- Email: [email protected]
Clinical implementation and range evaluation of in vivo PET dosimetry for particle irradiation in patients with primary glioma
Radiotherapy and Oncology 115 (2015) 179–185
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
in vivo PET dosimetry
Clinical implementation and range evaluation of in vivo PET dosimetry for particle irradiation in patients with primary glioma Sebastian P. Nischwitz a,⇑,1, Julia Bauer b,1, Thomas Welzel a, Harald Rief a, Oliver Jäkel b,c, Thomas Haberer b, Kathrin Frey d, Jürgen Debus a, Katia Parodi a,d, Stephanie E. Combs e,2, Stefan Rieken a,2 a University Hospital of Heidelberg, Department of Radiation Oncology; b Heidelberg Ion-Beam Therapy Centre; c German Cancer Research Center (DKFZ), Department of Medical Physics in Radiation Oncology, Heidelberg; d Ludwig-Maximilians-University of Munich, Department of Medical Physics, Garching; and e University Hospital Klinikum rechts der Isar, Department of Radiation Oncology, Munich, Germany
a r t i c l e
i n f o
Article history: Received 16 September 2014 Received in revised form 18 March 2015 Accepted 21 March 2015 Available online 2 April 2015 Keywords: PET In vivo dosimetry Ion therapy Monte Carlo simulation Glioblastoma Range verification
a b s t r a c t Purpose: The physical and biological properties of ion-beams offer various advantages in comparison to conventional radiotherapy, though uncertainties concerning quality assurance are still left. Due to the inverted depth dose profile, range accuracy is of paramount importance. We investigated the range deviations between planning simulation and post-fractional PET/CT measurement from particle therapy in primary glioblastoma. Methods and materials: 20 patients with glioblastoma undergoing particle therapy at our institution were selected. 10 received a proton-boost, 10 a carbon-ion-boost in addition to standard treatment. After two fractions, we performed a PET/CT-scan of the brain. We compared the resulting range deviation based on the Most-likely-shift method between the two measurements, and the measurements with corresponding expectations, calculated with the Monte-Carlo code FLUKA. Results: A patient’s two measurements deviated by 0.7 mm (±0.7 mm). Overall comparison between measurements and simulation resulted in a mean range deviation of 3.3 mm (±2.2 mm) with significant lower deviations in the 12C-arm. Conclusion: The used planning concepts display the actual dose distributions adequately. The carbon ion group’s results are below the used PTV safety margins (3 mm). Further adjustments to the simulation are required for proton irradiations. Some anatomical situations require particular attention to ensure highest accuracy and safety. Ó 2015 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 115 (2015) 179–185
Despite all clinical effort, patients with glioma still carry a bad prognosis; the 5-year-overall-survival of patients treated with standard of care is 9.8% [1,2]. Thus, the urge to find better treatment for these patients is strong. Within the last years, a new modality has become available as a promising option to fight cancer in radiotherapy – the use of protons and carbon ions; the CLEOPATRA study by Combs et al. is currently investigating the effect of particle therapy on glioblastoma [3]. The main advantage of particles is their socalled inverted depth dose profile related to the ‘‘Bragg–Peak’’ [4]. In addition, carbon ions offer an increased biological effectiveness (RBE), resulting in even more severe cell damage of the pathologic tissue [5]. It has been shown that especially in glioblastoma cell lines, carbon ion irradiation in combination with temozolomide ⇑ Corresponding author. 1 2
E-mail address: [email protected] (S.P. Nischwitz). Joint first authors. Contributed equally to this work.
http://dx.doi.org/10.1016/j.radonc.2015.03.022 0167-8140/Ó 2015 Elsevier Ireland Ltd. All rights reserved.
shows a high damage potential [6]. It is now possible to administer escalated doses in a very narrow window to protect the non-affected tissue in the entrance- and exit-channel of the irradiated area. Glioma patients benefit from the thus possible relative local dose escalation [7–9]. Of course, this benefit brings along a risk of irradiating the intended-to-spare tissue and missing the intended-to-treat tissue in cases of inadequate dose deposition or insufficient position verification. Thereby, it is of utmost importance to hit the intended area exactly where planned. This puts additional emphasis on treatment planning and verification of positioning. The clinical benefit of that accuracy still lacks evidence, because generous safety margins of up to 10 mm are being employed due to uncertainties in the knowledge of the beam range; the MIRANDA study by Combs et al. uses positron-emission-tomography (PET) to monitor patients treated with particle therapy in general [10]. PET is the only way to monitor the in vivo activity induced by particle therapy to be clinically tested; at our institution, we use off-line PET, meaning the PET-scanner is located in a separate room
180
PET dosimetry in particle radiotherapy
close by the irradiation units, in contrast to in-beam (physically combined PET scanner and radiation application device) or inroom (PET scanner in the treatment room equipped with robotic patient transportation system) scanners [11–13]. Off-line PET delivers very promising results using only one radiation field and a minimum delay, it is the method of choice when the PET/CTscanner is already installed and a practical solution is sought, although the optimum compromise in terms of imaging performances and cost-effectiveness seems to be the in-room scanner nowadays [14]. The activity induction of carbon ions differs from the one of protons: protons only activate the tissue nuclei they hit, leading to a target-activation, carbon ions can also fragmentate themselves, leading to detectable b+-activity from 12C-to-11C-reactions in addition to the target-activation [15]. Some research groups in Japan [16,17], the USA [18,19] and at GSI Darmstadt, Germany [12,20] have already gathered some experience concerning the in vivo monitoring and several authors (Frey et al. [21], Knopf et al. [22], Min et al. [23], Helmbrecht et al. [24], Kuess et al. [25]) have investigated strategies to evaluate the beam range and thereby to validate the accuracy of particle therapy. In the present manuscript, we utilised the most-likely-shift (MLS) approach [21] to perform an in vivo evaluation of the beam range in 20 patients with glioma treated with protons or carbon ions. Methods and materials
carbon ions, shortly before the patient’s first appointment at HIT; safety margins of 3 mm were added according to protocol [3]. Treatment The beam application itself is conducted using the intensitycontrolled rasterscanning technique [27]. On two of the radiation applications, we performed a PET/CT scan right after the treatment, one early and one late during the radiation course. For that purpose, there is a commercial full-ring PET/CT scanner (Siemens Biograph mCTÒ) located in a room next to the irradiation units at HIT. The time of the patient’s irradiation (tirr) was noted, together with the time bridging the gap between irradiation of the last field and start of the PET scan (Dt), as well as the time during which the PET scan was running (30 min, according to protocol MIRANDA [10]). The irradiation itself lasted between 1 min and 3 min per field being rounded to the full minute. The patient was walked to the PET scanner right after the irradiation of the last field was completed and then repositioned using the same mask to have him accurately immobilized in the same position as during irradiation; immobilizations achieve accuracy of around 1–3 mm. A CT with a voxel size of 1 mm and a slice-thickness of 3 mm was performed prior to the actual PET scan to verify the positioning and to allow later handling (registration, scatter corrections and attenuation) of the PET scan, so that Dt ranged between 4 min and 11 min.
