- Email: [email protected]
electron spin resonance (ESR) dosimetry in radiotherapy of prostate cancer: A feasibility study
Radiotherapy and Oncology 88 (2008) 140–147 www.thegreenjournal.com
Dosimetry
In vivo alanine/electron spin resonance (ESR) dosimetry in radiotherapy of prostate cancer: A feasibility study Daniela Wagnera,*, Mathias Antonb, Hilke Vorwerka, Tammo Gsa ¨ngera, Hans Christiansena, Bjoern Poppec, Clemens Friedrich Hessa, Robert Michael Hermanna a
Department of Radiotherapy and Radiooncology, University Hospital Goettingen, Germany, bPhysikalisch-Technische Bundesanstalt, Braunschweig, Germany, cDepartment of Radiooncology, Pius Hospital Oldenburg, Germany
Abstract Purpose: We have developed a device to evaluate the potential of alanine/electron spin resonance (ESR) dosimetry for quality assurance in 3D conformal radiotherapy for prostate cancer. It consists of a rectal balloon carrying eight alanine dosimeter probes and two metal markers to document the exact position of the balloon. We measured the effects of an air-filled rectal balloon on the dose at the rectal wall and compared these results with the applied dose distribution of the treatment planning system. Materials and methods: During 10 fractions with 2.0 Gy per fraction, the accumulated doses were measured in 3 patients. The results of the ESR measurements were compared to the applied doses. Results: It was possible to insert the device without clinical complications and without additional rectal discomfort for the patients. The measurements of the dose accumulated at the anterior and the posterior rectal wall agreed with the applied dose within a mean deviation of 1.5% (overestimation of the dose) and 3.5% (underestimation of the dose), respectively. However, clinically significant differences between applied and measured rectal doses were seen in a patient with a hip prosthesis. In this case, the dose at the anterior rectal wall was overestimated by the TPS by about 11% and the dose at the posterior rectal wall was underestimated by approximately 7%. Conclusion: The method presented in this study is useful for quality control of irradiations in vivo. c 2008 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 88 (2008) 140–147.
Keywords: Radiation therapy; Alanine; Electron spin resonance; In vivo; Dosimetry; Prostate cancer
3D conformal radiotherapy for prostate cancer is a curative treatment option. Due to the close anatomical relation between the prostate and the rectum, proctitis as acute and late toxicity of radiotherapy is of major concern. In order to reduce the dose to the posterior wall due to scattered radiation, to enlarge the distance between the prostate and the major parts of the rectum and to fix the prostate gland in the pelvis, several studies tested and recommended the insertion of rectal balloons inflated with air during radiotherapy [1]. These balloons are placed into the rectum before each single fraction. However, the placement of an air-filled rectal balloon directly behind the target volume raises several questions concerning the dose distribution. Comparable geometric set-ups were studied with film dosimetry in a water phantom [2]. A balloon inflated with 100 cc of air induced a dose reduction at the air-tissue interface of 60%, compared with the same geometry without the air cavity for a 15-MV photon beam and a field size of 2 cm · 2 cm. The dose beyond the interface recovered quickly and the dose reductions
due to the air cavity were 50%, 28%, 11% and 1% at 2, 5, 10 and 15 mm, respectively, from the air-tissue interface [2]. Evaluation of the dose distribution using more portal films still showed a dose reduction of about 15% at the air-tissue interface with a rapid dose build-up at 1 and 2 mm (8% and 5% reduction) [2]. In this context, we performed in vivo measurements of the doses at the anterior and the posterior rectal wall during 3D conformal radiotherapy for prostate cancer with rectal balloons inflated with 60 cc of air, and compared the results with the doses calculated by the treatment planning system (TPS). The effectiveness of in vivo dosimetry has been reported before by MOSFET [3]. In vivo dosimetry is useful and feasible for quality assurance [4–6]. We equipped rectal balloons with alanine dosimeter probes. We chose alanine/ESR dosimetry because of the good water equivalence and the small size of its probes and because of the very weak dependence of its response on the radiation quality of therapeutic photon and electron beams (see, for example, [7]). No data or electric cables
0167-8140/$ - see front matter c 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2008.03.017
D. Wagner et al. / Radiotherapy and Oncology 88 (2008) 140–147
141
were needed. Its signal stability and dose rate independence are particularly suitable for in vivo measurements. Integrated dose measurements spread over some treatment fractions are possible and increase its accuracy. Alanine/ ESR is an accurate dosimetry system from high doses (0.1–100 kGy) to radiotherapy-relevant low dose levels (1–10 Gy) [8]. The dosimeters are read out in a non-destructive manner, different from TLD. The rate of recombination of radiation-induced free radicals (fading) is low. If the probes are stored in a dry environment, the dose information is stable for more than a year. A further advantage for medical use is the non-toxicity of the alanine dosimeter probes. The feasibility of alanine dosimetry in daily clinical practice has been shown before [9].
gland and seminal vesicles and contained 1 cm margins for patient positioning errors and internal organ motion. For treatment planning, all patients received a CT scan in supine position with a rectal balloon that was equipped with metal markers for localisation purposes. On the basis of this data set, four fields were planned with beam angles of 0, 90, 180, and 270, for all three patients. Each field was treated with 20 MV X-ray beams, weighting 1:1:1:1. The dose distribution was calculated with Eclipse Version 6.5 (Varian Medical Systems, Helsinki, Finland), using the anisotropic analytical algorithm (AAA) with a grid size of 0.5 cm · 0.5 cm · 0.5 cm. The AAA is a 3D pencil beam convolution/superposition algorithm that uses separate Monte Carlo derived modelling for primary photons, scattered extra-focal photons, and electrons scattered from the beam limiting devices [10].
Materials and methods
In vivo dosimetry
Patient treatment plans Three patients with prostate cancer who received 3D conformal radiotherapy with prostate fixation by means of rectal balloons (60 cc of air) were included in this study. One patient had a hip prosthesis. All patients gave informed consent to quality assurance measurements. The prescribed dose was 60 Gy, with 2.0 Gy per fraction. The primary target volume (PTV) included the prostate
For in vivo measurements, the rectal balloons were equipped with alanine dosimeter probes (probe size 0.5 cm · 0.5 cm · 0.2 cm). Four alanine probes were placed on one side of the balloon, four on the opposite side (Fig. 1). The probes were shrink-wrapped in 0.18 mm thin polyethylene foil in order to keep them dry and to be able to affix them to the balloon. For reproducible positioning of the balloons, a special device was developed (Fig. 1). It consisted
Fig. 1. (a) Sketch of the used rectal balloon equipped with alanine dosimeters. Four alanine dosimeter probes were placed at the anterior rectal wall (1–4) and: four alanine dosimeters at the posterior rectal wall (5–8). Note the visualisation device on top of the balloon consisting of two metal markers. (b) Picture of one rectal balloon equipped with alanine dosimeters. For illustration, the positioning device was inserted. At the top of the rectal balloon, the visualisation device with metal markers was added.
142
Alanine dosimetry in prostate radiotherapy
of an inflexible rod that was inserted into the balloon prior to placement in the patient. It reduced the rotational deviation of the balloon during the positioning procedure.
Daily identical positioning control of alanine dosimeter probes in the patient The positioning of the alanine dosimeter probes in the patient was verified daily using two verification images, one ventral–dorsal and one lateral. Alanine probes are nearly soft-tissue-equivalent which makes them invisible on digital radiographs. Therefore, we developed a visualisation device consisting of two metal markers. It was affixed to the top of the balloon at an adequate distance of 3 cm from the alanine probes to avoid perturbations due to scattering. With the aid of the visualisation device and the daily verification images (for the last two patients), we were able to determine the positions of the alanine dosimeter probes in x-, y- and z-direction (Fig. 2).
