- Email: [email protected]
Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?
Radiotherapy and Oncology xxx (2018) xxx–xxx
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
Original article
Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect? Clarissa Gillmann a, Antony J. Lomax b, Damien C. Weber b,c, Oliver Jäkel d,e,f, Christian P. Karger d,e,⇑ a Institute of Radiology, University Hospital Erlangen, Germany; b Center for Proton Therapy, Paul Scherrer Institute, ETH Domain, PSI Villigen; c Department of Radiation Oncology, University Hospital of Bern, Switzerland; d Department of Medical Physics in Radiation Oncology, German Cancer Research Center (DKFZ), Heidelberg; e National Center for Radiation Research in Oncology (NCRO), Heidelberg Institute for Radiation Oncology (HIRO); and f Heidelberg Ion Beam Therapy Center (HIT), Heidelberg University Hospital, Germany
a r t i c l e
i n f o
Article history: Received 7 August 2017 Received in revised form 22 January 2018 Accepted 29 January 2018 Available online xxxx Keywords: Proton and carbon ion radiotherapy Relative biological effectiveness Local effect model Temporal lobe tolerance dose
a b s t r a c t Purpose: To derive the dose–response curve for temporal lobe reactions (TLRs) after proton therapy and to compare the resulting relative biological effectiveness (RBE)-weighted tolerance doses based on an RBE of 1.1 with published values for carbon ions, which were calculated by the two versions of the local effect model (LEM I or IV). Methods and materials: 62 patients treated with protons for skull base tumors were analyzed for TLRs using magnetic resonance imaging. Within the mean follow-up time of 38 months, TLRs were observed in six patients. Dose–response curves based on the RBE-weighted maximum dose, excluding the 1 cm3volume with the highest dose, were derived and compared to previously published dose–response curves for carbon ions, which were obtained using LEM I or IV, respectively. Results: The dose–response curves for protons and LEM I were found to be almost identical while the curve of LEM IV was shifted toward higher doses. The resulting tolerance doses at the 5% effect level were 3:9 5:0 68:25:6 þ2:7 , 68:6þ3:0 and 78:3þ3:8 Gy (RBE), respectively. Conclusions: The RBE-weighted dose prescription for protons leads to the same RBE-weighted dose–response curve for TLR as the one for LEM I-based carbon ions, while LEM IV predicts clinically significant higher tolerance doses. Ó 2018 Elsevier B.V. All rights reserved. Radiotherapy and Oncology xxx (2018) xxx–xxx
Radiation therapy of skull-base tumors with protons and carbon ions has gained increasing interest throughout the last two decades. One major advantage of particle irradiations over photon treatments is the higher degree of target conformity and in the case of carbon ions, the increased relative biological effectiveness (RBE), defined as the ratio of a photon and an isoeffective ion dose [1]. The RBE of carbon ions is highly variable and rises with increasing penetration depth of the ions. Therefore, carbon ion therapy uses detailed biophysical models, such as the mixed beam model [2], the local effect model (LEM) [3] or the microdosimetric kinetic model [4] to consider these dependencies. In contrast, the RBE of protons varies less over the irradiated volume, however, recent in vitro measurements demonstrated a significant increase at the distal edge of the treatment field, where RBE-values of more than 2 were found [5–7]. In spite of these experimental findings, a
⇑ Corresponding author at: Department of Medical Physics in Radiation Oncology, German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. E-mail address: [email protected] (C.P. Karger).
fixed RBE of 1.1 is still used for patient treatments and a discussion whether this is appropriate is ongoing [8–11]. In the previous carbon ion therapy project at the Helmholtz Center for Heavy Ion Research (GSI) and at the Heidelberg Ion Beam Therapy Center (HIT), the local effect model in its initial version (LEM I) [3] has been applied up to now. Recently, a further developed version (LEM IV) [12] became also available. Both models use the microscopic dose distribution around the ion track together with the photon survival curve to calculate the RBE [3]. At present, it is an open question which LEM-version predicts the RBE more accurately in patients. Preclinical in vivo testing of both models in animals suggests a better agreement of LEM IV in the high- but not in the low- linear energy transfer (LET) region [13–15], but the available data do not allow a definitive assessment and further investigations are required. In parallel to these experimental studies, clinical investigations in a very homogeneous collective of chordoma and chondrosarcoma patients, which were treated within a dose escalation phase at GSI, were performed [16,17]. In this collective, 10 out of 59 patients experienced reactions of the temporal lobes (5 unilateral and 5 bilateral) detected
https://doi.org/10.1016/j.radonc.2018.01.018 0167-8140/Ó 2018 Elsevier B.V. All rights reserved.
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
2
TL tolerance for proton and 12C-ion RT
in contrast-enhanced T1-weighted magnetic resonance imaging (MRI) during follow-up. In a first treatment planning analysis, a dose–response curve could be established using the maximum RBE-weighted dose in the temporal lobe as dosimetric variable, without considering the 1 cm3-volume receiving the highest dose (Dmax,V-1cm3) [16]. In a second analysis, the RBE-weighted dose distributions of the patients were recalculated based on the absorbed dose distributions, now using LEM IV instead of LEM I. Again, the corresponding dose–response curve was calculated [17]. A comparison of both curves revealed a clinically significant difference of 9.5 Gy (RBE) for the tolerance dose relating to a 5 % complication probability (TD5) [17]. Ideally, in order to determine which version of the LEM predicts the RBE most accurately, a comparison of these curves with the respective dose–response curve of photon treatments is required. Unfortunately, no such photon data are currently available and tolerance data from the Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) study [18] did not help with this respect due to the large confidence interval resulting from the analysis of the prescribed, rather than the local dose, in the underlying patient collectives. Instead, in the present study, we analyze the dose-volume histograms of a patient collective with skull base tumors that was treated with protons at the Paul Scherrer Institute (PSI), using exactly the same methodologies and dosimetric variables as in the previous studies [16,17]. The patient collective and the experienced side effects have been reported in a previous study [19], however, this study focused on the correlation of dose volume parameters with toxicity grade rather than on establishing a dose–response curve. Assuming current clinical practice of prescribing a fixed proton RBE of 1.1 over the whole irradiated volume, the derived dose–response curve for protons is used to assess the prescribed carbon ion doses when using the two LEM-versions. Materials and methods Details of patient characteristics as well as the proton treatments and the dose–response analysis have been described previously [16,17,19] and therefore only a brief description is given here.
Patient treatments and follow-up In total, 62 patients with chordomas or chondrosarcomas of the skull base were initially included in the study. One patient that was included in the previous analysis of Pehlivan et al. [19] was excluded here, as it retrospectively turned out that the MRI changes were related to histologically proven tumor progression rather than changes in the normal brain. Patients were treated using scanned pencil beam proton therapy at the Paul Scherrer Institute (PSI), Villigen, Switzerland between 1998 and 2005 [19]. Treatment planning as well as the analysis in this study was performed using a fixed RBE of 1.1 (relative to 60Co) [20], which is the current standard in all proton
therapy centers. LET-variation was not considered. Median prescribed dose to the planning target volume (PTV) was 71.7 Gy (RBE) (range, 63–74 Gy (RBE)). Treatments were delivered in 33 to 39 fractions with 1.8–2.7 Gy (RBE) per fraction, although no patients in this series received this highest fraction dose for more than three fractions of their total treatment. After completion of radiation therapy, all patients underwent regular clinical follow-up examinations as detailed previously [21]. Mean follow-up time was 38 months (range, 14–92). For the purpose of this study, local contrast enhancement in the temporal lobe detected on T1-weighted post-gadolinium MR-images was considered as clinical endpoint and is referred to as temporal lobe reaction (TLR) in the following. TLR were detected in six (9.7%) patients (3 unilateral and 3 bilateral). TLR were classified according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE, v3.0) grading system [22]. Four patients (3 unilateral, 1 bilateral) were diagnosed with asymptomatic TLR, that are not interfering with activities of daily life. Two patients (2 bilateral) had clinical symptoms of impaired short term memory and disorientation and were thus diagnosed with Grade 3 [19]. For the purpose of this analysis, TLR (independent of toxicity grade) was used as clinical endpoint and reactions of both temporal lobes were considered as independent events resulting in an incidence of 9 events out of 122 temporal lobes (7.4%). Table 1 summarizes patient characteristics and treatment related parameters for the proton treatments in comparison to the previously reported carbon ion treatments.