Patient data and examination protocol We selected a group of 20 patients with malignant glioma, who qualified for treatment according to previously published protocols [3]. The group consisted of two equally sized sub-groups, whereof one was treated with 12C-ions (12C) and the other one with protons (p+). Diagnosis was glioblastoma (GBM) in 18 of 20 cases; also there was one pleomorphic xanthoastrocytoma (PXA) and one anaplastic astrocytoma (AA). Patients’ age ranged from 32 to 69 years with a median age of 59.5. There were 15 male (75%) and 5 female (25%) patients. Except for one case, they all received a photon irradiation of 48–50 Gy in combination with temozolomide according to standard of care [1] prior to the treatment with a particle boost using one (6 cases, 30%) or two fields (14 cases, 70%). The one patient with AA did not receive any irradiation prior to his boost. The particle boost was set to be 6 3 Gy(RBE) (12C) or 5 2 Gy(RBE) (p+), thus the total dose adds up to 60 Gy(RBE) or 68 Gy(RBE), considering the higher RBE of 12C-ions [3,26]. Proton irradiation was administered at this dose to mimic conventional photon doses, while in carbon ion treatments, dose escalation was investigated in the patient cohort [3]. 13 patients (65%) received the combined radiochemotherapy (RCT) as an adjuvant therapy after (partial) resection, whereas it was the primary therapy after stereotactic biopsy in order to assure the diagnosis for the other seven patients (35%). Treatment planning Particle therapy of the brain takes place at our institution (HIT, University of Heidelberg, Germany) using individually manufactured polycaprolactone-based thermoplastic fibre masks with a density of 1.13 g/cm3 for immobilization. The mask was produced and a native CT with 3 mm slices and a voxel size of 1 mm and an MRI were performed to plan the boost therapy with the treatment planning software PT-Planning (Siemens, Erlangen, Germany) for both qualities, protons and
Simulation The reconstructed images from the measured PET data were compared to a simulation of the irradiation-induced b+-activity and to another (former) examination to investigate consistency and possible anatomical changes or other factors influencing beam delivery. For better correlation and valid comparability, we calculated expected spatial b+ activity patterns with the Monte-Carlo code FLUKA [28,29], assuming a correct dose delivery based on the repeated random sampling algorithm Monte-Carlo simulations provide. The specific treatment plan and the treatment planning CT for anatomical and elemental composition are input. The elemental composition is by the anatomical correlate represented by Hounsfield Units (HU) being allocated to each a different composition [30], treating the brain as one, due to the low contrasts of soft tissues in CT. The b+-isotope distribution is calculated for only the most relevant b+-active isotopes of rather long half-lives, such being 15 O (t½ = 2.03 min), 11C (t½ = 20.38 min), 13N (t½ = 9.96 min), 38K (t½ = 7.6 min), 30P (t½ = 2.5 min) and 34Cl (t½ = 32.0 min) [11,31]. To display the actual expected activity pattern, the measured times approximated as integer numbers (cf. Treatment) have to be taken into account. The biological washout [18,32–34] (perfusion), that has been deduced from a rabbit model and decay analysis of patient data [18,35] and the physical decay, as well as the distribution itself depend on how long the matter was irradiated, how much time passed until the recording started and how long the recording lasted at all [11]. Data processing The data set was processed using a specially designed module within the application framework MeVisLab (MeVis Medical Solutions AG, Fraunhofer MeVis, Bremen, Germany) [36]. With this tool, it is possible to visualise and evaluate the range difference between two data sets by the so-called most-likely-shift approach;
181
S.P. Nischwitz et al. / Radiotherapy and Oncology 115 (2015) 179–185
this approach is based on a minimization of the absolute difference between two activity profiles in the distal part of an irradiated area, shifted against each other with an automatic identification of the distal area of interest [21]. This range shift is calculated for each pencil beam to acquire an average result for the whole irradiated volume. In comparison to earlier approaches, this method shows more robust results considering uncertainties caused by patient geometry, PET imaging concepts, counting statistics and more [21]. We created a result report for the range difference between the simulation and the first measured PET scan (simulation vs. m1), the simulation and the second measurement (simulation vs. m2) and between the two measured scans themselves (m1 vs. m2); the latter to verify reproducibility of the treatment. The purpose of the comparison between the two measurements was to validate the quality and accuracy of the PET measurement and the used algorithm, while a comparison of simulation and measurement allowed a statement about the actual irradiation range and used planning concepts.
10 received a proton-boost and 10 received a boost with carbon ions after their initial photon treatment. The acquired results can be regarded as valid, since the CT deviations all ranged below 1 mm.
Comparison of a patient’s two measurements Comparing the two PET measurements with each other, we could assess a strong consistency throughout the treatment field. The average deviation in range between the first (U1) and the second PET-scan (U2) was measured 0.7 mm (range: 0.0–2.1 mm) with a standard deviation of 0.7 mm. The average deviations for 12C and p+ are 0.7 mm (range: 0.3– 1.9 mm) and 0.7 mm (range: 0.0–2.1 mm) with standard deviations of 0.6 mm and 0.7 mm, respectively. The difference between 12 C-comparisons and p+-comparisons is not significant (p = 0.39). Individual results are displayed in (Table 2).
Comparison of a patient’s simulation with the corresponding measurement
Statistical methods To obtain valid results, the software also creates a proximal range difference connected to the CT, aiming to display the deviation between the planning CT and the one acquired prior to the PET scan, considering the presented HU in the proximal slices (in beam direction). It is necessary for it to be close to 0 mm, to avoid differences in the beam range caused by improper positioning. The arithmetic mean in range difference of each voxel within the irradiated area as absolute value and the resulting standard deviation were used to receive a valid result of each measurement. A calculation of the arithmetic means of the results was done to compare the groups. Statistical significance was determined using Student’s t-test; the significance level was set to be p = 0.05. Results Patients’ characteristics are summarized in (Table 1). In summary, 20 patients with malignant glioma were included, whereof
The average range difference of all the simulations vs. scan analyses, including both protons and carbon ions, was measured 3.3 mm (range: 0.2–7.2 mm) with a standard deviation of 2.2 mm. Detailed results are displayed in (Table 2). Regarding only the group with the patients treated with carbon ions, we detected an average range difference of 2.3 mm (range: 0.2–6.4 mm) with a standard deviation of 1.7 mm between simulation and measurement. The relative range difference was 2.2 mm (range: 1.0–6.4 mm) with a standard deviation of 1.9 mm; which signifies that the irradiation ranges were in average 2.2 mm further than simulated. Regarding only the group with the patients treated with protons, we observed an average range difference of 4.2 mm (range: 0.3–7.2 mm) with a standard deviation of 2.2 mm between simulation and measurement. Relative deviations correspond to the absolute ones; all deviations were positive, hence the actual beam range was higher than simulated. The difference between the two groups (12C vs. p+) is statistically significant (p = 0.005).
Table 1 Patients’ characteristics. ID
Particle
Age during boost
Gender
Diagnosis
Site
CTV [ml]
RCT
Photon irradiation prior to boost
Single dose (ion boost)
#Fractions (boost)
Total dose
#Fields (boost)
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10
12
61 65 49 58 46 45 67 52 60 55
# # # # # # $ # # $
PXA GBM
p. ri. po. le. f. ri. t. le. p. le. c. ri. f. le. t. ri. f. le. t. le.
60.17 94.61 26.44 65.55 73.55 19.13 94.18 49.76 115.61 106.80
Primary Primary Adjuvant Primary Primary Adjuvant Adjuvant Adjuvant Adjuvant Adjuvant
25 2 Gy
3 GyE
6
18 GyE + 50 Gy
1 1 2 2 2 1 2 1 2 2
P11 P12 P13 P14 P15 P16 P17 P18 P19 P20
p+
64 69 61 61 51 61 32 62 59 43
$ $ # $ # # # # # #
GBM
fp. le. c. ri. ph. ri. t. le. t. le. po. re. f. ri. bf. p. le. po. ri.
43.92 66.85 95.90 44.95 168.83 28.69 31.26 178.59 24.09 148.36
Adjuvant Primary Primary Adjuvant Adjuvant Adjuvant Adjuvant Primary Adjuvant Adjuvant
2 GyE
5
10 GyE + 50 Gy
1 2 2 2 2 2 2 2 2 1
C
AA
24 2 Gy 25 2 Gy
None
10 GyE + 48 Gy 10 GyE + 50 Gy
30
60 GyE
CTV = Clinical Target Volume, PXA = pleomorphic xanthoastrocytoma, GBM = glioblastoma, AA = anaplastic astrocytoma, ri = right, le = left, p = parietal, po = parietooccipital, f = frontal, bf = bifrontal, fp = frontoparietal, t = temporal, c = central, ph = panhemispheric.
182
PET dosimetry in particle radiotherapy
SD [mm]
a standard deviation of 3.1 mm. P15#2 is a large tumour of complex, irregular shape with a central necrosis and peripheral contrast enhancement, leading to a range deviation of 7.2 mm with a standard deviation of 5.5 mm.
0.39
2.95
Discussion
0.68
1.93
1.89
4.67
0.28
1.86
1.71
3.55
0.26
2.26
0.36
3.14
0.94
2.29
0.45
3.06
0.39
2.83
Table 2 Detailed results – range deviation m1/2 = measurement1/2. ID
Dt [min]
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 All
P13 P14 P15 P16 P17 P18 P19 P20
*
3.33 4.43 4.22 4.55 4.34 5.20 2.74 3.03 4.34 4.34 5.25 2.19 4.23 3.91 5.56 3.98 4.75 5.24 4.81 3.53
m1 vs. m2 ø range dev.[mm]*
7 7 7 6 9 7 4 5 7 9 6 4 6 6 7 6 5 4 6 7
2.10 0.99 1.74 1.04 0.73 2.28 1.24 1.26 2.77 1.47 2.00 1.91 4.56 3.32 0.19 0.23 5.73 6.40 2.95 3.73 2.33
1.71
0.74
0.60
#1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2
6 6 8 8 7 7 5 7 11 7 6 5 5 6 6 6 5 6 6 6
6.77 7.00 4.52 3.16 5.13 6.37 0.30 2.08 5.67 7.23 1.01 1.27 5.29 6.50 2.03 2.23 4.40 2.68 5.96 4.90
4.01 4.58 6.33 5.74 5.17 5.43 3.05 4.53 4.79 5.46 0.50 0.50 5.22 5.10 5.29 5.92 3.36 2.71 4.23 4.22
0.04
0.76
1.61
1.19
0.59
1.24
0.91
1.93
2.09
2.10
0.00
0.73
0.03
1.25
0.65
1.26
1.16
1.17
0.06
1.10
C
P12
SD [mm]
#1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2
12
P11
Simulation vs. measurement ø range dev. [mm]*
All p+
4.23
2.15
0.71
0.73
Overall (12C, p+)
3.28
2.17
0.73
0.65
All values are transformed into natural numbers.
