Radiol Oncol 2005; 39(3): 211-7. Setup error and its effect on safety margin in conformal radiotherapy of the prostate Borut Kragelj Department of Radiation Oncology, Institute of Oncology, Ljubljana, Slovenia Background. In radiotherapy, setup errors in positioning the patients influence the size of safety margin and thereby also the size of irradiation field and toxicity of radiotherapy. Methods. The setup errors were calculated by evaluating the deviations from the measured distance be-tween the irradiation field margin and the bony pelvis. Results. The research was performed on 23 patients. With respect to lateral, craniocaudal and anteropos-terior axis, the observed systemic error ranged from -5 to + 9 mm, -4 to +5 mm, and from -4 to +4 mm, re-spectively, whereas the observed random error ranged from 0 to 7.5 mm, 0 to 3.6 mm, and from 0 to 4.2 mm, respectively. The safety margin, with the 90% probability to cover clinical target volume (CTV) and al-lowing for the prostate position variability, measured 9 mm, 9.5 mm, 7 mm, and 10 mm in the respective lateral, craniocaudal, anterior and dorsal direction. Conclusions. Irradiation of the prostate with a 7 mm dorsal safety margin, allowing for 90% coverage prob-ability of CTV, was feasible in 22/23 patients on condition that the gross systemic error (>3mm) was elim-inated. Key words: prostatic neoplasms; radiotherapy, conformal Introduction Due to acute and particularly chronic postirradiation complications in radiotherapy of the prostate, the dose application, and conse-quently also the irradiation effect, are re-Received 15 August Accepted 31 August Correspondence to: Borut Kragelj, MD, PhD, Department of Radiation Oncology, Institute of Oncology, Zaloška 2, SI-1000 Ljubljana, Slovenia; Phone: +386 1 587 9489; Fax: +386 1 587 9400; E-mail: bkragelj@onko-i.si stricted. The incidence and grade of acute and chronic complications depend upon the dose and volume of the surrounding organs involved in the irradiation area.1-4 As the restrictions on dose application should in no way be disregarded, the irradiation of the prostate can be performed on con-dition that the beam is aimed as accurately at the target volume as possible. The size of the irradiation field is dependent on the width of safety margin around the target volume that ensures that the area to be irradiated is actu-ally or most probably irradiated. However, the more the safety margin is extended, the 212 Kragelj B / Conformal radiotherapy of the prostate greater is the exposure of the surrounding organs to irradiation and consequently also the risk of therapy-related complications. In the radiotherapy of prostate cancer, the safety margin width is dependent upon the prostate motion and patient positioning er-rors. The aim of our research was therefore to estimate the setup errors, and from the ob-tained results, determine an optimal safety margin. Methods Patients The patients who were irradiated at our Institute in the period from September 2004 to March 2005 were included into the study. These patients were given radiotherapy as a unique treatment or as adjuvant treatment after primary prostatectomy. Irradiation technique Irradiation was performed by a 15-MeV linear accelerator (Varian, Clinac 2100 C/D), using the four-field technique, at the angles of 0, 90, 180, and 270 degrees in the sense of 3D con-formal radiotherapy. Irradiation simulation was carried out on Philips CT MX 8000 multi-slice simulator. Target volumes were mapped using the program CMS Focal, whereas the program CMS Xio 4.2.0 was used for irradiation planning that was carried out by beam-eye-view technique. The fields were framed by multileaf collimator. The patients were irradiated in supine position with the feet resting on a support cush-ion (Sinmed Feetfix support cushion) and the knees provisionally supported. The isocenter was defined by three spots – one on the abdominal wall and two lateral spots. The position of each patient was additionally marked on the isocenter plane by four lines (with re-spect to the patient’s axis, three longitudinal lines and one