Dose–response analysis While the clinical impact of different dose-volume parameters has been investigated previously [19], this study focuses on a dose–response analysis. The evaluation was based on the RBEweighted dose volume histograms (DVH) of the temporal lobes. From the DVH, we selected the maximum dose excluding the 1 cm3 volume with the highest dose (Dmax,V-1cm3) as independent variable, since a Principle Component Analysis (PCA) performed in our previous study [16] had identified this parameter to correlate best with the occurrence of TLR. Using this parameter, we calculated the dose–response curve for the proton cohort. As patients were treated with different fractionation schedules and since the fractional dose was changed during the treatment for some patients, the DVHs had to be determined for a standard fraction size of 2 Gy (RBE) for all fractions. This was done using the linear-quadratic model (LQM) [23] by (i) recalculating the fractional dose distribution to a treatment with 2 Gy (RBE) (see Appendix 1), (ii) adding up the resulting fractional dose distributions, and (iii) generating a new DVH, now referring to a 2 Gy (RBE) per fraction treatment. To be consistent with our previous analysis of TLR, an a=b-value of 2 Gy was used as input for the LQM-calculation [16,17]. As a=b = 3 Gy might be considered as more appropriate for brain tissue, the conversion of the actual fractionation schedule to the standard 2 Gy (RBE) fractionation schedule was repeated for protons as well as for carbon ions using this value.
Table 1 Characteristics of analyzed patients.
Number of patients (chordoma/chondrosarcoma) Mean age [years] Median prescribed dose [Gy (RBE)] Median prescribed dose rescaled to 2 Gy per fraction [Gy (RBE)] Mean follow-up time [months] Number of patients with TLR [%] (unilateral/bilateral) Overall TLR incidence
Protons (PSI)
Carbon ions (GSI)
61 (40/21) 45 (12–74) 71.7 (63–74) 74 (53.9–75.5) 38 (14–92) 6 [9.7%] (3/3) 9 out of 122
59 (40/19) 50 (16–79) 60 (60–75) LEM I: 75 (75–82.5), LEM IV: 81.8 (81.8–89.3) 34 (4–79) 10 [16.9%] (5/5) 15 out of 118
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
C. Gillmann et al. / Radiotherapy and Oncology xxx (2018) xxx–xxx
Each temporal lobe was classified according to the occurrence or non-occurrence of TLR respectively. Using this binary classification, a logistic dose–response model was adjusted using the converted Dmax,V-1cm3 as an independent dosimetric variable. From the resulting dose–response curve, tolerance doses (TD) were determined for different effect probabilities. For more details on the analysis, the reader is referred to Schlampp et al. [16]. Statistics The logistic response model was adjusted with the statistic package GNU R [24] using a maximum-likelihood fitting procedure. Results Fig. 1 displays an MR-image of a patient exhibiting a TLR. The distributions of Dmax,V-1cm3, for the clinically applied proton and carbon ion (LEM I) treatment plans as well as for the recalculated LEM IV-dose distributions are shown in Fig. 2. While the distributions are well comparable for patients without TLR, higher median doses were found for LEM I and especially for LEM IV for patients with TLR. The corresponding dose–response curves for radiationinduced TLR with Dmax,V-1cm3 as independent dosimetric variable are shown in Fig. 3. It can be seen that the dose–response curves for protons and LEM I-based carbon ions are almost identical, while the curve for LEM IV is shifted toward higher doses by almost 10 Gy (RBE). The RBE-weighted tolerance doses of protons as well as LEM I- and LEM IV-based carbon ions are summarized in Table 2 for different effect levels. For a/b = 2 Gy (RBE), the value of TD5 for protons (68:25:6 þ2:7 Gy (RBE)) is almost identical to that of LEM I-based
carbon ions (68:63:9 þ3:0 Gy (RBE)), but markedly lower than for LEM
IV-based carbon ions (78:35:0 þ3:8 Gy (RBE)). Similarly, TD50 increases
2:5 from 86:23:9 þ14:8 (protons) and 87:0þ3:2 Gy (RBE) (LEM I-based carbon
ions) to 99:83:0 þ3:7 Gy (RBE) for LEM IV-based carbon ions. Tolerance doses of protons and LEM I-based carbon ions differ at most by 1.2 Gy (RBE) at any effect level, while carbon ion tolerance doses of LEM IV are generally increased by 10–15 Gy (RBE). Using a/b = 3 Gy (RBE) instead of 2 Gy (RBE), the tolerance doses for protons remained nearly the same (mean shift 0.7%, range 1.6% to 0.7%), while the shift for carbon ions was slightly larger (-5.1% (range 6.1% to 3.9%) and 5.1% (range 6.0% to 3.8%) for LEM I and LEM IV, respectively.
3
Discussion For ion beam radiotherapy, the calculation of the RBE-weighted dose generally depends on the RBE-model used in treatment planning. Several models have been proposed for proton [25–27] and carbon ion [2–4] beams. While the clinical necessity of using a variable RBE for protons is still under discussion, the RBE-models for carbon ions are clinically employed despite their limitations. With respect to the LEM [12,16], the situation is further complicated by the fact that the two versions I and IV differ not only in the model design, but also use different input values, two aspects, which cannot be separated by our analysis. These parameter values are therefore intrinsically included in the presented analysis. In previous studies, we established dose–response curves for TLR for LEM I- and LEM IV-based RBE-weighted dose distributions and found a clinically significant difference of 9.5 Gy (RBE) for the tolerance dose TD5 [17]. Unfortunately, as no detailed data are available for dose–response of the temporal lobe after photon treatments, a direct comparison of the LEM prediction to photon tolerance doses is not possible. Although some information on photon tolerance doses has been published in the literature, the uncertainties are too large to allow a definitive answer to the question which version of the LEM predicts the RBE for TLR more accurately (for a detailed discussion of the available photon data, see Gillmann et al. [17]). In the present analysis, we have established a dose–response curve for TLR after proton irradiation with the intention of comparing the response curves of the LEM I and LEM IV models with those resulting from proton irradiations using a fixed RBE of 1.1. The strength of the present study is that it uses the same methodologies with the same dosimetric variable in comparable patient cohorts. Nevertheless, the analysis involves some uncertainties, which are discussed below. The tolerance doses for protons found in the present study are relatively high as compared to the data published by Emami et al. [28]. However, due to the higher degree of conformity of proton therapy, irradiated normal tissue volumes are generally reduced and it is the experience from several proton therapy groups that delivery of 70–76 Gy (RBE) is associated with a clinically acceptable percentage of brain necrosis [21,29–31]. In this context, the relatively high value of TD5 in our study appears plausible. Our comparison of the dose–response curves is based on two independent patient collectives that were treated for skull-base tumors either with protons at PSI or with carbon ions at GSI. While
Fig. 1. Co-registered contrast-enhanced T1-weighted spin-echo MR-images at 1 year (a) and 3 years (b) after completion of proton therapy. A disruption of the blood–brain barrier at the frontal part of the left temporal lobe is clearly visible.
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
4
TL tolerance for proton and 12C-ion RT
Fig. 2. Distributions of Dmax,V-1cm3 for proton- as well as LEM I- and LEM IV-based carbon ion dose distributions for temporal lobes without (no TLR) and with (TLR) temporal lobe reactions determined with a/b = 2 Gy (RBE) (Solid line: median; box: first and third quartiles; whiskers: 5% and 95% quantiles; circles: outliers).