Case reports In the following, we would like to report on 4 specific cases, which represent the overall impression. We chose P6#2 and P9#2 as representative studies of the carbon ion group, and P14#1 and P15#2 as representatives of the proton-group. While P6#2 and P14#1 show good concordance, meaning the range deviation is low (<2 mm), P9#2 and P15#2 show larger range shifts (>5 mm). In Fig. 1 we show range deviations for the two carbon ion studies, appertaining CT-images with planning-target-volume (PTV) dose coverage and correlating MRI-images. P6#2 is a small tumour with homogeneous contrast enhancement; it leads to an average deviation of 1.9 mm with a standard deviation of 2.2 mm. P9#2 is a large tumour of complex, irregular shape with a peripheral contrast enhancement and a central necrosis resulting in an average deviation of 6.4 mm with a standard deviation of 5.2 mm. (Fig. 2) displays the proton studies. P14#1 is a small tumour with homogeneous contrast enhancement and convex-shaped, clear tumour borders. It is completely surrounded by brain parenchyma. The measured average range deviation is 0.3 mm with
In the present manuscript we assessed the beam range of proton and carbon ion irradiation in 20 patients with primary glioma during 2 fractions, early and late during the irradiation course, each using a FLUKA generated simulation. In a comparison of a patient’s two measurements we found a strong consistency with an average deviation of 0.7 mm with 75% of the analyses deviating less than a millimetre; this supports the consistent quality of our methodology. No significant result was found concerning the difference between 12C- and p+-irradiated patients. The observed minor deviations could be caused by minimal movements of the patient within the mask of up to 3 mm, potential morphological changes of the tumour itself and/or the irradiated area between the measurements or by changes concerning the biological washout (due to e.g., higher blood flow). Another possible option is it just being coincidental statistical spreading of the b+activity. Frey et al. [24] discuss some uncertainties of the MLS-method dividing them in external and internal factors; the external factors address the comparability of two activity distributions including registration of the two CT-scans, while internal factors handle intrinsic events related to washout or reaction cross-sections [21]. The comparison of simulation and measurement delivered in general satisfying results. The average deviation was 3.3 mm with a significant difference between 12C- (2.3 mm) and p+-irradiated (4.2 mm) patients. Even though we analysed a quite small number of cases, we are convinced that these results are valid and representative for particle therapy in the central nervous system. Regarding the 12C-group, we believe that the results, being below the used safety margins of commonly 3-–5 mm, allow a solid prediction of the beam’s range. This confirms the role of Monte-Carlo simulation and PET monitoring in the context of 12C therapy planning. A used safety margin of 3 mm with possible individual expansion in complex cases appears to be a reasonable value to cover the majority of the irradiation cases in the future. We would like to point out, that altogether in most of the cases we received results approximating a Gaussian distribution with some outliers, and activity profiles similar to the simulated ones. Transferring that to the constructed map of the range deviation, most of the covered area presents a beam range very close to the predicted one with only a few spots of nonconformity, resulting in shifted means. As for the p+-group, the measured results differ somewhat more from the simulated ones. This might lead to the conclusion that safety margins should be more generous than in 12C treatment, which should not be overrated, hence it is a theoretical appearance induced by the simulation modelling. On the other hand, one may feel that this range verification tool is not suitable for proton irradiations, which will be determined meticulously in future analyses. Another possible explanation might be the fact that a carbon ion irradiation delivers a more favourable spatial distribution of the activity with localised maxima towards the end of range due to the 11C projectile fragments, leading to a lower range deviation. We are aware of our study’s limitations, one being a nonconsideration of general differences in the positron-induction mechanisms of protons and carbon ions, that is mainly responsible
S.P. Nischwitz et al. / Radiotherapy and Oncology 115 (2015) 179–185
183
Fig. 1. Visualisation of two exemplary carbon ion studies. (a) Low range deviation in P6#2, from left to right: sagittal CT with planned irradiation area, sagittal MR (T1w, Gd) displaying the small rather homogeneously enhancing tumour, mapping of range deviation (irradiated area) with histogram. (b) Suboptimal range deviation in P9#2, from left to right: axial CT with planned irradiation area, axial MR (T1w, Gd) displaying the large complex-shaped tumour with central necrosis, peripheral enhancement and surrounding oedema, mapping of range deviation (irradiated area) with histogram.
for the more differing results in proton irradiation; while in protons only the irradiated target emits positrons, there is an activity-induction by the projectile in carbon ions additionally [37]. Furthermore it has been shown that there is also a difference concerning the washout properties depending on the ion species [38]. Experiments have shown that the washout itself is responsible for a few millimetres inaccuracy that could be reduced by suppressing the washout processes in phantom irradiation [39]. This calls for adaptation of our simulation framework in further studies to investigate clear clinical feasibility of range verification in proton irradiated glioma [40]. Currently, we are performing such modifications at our institution, including a more differentiated structure analysis, considering the distinct elemental compositions of the particular brain matters [41]. When we take a look at our presented cases, we can derive that in homogeneous tumours, the beam range is more accurate than in inhomogeneous ones. The middle pictures in Figs. 1a and 2a show T1-weighted (T1w) MR-images with contrast media (Gd) of patients 6#2 and 14#1 which are located in the distal reading range, where the tumour can be described as evenly contrast-enhanced structures. This is reflected in the mappings to the right, where we have a depiction just as evenly plotted. The MR-images (T1w, Gd) in Figs. 1b and 2b are slices of the distal reading range of studies P9#2 and P15#2. In them, we can see that both tumours are only partially contrast enhancing. The structures appear diffuse and irregular. In the corresponding beam
range mappings, we get patchy and multi-coloured representations of the latters. Here from we draw the conclusion, that anatomical and elemental inhomogeneous structures, such as peri-/intratumoural bleedings, oedema, dense bands of connective tissue such as the cerebral falx and exogenous surgical patches, but equally washout-altering factors like junctions to the cerebrospinal fluid or blood vessels, might cause mismatches between range simulation and verification. This is especially pronounced for proton-induced activation, which is limited to target fragments and thus more sensitive to the tissue anatomical features. The statistical fluctuation of the Monte-Carlo simulation was below the voxel size that can be distinguished with MLS (<0.6 mm), hence without decisive influence on the results. Timing of a patient’s PET and number of fields did not affect our data either, as the results were comparable, regardless of those values, thus they should not play a major role, but should be taken into consideration in future investigations with a larger number of patients in each group. Another location, where range verification can also play a decisive role, is the pelvis or the sacral area. Tumours of the pelvis are a special challenge for radiotherapy, since the air-filled rectum and differing bladder volumes are potential factors in the radiation field that vary in between the planning and irradiations. The large density differences between bone and air can cause high deviations of planned and actually applied radiation. Future investigations should focus on this topic.
184
PET dosimetry in particle radiotherapy
Fig. 2. Visualisation of two exemplary proton studies. (a) Low range deviation in P14#1, from left to right: axial CT with planned irradiation area, axial MR (T1w) of small, homogeneously enhancing tumour, mapping and histogram of range deviation (irradiated area) with very low scattering around 0 mm. (b) Suboptimal range deviation in P15#2, from left to right: sagittal CT with planned irradiation area, sagittal MR (T1w, Gd) displaying a large, irregularly shaped tumour with central necrosis and peripheral contrast enhancement, corresponding patchy mapping of range deviation with histogram (irradiated area).
Conclusion Commonly used particle irradiation planning concepts display the actual dose distribution adequately. The offline PET-in vivomonitoring is a valid mean to receive data on the beam range. To confirm the same accuracy for 1H-irradiations as for 12Cirradiations, further studies are necessary and being performed. To ensure the highest possible precision, individual anatomical features need to be further investigated and considered when planning therapeutic irradiations with protons or carbon ions. Conflict of interest statement Dr. Bauer reports grants from European Union, grants from German Federal Ministry of Research and Education during the conduct of the study. Acknowledgements We would like to thank Thomas Tessonnier from the University Hospital of Heidelberg, Department of Radiation Oncology for fruitful discussion. References [1] Stupp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 2005;352:987–96.