Dose measurements by means of alanine/ESR dosimetry During 10 irradiations, the delivered dose was accumulated in the dosimeter probes. One specially trained physician positioned the rectal balloons. Patients were irradiated at a Clinac 2300C/D equipped with an on-board
imager (Varian Medical Systems, Palo Alto, CA, USA). The patients’ rectal temperature was assumed to be 37 C ± 0.5 C. Knowledge of the temperature and its uncertainty is required in order to correct the irradiation temperature dependence of the alanine dosimeter response. The time required for positioning the patient and setting up the treatment machine was sufficient for the adjustment of the alanine probe temperature from room temperature to patient temperature. After patient positioning, two verification images were taken as described above. After delivery of all treatment fields, the rectal balloons were stored outside the treatment room in a box which also contained ‘‘controls’’ and reference probes (see next section). After 10 irradiations, the alanine probes were sent to the Physikalisch-Technische Bundesanstalt Braunschweig, Germany (PTB). This institution performs high accuracy measurements with the ESR method. This method has been described in detail elsewhere [1,2].
Reference alanine dosimeter probe irradiation In order to detect possible accidental irradiation of the probes, the box which contained the balloon-mounted probes also contained a set of unirradiated probes. In addition, three sets of probes intended for reference irradiations were also provided. Each set contained four alanine
Fig. 2. DRRs (left side) and verification images (right side) of one patient, beam angle (a) 0 and (b) 270. For optimal visualisation, in the DRRs the bones and structures with a high HU number were emphasised. During CT data generation, a rectal balloon equipped with metal markers instead of alanine dosimeters was inserted. On the right side, the corresponding verification image is illustrated. Because of the near softtissue equivalence of alanine dosimeter probes, only the visualisation device at the top of the balloon (red borders) is visible.
D. Wagner et al. / Radiotherapy and Oncology 88 (2008) 140–147
pellets. The reference probes were contained in a holder made of polymethylmethacrylate (PMMA) which fitted exactly in a watertight sleeve for an ionization chamber (IC, applicator for type PTW 23332). The reference probes were irradiated in a 28 cm · 28 cm · 28 cm water phantom. The probes were placed in the water phantom 10 min before the irradiation started to adjust from room temperature to water temperature. The water temperature was measured and documented in order to be used for the irradiation temperature correction of the alanine response. The irradiations were performed at the same accelerator that was used for the treatments. Using a photon beam of 20 MV X-rays, a dose of 50.0 Gy was applied. Three irradiations were carried out: the first one on the first treatment day, the second one after the fifth fraction and the third one after the 10th fraction. With these irradiated reference dosimeter probe sets, it was possible for the PhysikalischTechnische Bundesanstalt to check the applied dose under reference conditions on the treatment machine. At the same time, these probes were used to detect unexpected fading effects since they were kept under the same storage conditions as the balloon-mounted dosimeter probes.
Analysis To correct the applied dose for daily positioning inaccuracies, digital reference radiographs (DRR) were calculated from the planning CT data set. Furthermore, the position of the visualisation device was analysed in the verification images for every treatment day. By matching the DRR of the planning CT with the verification image, the x-, y- and z-shifts of the alanine probes were assessed. Thereby, the dosimeter probe positions could be reconstructed into the calculated dose distribution which resulted in an estimated uncertainty of 1.5% for the applied dose. It was impossible to discern on the digital portal images whether the dosimeter probes were shifted to the right or
143
to the left side. However, as the dose distribution was symmetric due to the planning target volume symmetry of the prostate, the 2D verification of the dosimeter probe positioning was adequate. After assessing the x-, y- and z-shifts, the isocentre was adjusted. Thereby it could be assured that the air-soft tissue interface was analyzed. Dose distributions for the adjusted isocentre were calculated for all 10 fractions. Every single dose was summed up and compared to the measured dose. Additional doses for the verification images were taken into account as well. After these corrections, the integrated doses measured by means of the alanine dosimeters were compared to the applied doses from the TPS.
Statistical considerations The uncertainties have been determined according to the guidelines given by the International Organization of Standardisation (ISO) the ‘‘Guide to the expression of uncertainty in measurement (GUM)’’ [13]. According to the classification given in the GUM, all uncertainty components tabulated in Table 1 were so-called type B uncertainties. This means, that we could not obtain repeated measurements in the patients (as required for type A uncertainties). Instead, the uncertainties were determined by assigning probability distributions with the corresponding standard uncertainties (square root of the variance) to the measured values. The parameters of these probability distributions were obtained from previous measurements or scientific judgement or experience [11,12]. As n = 3 patients were measured, we could not perform a meaningful statistic comparison of the measured values with the applied ones. However, the visual comparison of the differences combined with the in vitro data indicates systematic under- and overestimations of the dose by the TPS.
Table 1 Uncertainty budget Applied dose
Dose measured by alanine/ESR
Component for the applied dose
Relative standard uncertainty in %
Component for alanine ESR measurements
Relative standard uncertainty in %
Monitor output fluctuations of treatment machine
0.75
Primary standard Mass of the probes Irradiation temperature correction
0.20 0.09 0.36
Dose calculation of the treatment planning system
1.0
ESR amplitude at 10 Gy (20 Gy)
0.40 (0.20)
Positioning of the alanine dosimeters
1.5
Radiation quality correction
0.35
Basic data measurements affected the anterior rectal wall (8 cm tissue depth) (absolute uncertainty)
0.3 (3 mm)
Fading correction Correction for using different holders for calibration and measurement
0.31 0.29
Basic data measurements affected the posterior rectal wall (4 cm tissue depth) (absolute uncertainty)
0.25 (3 mm)
Uncertainty due to inhomogeneous environment (air/tissue interface)
0.57
Total at posterior (anterior)
3.5 (3.6)
Total at 10 Gy (20 Gy)
0.98 (0.92)
144
Alanine dosimetry in prostate radiotherapy
Results Reference alanine dosimeter probes irradiation For the first patient, the reference alanine dosimeter probes were irradiated with 25.0 Gy on the first day of the treatment. The dose determined with the secondary standard measurement system of the PTB was 25.14 Gy ± 0.11 Gy, which is 0.6% higher than the stated dose of 25.0 Gy. For the next two patients, the reference dose was increased to 50.0 Gy. For the second and third patient, the mean determined dose for the reference irradiations was 49.67 Gy ± 0.22 Gy and 49.78 Gy ± 0.22 Gy, respectively. Both values agreed very well with the stated dose of 50.0 Gy. The relative deviation between the determined dose and the stated dose was always less than 1% in reference to the national standard.
Clinical aspects The rectal balloons could be introduced with the aid of the positioning device without additional rectal discomfort for the patients. No patient refused the application of the balloons during the course of the measurements. Additional time required for the verification images before each fraction was about 30 s with a modern amorphous silicon flat panel and a highly trained team. From a clinical point of view, the insertion and realisation of these measurements was easy to perform and feasible.
Daily identical positioning control of alanine dosimeter probes in the body The matching of the DRRs and the verification image was performed with the visualisation device as described above. For all three patients, the mean shift of the alanine dosimeter probes in x-direction was 0.5 cm (range: 0.1– 1.1 cm), in y-direction was 0.1 cm (range: 0.9–0.8 cm), and in z-direction was 0.6 cm (range: 1.9–0.7 cm), respectively.