Fig. 3. Dose–response curves for radiation-induced TLR for proton- as well as for LEM I- and LEM IV-based carbon ions. Data points and numbers indicate the incidence rates x/n (x out of n temporal lobes exhibit TLR) after proton treatments. These incident rates were determined in 10-Gy (RBE) intervals and assigned to the respective mean dose over all patients. Note that the incident rates are displayed only for visualization and that the response curves were calculated based on the individual temporal lobe doses. The incidence rates for carbon ions are shown in Gillmann et al. [17].
both cohorts are comparable in terms of the number and characteristics of patients, treatment planning approaches and clinical endpoint, the mean follow-up time was four months longer at PSI compared to GSI. This might have biased our analysis to some extent as the overall incidence of TLR could be underestimated in the GSI collective. On the other hand, experimental high LETirradiations have been reported to be associated with a reduced latency time [14,15]. This may reduce the probability of a potential bias originating from the shorter follow-up time in the GSI collective. A more detailed examination of the GSI data also showed that the observed TLR were not related to prolonged follow-up periods [16]. The last TLR in the GSI collective was observed at 25 months, which lies well within the mean follow-up time of 34 months. Thus both patient collectives appear comparable with respect to the probability of TLR detection. In our analysis, the temporal lobes of one patient were considered as being independent of each other as it seemed plausible that local dose hot spots primarily lead to local reactions. How-
ever, an interaction between the temporal lobes cannot be completely ruled-out and if there actually was such an interaction, the radiation tolerance is expected to decrease similarly for protons and carbon ion as the fraction of unilateral versus bilateral responding patients is the same in the proton and carbon ion cohorts (Table 1). For the patients with TLR, the RBE-weighted doses at the temporal lobe were lower at PSI as compared to GSI (Fig. 2) and accordingly, no data are available at the upper part of the dose–response curve. This results in a larger uncertainty for the comparison with the carbon ion curves above an effect level of 20%. Clinically, however, the tolerance doses at low effect levels are most important and at the investigated 5%-level, the uncertainty is comparable to that of carbon ions. Due to the lower doses at PSI, the temporal lobe volume treated with more than 85 Gy (RBE), V,D > 85 Gy (RBE), was zero in all patients. It was therefore not possible to relate this parameter with the effect probability and perform comparisons with our previous studies [16,17].
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
C. Gillmann et al. / Radiotherapy and Oncology xxx (2018) xxx–xxx Table 2 RBE-weighted tolerance doses with standard errors (SE) for radiation-induced TLR for proton-, LEM I- and LEM IV-based carbon ion dose distributions using a/b = 2 Gy and a/b = 3 Gy. Tolerance dose
Dmax,V-1cm3 ± 1 SE [Gy (RBE)] Protons
Carbon ions (LEM I)
Carbon ions (LEM IV)
a/b = 2 Gy 68:25:6 þ2:7 72:23:1 þ2:3
68:63:9 þ3:0
78:35:0 þ3:8
TD10 TD30
81:02:7 þ5:4 86:23:9 þ14:8 94:76:2 þ14:8 99:77:7 þ18:3 8:9 104:2þ21:7
81:72:3 þ2:6 2:5 87:0þ3:2 95:73:5 þ4:9 100:74:1 þ6:2 4:9 105:4þ7:3
93:62:9 þ3:1
TD5
68:35:2 þ2:7
65:93:5 þ2:7
75:34:7 þ3:4
TD10
72:72:9 þ2:1 80:52:5 þ5:0 85:43:6 þ8:2 93:55:8 þ13:7 98:27:1 þ17:0 8:3 102:5þ20:1
2:8 70:1þ2:3 77:72:1 þ2:3 82:52:3 þ2:9 3:1 90:2þ4:5 94:83:7 þ5:6 4:4 99:0þ6:6
80:23:6 þ2:9
TD5
TD50 TD80 TD90 TD95
73:33:1 þ2:5
83:84:0 þ3:2 99:83:0 þ3:7 109:94:0 þ5:7
4:8 115:8þ7:2 5:6 121:2þ8:6
a/b = 3 Gy
TD30 TD50 TD80 TD90 TD95
89:12:5 þ2:7 94:62:6 þ3:4
103:73:6 þ5:2 109:04:3 þ6:5
5:1 113:9þ7:8
In accordance with our previous studies [16,17], we selected Dmax,V-1cm3 as independent dosimetric parameter for the dose– response analysis. As the exact value of the tolerance doses depends on this selection, the question arises, which dosimetric variable correlates best with the occurrence of the clinical endpoint. Regarding TLR, it can be expected that its occurrence correlates with high local doses, which suggests using the maximum dose. On the other hand, excluding a small volume with the highest dose increases the robustness of the analysis as accidentally high doses in single voxels originating from weaknesses of the dose calculation algorithm are not considered. Typically, such hotspots in ion therapy are smaller than 1 cm3 and we therefore excluded this volume with the highest dose. This is of special importance for the recalculated LEM IV-dose distributions as they are highly nonuniform as compared to those of LEM I. It has also to be noted that the use of Dmax,V-1cm3 as a single representative parameter neglects the potential influence of the irradiated volume on the response, which may be slightly different for protons and carbon ions. With this respect, however, the dose distribution of carbon ions in the temporal lobe can be expected to be much better comparable with that of protons than with that of photons. Our analysis revealed an almost perfect agreement between the dose–response curves for protons and LEM I-based carbon ions. This finding, however, relies on the assumption of a LETindependent proton RBE of 1.1. The validity of this assumption is currently being discussed and there is evidence from several in vitro experiments [5–7] as well as from a few clinical studies [10,11] that the RBE increases significantly at the distal end of proton beams. If this was confirmed also in vivo, the RBE-weighted proton doses as well as the dose–response curve may be shifted to higher doses. At present, however, there is still a lack of evidence from dedicated in vivo experiments as well as from clinical studies to definitely confirm and quantify a clinically relevant increase of the proton RBE. Indeed, the present study showed that the same RBE-weighted doses of protons and LEM I-based carbon ions lead to the same clinical effects in the temporal lobe when a global proton RBE of 1.1 is applied. In contrast, LEM IV predicts clinically significant higher tolerance doses. This finding is based on TLR as a highly relevant clinical endpoint in combination with Dmax,V-1cm3 as dosimetric parameter and is thus of great practical relevance for safe treatment delivery in carbon ion therapy. Under the
5
premise that RBE-weighted dose prescription in proton therapy is actually photon-equivalent, LEM I-based RBE-weighted doses can also be considered as approximately photon-equivalent. Finally, an intrinsic parameter of our analysis is the a=b-value, which was used to convert the irradiations to treatments with 2 Gy (RBE) per fraction. For this, we selected a=b = 2 Gy to be consistent with our previous carbon ion studies [16,17], where a=b entered not only into the conversion of fractionation schedules but also as a parameter of the LEM. However, as a=b = 3 Gy might be considered as more appropriate for brain tissue [18], we additionally converted the proton and carbon ion doses by using this a=b-value and repeated the dose response analysis. The slightly larger decrease of the tolerance doses for carbon ions as compared to protons (Table 2) can be attributed to the fact that all fractional carbon ion doses were larger than 2 Gy (RBE) while the fractional doses for protons were already 2 Gy (RBE) in most patients and in the remaining patients, some doses were larger and some were smaller. As the difference between the shifts of the proton and carbon ion tolerance doses is still small as compared to the 14%difference of the tolerance doses of LEM I and LEM IV, the selection of a=b does not decisively affect the conclusion of our study. To summarize, a dose–response curve for temporal lobe reactions after proton irradiation was established based on a fixed proton RBE of 1.1 as the current clinical standard. This curve was used as a reference to analyze whether LEM I or LEM IV predict RBE-weighted tolerance doses better. As a result, the RBEweighted dose prescription for protons leads to the same response of the temporal lobe as for LEM I-based carbon ion treatments, while LEM IV predicts 10–15 Gy (RBE) higher tolerance doses. Further studies in larger patient cohorts are necessary to investigate the spatial dependence of the proton RBE, the accuracy of related models as well as the impact on the comparison of RBE-weighted tolerance doses for protons with those of carbon ions.
