Organ motion, set-up variation and treatment margins in radical radiotherapy of urinary bladder cancer

Radiotherapy and Oncology 69 (2003) 291–304 www.elsevier.com/locate/radonline

Organ motion, set-up variation and treatment margins in radical radiotherapy of urinary bladder cancer Ludvig Paul Murena,b,*, Rune Smaalanda, Olav Dahlb a Department of Oncology and Medical Physics, Haukeland University Hospital, N-5021 Bergen, Norway Section of Oncology, Institute of Medicine, Medical Faculty, University of Bergen, N-5021 Bergen, Norway

b

Received 27 May 2002; received in revised form 10 June 2003; accepted 26 June 2003

Abstract Background and purpose: A major challenge in conformal radiotherapy of bladder cancer is to determine adequate treatment margins. For this purpose, we therefore quantified the internal motion of the urinary bladder as well as the external patient set-up variation during a course of fractionated radiotherapy. In the light of the recently introduced ICRU-62 concept, the planning organ at risk volume, we also studied the internal motion of nearby organs at risk, the rectum and intestine. Material and methods: Weekly CT scans and electronic portal images (EPIs) were sampled from 20 patients during radical, conformal bladder irradiation (60 – 64 Gy/2 Gy in five fractions weekly). The planning scans were acquired with 70 ml of bladder contrast instilled, and patients were instructed to void before the treatment/repeat scanning sessions. Internal motion of the bladder, rectum and intestine was measured by 3-D image matching of the repeat scans to the patients’ planning scans. Internal margins (CTV-to-ITV) were determined using both a direct empirical approach and an analytically derived margin recipe. The external patient set-up variability was determined by 2-D matching of front and lateral EPIs to corresponding digitally reconstructed radiographs. Results: A total of 149 CT scans (20 for planning, 129 during the treatment course) and 133 sets of EPIs were analysed. Bladder volumes were smaller during treatment than in the planning situation in 85% of the repeat scans. Nevertheless, we found the repeat scan bladder volumes to extend outside the planning scan bladder contours in 89% of the scans, on average with 9% of the volume (range: 0 –47%). Eight patients (40%) had at least one repeat scan (25 scans in total) where displacements .15 mm were observed at one or more sides of the bladder. CTV-to-ITV margins of 10 mm inferior, 20 mm superior, 11 mm left, 8 mm right, 20 mm anterior and 14 mm posterior were required to simultaneously encompass all bladder deflections except for the largest outward deflection in all directions in 84% of the patients. Including patient set-up variation (CTV-to-PTV), we found that an additional safety margin of 2 – 6 mm had to be added in the various directions. The rectum expanded outside the planning contours in all repeat scans, on average with 24% of the volume (range: 2 – 69%). The volume of intestine found close to the bladder were significantly and negatively correlated to the bladder volume in almost half of the patients. Conclusion: This study documented both a large internal motion of the bladder and a substantial patient set-up variation. Our current treatment margins have been adjusted according to the findings of this study. Considerable variation in position and volume of the rectum and intestine was also documented. q 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Urinary bladder cancer; Organ motion; Set-up uncertainty; Internal margins; Set-up margins

1. Introduction The survival rate after radical radiotherapy (RT) of urinary bladder cancer has traditionally been regarded as inferior of the survival after radical surgery, i.e. cystectomy. However, radical RT has often been prescribed for patients * Corresponding author. Section of Medical Physics, Department of Oncology and Medical Physics, Haukeland University Hospital, N-5021 Bergen, Norway. 0167-8140/$ - see front matter q 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0167-8140(03)00246-9

that for medical reasons were not suited for cystectomy. This has made randomised trials difficult to perform. Evidence for a radiation dose –response relation for bladder tumours has been presented, since the delivery of higher radiation doses has been reported to increase both long-term local control [3,22,26,33,35] and survival [26,33,35]. Addition of cisplatin-based chemotherapy regimens to local RT has shown promising radiation-sensitising effects [2,11,28,41,42]. Multi-modality organ-sparing treatment for

292

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

urinary bladder cancer, in which RT is a major part of the local treatment, therefore appears as a conservative, yet effective alternative to radical surgery [9,28,41]. In the conformal RT (CRT) approach individually shaped radiation fields are used to direct high doses of radiation to the tumour. The aim of CRT is to sterilise all malignant tumour cells and thereby to obtain local tumour control without risking normal tissue morbidity [5]. The balance between hitting the tumour in each fraction and at the same time avoiding as much as possible of the involved organs at risk (ORs) in practice becomes a question of the size of the safety margins around the clinical target volume (CTV). This conflict is particularly problematic in the treatment of tumour sites such as the urinary bladder where large internal organ motion is expected, where sensitive normal tissues are located close to the tumour, and where patient positioning uncertainties to a certain extent are unavoidable. Previously we have presented the background of a dose conformation technique particularly suitable for bladder irradiation, so-called partially wedged beams [25]. In a treatment planning study we documented that this technique both improved the dose homogeneity in the bladder target and reduced intestine and rectum peak doses [24]. Since bladder cancer patients are subject to potentially severe radiation-induced intestinal adverse effects [8,12,30,31,37] and currently applied tumour doses fail to control a large proportion of the patients, there is a potential for improved CRT techniques for these patients. Currently, we run a clinical trial of radiation dose escalation for bladder tumours using this dose conformation technique. Radiation dose escalation depends on secure definitions of the tumour and ORs. The recent ICRU 62 report provided guidelines and a framework for studies on internal motion and set-up variability for determination of target volume margins [15], based on the work of the Nordic Association of Clinical Physics (NACP) [1]. By measuring the internal anatomical variability of the CTV during a course of fractionated RT, one can determine the appropriate internal margins (IMs), which extend the CTV into the internal target volume (ITV). From the NACP point of view, the ITV is the most relevant target volume to which target doses should be prescribed and from which target coverage should be evaluated and reported [1]. The next step in the chain involves the quantification of the reproducibility of the patient set-up on the treatment table. This variability requires the inclusion of a further safety margin—the setup margin (SM)—to give the planning target volume (PTV) in the ICRU terminology. There has been little attention in the literature on how to convert organ motion data into treatment margins for nonsolid organs. Most of the ‘margin recipes’ presented for the determination of treatment margins from repeat CT and electronic portal image (EPI) studies have been developed for solid CTVs, e.g. the prostate [20,27,34,40]. In this study we therefore present a more direct, empirical approach for analysis of organ motion of a hollow organ. The internal

margins (CTV-to-ITV) resulting when adapting the margin recipe of van Herk et al. [20,27,40] to the present situation were compared with the internal margins resulting from this empirical method. A new concept introduced in the ICRU 62 report was the planning organ at risk volume (PRV) [15,19]. This volume contains a specific OR and a safety margin around it to account for anatomical and geometrical variability of the organ, in analogy with the target volume definition of the same report. In this study we therefore quantified the motion of both the rectum and the intestine. For the rectum we considered the possibility of using a delineation margin around this volume to obtain a volume that included some of the observed rectum variation. The majority of studies on organ motion and set-up uncertainties in pelvic irradiation have focused on prostate irradiation [7,17,23,39]. Only a few published studies have been carried out on patients with bladder cancer [13,21,36,38]. Three of these studies focused on organ motion only, and they all included a limited number of repeat CT scans (one to three repeat scans) per patient. The aim of this study was therefore to quantify both organ motion and set-up variation in bladder irradiation in order to determine the appropriate treatment margins for this group of patients. Weekly imaging examinations (CT scans, sets of EPIs) were performed on each patient. Additionally, in the light of the recently introduced ICRU 62 concept, the PRV, we studied the internal motion of nearby organs at risk, the rectum and intestine.

