Original article Strahlenther Onkol 2014 DOI 10.1007/s00066-014-0638-9 Received: 22 August 2013 Accepted: 5 February 2014

Piernicola Pedicini1,5 · Alba Fiorentino2 · Vittorio Simeon3 · Paolo Tini4 · Costanza Chiumento5 · Luigi Pirtoli4 · Marco Salvatore6 · Giovanni Storto1

© Springer-Verlag Berlin Heidelberg 2014

2Radiation Oncology Department, Sacro Cuore – Don Calabria Hospital, Negrar, Verona, Italy

1Unit of Nuclear Medicine, Department of Radiation and Metabolic Therapies,

I.R.C.C.S.-Regional-Cancer-Hospital-C.R.O.B, Rionero-in-Vulture, Italy 3Laboratory of Preclinical and Translational Research, I.R.C.C.S.-Regional-

Cancer-Hospital-C.R.O.B, Rionero-in-Vulture, Italy 4Unit of Radiation Oncology, Department of Medicine Surgery and Neurological

Sciences, University of Siena and Tuscany Tumor Institute, Siena, Italy 5Unit of Radiotherapy, Department of Radiation and Metabolic Therapies,

I.R.C.C.S.-Regional-Cancer-Hospital-C.R.O.B, Rionero-in-Vulture, Italy 6Unit of Nuclear Medicine, I.R.C.C.S. SDN Foundation, Napoli, Italy

Clinical radiobiology of glioblastoma multiforme Estimation of tumor control probability from various radiotherapy fractionation schemes

Glioblastoma (GB) is a devastating brain malignancy with an incidence of 3.1/100,000 in the USA. The median survival times are presently 12–15 months [28]. Maximal tumor debulking is considered to be an effective treatment modality, while tumor size, performance status, and the patient’s age are also demonstrated to be prognostic factors for improved survival [28]. The first promising data supporting the use of adjuvant radiotherapy (RT) to increase overall survival (OS) and progression-free survival (PFS) were reported in 1976 [32]. These studies showed the benefit of adjuvant RT alone or in conjunction with chemotherapy with bischloroethylnitrosoure (BCNU), demonstrating a median survival of 37 weeks and 40.5 weeks, respectively. In subsequent years, despite improved RT practices and technologies [1, 2], only the introduction of the integrated use of chemotherapy [mainly concurrent adjuvant temozolomide (TMZ)], adopted in the EORTC-NCIC phase III randomized trial, showed a median survival time increase of a few months: The 2-year survival and PFS rates were 27 and 10.7 %, re-

spectively [28]. Considering that GB has a high incidence of recurrence within 2 cm of the original tumors and that nearly all the patients affected by GB still succumb to local disease progression, the main intent of the multimodality approach should be improved local disease control [14]. Therefore, postoperative RT modalities, including total dose and new fractionation schedules [2], should be the subject of a critical reappraisal for these patients. However, the introduction of intensity-modulated radiotherapy (IMRT) with target volumes that precisely fit the tumor extension and spare the surrounding organs at risk did not achieve the expected improvement in local control and survival advantage [6]. This is most likely due to the interpatient and intratumor heterogeneity of tumor cell sensitivity to ionizing radiation of GB [26]; therefore, an improvement in our understanding of the radiobiological characteristics of GB appears to be necessary for determining which fractionation and total dose schedule will provide a significant therapeutic benefit [13]. To achieve this aim, an analytical/ graphical method that had previous-

ly been introduced and used for prostate and head and neck cancers was studied [22–25]. We used this method to identify the value of the GB α/β ratio, whose current estimate is approximately 10 Gy in the radiotherapy community, and to assess the intrinsic whole tumor radiosensitivity (α), repair capability (β), repopulation doubling time (Td), and kick-off time for accelerated proliferation (Tk) of GB. An appropriate estimate of such parameters and a critical analysis of their confidence intervals (CI95 %) would help radiation oncologists to identify the most effective RT treatment schedule. The model employed in this study might fulfill this requirement.

Radiobiological model The radiobiological method employed to fit the clinical data in this work is based on the linear quadratic model (LQ) that has been largely used to describe the surviving fraction of cells (S) in the tissue exposed to a total radiation dose D = n · d (n being the number of fractions and d the dose per fraction). When corrected for repopulation effect, S can be expressed as follows: Strahlentherapie und Onkologie X · 2014 

| 1

Original article α/β (Gy)

α/β (Gy)

16 6

16

4 14

14

12 2

12

10 0

10

8

8

6

6

4

4

2

2

0

0

a

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0 α (Gy-1)

b

0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2 α (Gy-1)

Fig. 1 8 The relationship between α and α/β for glioblastoma multiforme. The black curves were obtained from Eq. 2 using all possible couples of clinical data (upper area of . Table 1) and Td as a free parameter by varying its value up to the coincidence for all curves. The intersections of the curves represent the best estimate of α, α/β, and Td (a). The gray curves represent the 95 % confidence interval (only three curves shown) and the shaded area indicates the overall range of uncertainties (b)



S=e

− D ⋅(α + β ⋅ d ) + γ (T − Tk )