Uncertainty estimation Since the ratio of measured to applied dose was considered as the end result, the following sources contributing to the overall uncertainty of the result were identified: • • • •
alanine/ESR measurement method monitor output fluctuation dose calculation of the treatment planning system positioning of the alanine dosimeter probes
A detailed uncertainty budget for the ESR measurement method has been published elsewhere [12]. In our study, additional components of the combined uncertainty had to be taken into account due to the higher irradiation temperature of 37 C, the radiation quality and fading corrections, a correction due to the use of different holders for calibration and measurement, and an additional component for the inhomogeneity of the environment during irradiation (air/ tissue interfaces). In addition, the uncertainty of the ESR amplitude doubled with respect to previously published data, since only one probe per dose was used instead of four as in the cited publication. However, the combined relative
standard uncertainty of the dose measurement inside the patient using alanine/ESR was only 1%, including the uncertainty of the primary standard. The different contributions are listed in Table 1. As mentioned above, the positions of the alanine dosimeter probes in vivo differed from fraction to fraction. Although we tried to reproduce the exact position for each irradiation fraction, we had to take an uncertainty of 1.5% into account due to the following. An exact shift of the isocentre was not always possible because of the distance of 5 mm between the slices of the planning CT. Furthermore, the air-filled balloons changed in diameter, even though they were filled daily with 60 cc of air. Both had a direct effect on the alanine dosimeter probe position in the patient and, therefore, on the applied dose. The daily monitor output fluctuation of the treatment machine varies up to 1.5% (daily measurements with ionization chamber). For the accuracy of the dose calculation, the manufacturer specifies 1.0%. The uncertainty of the treatment machine’s basic data measurements had to be taken into consideration within 3 mm. The consideration of 3 mm contained the exact positioning of the IC during basic data measurements for the TPS before clinical operation. As a result, an uncertainty of the applied dose of 3.6% for the anterior rectal wall and of 3.5% for the posterior rectal wall was obtained, assuming a distance of 4 cm between the patient surface and the posterior alanine dosimeter probes, and a distance of 8 cm between the patient surface and the anterior alanine dosimeter probes. The components of the combined uncertainty for the applied dose are shown in Table 1. All uncertainties are stated as k = 1 (1r) standard uncertainties.
Dose calculations using the TPS For all patients, the applied dose of every fraction was acquired, taking the day-to-day variations in the position into account. For the first patient, only three verification images were available. In that case, the actual position could not be considered accurately. Therefore, it was assumed that the alanine dosimeter probes at the anterior and posterior rectal wall were located at their planned positions. For the first patient, the mean applied dose at the anterior rectal wall amounted to 19.97 Gy ± 0.72 Gy (range: 19.70–20.13 Gy), and at the posterior rectal wall to 11.45 Gy ± 0.40 Gy (range: 11.35–11.55 Gy) (Table 2). For the last two patients, verification images were available for every treatment day. Taking the day-to-day shift also into account, the mean applied dose for the second patient at the anterior rectal wall amounted to 19.23 Gy ± 0.70 Gy (range: 19.03–19.42 Gy), and at the posterior rectal wall to 14.98 Gy ± 0.52 Gy (range: 14.41–15.00 Gy), and, finally, for the third patient at the anterior rectal wall to 19.56 Gy ± 0.70 Gy (range: 19.20–19.75 Gy), and at the posterior rectal wall to 11.37 Gy ± 0.40 Gy (range: 10.92– 12.25 Gy). For all the applied dose values, an uncertainty of 1.0% and 3 mm was taken into account for the accuracy of the treatment planning system and for the accuracy of the basic beam data measurements, respectively. Also, the uncertainty of the reconstruction of the alanine probe
D. Wagner et al. / Radiotherapy and Oncology 88 (2008) 140–147
145
Table 2 Applied and measured dose for all patients for the anterior and the posterior rectal wall Patient
1
Position
1 2 3 4
Mean 2
9 10 11 12
Mean 3
Mean
17 18 19 20
Anterior rectal wall
Position
Applied dose Gy
Measured dose Gy
19.70 20.04 20.13 19.99
19.28 19.73 19.81 19.72
19.97
19.64
19.42 19.12 19.03 19.34
19.15 18.82 18.78 19.12
19.23
18.97
19.20 19.75 19.65 19.62
16.28 17.38 17.88 18.78
19.56
17.58
5 6 7 8
13 14 15 16
21 22 23 24
Posterior rectal wall Applied dose Gy
Measured dose Gy
11.39 11.35 11.50 11.55
11.88 11.85 11.95 11.98
11.45
11.92
15.86 14.65 14.41 15.00
16.37 15.14 14.92 15.49
14.98
15.48
10.92 11.10 11.22 12.25
11.60 11.86 12.14 13.15
11.37
12.19
positions during every fraction (isocentre shift) contributed a further component of 1.5%, as explained above.
Dose measurements with alanine/ESR dosimetry The integrated measured doses for all 10 fractions per patient are listed in Table 2. An integrated dose between 11 and 20 Gy for the alanine dosimeters was obtained. First estimations showed that the dosimeter probes at the anterior rectal wall received approximately 100% of the delivered dose, whereas the dosimeter probes located at the posterior rectal wall received about 60% of the delivered dose. For the first patient, the measured dose at the anterior rectal wall amounted to 19.64 Gy ± 0.18 Gy (range: 19.28– 19.81 Gy), and at the posterior rectal wall to 11.92 Gy ± 0.12 Gy (range: 11.85–11.98 Gy). For the second patient, we measured a dose at the anterior rectal wall of 18.97 Gy ± 0.17 Gy (range: 18.78–19.15 Gy), and at the posterior rectal wall of 15.48 Gy ± 0.15 Gy (range: 14.92– 16.37 Gy). For the third patient, a dose at the anterior rectal wall of 17.58 Gy ± 0.16 Gy (range: 16.28–18.78 Gy), and at the posterior rectal wall of 12.19 Gy ± 0.12 Gy (range: 11.60–13.15 Gy) was obtained, respectively.
Comparison of measured dose and applied dose The relative deviation between applied dose and alanine measurement result is shown in Fig. 3. For the first patient, the mean relative difference at the anterior rectal wall was 1.68% (range: 2.18% to 1.37%), for the second patient 1.37% (range: 1.41% to 1.15%) and for the third patient with the hip prosthesis 11.23% (range: 17.94% to 4.47%). For the first patient, the mean relative difference for the posterior rectal wall was 3.92% (range: 3.59–4.22%), for the second 3.23% (range: 3.12–3.43%), and for the third patient 6.69% (range: 5.86–7.58%). The result of this com-
Fig. 3. Ratio of measured to applied dose for all three patients. From left to right: patient 1 (blue symbols), patient 2 (magenta), patient 3 (red). The values for the anterior wall are indicated by the filled circles, whereas the values for the posterior wall are represented by the triangles. Error bars represent the standard uncertainty of the ratio.
parison is presented in Fig. 3, where the ratio of the measured to the calculated dose is shown as a function of the probe/measurement number (coverage factor k = 1). There appears to be no significant difference between measured and applied dose for the anterior wall of patients one and two (Fig. 3). For the posterior rectal wall, the underestimation of the dose by the treatment planning system may already be considered as significant (Fig. 3). For patient three, the deviations are obvious and clearly outside
146
Alanine dosimetry in prostate radiotherapy
the limits of uncertainty. The dose at the anterior rectal wall is systematically overestimated and the dose at the posterior rectal wall is systematically underestimated by the TPS (Fig. 3).
