Effect of image-guidance frequency on geometric accuracy and setup margins in radiotherapy for locally advanced lung cancer.

PURPOSE: To assess the relative effectiveness of five image-guidance (IG) frequencies on reducing patient positioning inaccuracies and setup margins for locally advanced lung cancer patients. METHODS AND MATERIALS: Daily cone-beam computed tomography data for 100 patients (4,237 scans) were analyzed. Subsequently, four less-than-daily IG protocols were simulated using these data (no IG, first 5-day IG, weekly IG, and alternate-day IG). The frequency and magnitude of residual setup error were determined. The less-than-daily IG protocols were compared against the daily IG, the assumed reference standard. Finally, the population-based setup margins were calculated.
RESULTS: With the less-than-daily IG protocols, 20-43% of fractions incurred residual setup errors ≥ 5 mm; daily IG reduced this to 6%. With the exception of the first 5-day IG, reductions in systematic error (∑) occurred as the imaging frequency increased and only daily IG provided notable random error (σ) reductions (∑ = 1.5-2.2 mm, σ = 2.5-3.7 mm; ∑ = 1.8-2.6 mm, σ = 2.5-3.7 mm; and ∑ = 0.7-1.0 mm, σ = 1.7-2.0 mm for no IG, first 5-day IG, and daily IG, respectively. An overall significant difference in the mean setup error was present between the first 5-day IG and daily IG (p < .0001). The derived setup margins were 5-9 mm for less-than-daily IG and were 3-4 mm with daily IG.
CONCLUSION: Daily cone-beam computed tomography substantially reduced the setup error and could permit setup margin reduction and lead to a reduction in normal tissue toxicity for patients undergoing conventionally fractionated lung radiotherapy. Using first 5-day cone-beam computed tomography was suboptimal for lung patients, given the inability to reduce the random error and the potential for the systematic error to increase throughout the treatment course.