[2] Stupp R, Hegi ME, Mason WP, et al. Effects of radiotherapy with concomitant and adjuvant temozolomide versus radiotherapy alone on survival in glioblastoma in a randomised phase III study: 5-year analysis of the EORTCNCIC trial. Lancet Oncol 2009;10:459–66. [3] Combs SE, Kieser M, Rieken S, et al. Randomized phase II study evaluating a carbon ion boost applied after combined radiochemotherapy with temozolomide versus a proton boost after radiochemotherapy with temozolomide in patients with primary glioblastoma: the CLEOPATRA trial. BMC Cancer 2010;10:478. [4] Wilson RR. Radiological use of fast protons. Radiology 1946;47:487–91. [5] Elsässer T, Weyrather WK, Friedrich T, et al. Quantification of the relative biological effectiveness for ion beam radiotherapy: direct experimental comparison of proton and carbon ion beams and a novel approach for treatment planning. Int J Radiat Oncol Biol Phys 2010;78:1177–83. [6] Combs SE, Bohl J, Elsässer T, et al. Radiobiological evaluation and correlation with the local effect model (LEM) of carbon ion radiation therapy and temozolomide in glioblastoma cell lines. Int J Radiat Biol 2009;85:126–37. [7] Hauswald H, Rieken S, Ecker S, et al. First experiences in treatment of lowgrade glioma grade I and II with proton therapy. Radiat Oncol 2012;7:189. [8] Rieken S, Habermehl D, Haberer T, Jaekel O, Debus J, Combs SE. Proton and carbon ion radiotherapy for primary brain tumors delivered with active raster scanning at the Heidelberg Ion Therapy Center (HIT): early treatment results and study concepts. Radiat Oncol 2012;7:41. [9] Mizoe JE, Tsujii H, Hasegawa A, et al. Phase I/II clinical trial of carbon ion radiotherapy for malignant gliomas: combined X-ray radiotherapy, chemotherapy, and carbon ion radiotherapy. Int J Radiat Oncol Biol Phys 2007;69:390–6. [10] Combs SE, Bauer J, Unholtz D, et al. Monitoring of patients treated with particle therapy using positron-emission-tomography (PET): the MIRANDA study. BMC Cancer 2012;12:133. [11] Bauer J, Unholtz D, Sommerer F, et al. Implementation and initial clinical experience of offline PET/CT-based verification of scanned carbon ion treatment. Radiother Oncol 2013;107:218–26. [12] Crespo P, Shakirin G, Enghardt W. On the detector arrangement for in-beam PET for hadron therapy monitoring. Phys Med Biol 2006;51:2143–63.
S.P. Nischwitz et al. / Radiotherapy and Oncology 115 (2015) 179–185 [13] Nishio T, Miyatake A, Ogino T, Nakagawa K, Saijo N, Esumi H. The development and clinical use of a beam ON-LINE PET system mounted on a rotating gantry port in proton therapy. Int J Radiat Oncol Biol Phys 2010;76:277–86. [14] Shakirin G, Braess H, Fiedler F, et al. Implementation and workflow for PET monitoring of therapeutic ion irradiation: a comparison of in-beam, in-room, and off-line techniques. Phys Med Biol 2011;56:1281–98. [15] Schardt D, Elsässer T, Schulz-Ertner D. Heavy-ion tumor therapy: physical and radiobiological benefits. Rev Mod Phys 2010;82:383–425. [16] Nishio T, Miyatake A, Inoue K, et al. Experimental verification of proton beam monitoring in a human body by use of activity image of positron-emitting nuclei generated by nuclear fragmentation reaction. Radiol Phys Technol 2008;1:44–54. [17] Hishikawa Y, Kagawa K, Murakami M, Sakai H, Akagi T, Abe M. Usefulness of positron-emission tomographic images after proton therapy. Int J Radiat Oncol Biol Phys 2002;53:1388–91. [18] Parodi K, Paganetti H, Shih HA, et al. Patient study of in vivo verification of beam delivery and range, using positron emission tomography and computed tomography imaging after proton therapy. Int J Radiat Oncol Biol Phys 2007;68:920–34. [19] Hsi WC, Indelicato DJ, Vargas C, Duvvuri S, Li Z, Palta J. In vivo verification of proton beam path by using post-treatment PET/CT imaging. Med Phys 2009;36:4136–46. [20] Enghardt W, Parodi K, Crespo P, Fiedler F, Pawelke J, Ponisch F. Dose quantification from in-beam positron emission tomography. Radiother Oncol 2004;73:S96–8. [21] Frey K, Unholtz D, Bauer J, et al. Automation and uncertainty analysis of a method for in vivo range verification in particle therapy. Phys Med Biol 2014;59:5903–19. [22] Knopf A, Parodi K, Paganetti H, Cascio E, Bonab A, Bortfeld T. Quantitative assessment of the physical potential of proton beam range verification with PET/CT. Phys Med Biol 2008;53:4137–51. [23] Min CH, Zhu X, Winey BA, et al. Clinical application of in-room positron emission tomography for in vivo treatment monitoring in proton radiation therapy. Int J Radiat Oncol Biol Phys 2013;86:183–9. [24] Helmbrecht S, Santiago A, Enghardt W, Kuess P, Fiedler F. On the feasibility of automatic detection of range deviations from in-beam PET data. Phys Med Biol 2012;57:1387–97. [25] Kuess P, Birkfellner W, Enghardt W, Helmbrecht S, Fiedler F, Georg D. Using statistical measures for automated comparison of in-beam PET data. Med Phys 2012;39:5874–81. [26] Tsujii H, Kamada T. A review of update clinical results of carbon ion radiotherapy. Jpn J Clin Oncol 2012;42:670–85. [27] Haberer T, Becher W, Schardt D, Kraft G. Magnetic scanning system for heavy ion therapy. Nucl Instrum Meth A 1993;330:296–305.
185
[28] Battistoni G, Muraro S, Sala PR, et al. The FLUKA code: Description and benchmarking. In: Albrow M and Raja R, editors. Proceedings of the Hadronic Shower Simulation Workshop, 2006, Fermilab, 6–8 September. p. 200631–49. [29] Ferrari A, Sala PR, Fassò A, Ranft J. FLUKA: a multi-particle transport code, CERN-2005-10 (2005), INFN/TC_05/11, SLAC-R-773, 2005. [30] Schneider W, Bortfeld T, Schlegel W. Correlation between CT numbers and tissue parameters needed for Monte Carlo simulations of clinical dose distributions. Phys Med Biol 2000;45:459–78. [31] Parodi K, Ferrari A, Sommerer F, Paganetti H. Clinical CT-based calculations of dose and positron emitter distributions in proton therapy using the FLUKA Monte Carlo code. Phys Med Biol 2007;52:3369–87. [32] Helmbrecht S, Enghardt W, Parodi K, et al. Analysis of metabolic washout of positron emitters produced during carbon ion head and neck radiotherapy. Med Phys 2013;40:091918. [33] Fiedler F, Priegnitz M, Julich R, et al. In-beam PET measurements of biological halflives of 12C irradiation induced beta+-activity. Acta Oncol 2008;47:1077–86. [34] Parodi K, Bortfeld T, Haberer T. Comparison between in-beam and offline positron emission tomography imaging of proton and carbon ion therapeutic irradiation at synchrotron- and cyclotron-based facilities. Int J Radiat Oncol Biol Phys 2008;71:945–56. [35] Mizuno H, Tomitani T, Kanazawa M, et al. Washout measurement of radioisotope implanted by radioactive beams in the rabbit. Phys Med Biol 2003;48:2269–81. [36] Unholtz D, Sommerer F, Bauer J, et al. Post-therapeutical b+-activity measurements in comparison to simulations towards in-vivo verification of ion beam therapy. Poster presented at: Conference Record of the IEEE Nuclear Science Symposium and Medical Imaging Conference; Valencia, Spain. [37] Paganetti H. Range uncertainties in proton therapy and the role of Monte Carlo simulations. Phys Med Biol 2012;57:R99–R117. [38] Ammar C, Frey K, Bauer J, et al. Comparing the biological washout of beta+activity induced in mice brain after 12C-ion and proton irradiation. Phys Med Biol 2014;59:7229–44. [39] Tessonnier T, Rinaldi I, Liebl J, Parodi K. Assessment of the predictive power of PET-based range verification in ion therapy. Poster presented at: Workshop on Range Assessment and Dose Verification in Particle Therapy; 29–30 September 2014, Dresden, Germany. [40] Bauer J, Unholtz D, Kurz C, Parodi K. An experimental approach to improve the Monte Carlo modelling of offline PET/CT-imaging of positron emitters induced by scanned proton beams. Phys Med Biol 2013;58:5193–213. [41] Bauer J, Unholtz D, Sommerer F, et al. Experimental validation of simulated positron emitter yields and refined tissue classification for improved Monte Carlo modeling towards offline PET/CT-based verification of proton therapy. Poster presented at: 3rd European Workshop on Monte Carlo Treatment Planning of the European Workgroup on MCTP; Sevilla, Spain.