Discussion The reference alanine dosimeter probe irradiation turned out to be good absolute dose verification of our basic data contained in TPS and of the stability of our treatment machine output. Under reference conditions, the relative deviation between our stated dose for a 20 MV-X beam and the dose determined by the PTB was less than 0.9%. The uncertainties included (1) uncertainty of the alanine/ESR measurement method; (2) monitor output fluctuation of the treatment machine; (3) uncertainty of the dose calculation by the treatment planning system; and (4) daily alanine dosimeter probe positioning in the patient’s body. The estimation of the uncertainty of the applied dose yielded 3.6% for the anterior rectal wall and 3.5% for the posterior rectal wall (Table 1). For the ESR measurement technique, an uncertainty of 0.92% for the anterior rectal wall and 0.98% for the posterior rectal wall were stated (Table 1). The main sources of differences between the measured and applied doses were attributed to the uncertainty of the detector position inside the patient’s body. Furthermore, changes in the detector position during treatment application could not be controlled. The main obstacle was to reconstruct the proper measurement point of the alanine dosimeter probes to the corresponding CT slice in the TPS. The effect of the uncertainty in the determination of the detector position inside the patient’s body and uncontrolled changes in the detector position during the treatment was also described from Schultka et al. They detected a relative deviation between planned and calculated doses ranging from 23 to 14% [14]. If a phantom is used and, therefore, a fixed measurement setup a relative deviation within 5% can be achieved [15]. The measured dose values were, on the average, 1.5% lower than the applied dose calculated by the TPS for the anterior rectal wall and about 3.5% higher than the dose calculated for the posterior rectal wall. This statement holds for the two patients without a hip prosthesis. The overestimation of the dose at the anterior rectal wall had been predicted by a phantom study [2]. As cited above, the dose build-up is significantly lower in air than in water-equivalent soft tissue. As we measured directly at the air-tissue interface, we could support and reproduce these results in vivo. However, it is doubtful whether this dose overestimation of the TPS has any clinical relevance. An approximately 1.5% lower dose at the anterior rectal mucosa accounts for about 1 Gy, when treating a patient above 60 Gy. Furthermore, in the water phantom a rapid dose build-up was demonstrated in 1 and 2 mm water depths [2]. The underestimation of the doses at the posterior rectal wall may be explained by the larger range of scattered photons and electrons in air than in soft tissue. In order to draw more profound conclusions, Monte Carlo simulations might be helpful. Osteras et al. have performed Monte Carlo simula-
tions and compared these simulations to alanine/ESR measurements and treatment planning system calculation [16]. A chance observation was the determination of the accuracy of our TPS concerning the calculation of the dose distribution in a patient with a hip prosthesis that was made of steel. The effective atomic number of water is approximately 7; the one of steel is about 26. Not only the inaccuracy of the dose calculation due to the high atomic number and therefore the high dose absorption, but also the artefacts that already occurred during CT data generation significantly deteriorated the quality of the dose calculation results. An absolute difference between Monte Carlo Simulations, ionization chamber measurements, and alanine film measurements in the order of 1–15% in inhomogeneous phantoms are published [16]. Due to the metal hip implant and the perturbed treatment field it cannot be kept apart if the over- and under-estimation is an effect of the alanine/ESR measurements or the wrong calculated dose distribution of TPS. Both effects should be taken into account. For the posterior rectal wall, the measured dose was underestimated by 6% to 7%. The measured dose for the anterior rectal wall was dramatically overestimated by 4–18%. This is of major clinical concern, especially as this might also result in an alteration of the dose distribution in the PTV. Such observations have been published before [17].
Conclusion We have developed a device for in vivo measurements of the rectal dose in radiotherapy for prostate cancer. It was possible to insert this device without clinical complications and without additional rectal discomfort for the patients. For irradiations of alanine dosimeter probes under reference conditions, deviations of less than 1% in reference to the national standard were achieved. In the absence of metallic implants, the relative deviations between measured and applied dose values at the anterior rectal wall are less than or equal to 1.5% for the in vivo measurements. At the posterior rectal wall, relative deviations of up to 3.5% may occur. The dominant contribution to the overall uncertainty for the in vivo measurements was the positioning of the dosimeter probes in the patient’s body and their corresponding localisation in the CT data. Therefore it is expected that improving the probe positioning in the patient’s body – e.g. by an increased visibility in the radiographic images – will lead to more accurate results. The method presented in this study turned out to be useful for in vivo quality control of the irradiations. The relative deviation between the dose determined by the ESR measurements and the planned dose determined by the TPS was shown to be within the 5% limit recommended by the ICRU for doses above 0.7 Gy.
Acknowledgements We thank B. Thu ¨ne, M. Lyssy, T. Hackel, D.M. Boche and C. Eckerleben for their technical assistance and H.-J. Selbach for a Monte Carlo simulation of the alanine response at an air/tissue interface.
D. Wagner et al. / Radiotherapy and Oncology 88 (2008) 140–147 * Corresponding author. Daniela Wagner, Department of Radiotherapy and Radiooncology, University Hospital Goettingen, RobertKoch-Str. 40, 37075 Goettingen, Germany. E-mail address: [email protected] Received 4 January 2008; received in revised form 7 March 2008; accepted 16 March 2008; Available online 15 April 2008
References [1] Hille A, Schmidberger H, To ¨ws N, Weiss E, Vorwerk H, Hess CF. The impact of varying volumes in rectal balloons on rectal dose sparing in conformal radiation therapy of prostate cancer. A prospective three-dimensional analysis. Strahlenther Onkol 2005;181:709–16. [2] Teh BS, Dong L, McGary JE, Mai WY, Grant 3rd W, Butler EB. Rectal wall sparing by dosimetric effect of rectal balloon used during intensity-modulated radiation therapy (IMRT) for prostate cancer. Med Dosim 2005;30:25–30. [3] Ciocca M, Piazza V, Lazzari R, Vavassori A, Luini A, Veronesi P, et al. Real-time in vivo dosimetry using micro MOSFET detectors during intraoperative electron beam radiation therapy in early-stage breast cancer. Radiother Oncol 2006;78:213–6. [4] Duch MA, Ginjaume M, Chakkor H, Ortega X, Jomet N, Ribas M. Thermoluminescence dosimetry applied to in vivo dose measurements for total body irradiation techniques. Radiother Oncol 1998;47:319–24. [5] Lanson JH, Essers M, Meijer GJ, Minken AW, Uiterwaal GJ, Mijnheer BJ. In vivo dosimetry during conformal radiotherapy: requirements for and findings of a routine procedure. Radiother Oncol 1999;52:51–9. [6] Cozzi L, Fogliata-Cozzi A. Quality assurance in radiation oncology. A study of feasibility and impact on action level of an in vivo dosimetry program, during breast cancer irradiation. Radiother Oncol 1998;47:29–36.
147
[7] Bergstrand ES, Shortt KR, Ross CK, Hole EO. An investigation of the photon energy dependence of the EPR alanine dosimetry system. Phys Med Biol 2003;48:1753–71. [8] Sharpe PHG, Rajendran K, Sephton JP. Progress towards an alanine/ESR therapy level reference dosimetry service at NPL. Appl Radiat Isot 1996;47:1171–5. [9] Ciesielski B, Schultka K, Kobierska A, Nowak R, Peimel-Stuglik Z. In vivo alanine/EPR dosimetry in daily clinical practice: a feasibility study. Int J Radiat Oncol Biol Phys 2003;56:899–905. [10] Ulmer W, Harder D. A triple Gaussian pencil beam model for photon beam treatment planning. Med Phys 1995;5:25–30. [11] Anton M. Uncertainties in alanine/ESR dosimetry at the Physikalisch-Technische Bundesanstalt. Phys Med Biol 2006;51:5419–40. [12] Anton M. Development of a secondary standard for the absorbed dose to water based on the alanine EPR dosimetry system. Appl Radiat Isot 2005;62:779–95. [13] ISO 1993 Guide to the Expression of Uncertainty in Measurement (GUM). 1st ed., 1993, corrected and reprinted 1995, International Organization for Standardization (ISO), Geneva, 1993. [14] Schultka K, Ciesielki B, Serkies K, Sawicki T, Tamawska Z, Jassem J. EPR/alanine dosimetry in LDR brachytherapy – a feasibility study. Radiat Prot Dosimetry 2006;120:171–5. [15] Olsson S, Bergstrand ES, Carlsson AK, Hole EO, Lund E. Radiation dose measurements with alanine/agarose gel and thin alanine films around a 192Ir brachytherapy source, using ESR spectroscopy. Phys Med Biol 2002;47:1333–56. [16] Osteras BH, Hole EO, Olsen DR, Malinen E. EPR dosimetry of radiotherapy photon beams in inhomogeneous media using alanine films. Phys Med Biol 2006;51:6215–328. [17] Ding GX, Yu CW. A study on beams passing through hip prosthesis for pelvic radiation treatment. Int J Radiat Oncol Biol Phys 2001;51:1167–75.