2. Materials and methods 2.1. Selection of patients Twenty patients (16 men, four women) were included from January 2000 to October 2001 after written consent was obtained. The age at the start of treatment ranged from 58 to 88 years (mean age: 75 years). All patients had muscle invading transitional cell tumours [24,33]. Patients were treated in supine position with a standard four-field CRT technique, and prescribed a dose of 60 –64 Gy (five weekly fractions of 2 Gy) to the whole bladder with 15 – 20 mm isotropic margins. 2.2. The CT scanning procedure The planning CT scans and the weekly repeat CT scans consisted of 10/10 slices (10-mm thick slices with 10-mm interval) from Th10 – Th11 to L5 –S1 and 5/5 slices from L5 – S1 to 2 cm inferior of the caudal pelvic floor, covering the whole intestine and rectum in addition to the tumourbearing urinary bladder. All patients were scanned in supine position with our dedicated CT scanner (ProSpeed SX Power, GE Medical Systems, Milwaukee, WI, USA). The number of axial CT slices per scan ranged from 56 to

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

72. A 70-ml contrast volume was instilled into the bladder before acquiring the planning scan. All patients were given both written and oral instructions to empty their bladder before each treatment/scanning session, and they were frequently reminded about this policy throughout the treatment period. The repeat scans were acquired as close as practically possible in time to the treatment session, preferably within 30 min. If the time period between scanning and treatment was longer than approximately 30 min, patients were instructed to void before the second (treatment or scanning) session. 2.3. The procedure applied for weekly electronic portal imaging Weekly EPIs were obtained with the Varian Portal Vision system (SW version 5.0, LC250 detector; Varian Ass, Palo Alto, CA, USA) the same day as the patients were scheduled for CT scanning. At each treatment session, two AP (08) images and two lateral (908) images were acquired using the ‘double exposure’ facility of the PortalVision system, i.e. for each gantry angle one portal image with extended field size and one image of the treatment field were recorded. The latter image was used for field shape verification, whereas the first was used for the actual patient set-up assessment. The responsible oncologist approved the set of EPIs acquired during the first treatment session. 2.4. Registration of CT scans and organ definitions The repeat scans were registered to the corresponding planning scan using the Advantage Fusion software (v. 1.15; GE Medical Systems, Milwaukee, WI, USA). The CT scans were first registered using an automatic procedure that primarily matched on bony anatomy. Before accepting the registration an arbitrary number of anatomical landmarks could be defined in both CT series to improve the registration. The automatic matching procedure performed very well in our CT series since they were all acquired with the same slice thickness and interval. However, to secure as high matching accuracy as possible, we indicated the position of six bony landmarks in both the planning/reference exam and the repeat scan. The landmarks included (1) the lower part of the coccyx, as defined in axial and sagittal cross-sections; (2) the cranial/dorsal point of the symphysis pubis, as defined in axial and coronal cross-sections; (3 and 4) the caudal/central part of both the left and right ischium tuberosity, as defined in axial and coronal sections; (5) the caudal/central part of S5, as defined in axial and coronal sections; and (6) the promontory, as defined in axial and sagittal cross-sections. We aimed at keeping the maximum discrepancy between the landmarks below 2 – 3 mm. Finally, the registration quality was checked visually using available software tools. After having completed the registration of the repeat scans to the planning scan, the bladder, rectum and intestine

293

were outlined in the repeat scans in the Advantage Fusion software. The bladder volume was defined as the volume within the outer bladder wall, and was contoured in all relevant slices from dome to apex. For the rectum, we adopted the definition used in a range of recent publications, using the first slice below the recto-sigmoid flexure as the superior/cranial limit and the first slice above the anal verge as the inferior/caudal limit [4,6,16,32]. Our rectum volumes included the rectal wall as well as the contents of the rectum. For the intestine, all segments of intestinal tissue were systematically outlined. To secure consistent organ definitions, all images were registered and all volumes outlined by one operator (LPM). In the Advantage Fusion software all organ delineations in the repeat scan were automatically transferred to the planning scan using the 3-D image registration transform. The contours of each organ in every repeat scan were saved as separate DICOM RT Structure Sets (DICOM RTSSs) linked to the planning scan. Subsequently, these DICOM RTSSs could be loaded into both the Helax-TMS treatment planning system (v. 6.0.2; Nucletron, Uppsala, Sweden) and the Advantage Sim software (v. 1.15; GE Medical Systems, Milwaukee, WI, USA) along with the planning scan. 2.5. Analysis of internal bladder motion and determination of internal margins (CTV-to-ITV) All bladder contours from the repeat scans were imported into the Helax-TMS planning system along with the corresponding planning scan using the standard image and structure import system of Helax-TMS. The bladder as seen in the planning scan was contoured in the Helax-TMS system, using the same window/level settings as in Advantage Fusion. Using available Helax-TMS software tools we measured the volume and height (Sup – Inf) of the bladder in both the planning and the repeat scans. As a measure of bladder displacements, we also outlined the repeat scan bladder volume that was outside the planning scan bladder volume. The displacement of different parts of the bladder throughout treatment was quantified with the 3-D margin tool of Helax-TMS. Using this tool, we measured the deflections of the six bladder walls in each of the repeat scans relative to the planning scan, taking bladder deflections in all slices into account. 2.5.1. Direct methods for margin determination The margin tool was subsequently used in two direct methods to determine the margins needed to account for the internal bladder motion seen in this patient material. Initially, we determined the margins that in each individual case encompassed all observed bladder structures at all sides of the bladder in the repeat scans (Method I). Secondly, we derived the margins required to envelope around all volumes where the bladder was located more than once, i.e. all bladder deflections except for the largest outward