transversal line). Radiol Oncol 2005; 39(3): 211-7. Irradiation area Clinical target volume (CTV) included the prostate, seminal vesicles, and in the patients at high risk, also regional lymph nodes. The safety margin determining the planned target volume (PTV) was 1.5 cm wide; in the closing phase of radiotherapy, in which only the prostate was exposed to irradiation, the safe-ty margin was reduced to 1 cm, and on the dorsal side, to 0.7 cm. Irradiation precision Irradiation precision was tested by amor-phous silicon-based portal imaging system (EPI). The image was additionally processed by Varian’s vision software, version 6.1 that allows a precise computerized reconstruction of irradiation field borders. The setup error in patient positioning was calculated by comparing the distance be-tween the field margin and selected bony pelvic structures of the digitally reconstruct-ed radiographs (DRRs) obtained from the planning CT data and EPI. The display and comparison of images as well as measure-ments were made by computer program Multiaccess, version 8,00J0, Impac Medical Systems. Methods The setup error in patient positioning was de-fined by the deviations along the craniocau-dal, anteroposterior and lateral axis. All devi-ations in the cranial direction, to the right and anterior were marked as positive and the de-viations in the caudal direction, to the right and dorsum were marked as negative. In order to eliminate errors in deviation measure-ments, more measurements were performed yielding a higher mean value. Therefore, in final evaluations of safety margins, the inaccu-racy of measurements was neglected. From the above measurements, a systemic (SE) and a random setup error (RE) with re- Kragelj B / Conformal radiotherapy of the prostate 213 spect to each of the three axes was calculated for each individual patient. Systemic error was defined as a mean deviation of a patient positioning from the isocenter and random error as a deviation of each individual meas-urement from the mean value. The size of random error was marked as 1SD. The sys-temic and random setup errors were then cal-culated for the entire group of patients. The systemic error for the entire group (SEeg) was defined as arithmetic mean of all individual systemic errors. In calculating the random error for the entire group (REeg), random deviations of indi-vidual SE from SEeg were also taken into account. The size of the random component of SE was estimated by 1 SD (SDse) and derived from an even distribution of individual SE around SEeg. The size of RE was also consid-ered and was derived from an even distribution of individual RE around arithmetic mean of individual RE (AM RE). The size of indi-vidual RE deviations from AM RE was also defined as 1 SD (SDam). Adding AM RE and 1.63 SDam, SDre was obtained with 95% probability that all RE were comprised in SDre. In the calculation of REeg, SDse and SDre were considered as independent parameters. REeg, expressed as 1 SD, was obtained by the square root of SDse2+ SDre2.5 From the obtained systemic (SEeg) and random (REeg) setup errors, the size of safety margin was calculated in order to compen-sate for the errors in positioning the patient along each of the axis. Because of relatively high prevalence and severity of late rectal radiation toxicity, the anterior and posterior safety margins along the anteroposterior axis were calculated separately. Table 1. The range of the set up errors and the systemic to the direction of the positioning deviation Deviation - direction Lateral Setup error-range (mm) -10 do +12 Systemic error (mm) -5.0 do +9.0 Random error (1SD in mm) 0 do 7.5 The safety margin calculation by adding SEeg and 1.5 SD REeg was based on Goitein’s estimation6 that 1.5 SD is a sensible compro-mise between the risk of underdosing the target volume and of excessive overdosing of the surrounding healthy tissue. This complies well with 90% confidence interval of random deviation in any of the directions. In further calculations of safety margin, the prostate position variability was taken into account in addition to