Volumetric image guidance using carina vs spine as registration landmarks for conventionally fractionated lung radiotherapy.

PURPOSE: To compare the relative accuracy of 2 image guided radiation therapy methods using carina vs spine as landmarks and then to identify which landmark is superior relative to tumor coverage. METHODS AND MATERIALS: For 98 lung patients, 2596 daily image-guidance cone-beam computed tomography scans were analyzed. Tattoos were used for initial patient alignment; then, spine and carina registrations were performed independently. A separate analysis assessed the adequacy of gross tumor volume, internal target volume, and planning target volume coverage on cone-beam computed tomography using the initial, middle, and final fractions of radiation therapy. Coverage was recorded for primary tumor (T), nodes (N), and combined target (T+N). Three scenarios were compared: tattoos alignment, spine registration, and carina registration.
RESULTS: Spine and carina registrations identified setup errors ≥ 5 mm in 35% and 46% of fractions, respectively. The mean vector difference between spine and carina matching had a magnitude of 3.3 mm. Spine and carina improved combined target coverage, compared with tattoos, in 50% and 34% (spine) to 54% and 46% (carina) of the first and final fractions, respectively. Carina matching showed greater combined target coverage in 17% and 23% of fractions for the first and final fractions, respectively; with spine matching, this was only observed in 4% (first) and 6% (final) of fractions. Carina matching provided superior nodes coverage at the end of radiation compared with spine matching (P=.0006), without compromising primary tumor coverage.
CONCLUSION: Frequent patient setup errors occur in locally advanced lung cancer patients. Spine and carina registrations improved combined target coverage throughout the treatment course, but carina matching provided superior combined target coverage.