IMRT of large fields: whole-abdomen irradiation.

PURPOSE: To assess the feasibility of inverse planning for whole-abdomen intensity-modulated radiation therapy (IMRT) with bone marrow and kidney sparing and to develop approaches to circumventing field size restrictions in the application of whole-abdomen IMRT using dynamic multileaf collimators (DMLC). METHODS AND MATERIALS: The entire peritoneal cavity as derived from serial computerized tomography scans was defined as the gross target volume, whereas the planning target volume (PTV) was defined as the gross target volume plus a 5-mm margin extending 1 cm superiorly and inferiorly. In 10 randomly selected patients, the PTV ranged from 5629 to 12578 cc (median 7935 cc), and the superior-inferior, lateral, and anterior-posterior dimensions of the PTV ranged from 37 to 46 cm (median 42.5 cm), 27 to 33 cm (median 29 cm), and 18 to 23 cm (median 20 cm), respectively. A single isocenter was defined for patients with field length <40 cm. For patients with fields >40 cm, two isocenters were defined: one in the abdominal region, and the other in the pelvis. For IMRT planning, five 15-MV intensity-modulated beams at gantry angles of 180 degrees, 105 degrees, 35 degrees, 325 degrees, and 255 degrees were used. Optimization was designed to spare kidneys and bones. To fully account for the significant scattered dose contributions, an iterative process for dose calculations was implemented in the optimization. To overcome the 15-cm field width limit of our DMLC delivery system, fields with a width >15 cm were split into two or more subfields. To minimize field match errors, adjacent subfields overlapped by at least 2 cm, with intensity "feathering" in the overlap region. For patients with two isocenters, fields were overlapped and feathered in the cephalad-caudad direction by at least 3 cm. For comparison, conventional anterior-posterior/posterior-anterior 6-MV photon beams with posterior kidney blocks at extended distance were also generated for each patient.
RESULTS: Treatment plan optimization calculations required 20-80 min on a 500-MHz DEC alpha workstation. Including beam splitting, an average of 16 DMLC beams was used per patient. Delivery of 150 cGy required, on average, 1442 monitor units. For the same dose constraints on the kidneys, whole-abdomen IMRT resulted in significant dose reduction to the bones and improved PTV coverage as compared to conventional treatment. For a prescription dose of 30 Gy, the volume of the pelvic bones receiving more than 21 Gy was reduced on average by almost 60% with IMRT, and the mean dose to all bones was reduced from 24.0 +/- 1.5 Gy to 18.5 +/- 1.0 Gy (p = 0.002). PTV coverage, as measured by V95 (the volume receiving 95% of the prescription dose), improved from 71.7 +/- 4.8% with conventional treatment to 83.5 +/- 3.9% with IMRT (p = 0.002), although small regions of underdose in areas near the kidneys could not be avoided completely. The high-dose regions within the PTV, as measured by D05 (the dose covering 5% of PTV volume), increased slightly from 31.2 +/- 0.6 Gy with conventional treatment to 32.8 +/- 0.2 Gy with IMRT.
CONCLUSION: We have developed a process to plan and deliver whole-abdomen IMRT using standard linear accelerators and DMLC. IMRT can achieve better PTV coverage with the same level of kidney sparing and improved sparing of the bone marrow. These methods may be applicable also to other sites requiring large-field irradiation.