those defined on EXP CT sets. The average INSP lung density was 8.0% (1SD: 3.4%; range: 1.1–8.8%) lower than the EXP lung density. On average, the FB and AVE total lung volumes were 6.3% and 8.8% (1SD: 3.5% and 5.4%) higher than the lung volume defined on the EXP CT. The average dose differences in the non-zero dose voxels in the lung were 12.3, 6.0, and 29.8 cGy, for the INSP, AVE, and FB CT data sets, respectively, relative to the reference dose distribution calculated on the EXP CT. Compared to the dose prescriptions of 6000 to 6600 cGy, the spatial dose variation was minimal when using different CTs for dose calculation. As a consequence, we found a strong volume effect for the dose-volume relationships using different lung volume definitions. The total lung V5, V20, V50, and EUD were higher in INSP CT than in EXP CT by 3.3%, 2.4%, 2.0%, and 150 cGy (1SD: 6.0%, 7.6%, 2.1%, and 84 cGy; range: −12.15% – +15.80%, −17.36% – +21.88%, −3.36% – +6.30%, and 7 cGy–285 cGy), respectively. We found that using a simple volume scaling factor can effectively convert the DVH results from one CT lung volume into another. The INSP and EXP V5, V20, and V50 values can be predicted more accurately (1SD: <1.4%) from dose distributions calculated on the AVE CT than those calculated from the FB CT (1SD: <3.2%).ConclusionsWe found large variations in reported DVH values when different lung volume definitions were used. However, population-based relationships among different lung volumes can be used to convert DVHs into a more standardized dose-volume definition. Purpose/Objective(s)With 4DCT imaging gradually moving into routine practice in the management of thoracic cancers, questions are being raised as to how best to standardize dose-volume reporting for various lung volumes defined using different CT data sets. The goals of this study were (1) to investigate various lung volume definitions and their inter-relationships; (2) to study the impact on lung dose volume histograms (DVHs); and (3) to propose a population-based model for converting one lung volume definition to another for prospective or retrospective dose reporting. With 4DCT imaging gradually moving into routine practice in the management of thoracic cancers, questions are being raised as to how best to standardize dose-volume reporting for various lung volumes defined using different CT data sets. The goals of this study were (1) to investigate various lung volume definitions and their inter-relationships; (2) to study the impact on lung dose volume histograms (DVHs); and (3) to propose a population-based model for converting one lung volume definition to another for prospective or retrospective dose reporting. Materials/Methods40 retrospectively selected stage III/IV non-small-cell lung cancer patients were included. Each patient had a 4DCT and a fast free-breathing (FB) helical CT scan. All lungs were contoured in the FBCT and in the 4DCT data sets at the end of expiration (EXP) and the end of inspiration (INSP). A special time-averaged CT was created by averaging CT numbers in the same spatial voxel over all reconstructed phases of the 4DCT. This represents the time-averaged lung density (AVE). We compared various lung volumes using EXP CT as the reference volume. In order to compare the dose variations and dose-volume effects for lung tissue, we calculated the dose distribution using the same treatment plan in EXP, INSP, AVE, and FB CT data sets in 15 of 40 patients. We compared the percentages of lung exposed to more than 5 (V5), 20 (V20), and 50 (V50) Gy. In addition, we considered an alternative DVH calculation method that used the same dose distribution calculated on a single CT but normalized to the lung volume on a different CT data set. Finally, we used the population-based lung volumetric relationships in various CT data sets to convert DVHs and compared the predictive accuracy. 