Conflict of interest None.
Appendix 1. To convert the delivered RBE-weighted fractional dose distribution, D1 ðx; y; zÞ, into that of a treatment with 2 Gy (RBE), D2 ðx; y; zÞ, we used the isoeffect relation
D2 ðx; y; zÞ ¼ D1 ðx; y; zÞ
ða=b þ d1 Þ ; ða=b þ d2 Þ
ð1Þ
obtained from equating the survival fractions of the LQM (23). In Eq. (1), d1 , and d2 are the RBE-weighted doses prescribed per fraction at the actual and the 2 Gy (RBE) treatment, respectively. The a=b-ratio refers to photons and was set to either 2 or 3 Gy. For the new dose distribution, D2 ðx; y; zÞ, the local TL-dose was determined from the DVH. References [1] Karger CP, Peschke P. RBE and related modeling in carbon-ion therapy. Phys Med Biol 2017. [2] Kanai T, Furusawa Y, Fukutsu K, Itsukaichi H, Eguchi-Kasai K, Ohara H. Irradiation of mixed beam and design of spread-out Bragg peak for heavy-ion radiotherapy. Radiat Res 1997;147:78–85. [3] Scholz M, Kellerer AM, Kraft-Weyrather W, Kraft G. Computation of cell survival in heavy ion beams for therapy. Radiat Environ Biophys 1997;36:59–66. [4] Kase Y, Kanai T, Matsumoto Y, Furusawa Y, Okamoto H, Asaba T, et al. Microdosimetric measurements and estimation of human cell survival for heavy-ion beams. Radiat Res 2006;166:629–38.
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
6
TL tolerance for proton and 12C-ion RT
[5] Britten RA, Nazaryan V, Davis LK, Klein SB, Nichiporov D, Mendonca MS, et al. Variations in the RBE for cell killing along the depth-dose profile of a modulated proton therapy beam. Radiat Res 2013;179:21–8. [6] Guan F, Bronk L, Titt U, Lin SH, Mirkovic D, Kerr MD, et al. Spatial mapping of the biologic effectiveness of scanned particle beams: towards biologically optimized particle therapy. Sci Rep 2015;18:9850. [7] Patel D, Bronk L, Guan F, Peeler CR, Brons S, Dokic I, et al. Optimization of Monte Carlo particle transport parameters and validation of a novel high throughput experimental setup to measure the biological effects of particle beams. Med Phys 2017;44:6061–73. [8] Paganetti H. Significance and implementation of RBE variations in proton beam therapy. Technol Cancer Res Treat 2003;2:413–26. [9] Paganetti H. Relative biological effectiveness (RBE) values for proton beam therapy. Variations as a function of biological endpoint, dose, and linear energy transfer. Phys Med Biol 2014;59:R419–72. [10] Peeler CR, Mirkovic D, Titt U, Blanchard P, Gunther JR, Mahajan A, et al. Clinical evidence of variable proton biological effectiveness in pediatric patients treated for ependymoma. Radiother Oncol 2016;121:395–401. [11] Buchsbaum JC. Are treatment toxicity issues in particle therapy a clarion call for biologic treatment planning overall? Int J Radiat Oncol Biol Phys 2017;97:1085–6. [12] Elsässer T, Weyrather WK, Friedrich T, Durante M, Iancu G, Krämer M, 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. [13] Karger CP, Peschke P, Sanchez-Brandelik R, Scholz M, Debus J. Radiation tolerance of the rat spinal cord after 6 and 18 fractions of photons and carbon ions: experimental results and clinical implications. Int J Radiat Oncol Biol Phys 2006;66:1488–97. [14] Saager M, Glowa C, Peschke P, Brons S, Grün R, Scholz M, et al. Split dose carbon ion irradiation of the rat spinal cord: dependence of the relative biological effectiveness on dose and linear energy transfer. Radiother Oncol 2015;117:358–63. [15] Saager M, Glowa C, Peschke P, Brons S, Grün R, Scholz M, et al. The relative biological effectiveness of carbon ion irradiations of the rat spinal cord increases linearly with LET up to 99 keV/lm. Acta Oncol 2016;55:1512–5. [16] Schlampp I, Karger CP, Jäkel O, Scholz M, Didinger B, Nikoghosyan A, et al. Temporal lobe reactions after radiotherapy with carbon ions: incidence and estimation of the relative biological effectiveness by the local effect model. Int J Radiat Oncol Biol Phys 2011;80:815–23. [17] Gillmann C, Jäkel O, Schlampp I, Karger CP. Temporal lobe reactions after carbon ion radiation therapy: comparison of relative biological effectiveness-
[18] [19]
[20]
[21]
[22]
[23] [24] [25]
[26]
[27]
[28]
[29] [30] [31]
weighted tolerance doses predicted by local effect models I and IV. Int J Radiat Oncol Biol Phys 2014;88:1136–41. Lawrence YR, Li XA, el Naqa I, Hahn CA, Marks LB, Merchant TE, et al. Radiation dose-volume effects in the brain. Int J Radiat Oncol Biol Phys 2010;76:S20–7. Pehlivan B, Ares C, Lomax AJ, Stadelmann O, Goitein G, Timmermann B, et al. Temporal lobe toxicity analysis after proton radiation therapy for skull base tumors. Int J Radiat Oncol Biol Phys 2012;83:1432–40. Paganetti H, Niemierko A, Ancukiewicz M, Gerweck LE, Goitein M, Loeffler JS, et al. Relative biological effectiveness (RBE) values for proton beam therapy. Int J Radiat Oncol Biol Phys 2002;53:407–21. Weber DC, Malyapa R, Albertini F, Bolsi A, Kliebsch U, Walser M, et al. Long term outcomes of patients with skull-base low-grade chondrosarcoma and chordoma patients treated with pencil beam scanning proton therapy. Radiother Oncol 2016;120:169–74. National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE, V3.0) grading system. Available at: http://www.eortc.be/services/doc/ ctc/czcaev3.pdf. Accessed Dec. 17, 2001. In. Fowler JF. The linear-quadratic formula and progress in fractionated radiotherapy. Br J Radiol 1989;62:679–94. The R project for Statistical Computing [Internet, cited 2013 Mar 28]. Available from http://www.r-project.org/. Carabe A, Moteabbed M, Depauw N, Schuemann J, Paganetti H. Range uncertainty in proton therapy due to variable biological effectiveness. Phys Med Biol. 2012;57:1159–72. Wedenberg M, Lind BK, Hårdemark B. A model for the relative biological effectiveness of protons: the tissue specific parameter a/b of photons is a predictor for the sensitivity to LET changes. Acta Oncol 2013;52:580–8. McNamara AL, Schuemann J, Paganetti H. A phenomenological relative biological effectiveness (RBE) model for proton therapy based on all published in vitro cell survival data. Phys Med Biol 2015;60:8399–416. Emami B, Lyman J, Brown A, Coia L, Goitein M, Munzenrider JE, et al. Tolerance of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys 1991;21:109–22. Munzenrider JE, Liebsch NJ. Proton therapy for tumors of the skull base. Strahlenther Onkol 1999;175:57–63. Hug EB, Slater JD. Proton radiation therapy for chordomas and chondrosarcomas of the skull base. Neurosurg Clin N Am 2000;11:627–38. Noël G, Habrand JL, Mammar H, Pontvert D, Haie-Méder C, Hasboun D, et al. Combination of photon and proton radiation therapy for chordomas and chondrosarcomas of the skull base: the Centre de Protonthérapie D’Orsay experience. Int J Radiat Oncol Biol Phys 2001;51:392–8.