Dosimetry
In vivo alanine/electron spin resonance (ESR) dosimetry in radiotherapy of prostate cancer: A feasibility study Daniela Wagnera,*, Mathias Antonb, Hilke Vorwerka, Tammo Gsa ¨ngera, Hans Christiansena, Bjoern Poppec, Clemens Friedrich Hessa, Robert Michael Hermanna a
Department of Radiotherapy and Radiooncology, University Hospital Goettingen, Germany, bPhysikalisch-Technische Bundesanstalt, Braunschweig, Germany, cDepartment of Radiooncology, Pius Hospital Oldenburg, Germany
Abstract Purpose: We have developed a device to evaluate the potential of alanine/electron spin resonance (ESR) dosimetry for quality assurance in 3D conformal radiotherapy for prostate cancer. It consists of a rectal balloon carrying eight alanine dosimeter probes and two metal markers to document the exact position of the balloon. We measured the effects of an air-filled rectal balloon on the dose at the rectal wall and compared these results with the applied dose distribution of the treatment planning system. Materials and methods: During 10 fractions with 2.0 Gy per fraction, the accumulated doses were measured in 3 patients. The results of the ESR measurements were compared to the applied doses. Results: It was possible to insert the device without clinical complications and without additional rectal discomfort for the patients. The measurements of the dose accumulated at the anterior and the posterior rectal wall agreed with the applied dose within a mean deviation of 1.5% (overestimation of the dose) and 3.5% (underestimation of the dose), respectively. However, clinically significant differences between applied and measured rectal doses were seen in a patient with a hip prosthesis. In this case, the dose at the anterior rectal wall was overestimated by the TPS by about 11% and the dose at the posterior rectal wall was underestimated by approximately 7%. Conclusion: The method presented in this study is useful for quality control of irradiations in vivo. c 2008 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 88 (2008) 140–147.
Keywords: Radiation therapy; Alanine; Electron spin resonance; In vivo; Dosimetry; Prostate cancer
3D conformal radiotherapy for prostate cancer is a curative treatment option. Due to the close anatomical relation between the prostate and the rectum, proctitis as acute and late toxicity of radiotherapy is of major concern. In order to reduce the dose to the posterior wall due to scattered radiation, to enlarge the distance between the prostate and the major parts of the rectum and to fix the prostate gland in the pelvis, several studies tested and recommended the insertion of rectal balloons inflated with air during radiotherapy [1]. These balloons are placed into the rectum before each single fraction. However, the placement of an air-filled rectal balloon directly behind the target volume raises several questions concerning the dose distribution. Comparable geometric set-ups were studied with film dosimetry in a water phantom [2]. A balloon inflated with 100 cc of air induced a dose reduction at the air-tissue interface of 60%, compared with the same geometry without the air cavity for a 15-MV photon beam and a field size of 2 cm · 2 cm. The dose beyond the interface recovered quickly and the dose reductions
due to the air cavity were 50%, 28%, 11% and 1% at 2, 5, 10 and 15 mm, respectively, from the air-tissue interface [2]. Evaluation of the dose distribution using more portal films still showed a dose reduction of about 15% at the air-tissue interface with a rapid dose build-up at 1 and 2 mm (8% and 5% reduction) [2]. In this context, we performed in vivo measurements of the doses at the anterior and the posterior rectal wall during 3D conformal radiotherapy for prostate cancer with rectal balloons inflated with 60 cc of air, and compared the results with the doses calculated by the treatment planning system (TPS). The effectiveness of in vivo dosimetry has been reported before by MOSFET [3]. In vivo dosimetry is useful and feasible for quality assurance [4–6]. We equipped rectal balloons with alanine dosimeter probes. We chose alanine/ESR dosimetry because of the good water equivalence and the small size of its probes and because of the very weak dependence of its response on the radiation quality of therapeutic photon and electron beams (see, for example, [7]). No data or electric cables
0167-8140/$ - see front matter c 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.radonc.2008.03.017
D. Wagner et al. / Radiotherapy and Oncology 88 (2008) 140–147
141
were needed. Its signal stability and dose rate independence are particularly suitable for in vivo measurements. Integrated dose measurements spread over some treatment fractions are possible and increase its accuracy. Alanine/ ESR is an accurate dosimetry system from high doses (0.1–100 kGy) to radiotherapy-relevant low dose levels (1–10 Gy) [8]. The dosimeters are read out in a non-destructive manner, different from TLD. The rate of recombination of radiation-induced free radicals (fading) is low. If the probes are stored in a dry environment, the dose information is stable for more than a year. A further advantage for medical use is the non-toxicity of the alanine dosimeter probes. The feasibility of alanine dosimetry in daily clinical practice has been shown before [9].
gland and seminal vesicles and contained 1 cm margins for patient positioning errors and internal organ motion. For treatment planning, all patients received a CT scan in supine position with a rectal balloon that was equipped with metal markers for localisation purposes. On the basis of this data set, four fields were planned with beam angles of 0, 90, 180, and 270, for all three patients. Each field was treated with 20 MV X-ray beams, weighting 1:1:1:1. The dose distribution was calculated with Eclipse Version 6.5 (Varian Medical Systems, Helsinki, Finland), using the anisotropic analytical algorithm (AAA) with a grid size of 0.5 cm · 0.5 cm · 0.5 cm. The AAA is a 3D pencil beam convolution/superposition algorithm that uses separate Monte Carlo derived modelling for primary photons, scattered extra-focal photons, and electrons scattered from the beam limiting devices [10].
Materials and methods
In vivo dosimetry
Patient treatment plans Three patients with prostate cancer who received 3D conformal radiotherapy with prostate fixation by means of rectal balloons (60 cc of air) were included in this study. One patient had a hip prosthesis. All patients gave informed consent to quality assurance measurements. The prescribed dose was 60 Gy, with 2.0 Gy per fraction. The primary target volume (PTV) included the prostate
For in vivo measurements, the rectal balloons were equipped with alanine dosimeter probes (probe size 0.5 cm · 0.5 cm · 0.2 cm). Four alanine probes were placed on one side of the balloon, four on the opposite side (Fig. 1). The probes were shrink-wrapped in 0.18 mm thin polyethylene foil in order to keep them dry and to be able to affix them to the balloon. For reproducible positioning of the balloons, a special device was developed (Fig. 1). It consisted
Fig. 1. (a) Sketch of the used rectal balloon equipped with alanine dosimeters. Four alanine dosimeter probes were placed at the anterior rectal wall (1–4) and: four alanine dosimeters at the posterior rectal wall (5–8). Note the visualisation device on top of the balloon consisting of two metal markers. (b) Picture of one rectal balloon equipped with alanine dosimeters. For illustration, the positioning device was inserted. At the top of the rectal balloon, the visualisation device with metal markers was added.
142
Alanine dosimetry in prostate radiotherapy
of an inflexible rod that was inserted into the balloon prior to placement in the patient. It reduced the rotational deviation of the balloon during the positioning procedure.
Daily identical positioning control of alanine dosimeter probes in the patient The positioning of the alanine dosimeter probes in the patient was verified daily using two verification images, one ventral–dorsal and one lateral. Alanine probes are nearly soft-tissue-equivalent which makes them invisible on digital radiographs. Therefore, we developed a visualisation device consisting of two metal markers. It was affixed to the top of the balloon at an adequate distance of 3 cm from the alanine probes to avoid perturbations due to scattering. With the aid of the visualisation device and the daily verification images (for the last two patients), we were able to determine the positions of the alanine dosimeter probes in x-, y- and z-direction (Fig. 2).