294

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

deflection in each direction were encompassed within these margins (Method II). Since the 3-D margin tool we applied used an ellipsoid expansion, this last approach implicitly also encompassed a large fraction of the volume where the bladder was located only once. In both these methods, bladder deflections in all slices were taken into account. From these data the overall margins encompassing the bladder motion in a certain fraction of patients were derived. A main advantage with this method was that margins could be derived both when considering all directions globally and when considering each direction separately. 2.5.2. The applicability of an analytical margin recipe We also considered the margin recipe presented by van Herk and colleagues [20,27,40]. Their analytical method was based on knowledge of the systematic and random standard errors of the involved geometrical uncertainties, and was developed for a solid CTV (e.g. the prostate) assumed to move by translations only. Since bladder motion probably is a combination of translations, rotations and changes in morphology (size and shape), it was unclear whether the van Herk et al. recipe could be adapted to this problem, i.e. whether margins around the bladder could be determined from the systematic and random components of the geometrical uncertainties, and how each of these components should contribute to the margins. For instance, translational bladder motion would call for larger margins to cover all parts of the bladder at the same time than the same magnitude of organ motion caused by expansion and contractions. We considered each bladder wall separately, and used the data on the displacements of the different sides of the bladder measured in individual scans. Since these displacements were measured using the 3-D margin-growing tool, only outward (i.e. positive) deflections relative to the planning scan contours could be properly measured. To apply the margin recipe we therefore assumed that the inward deflections mirrored the outward deflections. For each case with more than one non-positive deflection we assigned these a random negative deflection with absolute value between zero and the maximum observed (positive) deflection for this patient. From this combination of measured and assigned displacements, the resulting systematic and random standard errors of the organ motion were estimated and then entered into the van Herk et al. recipe with the McKenzie et al. correction [20,27,40]. Due to the complicated nature of the bladder motion we used the version of the recipe giving the widest margins (3-D errors) [40]. That is, margins that should give 95% minimum dose coverage in 90% of patients were determined for each direction independently as the sum of 2.5 times the systematic standard error and 0.7 times the random standard error in the inferior and superior directions, and as the sum of 2.5 times the systematic standard error and 0.6 times the random standard error in the left, right, anterior and posterior directions [20,27,40] (Method III). According to

the Amsterdam group, the S.D. of the systematic errors was corrected for limited sample size [27]. 2.6. Analysis of set-up motion and determination of set-up margins The set-up variation was measured by comparing the EPIs of the treatment fields with the corresponding digitally reconstructed radiographs (DRRs) generated from the planning scans [10]. Prior to image matching, the outlines of pelvic bony structures (the sacrum and the pubic symphysis) were added to the DRRs applying the tools inherent in the Portal Vision system. The resulting structures were automatically transferred to the corresponding portal image and correctly positioned with respect to the detected field edge. A Laplace-like filter for the enhancement of the ridges typically formed behind bony structures was used to guide correct positioning of the anatomical structures. The inferior – superior and lateral set-up deviations were determined from the AP images, while the anterior – posterior set-up deviations were measured from the lateral images. Rotations were not taken into account. For the first half of patients the average results of three observers with the Portal Vision system were used; for the remaining patients the images were analysed by one of these three observers only. The measured set-up variation data were separated into its systematic and random components, with the S.D. of the systematic errors corrected for limited sample size [27]. To determine the set-up margins we applied the van Herk et al. recipe with the McKenzie et al. correction [20,27,40], using both the approximate bladder motion data and the measured set-up data. The set-up margins were defined as the linear difference between the total margins found when using the van Herk et al. recipe (systematic and random error; both organ motion and set-up variation) and the margins determined with this recipe when only considering organ motion (systematic and random error, organ motion only). 2.7. Analysis of internal rectum motion To analyse the internal rectum motion, all rectum contours and the related planning scan for each patient were imported into Helax-TMS (same method as in the bladder analysis). Initially, the planning scan rectum volume was defined using the same window/level as in Advantage Fusion. Helax-TMS software tools were used to determine the volume and length (Sup – Inf) of the rectum in both the planning and the repeat scans. To quantify how representative the planning scan was for the position and shape of the rectum during treatment, we measured the volume of the repeat scan rectum that was outside the planning scan rectum contour. To quantify rectum displacements at various levels of the rectum, we again used the 3-D margin tool and determined the margins needed to encompass the left, right, anterior and posterior

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

295

three sets of EPIs, one patient had four, five patients had six, eight patients had seven and five patients had eight EPI sets.

rectum displacements. (Inferior and superior displacements of the rectum were not considered clinically relevant.) For the included male patients these measurements were performed at the bladder top, centre and bottom, and at the prostate top, centre and bottom, at the positions seen in the planning scan. For the female patients the rectum variation was measured at the three slices related to the bladder as well as 2 cm below the bladder bottom (the latter assumed equivalent to the level of the apex of the prostate).

3.1. Bladder movements during the treatment course and internal bladder margins (CTV-to-ITV) The average bladder volume in the planning scans was 206 cm3 (S.D.: 122 cm3; range: 134 –686 cm3) (Table 1). Bladder volumes were generally smaller in the repeat scans than in the planning scan, with population-average bladder volumes ranging from 125 to 171 cm3 and an overall average of 143 cm3 (Table 1). On average the ratio of the repeat scan bladder volume and the planning scan bladder volume was 0.70, but this ratio ranged from 0.27 to 1.78 (Fig. 1). The repeat scan bladder volume was larger than the planning scan volume in 19 repeat scans (15%) for eight patients. Four patients had increased bladder volume in at least two repeat scans. The bladder in the repeat scans extended outside the planning scan bladder volume in 115 repeat scans (89%). On average, 16 cm3 or 9% (S.D.: 10%; range: 0– 47%) of the repeat scan bladder volume was outside the planning scan bladder contours. The four patients with an increased bladder volume (compared to the planning scan) in at least two repeat scans had a higher average fraction of the repeat scan bladder volume displaced outside the planning scan bladder contours (17% vs. 7%; P , 0:002, independent samples t-test). Bladder displacements as large as 29 –36 mm were observed at the superior, left, anterior and posterior side of the bladder (Table 2). In general, bladder displacements throughout the treatment course were largest at these four sides of the bladder. Eight patients (40%) had bladder displacements . 15 mm at one or more sections of the bladder (in a total of 25 repeat scans). Four patients (in 11 repeat scans) had this degree of displacement at the superior side of the bladder only. The remaining four cases had bladder movements . 15 mm at two or three of the sides of the bladder throughout treatment, the anterior side being the most variable (in total 12 incidences). In three of the eight patients with bladder displacements . 15 mm this was observed at a tumour-bearing side of

2.8. Analysis of internal intestinal motion Due to the shape of the lower part of the intestine (with segments of intestine both anterior and posterior in the pelvic/abdomen), the analysis of intestinal organ motion had to be approached differently than for the bladder and rectum. We first outlined the intestine in the planning scan in the Advantage Sim system; again the same window/level settings as in Advantage Fusion were used. We then imported the intestine volumes from the repeat scans into the Advantage Sim system together with the planning scan. The total volumes of intestine and the volumes of intestine below the promontory were delineated in both the planning and the repeat scans. Additionally, we quantified the volume of intestine below the slice corresponding to (1) the bladder top in the planning scan; and (2) 2 cm above the bladder top in the planning scan. The Sup –Inf distance from the slice corresponding to the bladder top in the planning scan down to the most inferior intestinal structure was also recorded (i.e. no. of slices counted).