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
Original article
Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect? Clarissa Gillmann a, Antony J. Lomax b, Damien C. Weber b,c, Oliver Jäkel d,e,f, Christian P. Karger d,e,⇑ a Institute of Radiology, University Hospital Erlangen, Germany; b Center for Proton Therapy, Paul Scherrer Institute, ETH Domain, PSI Villigen; c Department of Radiation Oncology, University Hospital of Bern, Switzerland; d Department of Medical Physics in Radiation Oncology, German Cancer Research Center (DKFZ), Heidelberg; e National Center for Radiation Research in Oncology (NCRO), Heidelberg Institute for Radiation Oncology (HIRO); and f Heidelberg Ion Beam Therapy Center (HIT), Heidelberg University Hospital, Germany
a r t i c l e
i n f o
Article history: Received 7 August 2017 Received in revised form 22 January 2018 Accepted 29 January 2018 Available online xxxx Keywords: Proton and carbon ion radiotherapy Relative biological effectiveness Local effect model Temporal lobe tolerance dose
a b s t r a c t Purpose: To derive the dose–response curve for temporal lobe reactions (TLRs) after proton therapy and to compare the resulting relative biological effectiveness (RBE)-weighted tolerance doses based on an RBE of 1.1 with published values for carbon ions, which were calculated by the two versions of the local effect model (LEM I or IV). Methods and materials: 62 patients treated with protons for skull base tumors were analyzed for TLRs using magnetic resonance imaging. Within the mean follow-up time of 38 months, TLRs were observed in six patients. Dose–response curves based on the RBE-weighted maximum dose, excluding the 1 cm3volume with the highest dose, were derived and compared to previously published dose–response curves for carbon ions, which were obtained using LEM I or IV, respectively. Results: The dose–response curves for protons and LEM I were found to be almost identical while the curve of LEM IV was shifted toward higher doses. The resulting tolerance doses at the 5% effect level were 3:9 5:0 68:25:6 þ2:7 , 68:6þ3:0 and 78:3þ3:8 Gy (RBE), respectively. Conclusions: The RBE-weighted dose prescription for protons leads to the same RBE-weighted dose–response curve for TLR as the one for LEM I-based carbon ions, while LEM IV predicts clinically significant higher tolerance doses. Ó 2018 Elsevier B.V. All rights reserved. Radiotherapy and Oncology xxx (2018) xxx–xxx
Radiation therapy of skull-base tumors with protons and carbon ions has gained increasing interest throughout the last two decades. One major advantage of particle irradiations over photon treatments is the higher degree of target conformity and in the case of carbon ions, the increased relative biological effectiveness (RBE), defined as the ratio of a photon and an isoeffective ion dose [1]. The RBE of carbon ions is highly variable and rises with increasing penetration depth of the ions. Therefore, carbon ion therapy uses detailed biophysical models, such as the mixed beam model [2], the local effect model (LEM) [3] or the microdosimetric kinetic model [4] to consider these dependencies. In contrast, the RBE of protons varies less over the irradiated volume, however, recent in vitro measurements demonstrated a significant increase at the distal edge of the treatment field, where RBE-values of more than 2 were found [5–7]. In spite of these experimental findings, a
⇑ Corresponding author at: Department of Medical Physics in Radiation Oncology, German Cancer Research Center (DKFZ), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. E-mail address: [email protected] (C.P. Karger).
fixed RBE of 1.1 is still used for patient treatments and a discussion whether this is appropriate is ongoing [8–11]. In the previous carbon ion therapy project at the Helmholtz Center for Heavy Ion Research (GSI) and at the Heidelberg Ion Beam Therapy Center (HIT), the local effect model in its initial version (LEM I) [3] has been applied up to now. Recently, a further developed version (LEM IV) [12] became also available. Both models use the microscopic dose distribution around the ion track together with the photon survival curve to calculate the RBE [3]. At present, it is an open question which LEM-version predicts the RBE more accurately in patients. Preclinical in vivo testing of both models in animals suggests a better agreement of LEM IV in the high- but not in the low- linear energy transfer (LET) region [13–15], but the available data do not allow a definitive assessment and further investigations are required. In parallel to these experimental studies, clinical investigations in a very homogeneous collective of chordoma and chondrosarcoma patients, which were treated within a dose escalation phase at GSI, were performed [16,17]. In this collective, 10 out of 59 patients experienced reactions of the temporal lobes (5 unilateral and 5 bilateral) detected
https://doi.org/10.1016/j.radonc.2018.01.018 0167-8140/Ó 2018 Elsevier B.V. All rights reserved.
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
2
TL tolerance for proton and 12C-ion RT
in contrast-enhanced T1-weighted magnetic resonance imaging (MRI) during follow-up. In a first treatment planning analysis, a dose–response curve could be established using the maximum RBE-weighted dose in the temporal lobe as dosimetric variable, without considering the 1 cm3-volume receiving the highest dose (Dmax,V-1cm3) [16]. In a second analysis, the RBE-weighted dose distributions of the patients were recalculated based on the absorbed dose distributions, now using LEM IV instead of LEM I. Again, the corresponding dose–response curve was calculated [17]. A comparison of both curves revealed a clinically significant difference of 9.5 Gy (RBE) for the tolerance dose relating to a 5 % complication probability (TD5) [17]. Ideally, in order to determine which version of the LEM predicts the RBE most accurately, a comparison of these curves with the respective dose–response curve of photon treatments is required. Unfortunately, no such photon data are currently available and tolerance data from the Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) study [18] did not help with this respect due to the large confidence interval resulting from the analysis of the prescribed, rather than the local dose, in the underlying patient collectives. Instead, in the present study, we analyze the dose-volume histograms of a patient collective with skull base tumors that was treated with protons at the Paul Scherrer Institute (PSI), using exactly the same methodologies and dosimetric variables as in the previous studies [16,17]. The patient collective and the experienced side effects have been reported in a previous study [19], however, this study focused on the correlation of dose volume parameters with toxicity grade rather than on establishing a dose–response curve. Assuming current clinical practice of prescribing a fixed proton RBE of 1.1 over the whole irradiated volume, the derived dose–response curve for protons is used to assess the prescribed carbon ion doses when using the two LEM-versions. Materials and methods Details of patient characteristics as well as the proton treatments and the dose–response analysis have been described previously [16,17,19] and therefore only a brief description is given here.
Patient treatments and follow-up In total, 62 patients with chordomas or chondrosarcomas of the skull base were initially included in the study. One patient that was included in the previous analysis of Pehlivan et al. [19] was excluded here, as it retrospectively turned out that the MRI changes were related to histologically proven tumor progression rather than changes in the normal brain. Patients were treated using scanned pencil beam proton therapy at the Paul Scherrer Institute (PSI), Villigen, Switzerland between 1998 and 2005 [19]. Treatment planning as well as the analysis in this study was performed using a fixed RBE of 1.1 (relative to 60Co) [20], which is the current standard in all proton
therapy centers. LET-variation was not considered. Median prescribed dose to the planning target volume (PTV) was 71.7 Gy (RBE) (range, 63–74 Gy (RBE)). Treatments were delivered in 33 to 39 fractions with 1.8–2.7 Gy (RBE) per fraction, although no patients in this series received this highest fraction dose for more than three fractions of their total treatment. After completion of radiation therapy, all patients underwent regular clinical follow-up examinations as detailed previously [21]. Mean follow-up time was 38 months (range, 14–92). For the purpose of this study, local contrast enhancement in the temporal lobe detected on T1-weighted post-gadolinium MR-images was considered as clinical endpoint and is referred to as temporal lobe reaction (TLR) in the following. TLR were detected in six (9.7%) patients (3 unilateral and 3 bilateral). TLR were classified according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE, v3.0) grading system [22]. Four patients (3 unilateral, 1 bilateral) were diagnosed with asymptomatic TLR, that are not interfering with activities of daily life. Two patients (2 bilateral) had clinical symptoms of impaired short term memory and disorientation and were thus diagnosed with Grade 3 [19]. For the purpose of this analysis, TLR (independent of toxicity grade) was used as clinical endpoint and reactions of both temporal lobes were considered as independent events resulting in an incidence of 9 events out of 122 temporal lobes (7.4%). Table 1 summarizes patient characteristics and treatment related parameters for the proton treatments in comparison to the previously reported carbon ion treatments.