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
in vivo PET dosimetry
Clinical implementation and range evaluation of in vivo PET dosimetry for particle irradiation in patients with primary glioma Sebastian P. Nischwitz a,⇑,1, Julia Bauer b,1, Thomas Welzel a, Harald Rief a, Oliver Jäkel b,c, Thomas Haberer b, Kathrin Frey d, Jürgen Debus a, Katia Parodi a,d, Stephanie E. Combs e,2, Stefan Rieken a,2 a University Hospital of Heidelberg, Department of Radiation Oncology; b Heidelberg Ion-Beam Therapy Centre; c German Cancer Research Center (DKFZ), Department of Medical Physics in Radiation Oncology, Heidelberg; d Ludwig-Maximilians-University of Munich, Department of Medical Physics, Garching; and e University Hospital Klinikum rechts der Isar, Department of Radiation Oncology, Munich, Germany
a r t i c l e
i n f o
Article history: Received 16 September 2014 Received in revised form 18 March 2015 Accepted 21 March 2015 Available online 2 April 2015 Keywords: PET In vivo dosimetry Ion therapy Monte Carlo simulation Glioblastoma Range verification
a b s t r a c t Purpose: The physical and biological properties of ion-beams offer various advantages in comparison to conventional radiotherapy, though uncertainties concerning quality assurance are still left. Due to the inverted depth dose profile, range accuracy is of paramount importance. We investigated the range deviations between planning simulation and post-fractional PET/CT measurement from particle therapy in primary glioblastoma. Methods and materials: 20 patients with glioblastoma undergoing particle therapy at our institution were selected. 10 received a proton-boost, 10 a carbon-ion-boost in addition to standard treatment. After two fractions, we performed a PET/CT-scan of the brain. We compared the resulting range deviation based on the Most-likely-shift method between the two measurements, and the measurements with corresponding expectations, calculated with the Monte-Carlo code FLUKA. Results: A patient’s two measurements deviated by 0.7 mm (±0.7 mm). Overall comparison between measurements and simulation resulted in a mean range deviation of 3.3 mm (±2.2 mm) with significant lower deviations in the 12C-arm. Conclusion: The used planning concepts display the actual dose distributions adequately. The carbon ion group’s results are below the used PTV safety margins (3 mm). Further adjustments to the simulation are required for proton irradiations. Some anatomical situations require particular attention to ensure highest accuracy and safety. Ó 2015 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology 115 (2015) 179–185
Despite all clinical effort, patients with glioma still carry a bad prognosis; the 5-year-overall-survival of patients treated with standard of care is 9.8% [1,2]. Thus, the urge to find better treatment for these patients is strong. Within the last years, a new modality has become available as a promising option to fight cancer in radiotherapy – the use of protons and carbon ions; the CLEOPATRA study by Combs et al. is currently investigating the effect of particle therapy on glioblastoma [3]. The main advantage of particles is their socalled inverted depth dose profile related to the ‘‘Bragg–Peak’’ [4]. In addition, carbon ions offer an increased biological effectiveness (RBE), resulting in even more severe cell damage of the pathologic tissue [5]. It has been shown that especially in glioblastoma cell lines, carbon ion irradiation in combination with temozolomide ⇑ Corresponding author. 1 2
E-mail address: [email protected] (S.P. Nischwitz). Joint first authors. Contributed equally to this work.
http://dx.doi.org/10.1016/j.radonc.2015.03.022 0167-8140/Ó 2015 Elsevier Ireland Ltd. All rights reserved.
shows a high damage potential [6]. It is now possible to administer escalated doses in a very narrow window to protect the non-affected tissue in the entrance- and exit-channel of the irradiated area. Glioma patients benefit from the thus possible relative local dose escalation [7–9]. Of course, this benefit brings along a risk of irradiating the intended-to-spare tissue and missing the intended-to-treat tissue in cases of inadequate dose deposition or insufficient position verification. Thereby, it is of utmost importance to hit the intended area exactly where planned. This puts additional emphasis on treatment planning and verification of positioning. The clinical benefit of that accuracy still lacks evidence, because generous safety margins of up to 10 mm are being employed due to uncertainties in the knowledge of the beam range; the MIRANDA study by Combs et al. uses positron-emission-tomography (PET) to monitor patients treated with particle therapy in general [10]. PET is the only way to monitor the in vivo activity induced by particle therapy to be clinically tested; at our institution, we use off-line PET, meaning the PET-scanner is located in a separate room
180
PET dosimetry in particle radiotherapy
close by the irradiation units, in contrast to in-beam (physically combined PET scanner and radiation application device) or inroom (PET scanner in the treatment room equipped with robotic patient transportation system) scanners [11–13]. Off-line PET delivers very promising results using only one radiation field and a minimum delay, it is the method of choice when the PET/CTscanner is already installed and a practical solution is sought, although the optimum compromise in terms of imaging performances and cost-effectiveness seems to be the in-room scanner nowadays [14]. The activity induction of carbon ions differs from the one of protons: protons only activate the tissue nuclei they hit, leading to a target-activation, carbon ions can also fragmentate themselves, leading to detectable b+-activity from 12C-to-11C-reactions in addition to the target-activation [15]. Some research groups in Japan [16,17], the USA [18,19] and at GSI Darmstadt, Germany [12,20] have already gathered some experience concerning the in vivo monitoring and several authors (Frey et al. [21], Knopf et al. [22], Min et al. [23], Helmbrecht et al. [24], Kuess et al. [25]) have investigated strategies to evaluate the beam range and thereby to validate the accuracy of particle therapy. In the present manuscript, we utilised the most-likely-shift (MLS) approach [21] to perform an in vivo evaluation of the beam range in 20 patients with glioma treated with protons or carbon ions. Methods and materials
carbon ions, shortly before the patient’s first appointment at HIT; safety margins of 3 mm were added according to protocol [3]. Treatment The beam application itself is conducted using the intensitycontrolled rasterscanning technique [27]. On two of the radiation applications, we performed a PET/CT scan right after the treatment, one early and one late during the radiation course. For that purpose, there is a commercial full-ring PET/CT scanner (Siemens Biograph mCTÒ) located in a room next to the irradiation units at HIT. The time of the patient’s irradiation (tirr) was noted, together with the time bridging the gap between irradiation of the last field and start of the PET scan (Dt), as well as the time during which the PET scan was running (30 min, according to protocol MIRANDA [10]). The irradiation itself lasted between 1 min and 3 min per field being rounded to the full minute. The patient was walked to the PET scanner right after the irradiation of the last field was completed and then repositioned using the same mask to have him accurately immobilized in the same position as during irradiation; immobilizations achieve accuracy of around 1–3 mm. A CT with a voxel size of 1 mm and a slice-thickness of 3 mm was performed prior to the actual PET scan to verify the positioning and to allow later handling (registration, scatter corrections and attenuation) of the PET scan, so that Dt ranged between 4 min and 11 min.
