Purpose: Interfractional organ motion and patient positioning errors during prostate radiotherapy can have deleterious clinical consequences. It has become clinical practice to re-position the patient with image-guided translational position correction before each treatment to compensate for those errors. However, tilt errors can only be corrected with table corrections in six degrees of freedom or "full" adaptive treatment planning strategies. Organ shape deformations can only be corrected by "full" plan adaptation. This study evaluates the potential of instant treatment plan adaptation (fast isodose line adaptation with real-time dose manipulating tools) based on cone-beam CT (CBCT) to further improve treatment quality.
Methods and materials: Using in-house software, CBCTs were modified to approximate a correct density calibration. To evaluate the dosimetric accuracy, dose distributions based on CBCTs were compared with dose distributions calculated on conventional planning CTs (PCT) for four datasets (one inhomogeneous phantom, three patient datasets). To determine the potential dosimetric benefit of a "full" plan adaptation over translational position correction, dose distributions were re-optimized using graphical "online" dose modification tools for three additional patients' CT-datasets with a substantially distended rectum while the original plans have been created with an empty rectum (single treatment fraction estimates).
Results: Absolute dose deviations of up to 51% in comparison to the PCT were observed when uncorrected CBCTs were used for replanning. After density calibration of the CBCTs, 97% of the dose deviations were <or=3% (gamma index: 3%/3mm). Translational position correction restored the PTV dose (D(95)) to 73% of the corresponding dose of the reference plan. After plan adaptation, larger improvements of dose restoration to 95% were observed. Additionally, the rectal dose (D(30)) was further decreased by 42 percentage points (mean of three patient datasets).
Conclusions: An accurate dose calculation based on CBCT-datasets is possible when density distributions are corrected. The presented adaptive strategy has the potential to reduce dose delivery errors due to organ deformations to a minimum.