Dose–response analysis While the clinical impact of different dose-volume parameters has been investigated previously [19], this study focuses on a dose–response analysis. The evaluation was based on the RBEweighted dose volume histograms (DVH) of the temporal lobes. From the DVH, we selected the maximum dose excluding the 1 cm3 volume with the highest dose (Dmax,V-1cm3) as independent variable, since a Principle Component Analysis (PCA) performed in our previous study [16] had identified this parameter to correlate best with the occurrence of TLR. Using this parameter, we calculated the dose–response curve for the proton cohort. As patients were treated with different fractionation schedules and since the fractional dose was changed during the treatment for some patients, the DVHs had to be determined for a standard fraction size of 2 Gy (RBE) for all fractions. This was done using the linear-quadratic model (LQM) [23] by (i) recalculating the fractional dose distribution to a treatment with 2 Gy (RBE) (see Appendix 1), (ii) adding up the resulting fractional dose distributions, and (iii) generating a new DVH, now referring to a 2 Gy (RBE) per fraction treatment. To be consistent with our previous analysis of TLR, an a=b-value of 2 Gy was used as input for the LQM-calculation [16,17]. As a=b = 3 Gy might be considered as more appropriate for brain tissue, the conversion of the actual fractionation schedule to the standard 2 Gy (RBE) fractionation schedule was repeated for protons as well as for carbon ions using this value.
Table 1 Characteristics of analyzed patients.
Number of patients (chordoma/chondrosarcoma) Mean age [years] Median prescribed dose [Gy (RBE)] Median prescribed dose rescaled to 2 Gy per fraction [Gy (RBE)] Mean follow-up time [months] Number of patients with TLR [%] (unilateral/bilateral) Overall TLR incidence
Protons (PSI)
Carbon ions (GSI)
61 (40/21) 45 (12–74) 71.7 (63–74) 74 (53.9–75.5) 38 (14–92) 6 [9.7%] (3/3) 9 out of 122
59 (40/19) 50 (16–79) 60 (60–75) LEM I: 75 (75–82.5), LEM IV: 81.8 (81.8–89.3) 34 (4–79) 10 [16.9%] (5/5) 15 out of 118
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
C. Gillmann et al. / Radiotherapy and Oncology xxx (2018) xxx–xxx
Each temporal lobe was classified according to the occurrence or non-occurrence of TLR respectively. Using this binary classification, a logistic dose–response model was adjusted using the converted Dmax,V-1cm3 as an independent dosimetric variable. From the resulting dose–response curve, tolerance doses (TD) were determined for different effect probabilities. For more details on the analysis, the reader is referred to Schlampp et al. [16]. Statistics The logistic response model was adjusted with the statistic package GNU R [24] using a maximum-likelihood fitting procedure. Results Fig. 1 displays an MR-image of a patient exhibiting a TLR. The distributions of Dmax,V-1cm3, for the clinically applied proton and carbon ion (LEM I) treatment plans as well as for the recalculated LEM IV-dose distributions are shown in Fig. 2. While the distributions are well comparable for patients without TLR, higher median doses were found for LEM I and especially for LEM IV for patients with TLR. The corresponding dose–response curves for radiationinduced TLR with Dmax,V-1cm3 as independent dosimetric variable are shown in Fig. 3. It can be seen that the dose–response curves for protons and LEM I-based carbon ions are almost identical, while the curve for LEM IV is shifted toward higher doses by almost 10 Gy (RBE). The RBE-weighted tolerance doses of protons as well as LEM I- and LEM IV-based carbon ions are summarized in Table 2 for different effect levels. For a/b = 2 Gy (RBE), the value of TD5 for protons (68:25:6 þ2:7 Gy (RBE)) is almost identical to that of LEM I-based
carbon ions (68:63:9 þ3:0 Gy (RBE)), but markedly lower than for LEM
IV-based carbon ions (78:35:0 þ3:8 Gy (RBE)). Similarly, TD50 increases
2:5 from 86:23:9 þ14:8 (protons) and 87:0þ3:2 Gy (RBE) (LEM I-based carbon
ions) to 99:83:0 þ3:7 Gy (RBE) for LEM IV-based carbon ions. Tolerance doses of protons and LEM I-based carbon ions differ at most by 1.2 Gy (RBE) at any effect level, while carbon ion tolerance doses of LEM IV are generally increased by 10–15 Gy (RBE). Using a/b = 3 Gy (RBE) instead of 2 Gy (RBE), the tolerance doses for protons remained nearly the same (mean shift 0.7%, range 1.6% to 0.7%), while the shift for carbon ions was slightly larger (-5.1% (range 6.1% to 3.9%) and 5.1% (range 6.0% to 3.8%) for LEM I and LEM IV, respectively.
3
Discussion For ion beam radiotherapy, the calculation of the RBE-weighted dose generally depends on the RBE-model used in treatment planning. Several models have been proposed for proton [25–27] and carbon ion [2–4] beams. While the clinical necessity of using a variable RBE for protons is still under discussion, the RBE-models for carbon ions are clinically employed despite their limitations. With respect to the LEM [12,16], the situation is further complicated by the fact that the two versions I and IV differ not only in the model design, but also use different input values, two aspects, which cannot be separated by our analysis. These parameter values are therefore intrinsically included in the presented analysis. In previous studies, we established dose–response curves for TLR for LEM I- and LEM IV-based RBE-weighted dose distributions and found a clinically significant difference of 9.5 Gy (RBE) for the tolerance dose TD5 [17]. Unfortunately, as no detailed data are available for dose–response of the temporal lobe after photon treatments, a direct comparison of the LEM prediction to photon tolerance doses is not possible. Although some information on photon tolerance doses has been published in the literature, the uncertainties are too large to allow a definitive answer to the question which version of the LEM predicts the RBE for TLR more accurately (for a detailed discussion of the available photon data, see Gillmann et al. [17]). In the present analysis, we have established a dose–response curve for TLR after proton irradiation with the intention of comparing the response curves of the LEM I and LEM IV models with those resulting from proton irradiations using a fixed RBE of 1.1. The strength of the present study is that it uses the same methodologies with the same dosimetric variable in comparable patient cohorts. Nevertheless, the analysis involves some uncertainties, which are discussed below. The tolerance doses for protons found in the present study are relatively high as compared to the data published by Emami et al. [28]. However, due to the higher degree of conformity of proton therapy, irradiated normal tissue volumes are generally reduced and it is the experience from several proton therapy groups that delivery of 70–76 Gy (RBE) is associated with a clinically acceptable percentage of brain necrosis [21,29–31]. In this context, the relatively high value of TD5 in our study appears plausible. Our comparison of the dose–response curves is based on two independent patient collectives that were treated for skull-base tumors either with protons at PSI or with carbon ions at GSI. While
Fig. 1. Co-registered contrast-enhanced T1-weighted spin-echo MR-images at 1 year (a) and 3 years (b) after completion of proton therapy. A disruption of the blood–brain barrier at the frontal part of the left temporal lobe is clearly visible.
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
4
TL tolerance for proton and 12C-ion RT
Fig. 2. Distributions of Dmax,V-1cm3 for proton- as well as LEM I- and LEM IV-based carbon ion dose distributions for temporal lobes without (no TLR) and with (TLR) temporal lobe reactions determined with a/b = 2 Gy (RBE) (Solid line: median; box: first and third quartiles; whiskers: 5% and 95% quantiles; circles: outliers).