the setup error. For the assess-ment of the position variability of the prostate, Zelefsky’s data were used.8 The safety margin was calculated by adding the arithmetic mean of the prostate movements and SEeg and 1.5 SD of the combined random error, calculated according to Rudat’s recommendations, taking into account the random setup error as well as the random prostate movement error. The calculations of safety margins were made for each axis and separately for anterior and posterior direc-tions. Results The research was performed on 23 patients in whom altogether 95 measurements were made for evaluating the position of a patient during irradiation. In each patient, three to maximum five positioning measurements were carried out. The deviations along the lateral axis ranged from -10 to +12 mm, along the cranio-caudal axis from -7 to 6 mm, and along the anteroposterior axis from -11 to +5 mm. The systemic error along the lateral axis was with-in the range of -5 to 9 mm, along the cranio-d random components of the setup errors with respect Craniocaudal Anteroposterior -7 do +6 -11 do +5 -4.2 do +4.8 -4.4 do +4.2 0 do 3.6 0-4.2 Radiol Oncol 2005; 39(3): 211-7. 214 Kragelj B / Conformal radiotherapy of the prostate Table 2. Systemic (SEeg) and random (REeg) setup errors with respect to the entire group of patients (23) and the direction of pisitioning deviation by presenting SEeg, AMRE, SDam, SDre Deviation - direction Lateral (mm) Craniocaudal (mm) Anteroposterior (mm) Systemic error (SE eg) +0.53 +0.17 -0.87 SDse (1SD ) 2.9 2.3 2.5 Random error (REeg) 5.1 4.1 4.9 AMRE 2.6 1.9 2.3 SDam 1.6 1.0 1.2 SDre 4.3 3.5 4.2 caudal axis -4.2 to +4.8 mm, and along the an-teroposterior axis -4.4. to +4.2 mm. The random error with the size of 1SD varied along the lateral axis from 0 to 7.5 mm, along the craniocaudal axis from 0 to 3.6 mm, and along the anteroposterior axis from 0 to 4.2 mm (Table 1). The calculated systemic error for the entire group along the lateral, craniocaudal, and an-teroposterior axis was +0.57 mm, +0.17 mm, and -0.87 mm, respectively. The random error for the entire group in the lateral, craniocau-dal, and anteroposterior axis was 5.1 mm, 4.1 mm and 4.9 mm, respectively. The two errors (SEeg and REeg) as well as SDse, AM RE, SDam, and SDre are presented in Table 2. The safety margin (SM) that would, with a 90% probability, cover an inaccurate positioning of a patient with respect to the lateral and cran-iocaudal axis is 8.2 mm and 6.3 mm, respec-tively, and with respect to the anteroposterior axis, 6.5 mm towards the anterior and 8.3 mm posteriorily. The safety margin that would, in addition to the setup errors, compensate also for the prostate position variability (SM total) along the lateral and craniocaudal axis with the same confidence interval is 9.2 mm and 9.5 mm, respectively, and along the antero-posterior axis, 6.7 mm towards the anterior and 10.3 mm posteriorily (Table 3). Discussion The aim of this study was to determine an ad-equate safety margin that would allow an ac-ceptable exposure of a target area to irradiation. The difficulties of the prostate cancer patient radiotherapy that make this therapy unreliable are the prostate position variability as well as inaccuracy in all subsequent repo-sitionings of the patient in the initial pose. Monitoring the variability of prostate position is an exacting task, and so far, it has not been performed at our Institute. So, the size of safety margin, and thereby also the precision of irradiation, is dependent merely on the pre-cision in repositioning the patient in the initial irradiation position. In comparing our data on the patient repositioning precision to the pub-lished data, it may be concluded that our pre-cision was tolerably satisfactory. As reported by Rudat,5 the random error along the lateral, craniocaudal, and anteroposterior axis was 3.1 mm, 5.4 mm, and 4.9 mm, respectively. In our patients, the random error calculated by using a similar method, was 3.9 mm, 2.9 mm, and 3.5 mm, respectively. In the study by Song, the data on the deviations greater than 5 mm in 40% of