Patient data and examination protocol We selected a group of 20 patients with malignant glioma, who qualified for treatment according to previously published protocols [3]. The group consisted of two equally sized sub-groups, whereof one was treated with 12C-ions (12C) and the other one with protons (p+). Diagnosis was glioblastoma (GBM) in 18 of 20 cases; also there was one pleomorphic xanthoastrocytoma (PXA) and one anaplastic astrocytoma (AA). Patients’ age ranged from 32 to 69 years with a median age of 59.5. There were 15 male (75%) and 5 female (25%) patients. Except for one case, they all received a photon irradiation of 48–50 Gy in combination with temozolomide according to standard of care [1] prior to the treatment with a particle boost using one (6 cases, 30%) or two fields (14 cases, 70%). The one patient with AA did not receive any irradiation prior to his boost. The particle boost was set to be 6 3 Gy(RBE) (12C) or 5 2 Gy(RBE) (p+), thus the total dose adds up to 60 Gy(RBE) or 68 Gy(RBE), considering the higher RBE of 12C-ions [3,26]. Proton irradiation was administered at this dose to mimic conventional photon doses, while in carbon ion treatments, dose escalation was investigated in the patient cohort [3]. 13 patients (65%) received the combined radiochemotherapy (RCT) as an adjuvant therapy after (partial) resection, whereas it was the primary therapy after stereotactic biopsy in order to assure the diagnosis for the other seven patients (35%). Treatment planning Particle therapy of the brain takes place at our institution (HIT, University of Heidelberg, Germany) using individually manufactured polycaprolactone-based thermoplastic fibre masks with a density of 1.13 g/cm3 for immobilization. The mask was produced and a native CT with 3 mm slices and a voxel size of 1 mm and an MRI were performed to plan the boost therapy with the treatment planning software PT-Planning (Siemens, Erlangen, Germany) for both qualities, protons and
Simulation The reconstructed images from the measured PET data were compared to a simulation of the irradiation-induced b+-activity and to another (former) examination to investigate consistency and possible anatomical changes or other factors influencing beam delivery. For better correlation and valid comparability, we calculated expected spatial b+ activity patterns with the Monte-Carlo code FLUKA [28,29], assuming a correct dose delivery based on the repeated random sampling algorithm Monte-Carlo simulations provide. The specific treatment plan and the treatment planning CT for anatomical and elemental composition are input. The elemental composition is by the anatomical correlate represented by Hounsfield Units (HU) being allocated to each a different composition [30], treating the brain as one, due to the low contrasts of soft tissues in CT. The b+-isotope distribution is calculated for only the most relevant b+-active isotopes of rather long half-lives, such being 15 O (t½ = 2.03 min), 11C (t½ = 20.38 min), 13N (t½ = 9.96 min), 38K (t½ = 7.6 min), 30P (t½ = 2.5 min) and 34Cl (t½ = 32.0 min) [11,31]. To display the actual expected activity pattern, the measured times approximated as integer numbers (cf. Treatment) have to be taken into account. The biological washout [18,32–34] (perfusion), that has been deduced from a rabbit model and decay analysis of patient data [18,35] and the physical decay, as well as the distribution itself depend on how long the matter was irradiated, how much time passed until the recording started and how long the recording lasted at all [11]. Data processing The data set was processed using a specially designed module within the application framework MeVisLab (MeVis Medical Solutions AG, Fraunhofer MeVis, Bremen, Germany) [36]. With this tool, it is possible to visualise and evaluate the range difference between two data sets by the so-called most-likely-shift approach;
181
S.P. Nischwitz et al. / Radiotherapy and Oncology 115 (2015) 179–185
this approach is based on a minimization of the absolute difference between two activity profiles in the distal part of an irradiated area, shifted against each other with an automatic identification of the distal area of interest [21]. This range shift is calculated for each pencil beam to acquire an average result for the whole irradiated volume. In comparison to earlier approaches, this method shows more robust results considering uncertainties caused by patient geometry, PET imaging concepts, counting statistics and more [21]. We created a result report for the range difference between the simulation and the first measured PET scan (simulation vs. m1), the simulation and the second measurement (simulation vs. m2) and between the two measured scans themselves (m1 vs. m2); the latter to verify reproducibility of the treatment. The purpose of the comparison between the two measurements was to validate the quality and accuracy of the PET measurement and the used algorithm, while a comparison of simulation and measurement allowed a statement about the actual irradiation range and used planning concepts.
10 received a proton-boost and 10 received a boost with carbon ions after their initial photon treatment. The acquired results can be regarded as valid, since the CT deviations all ranged below 1 mm.
Comparison of a patient’s two measurements Comparing the two PET measurements with each other, we could assess a strong consistency throughout the treatment field. The average deviation in range between the first (U1) and the second PET-scan (U2) was measured 0.7 mm (range: 0.0–2.1 mm) with a standard deviation of 0.7 mm. The average deviations for 12C and p+ are 0.7 mm (range: 0.3– 1.9 mm) and 0.7 mm (range: 0.0–2.1 mm) with standard deviations of 0.6 mm and 0.7 mm, respectively. The difference between 12 C-comparisons and p+-comparisons is not significant (p = 0.39). Individual results are displayed in (Table 2).
Comparison of a patient’s simulation with the corresponding measurement
Statistical methods To obtain valid results, the software also creates a proximal range difference connected to the CT, aiming to display the deviation between the planning CT and the one acquired prior to the PET scan, considering the presented HU in the proximal slices (in beam direction). It is necessary for it to be close to 0 mm, to avoid differences in the beam range caused by improper positioning. The arithmetic mean in range difference of each voxel within the irradiated area as absolute value and the resulting standard deviation were used to receive a valid result of each measurement. A calculation of the arithmetic means of the results was done to compare the groups. Statistical significance was determined using Student’s t-test; the significance level was set to be p = 0.05. Results Patients’ characteristics are summarized in (Table 1). In summary, 20 patients with malignant glioma were included, whereof
The average range difference of all the simulations vs. scan analyses, including both protons and carbon ions, was measured 3.3 mm (range: 0.2–7.2 mm) with a standard deviation of 2.2 mm. Detailed results are displayed in (Table 2). Regarding only the group with the patients treated with carbon ions, we detected an average range difference of 2.3 mm (range: 0.2–6.4 mm) with a standard deviation of 1.7 mm between simulation and measurement. The relative range difference was 2.2 mm (range: 1.0–6.4 mm) with a standard deviation of 1.9 mm; which signifies that the irradiation ranges were in average 2.2 mm further than simulated. Regarding only the group with the patients treated with protons, we observed an average range difference of 4.2 mm (range: 0.3–7.2 mm) with a standard deviation of 2.2 mm between simulation and measurement. Relative deviations correspond to the absolute ones; all deviations were positive, hence the actual beam range was higher than simulated. The difference between the two groups (12C vs. p+) is statistically significant (p = 0.005).
Table 1 Patients’ characteristics. ID
Particle
Age during boost
Gender
Diagnosis
Site
CTV [ml]
RCT
Photon irradiation prior to boost
Single dose (ion boost)
#Fractions (boost)
Total dose
#Fields (boost)
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10
12
61 65 49 58 46 45 67 52 60 55
# # # # # # $ # # $
PXA GBM
p. ri. po. le. f. ri. t. le. p. le. c. ri. f. le. t. ri. f. le. t. le.
60.17 94.61 26.44 65.55 73.55 19.13 94.18 49.76 115.61 106.80
Primary Primary Adjuvant Primary Primary Adjuvant Adjuvant Adjuvant Adjuvant Adjuvant
25 2 Gy
3 GyE
6
18 GyE + 50 Gy
1 1 2 2 2 1 2 1 2 2
P11 P12 P13 P14 P15 P16 P17 P18 P19 P20
p+
64 69 61 61 51 61 32 62 59 43
$ $ # $ # # # # # #
GBM
fp. le. c. ri. ph. ri. t. le. t. le. po. re. f. ri. bf. p. le. po. ri.
43.92 66.85 95.90 44.95 168.83 28.69 31.26 178.59 24.09 148.36
Adjuvant Primary Primary Adjuvant Adjuvant Adjuvant Adjuvant Primary Adjuvant Adjuvant
2 GyE
5
10 GyE + 50 Gy
1 2 2 2 2 2 2 2 2 1
C
AA
24 2 Gy 25 2 Gy
None
10 GyE + 48 Gy 10 GyE + 50 Gy
30
60 GyE
CTV = Clinical Target Volume, PXA = pleomorphic xanthoastrocytoma, GBM = glioblastoma, AA = anaplastic astrocytoma, ri = right, le = left, p = parietal, po = parietooccipital, f = frontal, bf = bifrontal, fp = frontoparietal, t = temporal, c = central, ph = panhemispheric.
182
PET dosimetry in particle radiotherapy
SD [mm]
a standard deviation of 3.1 mm. P15#2 is a large tumour of complex, irregular shape with a central necrosis and peripheral contrast enhancement, leading to a range deviation of 7.2 mm with a standard deviation of 5.5 mm.
0.39
2.95
Discussion
0.68
1.93
1.89
4.67
0.28
1.86
1.71
3.55
0.26
2.26
0.36
3.14
0.94
2.29
0.45
3.06
0.39
2.83
Table 2 Detailed results – range deviation m1/2 = measurement1/2. ID
Dt [min]
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 All
P13 P14 P15 P16 P17 P18 P19 P20
*
3.33 4.43 4.22 4.55 4.34 5.20 2.74 3.03 4.34 4.34 5.25 2.19 4.23 3.91 5.56 3.98 4.75 5.24 4.81 3.53
m1 vs. m2 ø range dev.[mm]*
7 7 7 6 9 7 4 5 7 9 6 4 6 6 7 6 5 4 6 7
2.10 0.99 1.74 1.04 0.73 2.28 1.24 1.26 2.77 1.47 2.00 1.91 4.56 3.32 0.19 0.23 5.73 6.40 2.95 3.73 2.33
1.71
0.74
0.60
#1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2
6 6 8 8 7 7 5 7 11 7 6 5 5 6 6 6 5 6 6 6
6.77 7.00 4.52 3.16 5.13 6.37 0.30 2.08 5.67 7.23 1.01 1.27 5.29 6.50 2.03 2.23 4.40 2.68 5.96 4.90
4.01 4.58 6.33 5.74 5.17 5.43 3.05 4.53 4.79 5.46 0.50 0.50 5.22 5.10 5.29 5.92 3.36 2.71 4.23 4.22
0.04
0.76
1.61
1.19
0.59
1.24
0.91
1.93
2.09
2.10
0.00
0.73
0.03
1.25
0.65
1.26
1.16
1.17
0.06
1.10
C
P12
SD [mm]
#1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2 #1 #2
12
P11
Simulation vs. measurement ø range dev. [mm]*
All p+
4.23
2.15
0.71
0.73
Overall (12C, p+)
3.28
2.17
0.73
0.65
All values are transformed into natural numbers.
