Implementing Plan of the Day for Cervical Cancer: A Comparison of Target Volume Generation Methods

Owing to substantial interfraction motion in cervical cancer, plan-of-the-day (PotD) adaptive radiation therapy may be of benefit to patients. Implementation is limited by uncertainty over how to generate the planning target volumes (PTVs). We compared published methods on our own patients.

Forty patients each had 3 planning scans with variable bladder filling and daily cone beam computed tomographies (cone beam CTs) during radiation therapy; 5 to 11 cone beam CTs were selected to represent interfraction motion. Clinical target volumes (CTVs) and organs at risk were contoured following EMBRACE-II guidelines. A literature search identified 30 adaptive and nonadaptive solutions to PTV generation, which we applied to our patients. PTV sizes and mean coverage of the daily CTV were determined. For 11 patients, the clinically implemented, subjectively edited plan library was also investigated.

Eleven studies assessed 15 PotD strategies against nonadaptive comparators on a median of 14 patients (range, 9-23). Some PotD approaches applied margin recipes to the CTV on each planning scan, some modeled the CTV against bladder volume, and others applied incremental isotropic margins to the CTV with a single planning scan. Generally, coverage improved as PTV size increased. The fixed isotropic margin required to provide 100% coverage of all patients was 44 mm, with a mean PTV size of 3316 cm³. The PotD strategy with the best coverage was a 2-plan library formed by modeling the CTV against bladder volume with extrapolation; it provided 98% mean coverage with 1419-cm³ mean PTV size. A 3-plan library consisting of the CTV on each planning scan with 10-mm margin provided 96% mean coverage with 1346-cm³ mean PTV size. The clinically implemented solution that employed subjective extrapolation had mean 100% coverage and 1282-cm³ PTV size on the 11-patient subset. Coverage provided by the best nonadaptive strategies was not statistically superior to the best PotD strategy (P = .13), but PTVs were larger (P = .02).

We identified a modeled 2-plan method and a simple 3-plan method, both of which provided excellent coverage with small PTVs compared with nonadaptive strategies.