Fig. 3. Dose–response curves for radiation-induced TLR for proton- as well as for LEM I- and LEM IV-based carbon ions. Data points and numbers indicate the incidence rates x/n (x out of n temporal lobes exhibit TLR) after proton treatments. These incident rates were determined in 10-Gy (RBE) intervals and assigned to the respective mean dose over all patients. Note that the incident rates are displayed only for visualization and that the response curves were calculated based on the individual temporal lobe doses. The incidence rates for carbon ions are shown in Gillmann et al. [17].
both cohorts are comparable in terms of the number and characteristics of patients, treatment planning approaches and clinical endpoint, the mean follow-up time was four months longer at PSI compared to GSI. This might have biased our analysis to some extent as the overall incidence of TLR could be underestimated in the GSI collective. On the other hand, experimental high LETirradiations have been reported to be associated with a reduced latency time [14,15]. This may reduce the probability of a potential bias originating from the shorter follow-up time in the GSI collective. A more detailed examination of the GSI data also showed that the observed TLR were not related to prolonged follow-up periods [16]. The last TLR in the GSI collective was observed at 25 months, which lies well within the mean follow-up time of 34 months. Thus both patient collectives appear comparable with respect to the probability of TLR detection. In our analysis, the temporal lobes of one patient were considered as being independent of each other as it seemed plausible that local dose hot spots primarily lead to local reactions. How-
ever, an interaction between the temporal lobes cannot be completely ruled-out and if there actually was such an interaction, the radiation tolerance is expected to decrease similarly for protons and carbon ion as the fraction of unilateral versus bilateral responding patients is the same in the proton and carbon ion cohorts (Table 1). For the patients with TLR, the RBE-weighted doses at the temporal lobe were lower at PSI as compared to GSI (Fig. 2) and accordingly, no data are available at the upper part of the dose–response curve. This results in a larger uncertainty for the comparison with the carbon ion curves above an effect level of 20%. Clinically, however, the tolerance doses at low effect levels are most important and at the investigated 5%-level, the uncertainty is comparable to that of carbon ions. Due to the lower doses at PSI, the temporal lobe volume treated with more than 85 Gy (RBE), V,D > 85 Gy (RBE), was zero in all patients. It was therefore not possible to relate this parameter with the effect probability and perform comparisons with our previous studies [16,17].
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
C. Gillmann et al. / Radiotherapy and Oncology xxx (2018) xxx–xxx Table 2 RBE-weighted tolerance doses with standard errors (SE) for radiation-induced TLR for proton-, LEM I- and LEM IV-based carbon ion dose distributions using a/b = 2 Gy and a/b = 3 Gy. Tolerance dose
Dmax,V-1cm3 ± 1 SE [Gy (RBE)] Protons
Carbon ions (LEM I)
Carbon ions (LEM IV)
a/b = 2 Gy 68:25:6 þ2:7 72:23:1 þ2:3
68:63:9 þ3:0
78:35:0 þ3:8
TD10 TD30
81:02:7 þ5:4 86:23:9 þ14:8 94:76:2 þ14:8 99:77:7 þ18:3 8:9 104:2þ21:7
81:72:3 þ2:6 2:5 87:0þ3:2 95:73:5 þ4:9 100:74:1 þ6:2 4:9 105:4þ7:3
93:62:9 þ3:1
TD5
68:35:2 þ2:7
65:93:5 þ2:7
75:34:7 þ3:4
TD10
72:72:9 þ2:1 80:52:5 þ5:0 85:43:6 þ8:2 93:55:8 þ13:7 98:27:1 þ17:0 8:3 102:5þ20:1
2:8 70:1þ2:3 77:72:1 þ2:3 82:52:3 þ2:9 3:1 90:2þ4:5 94:83:7 þ5:6 4:4 99:0þ6:6
80:23:6 þ2:9
TD5
TD50 TD80 TD90 TD95
73:33:1 þ2:5
83:84:0 þ3:2 99:83:0 þ3:7 109:94:0 þ5:7
4:8 115:8þ7:2 5:6 121:2þ8:6
a/b = 3 Gy
TD30 TD50 TD80 TD90 TD95
89:12:5 þ2:7 94:62:6 þ3:4
103:73:6 þ5:2 109:04:3 þ6:5
5:1 113:9þ7:8
In accordance with our previous studies [16,17], we selected Dmax,V-1cm3 as independent dosimetric parameter for the dose– response analysis. As the exact value of the tolerance doses depends on this selection, the question arises, which dosimetric variable correlates best with the occurrence of the clinical endpoint. Regarding TLR, it can be expected that its occurrence correlates with high local doses, which suggests using the maximum dose. On the other hand, excluding a small volume with the highest dose increases the robustness of the analysis as accidentally high doses in single voxels originating from weaknesses of the dose calculation algorithm are not considered. Typically, such hotspots in ion therapy are smaller than 1 cm3 and we therefore excluded this volume with the highest dose. This is of special importance for the recalculated LEM IV-dose distributions as they are highly nonuniform as compared to those of LEM I. It has also to be noted that the use of Dmax,V-1cm3 as a single representative parameter neglects the potential influence of the irradiated volume on the response, which may be slightly different for protons and carbon ions. With this respect, however, the dose distribution of carbon ions in the temporal lobe can be expected to be much better comparable with that of protons than with that of photons. Our analysis revealed an almost perfect agreement between the dose–response curves for protons and LEM I-based carbon ions. This finding, however, relies on the assumption of a LETindependent proton RBE of 1.1. The validity of this assumption is currently being discussed and there is evidence from several in vitro experiments [5–7] as well as from a few clinical studies [10,11] that the RBE increases significantly at the distal end of proton beams. If this was confirmed also in vivo, the RBE-weighted proton doses as well as the dose–response curve may be shifted to higher doses. At present, however, there is still a lack of evidence from dedicated in vivo experiments as well as from clinical studies to definitely confirm and quantify a clinically relevant increase of the proton RBE. Indeed, the present study showed that the same RBE-weighted doses of protons and LEM I-based carbon ions lead to the same clinical effects in the temporal lobe when a global proton RBE of 1.1 is applied. In contrast, LEM IV predicts clinically significant higher tolerance doses. This finding is based on TLR as a highly relevant clinical endpoint in combination with Dmax,V-1cm3 as dosimetric parameter and is thus of great practical relevance for safe treatment delivery in carbon ion therapy. Under the
5
premise that RBE-weighted dose prescription in proton therapy is actually photon-equivalent, LEM I-based RBE-weighted doses can also be considered as approximately photon-equivalent. Finally, an intrinsic parameter of our analysis is the a=b-value, which was used to convert the irradiations to treatments with 2 Gy (RBE) per fraction. For this, we selected a=b = 2 Gy to be consistent with our previous carbon ion studies [16,17], where a=b entered not only into the conversion of fractionation schedules but also as a parameter of the LEM. However, as a=b = 3 Gy might be considered as more appropriate for brain tissue [18], we additionally converted the proton and carbon ion doses by using this a=b-value and repeated the dose response analysis. The slightly larger decrease of the tolerance doses for carbon ions as compared to protons (Table 2) can be attributed to the fact that all fractional carbon ion doses were larger than 2 Gy (RBE) while the fractional doses for protons were already 2 Gy (RBE) in most patients and in the remaining patients, some doses were larger and some were smaller. As the difference between the shifts of the proton and carbon ion tolerance doses is still small as compared to the 14%difference of the tolerance doses of LEM I and LEM IV, the selection of a=b does not decisively affect the conclusion of our study. To summarize, a dose–response curve for temporal lobe reactions after proton irradiation was established based on a fixed proton RBE of 1.1 as the current clinical standard. This curve was used as a reference to analyze whether LEM I or LEM IV predict RBE-weighted tolerance doses better. As a result, the RBEweighted dose prescription for protons leads to the same response of the temporal lobe as for LEM I-based carbon ion treatments, while LEM IV predicts 10–15 Gy (RBE) higher tolerance doses. Further studies in larger patient cohorts are necessary to investigate the spatial dependence of the proton RBE, the accuracy of related models as well as the impact on the comparison of RBE-weighted tolerance doses for protons with those of carbon ions.