Dose measurements by means of alanine/ESR dosimetry During 10 irradiations, the delivered dose was accumulated in the dosimeter probes. One specially trained physician positioned the rectal balloons. Patients were irradiated at a Clinac 2300C/D equipped with an on-board
imager (Varian Medical Systems, Palo Alto, CA, USA). The patients’ rectal temperature was assumed to be 37 C ± 0.5 C. Knowledge of the temperature and its uncertainty is required in order to correct the irradiation temperature dependence of the alanine dosimeter response. The time required for positioning the patient and setting up the treatment machine was sufficient for the adjustment of the alanine probe temperature from room temperature to patient temperature. After patient positioning, two verification images were taken as described above. After delivery of all treatment fields, the rectal balloons were stored outside the treatment room in a box which also contained ‘‘controls’’ and reference probes (see next section). After 10 irradiations, the alanine probes were sent to the Physikalisch-Technische Bundesanstalt Braunschweig, Germany (PTB). This institution performs high accuracy measurements with the ESR method. This method has been described in detail elsewhere [1,2].
Reference alanine dosimeter probe irradiation In order to detect possible accidental irradiation of the probes, the box which contained the balloon-mounted probes also contained a set of unirradiated probes. In addition, three sets of probes intended for reference irradiations were also provided. Each set contained four alanine
Fig. 2. DRRs (left side) and verification images (right side) of one patient, beam angle (a) 0 and (b) 270. For optimal visualisation, in the DRRs the bones and structures with a high HU number were emphasised. During CT data generation, a rectal balloon equipped with metal markers instead of alanine dosimeters was inserted. On the right side, the corresponding verification image is illustrated. Because of the near softtissue equivalence of alanine dosimeter probes, only the visualisation device at the top of the balloon (red borders) is visible.
D. Wagner et al. / Radiotherapy and Oncology 88 (2008) 140–147
pellets. The reference probes were contained in a holder made of polymethylmethacrylate (PMMA) which fitted exactly in a watertight sleeve for an ionization chamber (IC, applicator for type PTW 23332). The reference probes were irradiated in a 28 cm · 28 cm · 28 cm water phantom. The probes were placed in the water phantom 10 min before the irradiation started to adjust from room temperature to water temperature. The water temperature was measured and documented in order to be used for the irradiation temperature correction of the alanine response. The irradiations were performed at the same accelerator that was used for the treatments. Using a photon beam of 20 MV X-rays, a dose of 50.0 Gy was applied. Three irradiations were carried out: the first one on the first treatment day, the second one after the fifth fraction and the third one after the 10th fraction. With these irradiated reference dosimeter probe sets, it was possible for the PhysikalischTechnische Bundesanstalt to check the applied dose under reference conditions on the treatment machine. At the same time, these probes were used to detect unexpected fading effects since they were kept under the same storage conditions as the balloon-mounted dosimeter probes.
Analysis To correct the applied dose for daily positioning inaccuracies, digital reference radiographs (DRR) were calculated from the planning CT data set. Furthermore, the position of the visualisation device was analysed in the verification images for every treatment day. By matching the DRR of the planning CT with the verification image, the x-, y- and z-shifts of the alanine probes were assessed. Thereby, the dosimeter probe positions could be reconstructed into the calculated dose distribution which resulted in an estimated uncertainty of 1.5% for the applied dose. It was impossible to discern on the digital portal images whether the dosimeter probes were shifted to the right or
143
to the left side. However, as the dose distribution was symmetric due to the planning target volume symmetry of the prostate, the 2D verification of the dosimeter probe positioning was adequate. After assessing the x-, y- and z-shifts, the isocentre was adjusted. Thereby it could be assured that the air-soft tissue interface was analyzed. Dose distributions for the adjusted isocentre were calculated for all 10 fractions. Every single dose was summed up and compared to the measured dose. Additional doses for the verification images were taken into account as well. After these corrections, the integrated doses measured by means of the alanine dosimeters were compared to the applied doses from the TPS.
Statistical considerations The uncertainties have been determined according to the guidelines given by the International Organization of Standardisation (ISO) the ‘‘Guide to the expression of uncertainty in measurement (GUM)’’ [13]. According to the classification given in the GUM, all uncertainty components tabulated in Table 1 were so-called type B uncertainties. This means, that we could not obtain repeated measurements in the patients (as required for type A uncertainties). Instead, the uncertainties were determined by assigning probability distributions with the corresponding standard uncertainties (square root of the variance) to the measured values. The parameters of these probability distributions were obtained from previous measurements or scientific judgement or experience [11,12]. As n = 3 patients were measured, we could not perform a meaningful statistic comparison of the measured values with the applied ones. However, the visual comparison of the differences combined with the in vitro data indicates systematic under- and overestimations of the dose by the TPS.
Table 1 Uncertainty budget Applied dose
Dose measured by alanine/ESR
Component for the applied dose
Relative standard uncertainty in %
Component for alanine ESR measurements
Relative standard uncertainty in %
Monitor output fluctuations of treatment machine
0.75
Primary standard Mass of the probes Irradiation temperature correction
0.20 0.09 0.36
Dose calculation of the treatment planning system
1.0
ESR amplitude at 10 Gy (20 Gy)
0.40 (0.20)
Positioning of the alanine dosimeters
1.5
Radiation quality correction
0.35
Basic data measurements affected the anterior rectal wall (8 cm tissue depth) (absolute uncertainty)
0.3 (3 mm)
Fading correction Correction for using different holders for calibration and measurement
0.31 0.29
Basic data measurements affected the posterior rectal wall (4 cm tissue depth) (absolute uncertainty)
0.25 (3 mm)
Uncertainty due to inhomogeneous environment (air/tissue interface)
0.57
Total at posterior (anterior)
3.5 (3.6)
Total at 10 Gy (20 Gy)
0.98 (0.92)
144
Alanine dosimetry in prostate radiotherapy
Results Reference alanine dosimeter probes irradiation For the first patient, the reference alanine dosimeter probes were irradiated with 25.0 Gy on the first day of the treatment. The dose determined with the secondary standard measurement system of the PTB was 25.14 Gy ± 0.11 Gy, which is 0.6% higher than the stated dose of 25.0 Gy. For the next two patients, the reference dose was increased to 50.0 Gy. For the second and third patient, the mean determined dose for the reference irradiations was 49.67 Gy ± 0.22 Gy and 49.78 Gy ± 0.22 Gy, respectively. Both values agreed very well with the stated dose of 50.0 Gy. The relative deviation between the determined dose and the stated dose was always less than 1% in reference to the national standard.
Clinical aspects The rectal balloons could be introduced with the aid of the positioning device without additional rectal discomfort for the patients. No patient refused the application of the balloons during the course of the measurements. Additional time required for the verification images before each fraction was about 30 s with a modern amorphous silicon flat panel and a highly trained team. From a clinical point of view, the insertion and realisation of these measurements was easy to perform and feasible.
Daily identical positioning control of alanine dosimeter probes in the body The matching of the DRRs and the verification image was performed with the visualisation device as described above. For all three patients, the mean shift of the alanine dosimeter probes in x-direction was 0.5 cm (range: 0.1– 1.1 cm), in y-direction was 0.1 cm (range: 0.9–0.8 cm), and in z-direction was 0.6 cm (range: 1.9–0.7 cm), respectively.