Case reports In the following, we would like to report on 4 specific cases, which represent the overall impression. We chose P6#2 and P9#2 as representative studies of the carbon ion group, and P14#1 and P15#2 as representatives of the proton-group. While P6#2 and P14#1 show good concordance, meaning the range deviation is low (<2 mm), P9#2 and P15#2 show larger range shifts (>5 mm). In Fig. 1 we show range deviations for the two carbon ion studies, appertaining CT-images with planning-target-volume (PTV) dose coverage and correlating MRI-images. P6#2 is a small tumour with homogeneous contrast enhancement; it leads to an average deviation of 1.9 mm with a standard deviation of 2.2 mm. P9#2 is a large tumour of complex, irregular shape with a peripheral contrast enhancement and a central necrosis resulting in an average deviation of 6.4 mm with a standard deviation of 5.2 mm. (Fig. 2) displays the proton studies. P14#1 is a small tumour with homogeneous contrast enhancement and convex-shaped, clear tumour borders. It is completely surrounded by brain parenchyma. The measured average range deviation is 0.3 mm with
In the present manuscript we assessed the beam range of proton and carbon ion irradiation in 20 patients with primary glioma during 2 fractions, early and late during the irradiation course, each using a FLUKA generated simulation. In a comparison of a patient’s two measurements we found a strong consistency with an average deviation of 0.7 mm with 75% of the analyses deviating less than a millimetre; this supports the consistent quality of our methodology. No significant result was found concerning the difference between 12C- and p+-irradiated patients. The observed minor deviations could be caused by minimal movements of the patient within the mask of up to 3 mm, potential morphological changes of the tumour itself and/or the irradiated area between the measurements or by changes concerning the biological washout (due to e.g., higher blood flow). Another possible option is it just being coincidental statistical spreading of the b+activity. Frey et al. [24] discuss some uncertainties of the MLS-method dividing them in external and internal factors; the external factors address the comparability of two activity distributions including registration of the two CT-scans, while internal factors handle intrinsic events related to washout or reaction cross-sections [21]. The comparison of simulation and measurement delivered in general satisfying results. The average deviation was 3.3 mm with a significant difference between 12C- (2.3 mm) and p+-irradiated (4.2 mm) patients. Even though we analysed a quite small number of cases, we are convinced that these results are valid and representative for particle therapy in the central nervous system. Regarding the 12C-group, we believe that the results, being below the used safety margins of commonly 3-–5 mm, allow a solid prediction of the beam’s range. This confirms the role of Monte-Carlo simulation and PET monitoring in the context of 12C therapy planning. A used safety margin of 3 mm with possible individual expansion in complex cases appears to be a reasonable value to cover the majority of the irradiation cases in the future. We would like to point out, that altogether in most of the cases we received results approximating a Gaussian distribution with some outliers, and activity profiles similar to the simulated ones. Transferring that to the constructed map of the range deviation, most of the covered area presents a beam range very close to the predicted one with only a few spots of nonconformity, resulting in shifted means. As for the p+-group, the measured results differ somewhat more from the simulated ones. This might lead to the conclusion that safety margins should be more generous than in 12C treatment, which should not be overrated, hence it is a theoretical appearance induced by the simulation modelling. On the other hand, one may feel that this range verification tool is not suitable for proton irradiations, which will be determined meticulously in future analyses. Another possible explanation might be the fact that a carbon ion irradiation delivers a more favourable spatial distribution of the activity with localised maxima towards the end of range due to the 11C projectile fragments, leading to a lower range deviation. We are aware of our study’s limitations, one being a nonconsideration of general differences in the positron-induction mechanisms of protons and carbon ions, that is mainly responsible
S.P. Nischwitz et al. / Radiotherapy and Oncology 115 (2015) 179–185
183
Fig. 1. Visualisation of two exemplary carbon ion studies. (a) Low range deviation in P6#2, from left to right: sagittal CT with planned irradiation area, sagittal MR (T1w, Gd) displaying the small rather homogeneously enhancing tumour, mapping of range deviation (irradiated area) with histogram. (b) Suboptimal range deviation in P9#2, from left to right: axial CT with planned irradiation area, axial MR (T1w, Gd) displaying the large complex-shaped tumour with central necrosis, peripheral enhancement and surrounding oedema, mapping of range deviation (irradiated area) with histogram.
for the more differing results in proton irradiation; while in protons only the irradiated target emits positrons, there is an activity-induction by the projectile in carbon ions additionally [37]. Furthermore it has been shown that there is also a difference concerning the washout properties depending on the ion species [38]. Experiments have shown that the washout itself is responsible for a few millimetres inaccuracy that could be reduced by suppressing the washout processes in phantom irradiation [39]. This calls for adaptation of our simulation framework in further studies to investigate clear clinical feasibility of range verification in proton irradiated glioma [40]. Currently, we are performing such modifications at our institution, including a more differentiated structure analysis, considering the distinct elemental compositions of the particular brain matters [41]. When we take a look at our presented cases, we can derive that in homogeneous tumours, the beam range is more accurate than in inhomogeneous ones. The middle pictures in Figs. 1a and 2a show T1-weighted (T1w) MR-images with contrast media (Gd) of patients 6#2 and 14#1 which are located in the distal reading range, where the tumour can be described as evenly contrast-enhanced structures. This is reflected in the mappings to the right, where we have a depiction just as evenly plotted. The MR-images (T1w, Gd) in Figs. 1b and 2b are slices of the distal reading range of studies P9#2 and P15#2. In them, we can see that both tumours are only partially contrast enhancing. The structures appear diffuse and irregular. In the corresponding beam
range mappings, we get patchy and multi-coloured representations of the latters. Here from we draw the conclusion, that anatomical and elemental inhomogeneous structures, such as peri-/intratumoural bleedings, oedema, dense bands of connective tissue such as the cerebral falx and exogenous surgical patches, but equally washout-altering factors like junctions to the cerebrospinal fluid or blood vessels, might cause mismatches between range simulation and verification. This is especially pronounced for proton-induced activation, which is limited to target fragments and thus more sensitive to the tissue anatomical features. The statistical fluctuation of the Monte-Carlo simulation was below the voxel size that can be distinguished with MLS (<0.6 mm), hence without decisive influence on the results. Timing of a patient’s PET and number of fields did not affect our data either, as the results were comparable, regardless of those values, thus they should not play a major role, but should be taken into consideration in future investigations with a larger number of patients in each group. Another location, where range verification can also play a decisive role, is the pelvis or the sacral area. Tumours of the pelvis are a special challenge for radiotherapy, since the air-filled rectum and differing bladder volumes are potential factors in the radiation field that vary in between the planning and irradiations. The large density differences between bone and air can cause high deviations of planned and actually applied radiation. Future investigations should focus on this topic.
184
PET dosimetry in particle radiotherapy
Fig. 2. Visualisation of two exemplary proton studies. (a) Low range deviation in P14#1, from left to right: axial CT with planned irradiation area, axial MR (T1w) of small, homogeneously enhancing tumour, mapping and histogram of range deviation (irradiated area) with very low scattering around 0 mm. (b) Suboptimal range deviation in P15#2, from left to right: sagittal CT with planned irradiation area, sagittal MR (T1w, Gd) displaying a large, irregularly shaped tumour with central necrosis and peripheral contrast enhancement, corresponding patchy mapping of range deviation with histogram (irradiated area).
Conclusion Commonly used particle irradiation planning concepts display the actual dose distribution adequately. The offline PET-in vivomonitoring is a valid mean to receive data on the beam range. To confirm the same accuracy for 1H-irradiations as for 12Cirradiations, further studies are necessary and being performed. To ensure the highest possible precision, individual anatomical features need to be further investigated and considered when planning therapeutic irradiations with protons or carbon ions. Conflict of interest statement Dr. Bauer reports grants from European Union, grants from German Federal Ministry of Research and Education during the conduct of the study. Acknowledgements We would like to thank Thomas Tessonnier from the University Hospital of Heidelberg, Department of Radiation Oncology for fruitful discussion. References [1] Stupp R, Mason WP, van den Bent MJ, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 2005;352:987–96.