Conflict of interest None.
Appendix 1. To convert the delivered RBE-weighted fractional dose distribution, D1 ðx; y; zÞ, into that of a treatment with 2 Gy (RBE), D2 ðx; y; zÞ, we used the isoeffect relation
D2 ðx; y; zÞ ¼ D1 ðx; y; zÞ
ða=b þ d1 Þ ; ða=b þ d2 Þ
ð1Þ
obtained from equating the survival fractions of the LQM (23). In Eq. (1), d1 , and d2 are the RBE-weighted doses prescribed per fraction at the actual and the 2 Gy (RBE) treatment, respectively. The a=b-ratio refers to photons and was set to either 2 or 3 Gy. For the new dose distribution, D2 ðx; y; zÞ, the local TL-dose was determined from the DVH. References [1] Karger CP, Peschke P. RBE and related modeling in carbon-ion therapy. Phys Med Biol 2017. [2] Kanai T, Furusawa Y, Fukutsu K, Itsukaichi H, Eguchi-Kasai K, Ohara H. Irradiation of mixed beam and design of spread-out Bragg peak for heavy-ion radiotherapy. Radiat Res 1997;147:78–85. [3] Scholz M, Kellerer AM, Kraft-Weyrather W, Kraft G. Computation of cell survival in heavy ion beams for therapy. Radiat Environ Biophys 1997;36:59–66. [4] Kase Y, Kanai T, Matsumoto Y, Furusawa Y, Okamoto H, Asaba T, et al. Microdosimetric measurements and estimation of human cell survival for heavy-ion beams. Radiat Res 2006;166:629–38.
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018
6
TL tolerance for proton and 12C-ion RT
[5] Britten RA, Nazaryan V, Davis LK, Klein SB, Nichiporov D, Mendonca MS, et al. Variations in the RBE for cell killing along the depth-dose profile of a modulated proton therapy beam. Radiat Res 2013;179:21–8. [6] Guan F, Bronk L, Titt U, Lin SH, Mirkovic D, Kerr MD, et al. Spatial mapping of the biologic effectiveness of scanned particle beams: towards biologically optimized particle therapy. Sci Rep 2015;18:9850. [7] Patel D, Bronk L, Guan F, Peeler CR, Brons S, Dokic I, et al. Optimization of Monte Carlo particle transport parameters and validation of a novel high throughput experimental setup to measure the biological effects of particle beams. Med Phys 2017;44:6061–73. [8] Paganetti H. Significance and implementation of RBE variations in proton beam therapy. Technol Cancer Res Treat 2003;2:413–26. [9] Paganetti H. Relative biological effectiveness (RBE) values for proton beam therapy. Variations as a function of biological endpoint, dose, and linear energy transfer. Phys Med Biol 2014;59:R419–72. [10] Peeler CR, Mirkovic D, Titt U, Blanchard P, Gunther JR, Mahajan A, et al. Clinical evidence of variable proton biological effectiveness in pediatric patients treated for ependymoma. Radiother Oncol 2016;121:395–401. [11] Buchsbaum JC. Are treatment toxicity issues in particle therapy a clarion call for biologic treatment planning overall? Int J Radiat Oncol Biol Phys 2017;97:1085–6. [12] Elsässer T, Weyrather WK, Friedrich T, Durante M, Iancu G, Krämer M, 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. [13] Karger CP, Peschke P, Sanchez-Brandelik R, Scholz M, Debus J. Radiation tolerance of the rat spinal cord after 6 and 18 fractions of photons and carbon ions: experimental results and clinical implications. Int J Radiat Oncol Biol Phys 2006;66:1488–97. [14] Saager M, Glowa C, Peschke P, Brons S, Grün R, Scholz M, et al. Split dose carbon ion irradiation of the rat spinal cord: dependence of the relative biological effectiveness on dose and linear energy transfer. Radiother Oncol 2015;117:358–63. [15] Saager M, Glowa C, Peschke P, Brons S, Grün R, Scholz M, et al. The relative biological effectiveness of carbon ion irradiations of the rat spinal cord increases linearly with LET up to 99 keV/lm. Acta Oncol 2016;55:1512–5. [16] Schlampp I, Karger CP, Jäkel O, Scholz M, Didinger B, Nikoghosyan A, et al. Temporal lobe reactions after radiotherapy with carbon ions: incidence and estimation of the relative biological effectiveness by the local effect model. Int J Radiat Oncol Biol Phys 2011;80:815–23. [17] Gillmann C, Jäkel O, Schlampp I, Karger CP. Temporal lobe reactions after carbon ion radiation therapy: comparison of relative biological effectiveness-
[18] [19]
[20]
[21]
[22]
[23] [24] [25]
[26]
[27]
[28]
[29] [30] [31]
weighted tolerance doses predicted by local effect models I and IV. Int J Radiat Oncol Biol Phys 2014;88:1136–41. Lawrence YR, Li XA, el Naqa I, Hahn CA, Marks LB, Merchant TE, et al. Radiation dose-volume effects in the brain. Int J Radiat Oncol Biol Phys 2010;76:S20–7. Pehlivan B, Ares C, Lomax AJ, Stadelmann O, Goitein G, Timmermann B, et al. Temporal lobe toxicity analysis after proton radiation therapy for skull base tumors. Int J Radiat Oncol Biol Phys 2012;83:1432–40. Paganetti H, Niemierko A, Ancukiewicz M, Gerweck LE, Goitein M, Loeffler JS, et al. Relative biological effectiveness (RBE) values for proton beam therapy. Int J Radiat Oncol Biol Phys 2002;53:407–21. Weber DC, Malyapa R, Albertini F, Bolsi A, Kliebsch U, Walser M, et al. Long term outcomes of patients with skull-base low-grade chondrosarcoma and chordoma patients treated with pencil beam scanning proton therapy. Radiother Oncol 2016;120:169–74. National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE, V3.0) grading system. Available at: http://www.eortc.be/services/doc/ ctc/czcaev3.pdf. Accessed Dec. 17, 2001. In. Fowler JF. The linear-quadratic formula and progress in fractionated radiotherapy. Br J Radiol 1989;62:679–94. The R project for Statistical Computing [Internet, cited 2013 Mar 28]. Available from http://www.r-project.org/. Carabe A, Moteabbed M, Depauw N, Schuemann J, Paganetti H. Range uncertainty in proton therapy due to variable biological effectiveness. Phys Med Biol. 2012;57:1159–72. Wedenberg M, Lind BK, Hårdemark B. A model for the relative biological effectiveness of protons: the tissue specific parameter a/b of photons is a predictor for the sensitivity to LET changes. Acta Oncol 2013;52:580–8. McNamara AL, Schuemann J, Paganetti H. A phenomenological relative biological effectiveness (RBE) model for proton therapy based on all published in vitro cell survival data. Phys Med Biol 2015;60:8399–416. Emami B, Lyman J, Brown A, Coia L, Goitein M, Munzenrider JE, et al. Tolerance of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys 1991;21:109–22. Munzenrider JE, Liebsch NJ. Proton therapy for tumors of the skull base. Strahlenther Onkol 1999;175:57–63. Hug EB, Slater JD. Proton radiation therapy for chordomas and chondrosarcomas of the skull base. Neurosurg Clin N Am 2000;11:627–38. Noël G, Habrand JL, Mammar H, Pontvert D, Haie-Méder C, Hasboun D, et al. Combination of photon and proton radiation therapy for chordomas and chondrosarcomas of the skull base: the Centre de Protonthérapie D’Orsay experience. Int J Radiat Oncol Biol Phys 2001;51:392–8.
Please cite this article in press as: Gillmann C et al. Dose–response curves for MRI-detected radiation-induced temporal lobe reactions in patients after proton and carbon ion therapy: Does the same RBE-weighted dose lead to the same biological effect?. Radiother Oncol (2018), https://doi.org/10.1016/ j.radonc.2018.01.018