3. Results All patients were scheduled for weekly repeat CT scanning and EPI sessions, but due to practicalities related to the conduction of the study, we managed to acquire a total of 149 CT scans (20 planning and 129 repeat scans) and 133 sets of EPIs. One patient had three repeat CT scans, eight patients had six scans, 10 had seven scans and one patient had eight repeat scans. The majority of EPIs (89%) were acquired on the same days as the CT scans. One patient had

Table 1 Population averages of parameters describing bladder volume variation throughout the treatment course

Mean bladder volume (cm3) S.D. of bladder volume (cm3) Max bladder volume (cm3) Min bladder volume (cm3) Paired t-test of equality of means, relative to planning scan Mean bladder fraction outside planning scan contour (%) S.D. of bladder fraction outside planning scan contour (%)

Planning scan

Scan 1

Scan 2

Scan 3

Scan 4

Scan 5

Scan 6

Scan 7

206 121 686 134

171 108 514 66

159 124 583 63

140 81 388 67

134 69 282 61

125 79 335 46

134 94 430 52

144 65 266 80

,0.001

,0.001

–

0.02

,0.001

0.004

0.003

0.02

–

12

10

9

8

8

8

13

–

14

10

9

7

10

10

7

296

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

Fig. 1. The ratio of the repeat scan bladder volume and the planning scan bladder volumes for the bladder patient material. The bladder volume was smaller in the repeat scan than during planning in 110 (85%) of the scans.

the bladder, as documented at the cystoscopy examination prior to treatment. In 13 scans of five of these eight patients this bladder displacement coincided with an increase in the repeat scan bladder volume compared to the planning scan bladder volume. In two patients (in nine scans) these displacements were caused by the presence of intestinal loops, which in the planning scan shifted the bladder away from its ‘normal’ position, i.e. where it was seen in the repeat scans. The planning scan, with bladder contours from the repeat scans superimposed, for one of these patients is shown in Fig. 2. In the last case having pronounced bladder deflections (in three scans) the bladder top extended more cranially in the repeat scan than in the planning scan. In one of the repeat scans of this patient the bladder volume was also larger than in the planning scan. Further data on bladder

displacements, separated into the various directions are given in Table 2. Initially, the CTV-to-ITV internal bladder margins were determined with our two direct empirical margin approaches (Methods I and II), analysing each side of the bladder separately (Fig. 3). However, to adequately encompass the bladder motion, the margins at all sides of the bladder should be considered simultaneously. Using Method II with this requirement, the internal margins (CTV-to-ITV) had to be 10 mm inferior, 20 mm superior, 11 mm left, 8 mm right, 20 mm anterior and 14 mm posterior to encompass the variation in 16 of 19 patients (84%) (Table 3). The use of 70 ml of contrast in the planning scan and then treatment with empty bladder could in practice imply an

Table 2 Summary of bladder displacements as seen in individual repeat scans for all patients

Overall maximum (mm) No. of scans with displacements .10 mm No. of patients with at least one scan with displacements .10 mm No. of scans with displacements .15 mm No. of patients with at least one scan with displacements .15 mm No. of scans with displacements .20 mm No. of patients with at least one scan with displacements . 20 mm

Inferior

Superior

Left

Right

Anterior

Posterior

15

32

31

15

36

29

6

29

14

5

24

16

5

9

7

3

8

8

0

12

4

0

12

2

0

5

2

0

4

2

0

7

4

0

6

1

0

4

2

0

3

1

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

297

Fig. 2. A patient where a loop of intestine had displaced the bladder posterior at the time of the planning scan. Six repeat scans were acquired during the treatment schedule, and were registered with the reference/planning scan using 3-D matching. The contours of the bladder in the registered, repeat scans are shown superimposed on the planning scan.

298

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

Fig. 3. Margins that encompassed the internal bladder motion, each of the six sides of the bladder analysed separately. (a) Margins that encompassed all bladder variations, including all observed bladder deflections in the six directions. (b) Margins that encompassed the majority of bladder variations, i.e. all variations except for the largest outward deflection in the six directions. In this last analysis, 19 patients with six to eight repeat scans were included. The dotted lines indicate the margins that encompassed the bladder motion in 90% and 95% of all patients.

extra safety margin. We did, however, not find any correlation between the internal margins required for each individual case (average across the six directions) and the planning scan bladder volumes. The CTV-to-ITV internal margins determined with the van Herk et al. recipe deviated considerably (with at most 9 mm) in single directions from the various margin alternatives derived from our direct methods (Table 3). However, reasonably good agreement (in the range 1 –3 mm) was found between the margins from the van Herk et al. recipe and the margins that covered all bladder

deflections except for the largest outward deflection in each direction (Method II) in 89% of the patients, each direction considered separately (Table 3, row 2). Agreement within 4 mm was achieved from the margins that covered all bladder deflections except for the largest outward deflection in each direction (Method II) in 84% of the patients, all directions considered simultaneously (Table 3, row 4). The approximate random and systematic components of the bladder motion applied in our adaptation of the van Herk et al. methodology [20,27,40] are given in Table 4.

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

299

Table 3 Internal (CTV-to-ITV) margins (mm) required to encompass different degrees of the random and systematic organ motion of the bladder, as derived with various methods. The margins in the fourth row are the internal bladder margins we recommend if using similar imaging and treatment procedures as described in this report

Method I, each side analysed separately, 90% the of patients Method II, each side analysed separately, 89% of the patients Method II, all sides analysed simultaneously, 89% of the patients Method II, all sides analysed simultaneously, 84% of the patients Method III (adaptation of van Herk et al. recipe), each side analysed separately, 90% of the patients

Inferior

Superior

Left

Right

Anterior

Posterior

12

26

13

10

25

14

10

17

11

5

20

11

10

20

21

8

21

14

10

20

11

8

20

14

8

20

12

8

18

10

required in the inferior, superior, left, right, anterior and posterior directions, respectively. According to the NACP terminology (i.e. delineating the ITV and not the PTV), these are the margins that should be secured between the ITV and the 95% isodose. Using ICRU 50/62 terminology, the total CTV-to-PTV margin is given by adding these margins to the appropriate margins in Table 3.