Uncertainty estimation Since the ratio of measured to applied dose was considered as the end result, the following sources contributing to the overall uncertainty of the result were identified: • • • •
alanine/ESR measurement method monitor output fluctuation dose calculation of the treatment planning system positioning of the alanine dosimeter probes
A detailed uncertainty budget for the ESR measurement method has been published elsewhere [12]. In our study, additional components of the combined uncertainty had to be taken into account due to the higher irradiation temperature of 37 C, the radiation quality and fading corrections, a correction due to the use of different holders for calibration and measurement, and an additional component for the inhomogeneity of the environment during irradiation (air/ tissue interfaces). In addition, the uncertainty of the ESR amplitude doubled with respect to previously published data, since only one probe per dose was used instead of four as in the cited publication. However, the combined relative
standard uncertainty of the dose measurement inside the patient using alanine/ESR was only 1%, including the uncertainty of the primary standard. The different contributions are listed in Table 1. As mentioned above, the positions of the alanine dosimeter probes in vivo differed from fraction to fraction. Although we tried to reproduce the exact position for each irradiation fraction, we had to take an uncertainty of 1.5% into account due to the following. An exact shift of the isocentre was not always possible because of the distance of 5 mm between the slices of the planning CT. Furthermore, the air-filled balloons changed in diameter, even though they were filled daily with 60 cc of air. Both had a direct effect on the alanine dosimeter probe position in the patient and, therefore, on the applied dose. The daily monitor output fluctuation of the treatment machine varies up to 1.5% (daily measurements with ionization chamber). For the accuracy of the dose calculation, the manufacturer specifies 1.0%. The uncertainty of the treatment machine’s basic data measurements had to be taken into consideration within 3 mm. The consideration of 3 mm contained the exact positioning of the IC during basic data measurements for the TPS before clinical operation. As a result, an uncertainty of the applied dose of 3.6% for the anterior rectal wall and of 3.5% for the posterior rectal wall was obtained, assuming a distance of 4 cm between the patient surface and the posterior alanine dosimeter probes, and a distance of 8 cm between the patient surface and the anterior alanine dosimeter probes. The components of the combined uncertainty for the applied dose are shown in Table 1. All uncertainties are stated as k = 1 (1r) standard uncertainties.
Dose calculations using the TPS For all patients, the applied dose of every fraction was acquired, taking the day-to-day variations in the position into account. For the first patient, only three verification images were available. In that case, the actual position could not be considered accurately. Therefore, it was assumed that the alanine dosimeter probes at the anterior and posterior rectal wall were located at their planned positions. For the first patient, the mean applied dose at the anterior rectal wall amounted to 19.97 Gy ± 0.72 Gy (range: 19.70–20.13 Gy), and at the posterior rectal wall to 11.45 Gy ± 0.40 Gy (range: 11.35–11.55 Gy) (Table 2). For the last two patients, verification images were available for every treatment day. Taking the day-to-day shift also into account, the mean applied dose for the second patient at the anterior rectal wall amounted to 19.23 Gy ± 0.70 Gy (range: 19.03–19.42 Gy), and at the posterior rectal wall to 14.98 Gy ± 0.52 Gy (range: 14.41–15.00 Gy), and, finally, for the third patient at the anterior rectal wall to 19.56 Gy ± 0.70 Gy (range: 19.20–19.75 Gy), and at the posterior rectal wall to 11.37 Gy ± 0.40 Gy (range: 10.92– 12.25 Gy). For all the applied dose values, an uncertainty of 1.0% and 3 mm was taken into account for the accuracy of the treatment planning system and for the accuracy of the basic beam data measurements, respectively. Also, the uncertainty of the reconstruction of the alanine probe
D. Wagner et al. / Radiotherapy and Oncology 88 (2008) 140–147
145
Table 2 Applied and measured dose for all patients for the anterior and the posterior rectal wall Patient
1
Position
1 2 3 4
Mean 2
9 10 11 12
Mean 3
Mean
17 18 19 20
Anterior rectal wall
Position
Applied dose Gy
Measured dose Gy
19.70 20.04 20.13 19.99
19.28 19.73 19.81 19.72
19.97
19.64
19.42 19.12 19.03 19.34
19.15 18.82 18.78 19.12
19.23
18.97
19.20 19.75 19.65 19.62
16.28 17.38 17.88 18.78
19.56
17.58
5 6 7 8
13 14 15 16
21 22 23 24
Posterior rectal wall Applied dose Gy
Measured dose Gy
11.39 11.35 11.50 11.55
11.88 11.85 11.95 11.98
11.45
11.92
15.86 14.65 14.41 15.00
16.37 15.14 14.92 15.49
14.98
15.48
10.92 11.10 11.22 12.25
11.60 11.86 12.14 13.15
11.37
12.19
positions during every fraction (isocentre shift) contributed a further component of 1.5%, as explained above.
Dose measurements with alanine/ESR dosimetry The integrated measured doses for all 10 fractions per patient are listed in Table 2. An integrated dose between 11 and 20 Gy for the alanine dosimeters was obtained. First estimations showed that the dosimeter probes at the anterior rectal wall received approximately 100% of the delivered dose, whereas the dosimeter probes located at the posterior rectal wall received about 60% of the delivered dose. For the first patient, the measured dose at the anterior rectal wall amounted to 19.64 Gy ± 0.18 Gy (range: 19.28– 19.81 Gy), and at the posterior rectal wall to 11.92 Gy ± 0.12 Gy (range: 11.85–11.98 Gy). For the second patient, we measured a dose at the anterior rectal wall of 18.97 Gy ± 0.17 Gy (range: 18.78–19.15 Gy), and at the posterior rectal wall of 15.48 Gy ± 0.15 Gy (range: 14.92– 16.37 Gy). For the third patient, a dose at the anterior rectal wall of 17.58 Gy ± 0.16 Gy (range: 16.28–18.78 Gy), and at the posterior rectal wall of 12.19 Gy ± 0.12 Gy (range: 11.60–13.15 Gy) was obtained, respectively.
Comparison of measured dose and applied dose The relative deviation between applied dose and alanine measurement result is shown in Fig. 3. For the first patient, the mean relative difference at the anterior rectal wall was 1.68% (range: 2.18% to 1.37%), for the second patient 1.37% (range: 1.41% to 1.15%) and for the third patient with the hip prosthesis 11.23% (range: 17.94% to 4.47%). For the first patient, the mean relative difference for the posterior rectal wall was 3.92% (range: 3.59–4.22%), for the second 3.23% (range: 3.12–3.43%), and for the third patient 6.69% (range: 5.86–7.58%). The result of this com-
Fig. 3. Ratio of measured to applied dose for all three patients. From left to right: patient 1 (blue symbols), patient 2 (magenta), patient 3 (red). The values for the anterior wall are indicated by the filled circles, whereas the values for the posterior wall are represented by the triangles. Error bars represent the standard uncertainty of the ratio.
parison is presented in Fig. 3, where the ratio of the measured to the calculated dose is shown as a function of the probe/measurement number (coverage factor k = 1). There appears to be no significant difference between measured and applied dose for the anterior wall of patients one and two (Fig. 3). For the posterior rectal wall, the underestimation of the dose by the treatment planning system may already be considered as significant (Fig. 3). For patient three, the deviations are obvious and clearly outside
146
Alanine dosimetry in prostate radiotherapy
the limits of uncertainty. The dose at the anterior rectal wall is systematically overestimated and the dose at the posterior rectal wall is systematically underestimated by the TPS (Fig. 3).