40 retrospectively selected stage III/IV non-small-cell lung cancer patients were included. Each patient had a 4DCT and a fast free-breathing (FB) helical CT scan. All lungs were contoured in the FBCT and in the 4DCT data sets at the end of expiration (EXP) and the end of inspiration (INSP). A special time-averaged CT was created by averaging CT numbers in the same spatial voxel over all reconstructed phases of the 4DCT. This represents the time-averaged lung density (AVE). We compared various lung volumes using EXP CT as the reference volume. In order to compare the dose variations and dose-volume effects for lung tissue, we calculated the dose distribution using the same treatment plan in EXP, INSP, AVE, and FB CT data sets in 15 of 40 patients. We compared the percentages of lung exposed to more than 5 (V5), 20 (V20), and 50 (V50) Gy. In addition, we considered an alternative DVH calculation method that used the same dose distribution calculated on a single CT but normalized to the lung volume on a different CT data set. Finally, we used the population-based lung volumetric relationships in various CT data sets to convert DVHs and compared the predictive accuracy. ResultsOn average, the total lung volumes and masses defined on INSP CT sets were 13.9% and 4.6% (1SD: 5.1% and 1.8%) higher than those defined on EXP CT sets. The average INSP lung density was 8.0% (1SD: 3.4%; range: 1.1–8.8%) lower than the EXP lung density. On average, the FB and AVE total lung volumes were 6.3% and 8.8% (1SD: 3.5% and 5.4%) higher than the lung volume defined on the EXP CT. The average dose differences in the non-zero dose voxels in the lung were 12.3, 6.0, and 29.8 cGy, for the INSP, AVE, and FB CT data sets, respectively, relative to the reference dose distribution calculated on the EXP CT. Compared to the dose prescriptions of 6000 to 6600 cGy, the spatial dose variation was minimal when using different CTs for dose calculation. As a consequence, we found a strong volume effect for the dose-volume relationships using different lung volume definitions. The total lung V5, V20, V50, and EUD were higher in INSP CT than in EXP CT by 3.3%, 2.4%, 2.0%, and 150 cGy (1SD: 6.0%, 7.6%, 2.1%, and 84 cGy; range: −12.15% – +15.80%, −17.36% – +21.88%, −3.36% – +6.30%, and 7 cGy–285 cGy), respectively. We found that using a simple volume scaling factor can effectively convert the DVH results from one CT lung volume into another. The INSP and EXP V5, V20, and V50 values can be predicted more accurately (1SD: <1.4%) from dose distributions calculated on the AVE CT than those calculated from the FB CT (1SD: <3.2%). On average, the total lung volumes and masses defined on INSP CT sets were 13.9% and 4.6% (1SD: 5.1% and 1.8%) higher than those defined on EXP CT sets. The average INSP lung density was 8.0% (1SD: 3.4%; range: 1.1–8.8%) lower than the EXP lung density. On average, the FB and AVE total lung volumes were 6.3% and 8.8% (1SD: 3.5% and 5.4%) higher than the lung volume defined on the EXP CT. The average dose differences in the non-zero dose voxels in the lung were 12.3, 6.0, and 29.8 cGy, for the INSP, AVE, and FB CT data sets, respectively, relative to the reference dose distribution calculated on the EXP CT. Compared to the dose prescriptions of 6000 to 6600 cGy, the spatial dose variation was minimal when using different CTs for dose calculation. As a consequence, we found a strong volume effect for the dose-volume relationships using different lung volume definitions. The total lung V5, V20, V50, and EUD were higher in INSP CT than in EXP CT by 3.3%, 2.4%, 2.0%, and 150 cGy (1SD: 6.0%, 7.6%, 2.1%, and 84 cGy; range: −12.15% – +15.80%, −17.36% – +21.88%, −3.36% – +6.30%, and 7 cGy–285 cGy), respectively. We found that using a simple volume scaling factor can effectively convert the DVH results from one CT lung volume into another. The INSP and EXP V5, V20, and V50 values can be predicted more accurately (1SD: <1.4%) from dose distributions calculated on the AVE CT than those calculated from the FB CT (1SD: <3.2%). ConclusionsWe found large variations in reported DVH values when different lung volume definitions were used. However, population-based relationships among different lung volumes can be used to convert DVHs into a more standardized dose-volume definition. We found large variations in reported DVH values when different lung volume definitions were used. However, population-based relationships among different lung volumes can be used to convert DVHs into a more standardized dose-volume definition.