3.2. Set-up variation and set-up margins Of the total 133 sets of EPIs, the lateral images were of reduced image quality (or were not completely captured) in 14 sets (11%), and these could therefore not be used in the analysis. Overall, single set-up deviations up to 15 mm in the inferior –superior direction, up to 11 mm left – right, and up to 12 mm in the anterior –posterior direction were observed. Individual systematic errors up to 12 mm inferior –superior, 5 mm left – right and 9 mm anterior – posterior were disclosed. The systematic and random components of the measured set-up variation were derived [27] (Table 4). When comparing with the approximate organ motion data it was found that set-up variation was the dominating source to systematic deviations in the inferior, right and posterior directions, while organ motion dominated the systematic variations in the superior, left and anterior directions (Table 4). Neither the random nor the systematic components of the set-up variation in any direction were correlated to the size of the patient, quantified as the ‘axial extent’ of the patient (the area of the ellipse with axes equal to half the thickness (anterior – posterior) and half the width (left – right) of the patient in the planning CT slice through the promontory). Using the van Herk et al. recipe with the approximate organ motion data and the measured set-up data (Table 4), we found that set-up margins of 6, 3, 2, 3, 3 and 4 mm were

3.3. Rectum movements during the treatment course A total of 141 rectum volumes were outlined and analysed in 19 patients (one patient where the rectum was very difficult to separate from the small and large intestine was excluded from the analysis). There was no time trend in the rectum volumes as the population-average volume at each scan seemed to be relatively constant over time, varying from a maximum of 82 cm3 in the planning scan to 62 cm3 in the seventh scan (Table 5). Large individual variations in the rectum volumes throughout the treatment course were disclosed. Overall, the rectum volumes ranged from 30 to 164 cm3. The ratio of the repeat scan rectum volume and the planning scan rectum volume was on average 0.93, and it ranged from 0.29 to 2.77. In five patients (26%) the volume of the rectum during treatment was always smaller than in the planning scan, in one patient (5%) the rectum was larger in all repeat scans than in the planning scan. For the remaining 13 patients

Table 4 Random and systematic components (1 S.D., mm) of the organ motion of the bladder and of the patient set-up variation entered into the van Herk et al. margin recipe. The organ motion data are approximate as inward bladder wall displacements could not accurately be measured, but were assumed to mirror the directly measured outward displacements (see text for details)

Organ motion, random (s) Organ motion, systematic (S) Set-up variation, random (s) Set-up variation, systematic (S)

Inferior

Superior

Left

Right

Anterior

Posterior

4.6 2.1 1.9 3.8

7.9 5.7 1.9 3.8

5.1 3.7 2.5 2.5

3.5 2.2 2.5 2.5

5.4 5.8 2.7 3.3

6.0 2.7 2.7 3.3

300

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

Table 5 Population averages of parameters describing rectum volume variation throughout the treatment course

Mean rectum volume (cm3) S.D. of rectum volume (cm3) Max rectum volume (cm3) Min rectum volume (cm3) Paired t-test of equality of means, relative to planning scan Mean rectum fraction outside planning scan contour (%) S.D. of rectum fraction outside planning scan contour (%)

Planning scan

Scan 1

Scan 2

Scan 3

Scan 4

Scan 5

Scan 6

Scan 7

82 34 151 36

72 27 120 34

70 34 164 30

72 29 160 38

65 20 121 38

66 23 127 32

69 28 134 40

62 25 103 34

–

ns

ns

ns

ns

ns

ns

ns

–

24

22

25

21

23

25

29

–

10

14

16

15

16

20

14

(68%) the rectum volume both increased and decreased throughout treatment relative to the planning scan. The repeat scan rectum volume extended outside the planning scan rectum contours in all repeat scans, on average with 18 cm3 or 24% (S.D.: 15%; range 2– 67%). In eight scans (7%) of three patients more than half of the rectum volume had extended outside the planning scan contours, in 34 scans (28%) of nine patients more than onethird of the rectum had been displaced outside the planning contour. The rectum displacements did not vary significantly between different levels (in the Sup– Inf direction) of the rectum. When comparing the displacements across the various levels of the rectum, we found that the anterior and left displacements were generally larger than the right and posterior displacements. 3.4. Large and small intestinal movements during the treatment course The whole intestine was completely defined in 133 scans in 18 patients, while the intestine below the level of the promontory was delineated in 140 scans in 19 patients. The intestine volume was on average 2277 cm3 (S.D.: 562 cm3), ranging from 1373 to 3669 cm3. Intestine volumes during treatment were not different from the intestine volumes in the planning scan. Between different patients, the average intestine volume across the repeat scans varied substantially, and ranged from 1517 to 3270 cm3 (S.D. of averages: 540 cm3). The intestine volume was found to be significantly and positively correlated to the axial extent of the patient (defined in Section 3.2) with the intestine volume (in cm3) equal to 1.93 times the axial extent (in cm2) þ 1005 cm3 (Pearson coefficient: 0.52; P ¼ 0:03). In single repeat scans for individual patients the intestine volume deviated considerably from the volume in the planning scan, with a ratio in the range of 0.71 to 1.54. The volume was within ^ 10% of the planning scan volume in only one patient (6%), but was within ^ 20% in 14 of 18 patients (78%). The volume of intestine at the level of or below the promontory was on average 455 cm3 (S.D.: 266 cm3, range:

75– 1339 cm3). None of the population-averages in the repeat scans were significantly different from the average in the planning scan. A large inter-patient variation was observed since the average intestine volume at the promontory level or below varied from 140 to 1165 cm3 (S.D. of averages: 256 cm3). The ratio of this volume between the repeat scans and the planning scans ranged from 0.57 to 3.16. Fourteen of the 19 patients (74%) had scans where the volume of intestine in this region was both higher and lower than in the planning scan. The fraction of intestine volume that was at the level of the promontory or below was on average 21% (S.D.: 14%), and ranged from 4 to 62%. Population-averages at each scan remained fairly constant (range of averages: 19 – 24%; S.D.: 2%), whereas large individual variations were seen, with averages across the scans ranging from 8 to 57% (S.D.: 14%). The measurements of the amount of intestine below the slice corresponding to (i) the bladder top in the planning scan; and (ii) 2 cm above the bladder top in the planning scan both showed similar features (data not shown). Analysing individual patients separately, we found that the volume of intestine close to the bladder during the treatment course (as quantified with the two previous intestine volumes) correlated with the size of the bladder (Table 6). Bladder volumes showed stronger correlations than bladder heights. Except from one scan, the intestine extended below the slice corresponding to the top of the bladder in the planning scan. Overall, the lowest segment of small/large intestine extended on average 3.8 cm below the slice corresponding to the bladder top in the planning scan (S.D.: 1.8 cm), and at most with 8.5 cm. Deviations from the planning situation with as much as 2.5 cm caudal and 3.0 cm cranial were seen in individual patients. In 10 patients (50%) the deviations throughout treatment from the planning situation were within ^ 1 cm.