[2] Stupp R, Hegi ME, Mason WP, et al. Effects of radiotherapy with concomitant and adjuvant temozolomide versus radiotherapy alone on survival in glioblastoma in a randomised phase III study: 5-year analysis of the EORTCNCIC trial. Lancet Oncol 2009;10:459–66. [3] Combs SE, Kieser M, Rieken S, et al. Randomized phase II study evaluating a carbon ion boost applied after combined radiochemotherapy with temozolomide versus a proton boost after radiochemotherapy with temozolomide in patients with primary glioblastoma: the CLEOPATRA trial. BMC Cancer 2010;10:478. [4] Wilson RR. Radiological use of fast protons. Radiology 1946;47:487–91. [5] Elsässer T, Weyrather WK, Friedrich T, et al. Quantification of the relative biological effectiveness for ion beam radiotherapy: direct experimental comparison of proton and carbon ion beams and a novel approach for treatment planning. Int J Radiat Oncol Biol Phys 2010;78:1177–83. [6] Combs SE, Bohl J, Elsässer T, et al. Radiobiological evaluation and correlation with the local effect model (LEM) of carbon ion radiation therapy and temozolomide in glioblastoma cell lines. Int J Radiat Biol 2009;85:126–37. [7] Hauswald H, Rieken S, Ecker S, et al. First experiences in treatment of lowgrade glioma grade I and II with proton therapy. Radiat Oncol 2012;7:189. [8] Rieken S, Habermehl D, Haberer T, Jaekel O, Debus J, Combs SE. Proton and carbon ion radiotherapy for primary brain tumors delivered with active raster scanning at the Heidelberg Ion Therapy Center (HIT): early treatment results and study concepts. Radiat Oncol 2012;7:41. [9] Mizoe JE, Tsujii H, Hasegawa A, et al. Phase I/II clinical trial of carbon ion radiotherapy for malignant gliomas: combined X-ray radiotherapy, chemotherapy, and carbon ion radiotherapy. Int J Radiat Oncol Biol Phys 2007;69:390–6. [10] Combs SE, Bauer J, Unholtz D, et al. Monitoring of patients treated with particle therapy using positron-emission-tomography (PET): the MIRANDA study. BMC Cancer 2012;12:133. [11] Bauer J, Unholtz D, Sommerer F, et al. Implementation and initial clinical experience of offline PET/CT-based verification of scanned carbon ion treatment. Radiother Oncol 2013;107:218–26. [12] Crespo P, Shakirin G, Enghardt W. On the detector arrangement for in-beam PET for hadron therapy monitoring. Phys Med Biol 2006;51:2143–63.
S.P. Nischwitz et al. / Radiotherapy and Oncology 115 (2015) 179–185 [13] Nishio T, Miyatake A, Ogino T, Nakagawa K, Saijo N, Esumi H. The development and clinical use of a beam ON-LINE PET system mounted on a rotating gantry port in proton therapy. Int J Radiat Oncol Biol Phys 2010;76:277–86. [14] Shakirin G, Braess H, Fiedler F, et al. Implementation and workflow for PET monitoring of therapeutic ion irradiation: a comparison of in-beam, in-room, and off-line techniques. Phys Med Biol 2011;56:1281–98. [15] Schardt D, Elsässer T, Schulz-Ertner D. Heavy-ion tumor therapy: physical and radiobiological benefits. Rev Mod Phys 2010;82:383–425. [16] Nishio T, Miyatake A, Inoue K, et al. Experimental verification of proton beam monitoring in a human body by use of activity image of positron-emitting nuclei generated by nuclear fragmentation reaction. Radiol Phys Technol 2008;1:44–54. [17] Hishikawa Y, Kagawa K, Murakami M, Sakai H, Akagi T, Abe M. Usefulness of positron-emission tomographic images after proton therapy. Int J Radiat Oncol Biol Phys 2002;53:1388–91. [18] Parodi K, Paganetti H, Shih HA, et al. Patient study of in vivo verification of beam delivery and range, using positron emission tomography and computed tomography imaging after proton therapy. Int J Radiat Oncol Biol Phys 2007;68:920–34. [19] Hsi WC, Indelicato DJ, Vargas C, Duvvuri S, Li Z, Palta J. In vivo verification of proton beam path by using post-treatment PET/CT imaging. Med Phys 2009;36:4136–46. [20] Enghardt W, Parodi K, Crespo P, Fiedler F, Pawelke J, Ponisch F. Dose quantification from in-beam positron emission tomography. Radiother Oncol 2004;73:S96–8. [21] Frey K, Unholtz D, Bauer J, et al. Automation and uncertainty analysis of a method for in vivo range verification in particle therapy. Phys Med Biol 2014;59:5903–19. [22] Knopf A, Parodi K, Paganetti H, Cascio E, Bonab A, Bortfeld T. Quantitative assessment of the physical potential of proton beam range verification with PET/CT. Phys Med Biol 2008;53:4137–51. [23] Min CH, Zhu X, Winey BA, et al. Clinical application of in-room positron emission tomography for in vivo treatment monitoring in proton radiation therapy. Int J Radiat Oncol Biol Phys 2013;86:183–9. [24] Helmbrecht S, Santiago A, Enghardt W, Kuess P, Fiedler F. On the feasibility of automatic detection of range deviations from in-beam PET data. Phys Med Biol 2012;57:1387–97. [25] Kuess P, Birkfellner W, Enghardt W, Helmbrecht S, Fiedler F, Georg D. Using statistical measures for automated comparison of in-beam PET data. Med Phys 2012;39:5874–81. [26] Tsujii H, Kamada T. A review of update clinical results of carbon ion radiotherapy. Jpn J Clin Oncol 2012;42:670–85. [27] Haberer T, Becher W, Schardt D, Kraft G. Magnetic scanning system for heavy ion therapy. Nucl Instrum Meth A 1993;330:296–305.
185
[28] Battistoni G, Muraro S, Sala PR, et al. The FLUKA code: Description and benchmarking. In: Albrow M and Raja R, editors. Proceedings of the Hadronic Shower Simulation Workshop, 2006, Fermilab, 6–8 September. p. 200631–49. [29] Ferrari A, Sala PR, Fassò A, Ranft J. FLUKA: a multi-particle transport code, CERN-2005-10 (2005), INFN/TC_05/11, SLAC-R-773, 2005. [30] Schneider W, Bortfeld T, Schlegel W. Correlation between CT numbers and tissue parameters needed for Monte Carlo simulations of clinical dose distributions. Phys Med Biol 2000;45:459–78. [31] Parodi K, Ferrari A, Sommerer F, Paganetti H. Clinical CT-based calculations of dose and positron emitter distributions in proton therapy using the FLUKA Monte Carlo code. Phys Med Biol 2007;52:3369–87. [32] Helmbrecht S, Enghardt W, Parodi K, et al. Analysis of metabolic washout of positron emitters produced during carbon ion head and neck radiotherapy. Med Phys 2013;40:091918. [33] Fiedler F, Priegnitz M, Julich R, et al. In-beam PET measurements of biological halflives of 12C irradiation induced beta+-activity. Acta Oncol 2008;47:1077–86. [34] Parodi K, Bortfeld T, Haberer T. Comparison between in-beam and offline positron emission tomography imaging of proton and carbon ion therapeutic irradiation at synchrotron- and cyclotron-based facilities. Int J Radiat Oncol Biol Phys 2008;71:945–56. [35] Mizuno H, Tomitani T, Kanazawa M, et al. Washout measurement of radioisotope implanted by radioactive beams in the rabbit. Phys Med Biol 2003;48:2269–81. [36] Unholtz D, Sommerer F, Bauer J, et al. Post-therapeutical b+-activity measurements in comparison to simulations towards in-vivo verification of ion beam therapy. Poster presented at: Conference Record of the IEEE Nuclear Science Symposium and Medical Imaging Conference; Valencia, Spain. [37] Paganetti H. Range uncertainties in proton therapy and the role of Monte Carlo simulations. Phys Med Biol 2012;57:R99–R117. [38] Ammar C, Frey K, Bauer J, et al. Comparing the biological washout of beta+activity induced in mice brain after 12C-ion and proton irradiation. Phys Med Biol 2014;59:7229–44. [39] Tessonnier T, Rinaldi I, Liebl J, Parodi K. Assessment of the predictive power of PET-based range verification in ion therapy. Poster presented at: Workshop on Range Assessment and Dose Verification in Particle Therapy; 29–30 September 2014, Dresden, Germany. [40] Bauer J, Unholtz D, Kurz C, Parodi K. An experimental approach to improve the Monte Carlo modelling of offline PET/CT-imaging of positron emitters induced by scanned proton beams. Phys Med Biol 2013;58:5193–213. [41] Bauer J, Unholtz D, Sommerer F, et al. Experimental validation of simulated positron emitter yields and refined tissue classification for improved Monte Carlo modeling towards offline PET/CT-based verification of proton therapy. Poster presented at: 3rd European Workshop on Monte Carlo Treatment Planning of the European Workgroup on MCTP; Sevilla, Spain.