repositionings of his patients9 also speak in favor of the satisfactory preci-sion in repositioning of our patients. Table 3. Safety margin depending on the setup error (SM) in conjunction with the prostate position variability (SMtotal) Direction Lateral Craniocaudalni Anterior Posterior SM (mm) SM total (mm) 8.2 9.2 6.3 9.5 6.5 6.7 8.3 10.3 Radiol Oncol 2005; 39(3): 211-7. Kragelj B / Conformal radiotherapy of the prostate 215 Figure 1. Systemic and random setup errors (1.5 SD) along the lateral axis in the patient positioning and safety margin at which CTV should have been includ-ed in the irradiation field with 90% probability. Figure 3. Systemic and random setup errors (1.5SD) along the anteroposterior axis in the patient position-ing and safety margin at which CTV should have been included in the irradiation field with 90% probability. The safety margin was calculated from the measurements of setup errors (in positioning of our patients) and from the data published on the prostate position variability. Our cal-culated safety margin – approximately 1 cm in all directions – is similar to the margins re-ported by various authors of the studies on prostate irradiation, except for the margin on the dorsal side, which was allowed to be smaller than 1 cm.10 However, if the safety margin of less than 1 cm is not coordinated with the prostate position variability and setup error, the coverage of CTV may not be suf-ficient - according to Zelefsky, in the patient in prone position and at a safety margins of 1 cm in the anterior lateral and craniocaudal Figure 2. Systemic and random setup errors (1.5SD) along the craniocaudal axis in the patient positioning and safety margin at which CTV should have been in-cluded in the irradiation field with 90% probability directions and of 0.6 cm at the dorsal side, the coverage of CTV at the dorsal side is 85 % before and 96% after the corrections for setup error and prostate displacement.8 In our patients, the coverage of CTV can be evaluat-ed only indirectly, by assessing the involve-ment of CTV dependent exclusively on the setup errors. The results are shown in Figures 1-3. The 90% probability of CTV coverage of the irradiation field at the safety margin of 1 cm along the lateral axis is achieved in 91% of patients, and in cranial and caudal directions in 91 % and 87% of patients, respectively, and on the anterior and posterior side in 96 % and 78 % of patients, respectively on side. By re-ducing the dorsal margin to 0.7 cm, the 90% probability of CTV involvement in the irradiation field would be obtained only in 74% of patients. The irradiation precision may not be im-proved either by changing the patient position or by additional support using various positioning aids. The prone position during irradiation may helps to reduce the exposure of the rectum to irradiation;8,11 however, there is no data proving that this position can improve the positioning precision, given that it somehow deprives the patient of comfort during irradiation. Additional disadvantage of the prone position is that, in this position, Radiol Oncol 2005; 39(3): 211-7. 216 Kragelj B / Conformal radiotherapy of the prostate Table 4. Systemic and random setup errors in patients with less than 90 % probability of CTV cover-age and the direction of the setup errors Direction Patient’s no. Systemic Random error mm error mm Right 16 9 2 19 0.3 7.5 Left 15 5.0 5.8 19 0.3 7.5 Cranial 8 4.8 1.5 20 3.0 2.9 Caudal 6 3.8 2.9 14 1.8 3.3 23 3.7 2.3 Anterior 1 3.2 2.4 Posterior 5 6.4 3.5 15 3.5 4.2 20 4.4 4.2 Posterior 3 3.7 2.3 (0.7cm) 8 4.2 2.9 18 1.2 3.9 urethrography is hard to perform and, hence, the apex of the prostate cannot be reliably lo-cated. Considering the precision of the patient positioning, even the use of support cushions may be questionable. Comparing different support systems, Song reported that a similar percentage of most evident errors, approx. 