Risk factors related to interfractional variation in whole pelvic irradiation for locally advanced pelvic malignancies

Purpose. The goal of the present study was to demonstrate risk factors affecting the interfractional variation in whole pelvic irradiation. Patients and methods. Daily image acquisitions of 101 patients with locally advanced pelvic malignancy were undertaken using a kilo-voltage orthogonal on-board imager. The baseline deviation (the shift between the initial treatment and each fraction; Value _Base) and day-to-day variation (the shift between the previous treatment and each fraction; Value _DD) were measured. The standard deviations (SD) along the x- (right-left), y- (cranial-caudal), and z- (anterior-posterior) axes (SD[x], SD[y], and SD[z], respectively), the 3D vector of the SD (SD[3D]), and the mean of 3D shift (mean[3D]) were calculated in each patient. Various clinical factors, lumbar pelvic balance and rotation, and the shift of 5 consecutive fractions from the initial treatment (Value _5Fx) were investigated as risk factors. Results. The prone set-up showed a larger mean _Base[3D] than in the supine position (p =0 .063). A body mass index (BMI) ≥ 30 kg/m ² resulted in the largest mean _DD[3D] (p = 0.078) and SD _DD[3D] (p = 0.058). All the SD _5Fx along the x-, y-, and z-axes had moderate linear relationships with SD _Base and SD _DD (p < 0.001). The SD _5Fx[3D] also had a moderate linear relationship with the mean _Base[3D], mean _DD[3D], SD _Base[3D], and SD _DD[3D] (p < 0.001). In multivariate analysis, the SD _5Fxhad the same significant relationship with SD _Base and SD _DD (p < 0.001). A BMI ≥ 30 kg/m ² was associated with the largest SD _DD[x] (p = 0.003). Conclusion. Close surveillance through high-quality and frequent image guidance is recommended for patients with extensive variations of the initial five consecutive fractions or obesity.