Discussion The reference alanine dosimeter probe irradiation turned out to be good absolute dose verification of our basic data contained in TPS and of the stability of our treatment machine output. Under reference conditions, the relative deviation between our stated dose for a 20 MV-X beam and the dose determined by the PTB was less than 0.9%. The uncertainties included (1) uncertainty of the alanine/ESR measurement method; (2) monitor output fluctuation of the treatment machine; (3) uncertainty of the dose calculation by the treatment planning system; and (4) daily alanine dosimeter probe positioning in the patient’s body. The estimation of the uncertainty of the applied dose yielded 3.6% for the anterior rectal wall and 3.5% for the posterior rectal wall (Table 1). For the ESR measurement technique, an uncertainty of 0.92% for the anterior rectal wall and 0.98% for the posterior rectal wall were stated (Table 1). The main sources of differences between the measured and applied doses were attributed to the uncertainty of the detector position inside the patient’s body. Furthermore, changes in the detector position during treatment application could not be controlled. The main obstacle was to reconstruct the proper measurement point of the alanine dosimeter probes to the corresponding CT slice in the TPS. The effect of the uncertainty in the determination of the detector position inside the patient’s body and uncontrolled changes in the detector position during the treatment was also described from Schultka et al. They detected a relative deviation between planned and calculated doses ranging from 23 to 14% [14]. If a phantom is used and, therefore, a fixed measurement setup a relative deviation within 5% can be achieved [15]. The measured dose values were, on the average, 1.5% lower than the applied dose calculated by the TPS for the anterior rectal wall and about 3.5% higher than the dose calculated for the posterior rectal wall. This statement holds for the two patients without a hip prosthesis. The overestimation of the dose at the anterior rectal wall had been predicted by a phantom study [2]. As cited above, the dose build-up is significantly lower in air than in water-equivalent soft tissue. As we measured directly at the air-tissue interface, we could support and reproduce these results in vivo. However, it is doubtful whether this dose overestimation of the TPS has any clinical relevance. An approximately 1.5% lower dose at the anterior rectal mucosa accounts for about 1 Gy, when treating a patient above 60 Gy. Furthermore, in the water phantom a rapid dose build-up was demonstrated in 1 and 2 mm water depths [2]. The underestimation of the doses at the posterior rectal wall may be explained by the larger range of scattered photons and electrons in air than in soft tissue. In order to draw more profound conclusions, Monte Carlo simulations might be helpful. Osteras et al. have performed Monte Carlo simula-
tions and compared these simulations to alanine/ESR measurements and treatment planning system calculation [16]. A chance observation was the determination of the accuracy of our TPS concerning the calculation of the dose distribution in a patient with a hip prosthesis that was made of steel. The effective atomic number of water is approximately 7; the one of steel is about 26. Not only the inaccuracy of the dose calculation due to the high atomic number and therefore the high dose absorption, but also the artefacts that already occurred during CT data generation significantly deteriorated the quality of the dose calculation results. An absolute difference between Monte Carlo Simulations, ionization chamber measurements, and alanine film measurements in the order of 1–15% in inhomogeneous phantoms are published [16]. Due to the metal hip implant and the perturbed treatment field it cannot be kept apart if the over- and under-estimation is an effect of the alanine/ESR measurements or the wrong calculated dose distribution of TPS. Both effects should be taken into account. For the posterior rectal wall, the measured dose was underestimated by 6% to 7%. The measured dose for the anterior rectal wall was dramatically overestimated by 4–18%. This is of major clinical concern, especially as this might also result in an alteration of the dose distribution in the PTV. Such observations have been published before [17].
Conclusion We have developed a device for in vivo measurements of the rectal dose in radiotherapy for prostate cancer. It was possible to insert this device without clinical complications and without additional rectal discomfort for the patients. For irradiations of alanine dosimeter probes under reference conditions, deviations of less than 1% in reference to the national standard were achieved. In the absence of metallic implants, the relative deviations between measured and applied dose values at the anterior rectal wall are less than or equal to 1.5% for the in vivo measurements. At the posterior rectal wall, relative deviations of up to 3.5% may occur. The dominant contribution to the overall uncertainty for the in vivo measurements was the positioning of the dosimeter probes in the patient’s body and their corresponding localisation in the CT data. Therefore it is expected that improving the probe positioning in the patient’s body – e.g. by an increased visibility in the radiographic images – will lead to more accurate results. The method presented in this study turned out to be useful for in vivo quality control of the irradiations. The relative deviation between the dose determined by the ESR measurements and the planned dose determined by the TPS was shown to be within the 5% limit recommended by the ICRU for doses above 0.7 Gy.
Acknowledgements We thank B. Thu ¨ne, M. Lyssy, T. Hackel, D.M. Boche and C. Eckerleben for their technical assistance and H.-J. Selbach for a Monte Carlo simulation of the alanine response at an air/tissue interface.
D. Wagner et al. / Radiotherapy and Oncology 88 (2008) 140–147 * Corresponding author. Daniela Wagner, Department of Radiotherapy and Radiooncology, University Hospital Goettingen, RobertKoch-Str. 40, 37075 Goettingen, Germany. E-mail address: [email protected] Received 4 January 2008; received in revised form 7 March 2008; accepted 16 March 2008; Available online 15 April 2008
References [1] Hille A, Schmidberger H, To ¨ws N, Weiss E, Vorwerk H, Hess CF. The impact of varying volumes in rectal balloons on rectal dose sparing in conformal radiation therapy of prostate cancer. A prospective three-dimensional analysis. Strahlenther Onkol 2005;181:709–16. [2] Teh BS, Dong L, McGary JE, Mai WY, Grant 3rd W, Butler EB. Rectal wall sparing by dosimetric effect of rectal balloon used during intensity-modulated radiation therapy (IMRT) for prostate cancer. Med Dosim 2005;30:25–30. [3] Ciocca M, Piazza V, Lazzari R, Vavassori A, Luini A, Veronesi P, et al. Real-time in vivo dosimetry using micro MOSFET detectors during intraoperative electron beam radiation therapy in early-stage breast cancer. Radiother Oncol 2006;78:213–6. [4] Duch MA, Ginjaume M, Chakkor H, Ortega X, Jomet N, Ribas M. Thermoluminescence dosimetry applied to in vivo dose measurements for total body irradiation techniques. Radiother Oncol 1998;47:319–24. [5] Lanson JH, Essers M, Meijer GJ, Minken AW, Uiterwaal GJ, Mijnheer BJ. In vivo dosimetry during conformal radiotherapy: requirements for and findings of a routine procedure. Radiother Oncol 1999;52:51–9. [6] Cozzi L, Fogliata-Cozzi A. Quality assurance in radiation oncology. A study of feasibility and impact on action level of an in vivo dosimetry program, during breast cancer irradiation. Radiother Oncol 1998;47:29–36.
147
[7] Bergstrand ES, Shortt KR, Ross CK, Hole EO. An investigation of the photon energy dependence of the EPR alanine dosimetry system. Phys Med Biol 2003;48:1753–71. [8] Sharpe PHG, Rajendran K, Sephton JP. Progress towards an alanine/ESR therapy level reference dosimetry service at NPL. Appl Radiat Isot 1996;47:1171–5. [9] Ciesielski B, Schultka K, Kobierska A, Nowak R, Peimel-Stuglik Z. In vivo alanine/EPR dosimetry in daily clinical practice: a feasibility study. Int J Radiat Oncol Biol Phys 2003;56:899–905. [10] Ulmer W, Harder D. A triple Gaussian pencil beam model for photon beam treatment planning. Med Phys 1995;5:25–30. [11] Anton M. Uncertainties in alanine/ESR dosimetry at the Physikalisch-Technische Bundesanstalt. Phys Med Biol 2006;51:5419–40. [12] Anton M. Development of a secondary standard for the absorbed dose to water based on the alanine EPR dosimetry system. Appl Radiat Isot 2005;62:779–95. [13] ISO 1993 Guide to the Expression of Uncertainty in Measurement (GUM). 1st ed., 1993, corrected and reprinted 1995, International Organization for Standardization (ISO), Geneva, 1993. [14] Schultka K, Ciesielki B, Serkies K, Sawicki T, Tamawska Z, Jassem J. EPR/alanine dosimetry in LDR brachytherapy – a feasibility study. Radiat Prot Dosimetry 2006;120:171–5. [15] Olsson S, Bergstrand ES, Carlsson AK, Hole EO, Lund E. Radiation dose measurements with alanine/agarose gel and thin alanine films around a 192Ir brachytherapy source, using ESR spectroscopy. Phys Med Biol 2002;47:1333–56. [16] Osteras BH, Hole EO, Olsen DR, Malinen E. EPR dosimetry of radiotherapy photon beams in inhomogeneous media using alanine films. Phys Med Biol 2006;51:6215–328. [17] Ding GX, Yu CW. A study on beams passing through hip prosthesis for pelvic radiation treatment. Int J Radiat Oncol Biol Phys 2001;51:1167–75.