Respiratory Motion Changes of Lung Tumors Over the Course of Radiation Therapy Based on Respiration-Correlated Four-Dimensional Computed Tomography Scans

Purpose: To determine whether lung tumor respiratory excursion at simulation is predictive of excursion during radiation and whether phase offsets between tumor and surrogate markers are constant throughout treatment. Methods and Materials: Respiration-correlated CT scans and two rescans (using a Brilliance Big Bore spiral CT simulator; Philips, Inc.) were obtained from 20 patients at simulation. Gross tumor volume (GTV) was contoured on 10 phases of the respiratory cycle, and excursions were calculated. Diaphragm and xyphoid motion were quantified. Phase offsets, ΔΦ, were calculated for patients with a GTV motion of >3 mm. Interfraction differences in excursions between simulation and rescans and magnitudes of variation in phase offset between fractions were evaluated. Results: Mean GTV excursions at simulation in superior-inferior, anterior-posterior, and medial-lateral directions were 0.67, 0.29, and 0.21 cm, respectively. The magnitude of superior-inferior GTV excursion correlated with tumor location (upper vs. lower lobe, p = 0.011). GTV excursions between simulation and rescan 1 (p = 0.115) and between simulation and rescan 2 (p = 0.071) were stable. Fourteen patients were analyzed for variations in phase offsets. GTV-xyphoid phase offset changed significantly between simulation and rescan 1 (p = 0.007) and simulation and rescan 2 (p = 0.008), with mean ΔΦ values of 13.2% (rescan 1) and 14.3% (rescan 2). Xyphoid-diaphragm offset changed between simulation and rescan 1 (p = 0.004) and between simulation and rescan 2 (p = 0.012), with mean ΔΦ values of 14.5% (rescan 1) and 7.6% (rescan 2). Conclusions: Interfraction consistency in tumor excursion suggests tumor excursion at simulation may direct therapy. Significant variations in phase lag between GTV and other anatomic structures throughout treatment have important implications for techniques that rely on surrogate structures to predict tumor motion.