Application of Statistical Control of Processes to the Determination of CTV Expansion Margins in Prostate Cancer Patients Treated With VMAT: A Prospective Study in 57 Patients

Abstract

To assess the usefulness of the use of average and range control charts in the CTV expansion margins calculation from the data of daily setup deviations of prostate cancer patients treated with VMAT. Fifty-seven consecutive prostate cancer patients undergoing VMAT were prospectively enrolled in the study. Adenocarcinoma of the prostate was confirmed in all patients by prostate biopsy and the disease stage was determined according our institution protocol. For each patient, two daily Cone Beam CT were performed on day 1 to 4; one after daily conventional positioning before treatment an another after each fraction. The scans were compared online with the planning CT. The average of the displacements was determined and checked by another two CBCTs on day 5. Subsequently a weekly follow up with 2 CBCTs (before and after treatment) was performed. From the data obtained with the CBCTs of the first four days, the setup corrections and the initial parameters of the average and rang control charts (control limits) were determined. This allowed to monitor, for each patient, whether the variability is purely random or presents trends and periodicities From the 6th treatment fraction the patient set up is corrected before CBCT and the control charts are adapted to monitor also the new variability. At the end of treatment, the control charts were analyzed to detect those patients whose deviations in the setup have always been in a “state of statistical control”, that is, those whose graphs meet the requirements that ensure that the variability is adjusted to the distribution of probability of Gauss or Normal, hypothesis under which the formula of van Herk can be applied to establish the margins of correction by uncertainty of the CTV in the setup. Finally, the results of the calculated margins for all patients in the study and for those in “state of statistical control” were compared. For the overall of patients studied, the margins calculated were: lateral + 0.45cm, longitudinal +0.52cm and vertical +0.56cm. For the subgroup of patients in a state of statistical control the margins calculated were lateral +0.29 cm, longitudinal +0.35 cm and vertical +0.38 cm. The mean and range control charts are useful for detecting abnormalities in the variability of patient setup and applying the appropriate corrective actions. Maintaining the set up variability in the control state, significantly decreases the margins associated with the uncertainty of the set up applied to the CTV.