4. Discussion In this paper we have presented data on volume and shape variation of the bladder, rectum and intestine and on

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

301

Table 6 Correlation in bladder volume/height and volume of intestine below the slice corresponding to the bladder top in the planning scan and below the slice 2 cm above the bladder top in planning scan Bladder volume vs. Vintestine, bladder top No. of patients with negative correlation coefficient No. of patients with negative, statistically significant correlation coefficient Correlation coefficient range Average correlation coefficient for patients with negative correlation Average correlation coefficient for patients with negative, statistically significant correlation

20

Bladder volume vs. Vintestine, 2 cm above bladder top

18

Bladder height vs. Vintestine, bladder top

18

Bladder height vs. Vintestine, 2 cm above bladder top

17

9 20.98 to 20.13

7 20.94 to 0.26

6 20.91 to 0.12

5 20.90 to 0.17

20.70

20.57

20.64

20.54

20.88

20.84

20.80

20.81

patient set-up accuracy during CRT for bladder cancer. To our knowledge, this is the first study to present data quantifying all these issues simultaneously using weekly repeat CT scans and EPIs. The previously published studies on organ motion during bladder irradiation have been limited to one to three repeat scans per patient, and to a certain extent also to a low number of CT slices per scan [13,21,36,38]. The data presented here showed that there were substantial day-today variations in volume and position of the bladder. To quantify this variation, it is crucial to measure on an adequate number of treatment days. We quantified organ motion once a week, e.g. in every fifth treatment session, a ‘sampling frequency’ that was a compromise between the ideal situation (measurements at each treatment session) and the strain that we felt was acceptable for the patients as well as the available radiographer manpower and CT scanner time. Our data also indicated that the studies with limited number of slices per scan might not be expected to capture all bladder motion [36]. This was demonstrated by comparing displacements quantified with the methods presented here and displacements measured in a central slice and at the top and bottom of the bladder (data not shown). However, with the disparity in methodology in mind, our observations seem to be in line with the reports from earlier studies. Turner et al. [38] measured organ motion three times during treatment, and found that 33% of the patients had at least one incidence of outwards wall movements larger than 15 mm in any direction. In the study of Harris and Buchanan [13], bladder motion was measured half way and at the end of treatment. The bladder was found to extend outside of the treatment fields in 20% of cases (1 – 2 cm margin was applied). Sur et al. [36] assessed bladder motion once, half way through the treatment course, and found that the target volume had moved out of the treatment field in 13% of the patients (1 –2 cm margin was applied). In a recent study of Meijer and colleagues, bladder motion was analysed using a very sophisticated method that allowed 2-D maps of the organ motion measured perpendicularly at all points of the bladder wall to be derived [21].

However, in order to present treatment margin proposals that could easily be implemented clinically, they approximated this method to a 3-D ellipsoid expansion, similar to the one used in the present study. The resulting margins agreed well with those derived with our method, indicating a quite similar degree of bladder motion. In our study, we observed at least one incidence of outward bladder movements larger than 15 mm in any direction in eight of 20 patients (40%) and in a total of 25 scans (19%). The studies with the highest number of measurements during treatment (the studies of Turner et al. [38], Meijer et al. [21] and the present study) documented the largest and a quite similar extent of bladder motion. The bladder had extended outside the planning scan contours in the vast majority (89%) of the repeat scans, despite that the bladder volumes were smaller during treatment than in planning in 85% of the scans. This shows that a certain extent of bladder motion away from the planning contours may be present even when good adherence to a bladder emptying policy is achieved. However, 14 of the 19 scans with increased bladder size were among the 25 repeat scans with displacements . 15 mm observed at least at one side of the bladder. An increased bladder volume relative to the planning situation could explain the large bladder displacements in one or more scans in five of the eight patients. The large variations observed at the superior, anterior and posterior side of the bladder support our policy of instructing patients to void immediately before each treatment session. In two patients large bladder displacements were seen because of the presence of intestinal loops in the planning scan that temporarily shifted the bladder away from the position seen in the scans acquired during treatment. In these cases this led to a systematic treatment error. However, in clinical routine where treatment is based on one planning scan only, it is difficult to judge whether the anatomy as shown in the CT scan will be representative or not for the treatment situation. Presently, without knowledge of the actual day-to-day position of the CTV, adding treatment margins around the CTV has been the only

302

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

practical method to take anatomical and geometrical uncertainties into account. In the future methods for depositing the daily radiation dose with high spatial accuracy will have a potential for improving the outcome of RT of tumour sites with large anatomical and/or geometrical variation. For instance, the tomotherapy systems that allow CTV imaging immediately before treatment without moving the patient are now commercially available (Tomotherapy Inc, Middleton, US) [18]. The potential for such treatment strategies was also documented in this study. The ratio of the ITVs determined by our Method II and the volumes encompassing the span of actual bladder positions ranged from 1.1 to 2.3 (average: 1.6). On average, these ITVs included a volume of 128 cm3 (range: 26– 347 cm3) that was not seen to contain the bladder in any of the repeat scans. In this study we determined the size of the internal margins required at the different sides of the bladder primarily applying a direct, empirical approach (Methods I and II). Additionally we also tested the applicability of the analytically derived van Herk et al. margin recipe (Method III). Overall, considerable differences between the results of the van Herk et al. margin recipe and the direct empirical methods were seen (Table 3). However, reasonable agreement was found between the IMs from the van Herk et al. recipe and two of the margin alternatives derived with the empirical methods (Table 3). As discussed in Section 2.5.2, it is not obvious whether the van Herk et al. recipe can be applied to derive bladder margins or not since many of the underlying assumptions seem to be violated. A major advantage with the direct empirical methods presented here is that they include no assumptions about the nature of the bladder motion. Another advantage with the direct empirical approaches we applied was that the required IMs could readily be determined considering either the different sides of the bladder separately or simultaneously. If bladder deflections in different directions are totally independent, this may have a profound impact on the size of the margins required. One should therefore apply the margins that at the same time take bladder deflections at all sides into account (Table 3, row 3 or 4). With this requirement, the margins needed to encompass the variation in 84% and 89% of the patients were almost the same (except for at the left side). Using similar imaging and treatment procedures as described in this report, we therefore recommend using the internal bladder margins given in Table 3, row 4. Independent of the method used for deriving the margins, this study documented that an-isotropic internal margins should be applied in bladder irradiation. A striking feature of our data is the difference between the margins required at the left and right side of the bladder. Since this appears not to be an effect caused by statistical chance, it reflects that there actually is larger organ motion at the left side of the bladder. A likely reason for this asymmetry is that the sigmoid colon bends from the recto-sigmoid flexure over the bladder and to the left.

