Correlation of lung tumor motion with external surrogate indicators of respiration.

PURPOSE To assess the correlation of respiratory volume and abdominal displacement with tumor motion as seen with X-ray fluoroscopy. Measurements throughout the patient's treatment course allowed an assessment of the interfractional reproducibility of this correlation. METHODS AND MATERIALS Data were acquired from 11 patients; 5 were studied over multiple days. Measurements of respiratory volume by spirometry and abdominal displacement by a real-time position tracking system were correlated to simultaneously acquired X-ray fluoroscopy measurements of superior-inferior tumor displacement. The linear correlation coefficient was computed for each data acquisition. The phase relationship between the surrogate and tumor signals was estimated through cross-correlation delay analysis. RESULTS Correlation coefficients ranged from very high to very low (0.99-0.39, p < 0.0001). The correlation between tumor displacement and respiratory volume was higher and more reproducible from day to day than between tumor displacement and abdominal displacement. A nonzero phase relationship was observed in nearly all patients (-0.65 to +0.50 s). This relationship was observed to vary over inter- and intrafractional time scales. Only 1 of 5 patients studied over multiple days had a consistent relationship between tumor motion and either surrogate. CONCLUSIONS Respiratory volume has a more reproducible correlation with tumor motion than does abdominal displacement. If forming a tumor-surrogate prediction model from a limited series of observations, the use of surrogates to guide treatment might result in geographic miss.