Evaluation of the influence of breathing on the movement and modeling of lung tumors

Purpose: Respiration causes movement and potential shape change in lung tumors that are not fully appreciated using conventional free-breathing CT models for radiotherapy planning. Although target expansion has the potential to ensure proper tumor coverage in the face of motion on a free-breathing CT scan, large variations in how individual patients' tumors move may make such expansions difficult to uniformly define. In addition, excessive expansion may result in the unnecessary inclusion of normal lung in the treated volume. This study was designed to evaluate the influence of breathing movement on tumors and to assess the validity of the free-breathing CT scan for target delineation in the lung. Methods and Materials: Data from 16 consecutive lung cancer patients who underwent treatment planning CT scans at inhale and exhale and during free breathing on a fast helical CT scanner were analyzed. Gross tumor volumes (GTV) were defined on each scan. A composite GTV was created by combining the inhale and exhale GTVs (COMP). Two methods of expansion were used to compare COMP to the free-breathing GTV (FREE). First the free-breathing data set was expanded uniformly by 1 cm (FREE + 1). Next, a nonuniform expansion was generated in all 6 directions to ensure complete coverage of COMP with the minimal subtended volume (FREE + EXP). The amount of excess normal lung treated with these 2 expansions was compared. The volume of the COMP missed using the 1-cm expansion was determined. Results: There was a significant amount of excess normal lung tissue treated with the uniform 1-cm (FREE + 1) expansion, as well as with the nonuniform (FREE + EXP) expansion. In addition, there were also cases where this technique led to marginal miss of the tumor, including one case where 47% of the overall tumor was missed with this 1-cm (FREE + 1) expansion. An attempt to create a systematic model for expansion was not successful. Although the mean expansions in the anterior-posterior, superior-inferior, and right-left directions were reasonable (0.9, 1.0, and 0.8 cm, respectively), the large intrapatient variations (sigma 0.6 cm anterior-posterior, 0.7 cm superior-inferior, and 0.5 cm right-left) suggest difficulty in assigning a simple rule for population target expansion. Some extension of FREE outside of the borders of COMP was observed, suggesting the need for evaluation of reproducibility over multiple breathing states. Conclusions: Traditional methods of expanding the GTV to CTV by 1 cm are less than ideal. This method tends to include more normal lung than necessary and may lead to marginal miss. Interpatient tumor movement variations further prohibit defining a simple rule for nonuniform expansion that would minimize the volume of normal lung in the target. Although the development of target volumes by combining information from breath-hold CT scans at inhale and exhale states shows some promise in minimizing excess lung irradiated while maintaining adequate tumor coverage, further tests of breathing reproducibility need to be performed to provide a confident baseline for defining target expansions by this technique. (C) 2004 Elsevier Inc.