The set-up variation data presented here seem to be in relatively good agreement with previous studies on the setup reproducibility of pelvic irradiation. Our random deviations are within the generally accepted state of the art accuracy (2.5 mm for prostate, 3.0 mm for general pelvic), while our systematic deviations in the inferiorsuperior and anterior – posterior directions are somewhat higher than other previous reports [14]. Under the assumption that the approximation of the inward bladder wall displacements is valid, the set-up variation dominates the total systematic variation in the inferior, right and posterior directions (Table 4). The set-up margins were derived using the van Herk et al. recipe using both the approximate organ motion data and the (unmodified) set-up data (Table 4). However, as the set-up margins were determined as the difference between the margins accounting for both organ motion and set-up variation and the margins accounting for organ motion only, the impact of this approximation should be minimised. At least at the inferior and posterior side there appears to be a potential for future margin reductions by improving the patient set-up accuracy. This study has therefore initiated a more active use of portal imaging in bladder irradiation at our clinic, including implementation of a relative simple correction strategy. Hopefully, this will lead to lower systematic set-up errors, and future margin reduction. Reduction of the posterior set-up margin should reduce rectum doses and could translate into lower incidences of treatment-induced rectal adverse effects. In the present study we pointed out patients with large bladder motion that may be at higher risk for geographical misses and thereby local treatment failures. Set-up variation data (at least the systematic component) and the actual treatment plan should be included to more clearly identify the high-risk patients. The target volume margins determined in the present study do not include bladder delineation uncertainties. Intuitively, since the bladder is relatively easy to outline and due to the documented large organ motion and set-up variation, bladder delineation uncertainties should have only a minor impact on the margins in bladder irradiation. This has also been demonstrated by Meijer and co-workers, as shown in their report [21]. A substantial variation in size and position of the rectum was documented in this study. Organ motion of the rectum has been a topic also in several previous studies (e.g. [17,29]), but the present study is the first to quantify the amount of rectum that extends outside of the initial rectum contours of the planning scan. We found that on average 24% of the rectum extended outside of the planning scan contours. Based on this observation, one may question the validity and value of rectum DVHs. The finding should at least be taken into account in the discussion on whether the rectum should be defined as the wall, surface or the whole volume (contents included). Due to the large rectum variation and due to the small thickness of the rectum

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304

wall, it may at least geometrically seem more robust to delineate the whole rectum volume than the wall only. Defining the rectum as the whole volume (contents included) may also be more in line with the proposal of the ICRU 62 report of using delineation margins around the rectum to account for geometrical uncertainties [15]. Based on unpublished data, McKenzie et al. proposed that a margin of approximately 4 mm should be used at the anterior rectal wall facing the high-dose region [19]. The rectum variation seen in our material, however, suggested that margins in the range of approximately 10 – 15 mm should be used if the resulting expanded volume shall encompass the majority of rectum displacements (data not shown). The margins proposed by McKenzie et al. seem to encompass approximately the average displacement seen in our data. It was beyond the scope of this presentation to analyse the rectum motion in terms of the McKenzie et al. approach; this will be addressed in future investigations. Regarding delineation of the intestine and the PORV concept, a probability distribution approach seems more suitable for the intestine than a margin-based approach. Yet the bottom line for this discussion is that ORs are outlined when planning RT to get an estimate of the dose received in the organ and to use this to judge the patients risk for experiencing treatment-induced adverse effects. It still remains an open question whether inclusion of OR motion data improves the predictive power of present methods used for estimating risks for developing normal tissue late adverse effects after (pelvic) irradiation. This should be a topic for future research. The demonstrated correlation between the bladder volume and the volume of intestine below the slice that corresponds to the bladder top in the planning scan clearly underlines the importance of maintaining an invariant bladder volume during the treatment course. If the bladder volume is smaller during treatment than in the planning situation, there is certainly a good chance that the bladder CTV is adequately covered during treatment, but at the expense of irradiating a larger volume of intestine. Presently, lack of local control is the major challenge in treatment of bladder cancer, and this justifies the use of sufficient, and therefore, wide treatment margins. But it should be taken into account when judging the risk of intestinal morbidity that the amount of intestine seen close above the bladder in the planning scan may underestimate the realistic situation during treatment. The situation is further complicated as different intestinal loops can extend downward to the bladder at different treatment sessions, effectively reducing the peak total doses received in the intestine. On the contrary, in patients with fixed intestinal loops after prior surgery, the same intestinal segments may receive the highest doses in each treatment fraction, probably increasing the risk of intestinal late effects. In conclusion, the present study of conformal bladder irradiation has disclosed both large internal motion of the bladder and a substantial patient set-up variation. Our

303

current treatment margins for bladder irradiation have been adjusted to match these findings. Large variations in volume and shape of the rectum and intestine were also documented. Future work should aim at reducing both organ motion and set-up variation in bladder irradiation and at developing methods to take the organ motion of ORs properly into account in modelling and prediction of late effects after pelvic irradiation.

Acknowledgements Generous grants from the Norwegian Cancer Society, the L. Meltzer foundation of the University of Bergen, and the Michael Irgens Flocks legacy of Haukeland University Hospital supported this study. The authors want to express their gratitude to Bernt Rekstad, M.Sc., and Espen Vorland, M.Sc., for their contributions during the initial phase of the study. Karsten Eilertsen, M. Sc., The Norwegian Radium Hospital, Oslo, is acknowledged for providing the RT Navigator software and his image registration expertise. Medical physicist Einar Dale, Ph.D., The Norwegian Radium Hospital, Oslo and radiologist Dag Jensen, M.D., Dept. of Diagnostic Radiology, Haukeland University Hospital are acknowledged for useful advice on how the rectum should be defined. All involved radiographers at the radiotherapy department, and in particular Randi Ekerold, B.Sc, Frode Gald, B.Sc, Victor Pettersen, B.Sc, and Jane Bauge, B.Sc, are acknowledged for their efforts.

References [1] Aaltonen P, Brahme A, Lax I, et al. Specification of dose delivery in radiation therapy. Recommendations by the Nordic Association of Clinical Physics (NACP). Acta Oncol 1997;36:1–32. [2] Awwad HK, Khalil E, Zikri A. The possibility of bladder preservation (BP) in the radical treatment of a mixed group of transitional and squamous cell bladder cancer: A phase II study (abstr). Radiother Oncol 2000;56:S13. [3] Battermann JJ, Hart GAM, Breur K. Dose–effect relations for tumour control and complication rate after fast neutron therapy for pelvic tumours. Br J Radiol 1981;54:899–904. [4] Boersma LJ, van den Brink M, Bruce AM, et al. Estimation of the incidence of late bladder and rectum complications after high-dose (70–78 GY) conformal radiotherapy for prostate cancer, using dose-volume histograms. Int J Radiat Oncol Biol Phys 1998;41: 83–92. [5] Dahl O, Kardamakis D, Lind B, Rosenwald JC. Current status of conformal radiotherapy. Acta Oncol. 1996;35:41–57. [6] Dale E, Olsen DR, Fossa˚ SD. Normal tissue complication probabilities correlated with late effects in the rectum after prostate conformal radiotherapy. Int J Radiat Oncol Biol Phys 1999;43:385– 91. [7] Dawson LA, Mah K, Franssen E, Morton G. Target position variability throughout prostate radiotherapy. Int J Radiat Oncol Biol Phys 1998;42:1155–61. [8] Duncan W, Williams JR, Kerr GR, et al. An analysis of the radiation related morbidity observed in a randomized trial of

304

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18] [19]

[20]

[21]

[22] [23]

[24]

[25]

