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

Reviewer: Charles B. Simone, II, MD
Abramson Cancer Center of the University of Pennsylvania
Last Modified: October 30, 2007

Presenter: Yixiu Kang, MD
Presenter's Affiliation: M.D. Anderson Cancer Center
Type of Session: Plenary


  • Local control in patients with early stage non-small cell lung cancer (NSCLC) after conventional radiotherapy has historically been poor.  Doses of up to 70 Gy delivered with 3D conformal radiotherapy achieve local control in only half of the patients treated. However, administering higher doses via conventional conformal techniques results in increased treatment toxicity.
  • Four-dimensional CT (4DCT) imaging, or three-dimensional CT imaging at a sequence of respiratory phases, is increasingly being used to treat thoracic malignancies and allows for spatial and temporal information on shape and mobility to be acquired synchronously.
  • It is currently unknown how best to standardize dose-volume reporting for various lung volumes as determined by different CT data sets.
  • This study was conducted to investigate various lung volume definitions and determine their inter-relationships, evaluate the effect on lung dose volume histograms (DVHs), and determine a population-based model that would allow one lung volume definition to be converted into another for accurate dose reporting.

Materials and Methods

  • Forty retrospectively selected patients with stage III or IV NSCLC were examined, 20 of whom had right-sided malignancies, 16 with left-sided tumors, and 4 with bilateral disease. Prescribed doses among these patients ranged from 45 to 70 Gy.
  • Each patient had a 4DCT and a fast free-breathing (FB) helical CT scan.
  • Contouring of the lungs for both the FBCT and the 4DCT data sets were conducted at the end of expiration (EXP) and the end of inspiration (INSP).
  • The investigators averaged CT numbers in the same spatial voxel over all reconstructed phases of the 4DCT to represent a time-averaged lung density (AVE).
  • Various lung volumes were compared, with EXP CT serving as a reference volume.
  • The dose distribution using the same treatment plan in EXP, INSP, AVE, and FB CT data sets was calculated in 15 of 40 patients to compare the dose variations and dose-volume effects for lung tissue.
  • The volume of lung exposed to greater than five (V5), 20 (V20), and 50 (V50) Gy were compared.
  • Lung volumetric relationships were used to convert DVHs, allowing for the predictive accuracy to be compared.


  • The total lung volumes and masses based on INSP CT sets on average were 13.9% and 4.6% higher than those based on EXP CT sets, respectively.
  • Conversely, the INSP lung density was on average 8.0% less than the EXP lung density.
  • The FB and AVE total lung volume averages were 6.3% and 8.8% higher than the lung volume defined on the EXP CT, respectively.
  • Dose differences were 12.3, 6.0, and 29.8 cGy in the non-zero dose voxels in the lung for the INSP, AVE, and FB CT data sets, respectively, in relation to the reference dose distribution as determined from EXP CT.
  • The spatial dose variation was insignificant when different CTs for dose calculation were used, meaning that the variation was preserved during respiration.
  • A strong volume effect for dose-volume relationships with different lung volume definitions was determined. 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, respectively.
  • The authors determined an effective means to convert DVH results from one CT lung volume into another using a simple volume scaling factor.
  • Dose distributions based on AVE CT were found to be more predictive of INSP and EXP V5, V20, and V50 values when compared with dose distributions based on FB CT.

Author's Conclusions

  • Large variations were seen in reported DVH values depending on which lung volume definitions were used.
  • DVHs, however, can be converted to a standardized dose-volume definition using population-based relationships among different lung volumes.

Clinical/Scientific Implications

When treating patients with NSCLC, motion of tumor and normal organs is of great clinical importance, as this motion effects the delineation of the target and normal tissues, the required margins, and the dose distribution. Optimal target definition is imperative since the risk of radiation toxicity has previously been shown to correlate with the amount of normal lung tissue irradiated. Optimal target definition, therefore, may allow for radiation oncologists to administer higher doses of radiation to improve local control, while minimizing the risks of such serious complications as radiation pneumonitis and treatment-related death. Based on this study, it appears that dose distributions based on the AVE CT should be used in place of distributions based on FB CT. Additionally, findings from this study can allow prospective or retrospective comparisons of dose reporting. Future research will need to be conducted to determine if smoking is correlated with the results in this study and whether differences exist based on tumor location with the lung. 

Partially funded by an unrestricted educational grant from Bristol-Myers Squibb.