40%, was made in positioning the pa-tients by using any type of support as in posi-tioning them with no support at all.9 On the other hand, a better irradiation pre-cision may be achieved by eliminating most apparent systemic errors (Table 4). By deter-mining the systemic and random positioning errors, it is possible to asses an appropriate safety margin and, by eliminating the sys-temic error, the probability of CTV involve-ment into the irradiation field may increase. By correcting the systemic error, due to which the probability of involving CTV along one of the axes during the positioning was lower than 90%, an adequate CTV coverage may be obtained along the lateral axis in 96% of pa-tients, in the cranial and caudal directions, in Radiol Oncol 2005; 39(3): 211-7. 100% and 96%, respectively, and along the an-teroposterior axis, in 100% of patients. In cas-es when the dorsal safety margin is reduced to 0.7 cm, the correction of the most evident systemic error may help to achieve an ade-quate exposure of CTV to irradiation in as much as 96% of patients. In that case, the cor-rection of isocenter would be required in 10/23 patients. From the estimates of systemic and random error, it is possible to identify the pa-tients in whom CTV would have been includ-ed in the irradiation field with 90% probabili-ty provided that the systemic error had been eliminated and the patients in which the esti-mated exposure of CTV to irradiation would have been obtained only by increasing the safety margin. By redefining the isocenter, which would be required in almost half of our patients, an optimal exposure of the most critical part of the prostate to irradiation, i.e. of the dorsal part, would be obtained in 22/23 patients, even though the safety margin was reduced to 0.7 cm. References 1. Fiorino C, Sanguineti G, Cozzarini C, Fellin G, Foppiano F, Menegotti L, et al. Rectal dose-volume constraints in high-dose radiotherapy of localized prostate cancer. Int J Radiat Biol Phys 2003; 57: 953-62. 2. Bauman G, Rodrigues G. In regard to Fiorino et al. Rectal dose-volume constraints in high-dose radio-therapy of localized prostate cancer. Int J Radiat Biol Phys 2004; 59: 912-4. 3. Harsolia AR, Vargas CE, Kestin LL, Yan D, Brabbins DS, Lockman DM, et al. Predictors of chronic urinary toxicity following treatment of prostate cancer with 3D conformal radiotherapy: Dose volume analysis of a phase II dose escalation study. [abstract]. Int J Radiat Biol Phys 2004; 60(Suppl 1): S437-8 4. Wallner KE, Merrick GS, Benson ML, Butler WM, Maki J, Tollenaar BG. Penile bulb imaging. Int J Radiat Biol Phys 2002; 53: 928-33. 5. Rudat V, Schraube P, Oetzel D, Zierhut D, Flentje Kragelj B / Conformal radiotherapy of the prostate 217 M, Wannenmacher M. Combined error of patient positioning variability and prostate motion uncer-tainty in 3D conformal radiotherapy of localized prostate cancer. Int J Radiat Biol Phys 1996; 35: 1027-34. 6. Goitein M. Nonstandard deviations. Med Phys 1983; 10: 709-11. 7. Zelefsky MJ, Happersett L, Leibel SA, Burman CM, Schwartz L, Dicker AP, et al. The effect of treatment positioning on normal tissue dose in pa-tients with prostate cancer treated with three-di-mensional conformal radiotherapy. Int J Radiat Biol Phys 1997; 37: 13-9. 8. Zelefsky MJ, Crean D, Mageras GS, Lyass O, Happersett L, Ling CC, et al. Quantification and predictors of prostate position variability in 50 pa-tients with multiple CT scans during conformal ra-diotherapy. Radiother Oncol 1999; 50: 225-34. 9. Song PJ, Washigton M, Vaida F, Lyass O, Happersett, Ling CC, et al. A comparison of four patient immobilization devices in the treatment of prostate cancer patients with three dimensional conformal radiotherapy. Radiother Oncol 1999; 50: 225-34. 10. Zelefsky MJ, Valicenti RK, Goodman K, Perez CA. Prostate cancer. Perez CA, Brady LW, Halperin EC, Schmidt-Ullrich RK, editors. Principles and practice of radiation oncology. 4th edition. Philadephia: Lippincot, Williams and Wilkins; 2004. p. 1692-762. 11. Bayley AJ, Catton CN, Haycocks T, Kelly V, Alasti H, Bristow R, et al. A randomized trial of supine vs. Prone positioning in patients undergoing esca-lated dose conformal radiotherapy for prostate cancer. Radiother Oncol 2007; 70: 37-44. Radiol Oncol 2005; 39(3): 211-7.