L.P. Muren et al. / Radiotherapy and Oncology 69 (2003) 291–304 neutron therapy for bladder cancer. Int J Radiat Oncol Biol Phys 1986;12:2085 –92. Dunst J, Rodel C, Zietman A, Schrott KM, Sauer R, Shipley WU. Bladder preservation in muscle-invasive bladder cancer by conservative surgery and radiochemotherapy. Semin Surg Oncol 2001;20: 24–32. Eilertsen, K. An object oriented approach to radiotherapy image registration, storage and communication. Proceedings of the 5th International Workshop on Electronic Portal Imaging. 1998. Phoenix, Arizona. Giovinazzo G, Arcangeli G, Tirindelli Danesi D, Saracino B, Cruciani E. Organ preservation in muscle infiltrating bladder cancer by concomitant chemoradiation (abstr). Radiother Oncol 2000;56: S13. Gospodarowicz MK, Rider WD, Keen CW, et al. Bladder cancer: Long-term follow-up results of patients treated with radical radiation. Clin Oncol 1991;3:155– 61. Harris SJ, Buchanan RB. An audit and evaluation of bladder movements during radical radiotherapy. Clin Oncol (R Coll Radiol) 1998;10:262 –4. Hurkmans CW, Remeijer P, Lebesque JV, Mijnheer BJ. Set-up verification using portal imaging; review of current clinical practice. Radiother Oncol 2001;58:105–20. ICRU, Report 62: Prescribing, recording and reporting photon beam therapy (supplement to ICRU report 50). Bethesda: International Commission on Radiation Units and Measurements. 1999. Jackson A, Skwarchuk MW, Zelefsky MJ, et al. Late rectal bleeding after conformal radiotherapy of prostate cancer. II. Volume effects and dose-volume histograms. Int J Radiat Oncol Biol Phys 2001;49: 685–98. Lebesque JV, Bruce AM, Kroes AP, Touw A, Shouman RT, van Herk M. Variation in volumes, dose-volume histograms, and estimated normal tissue complication probabilities of rectum and bladder during conformal radiotherapy of T3 prostate cancer. Int J Radiat Oncol Biol Phys 1995;33:1109 –19. Mackie TR, Balog J, Ruchula K, et al. Tomotherapy. Semin Radiat Oncol 1999;9:108–17. McKenzie A, van Herk M, Mijnheer B. Margins for geometric uncertainty around organs at risk in radiotherapy. Radiother Oncol 2002;62:299 –307. McKenzie AL, van Herk M, Mijnheer B. The width of margins in radiotherapy treatment plans. Phys Med Biol 2000;45: 3331–42. Meijer GJ, Rasch C, Remeijer P, Lebesque JV. 3D analysis of delineation errors, setup errors and organ motion during radiation therapy of bladder cancer. Int J Radiat Oncol Biol Phys 2003;55: 1277–87. Morrison R. The results of treatment of cancer of the bladder—a clinical contribution to radiobiology. Clin Radiol 1975;26:67–75. Mubata CD, Bidmead AM, Ellingham LM, Thompson V, Dearnaley DP. Portal imaging protocol for radical dose-escalated radiotherapy treatment of prostate cancer. Int J Radiat Oncol Biol Phys 1998;40: 221–31. Muren LP, Hafslund R, Gustafsson A, Smaaland R, Dahl O. Partially wedged beams improve radiotherapy treatment of urinary bladder cancer. Radiother Oncol 2001;59:21–30. Muren LP, Hafslund R, Murman A, Valen H, Dahl O. Partially wedged radiation beams. Med Phys 2000;27:527–34.

[26] Quilty PM, Duncan W. Primary radical radiotherapy for T3 transitional cell cancer of the bladder: An analysis of survival and control. Int J Radiat Oncol Biol Phys 1986;12:853–60. [27] Remeijer P, Geerlof E, Ploeger L, Gilhuijs K, van Herk M, Lebesque JV. 3-D portal image analysis in clinical practice: an evaluation of 2-D and 3-D analysis techniques as applied to 30 prostate cancer patients. Int J Radiat Oncol Biol Phys 2000;46:1281– 90. [28] Rodel C, Grabenbauer GG, Kuhn R, et al. Organ preservation in patients with invasive bladder cancer: initial results of an intensified protocol of transurethral surgery and radiation therapy plus concurrent cisplatin and 5-fluorouracil. Int J Radiat Oncol Biol Phys 2002;52: 1303–9. [29] Roeske JC, Forman JD, Mesina CF, et al. Evaluation of changes in the size and location of the prostate, seminal vesicles, bladder, and rectum during a course of external beam radiation therapy. Int J Radiat Oncol Biol Phys 1995;33:1321–9. [30] Salminen E. External beam radiation treatment of urinary bladder carcinoma. An analysis of results in 203 patients. Acta Oncol 1990;29: 909 –14. [31] Skolysewski J, Reinfuss M, Weiss M. Radical external beam radiotherapy of urinary bladder carcinoma. An analysis of results in 500 patients. Acta Oncol 1994;33:561–5. [32] Skwarchuk MW, Jackson A, Zelefsky MJ, et al. Late rectal toxicity after conformal radiotherapy of prostate cancer (I): multivariate analysis and dose-response. Int J Radiat Oncol Biol Phys 2000;47: 103 –13. [33] Smaaland R, Akslen LA, Tønder B, Mehus A, Lote K, Albrektsen G. Radical radiation treatment of invasive and locally advanced bladder carcinoma in elderly patients. Br J Urol 1991;67:61–9. [34] Stroom JC, de Boer HC, Huizenga H, Visser AG. Inclusion of geometrical uncertainties in radiotherapy treatment planning by means of coverage probability. Int J Radiat Oncol Biol Phys 1999; 43:905–19. [35] Suit HD, Westgate SJ. Impact of improved local control on survival. Int J Radiat Oncol Biol Phys 1986;12:453– 8. [36] Sur RK, Clinkard J, Jones WG, et al. Changes in target volume during radiotherapy treatment of invasive bladder carcinoma. Clin Oncol 1993;5:30–3. [37] Tait DM, Nahum AE, Meyer LC, et al. Acute toxicity in pelvic radiotherapy; a randomised trial of conformal versus conventional treatment. Radiother Oncol 1997;42:121– 36. [38] Turner SL, Swindell R, Bowl N, et al. Bladder movement during radiotherapy for bladder cancer: implications for treatment planning. Int J Radiat Oncol Biol Phys 1997;39:355– 60. [39] van Herk M, Bruce A, Guus Kroes AP, Shouman T, Touw A, Lebesque JV. Quantification of organ motion during conformal radiotherapy of the prostate by three dimensional image registration. Int J Radiat Oncol Biol Phys 1995;33:1311– 20. [40] van Herk M, Remeijer P, Rasch C, Lebesque JV. The probability of correct target dosage: dose-population histograms for deriving treatment margins in radiotherapy. Int J Radiat Oncol Biol Phys 2000;47:1121–35. [41] Zapatero A, Martin de Vidales C, Marin A, et al. Invasive bladder cancer: a single-institution experience with bladder-sparing approach. Int J Cancer 2000;90:287–94. [42] Zouhair A, Ozsahin M, Schneider D, et al. Invasive bladder carcinoma: a pilot study of conservative treatment with accelerated radiotherapy and concomitant cisplatin. Int J Cancer 2001;96:350 –5.