(1)

where the exponent represents the biologically effective yield of lethal damage per cell, T the overall treatment time (OTT), Tk the kick-off time for tumor repopulation, and γ = ln2/Td quantifies the effective tumor repopulation rate, Td being the repopulation doubling time. The clinical PFS resulting from radiation treatment can be obtained from Eq. 1, using the tumor control probability (TCP) model based on the Poisson hypothesis: 

PFS = e − N ⋅e

− D(α + β ⋅ d )+

ln 2 ( T − Tk ) Td

(2)

where N represents the number of clonogens for a given regimen of fractionation (d, D, and T fixed), the value of PFS in Eq. 2 depends on five parameters (N, α, β, Td, and Tk). If a random selection of patients in both groups is assumed and whatever the value of Tk, Eq. 2, written for two different schedules of fractionation (i.e., a and b), provides the following:  ln 2 α ( Db − Da ) + β( Db db − Da d a ) − (T − T ) ln PFS a Td b a =e (3) ln PFSba where the dependence by cell number N and kick-off time for tumor repopulation Tk has disappeared. Moreover, Eq. 3 also allows one to eliminate the dependence of

2 |  Strahlentherapie und Onkologie X · 2014

findings on the choice of different conformal RTs between institutions [22]. Therefore, when a sufficient number of different schedules and a large num­ ber of patients are available (to reduce the stochastic fluctuations), by taking the nat­ ural logarithm and by rearranging Eq. 3, an estimation of the cellular parameters (α, β, and Td) can be made by the following equation: db Db − d a Da α = β   1 ln 2 (Tb − Ta ) − (Db − Da ) C + α Td 

(4) where  ln PFS a   (5) C = ln    ln PFSb  is named “clinical efficacy factor.” Equation 4 establishes an independent relationship between α and α/β from which it is possible to include and compare studies with different clinical outcomes (C≠0). The curves relating to different schedules were graphed with α/β as an independent function of α and using Td as a free parameter by varying its value until the coincidence of all curves was obtained. The intersection point provided an estimate of α, α/β, and Td. This expedient allows us to eliminate the dependence of findings on the values of N and Tk. The uncertainties arising

from assumptions about N and Tk strongly influence the results of the other models that depend on such parameters. Therefore, this is the main advantage of such a procedure. Moreover, Eq. 4 is also substantially independent of the impact of chemotherapy: The procedure described above allows us to minimize the effects due to different chemotherapies (i.e., TMZ and BCNU). In fact, while a direct estimate of α, α/β, and Td by Eq. 2 would include the global effect of chemotherapy (unknown in this analysis), the estimation of the same parameters by Eq. 4 is independent of such an effect (in the case in which the two schedules in Eq. 3 would have the same chemotherapy), or is dependent only on the differences in terms of radiosensitization between various chemotherapies, which are often negligible. Once the estimate of α, β, and Td was made, an estimation of Dprolif, the dose required to offset the repopulation occurring after 1 day, in a fraction of 2 Gy, was obtained by the following equation:  D = prolif

ln 2 Td ⋅ (α + 2 ⋅ β)

(6)

Subsequently, an estimation of Tk was obtained using the hypothesis of stem cell activation [26] by the following equation:  11 (7) Tk ≅ 2 ⋅ (α + 2 ⋅ β)

Abstract · Zusammenfassung Strahlenther Onkol 2014  DOI 10.1007/s00066-014-0638-9 © Springer-Verlag Berlin Heidelberg 2014 P. Pedicini · A. Fiorentino · V. Simeon · P. Tini · C. Chiumento · L. Pirtoli · M. Salvatore · G. Storto

Clinical radiobiology of glioblastoma multiforme. Estimation of tumor control probability from various radiotherapy fractionation schemes Abstract Background and purpose.  The aim of this study was to estimate a radiobiological set of parameters from the available clinical data on glioblastoma (GB). Patients and methods.  A number of clinical trial outcomes from patients affected by GB and treated with surgery and adjuvant radiochemotherapy were analyzed to estimate a set of radiobiological parameters for a tumor control probability (TCP) model. The analytical/graphical method employed to fit the clinical data allowed us to estimate the intrinsic tumor radiosensitivity (α), repair capability (β), and repopulation doubling time (Td) in a first phase, and subsequently the number of clonogens (N) and kick-off time for accelerated proliferation (Tk). The results were used to

formulate a hypothesis for a scheduleexpected to significantly improve local control. The 95 % confidence intervals (CI95 %) of all parameters are also discussed. Results.  The pooled analysis employed to estimate the parameters summarizes the data of 559 patients, while the studies selected to verify the results summarize data of 104 patients. The best estimates and the CI95 % are α = 0.12 Gy−1 (0.10–0.14), β = 0.015 Gy−2 (0.013–0.020), α/β = 8 Gy (5.0– 10.8), Td = 15.4 days (13.2–19.5), N = 1 · 104 (1.2 · 103–1 · 105), and Tk = 37 days (29–46). The dose required to offset the repopulation occurring after 1 day (Dprolif) and starting after Tk was estimated as 0.30 Gy/day (0.22– 0.39).

Conclusion.  The analysis confirms a high value for the α/β ratio. Moreover, a high intrinsic radiosensitivity together with a long kick-off time for accelerated repopulation and moderate repopulation kinetics were found. The results indicate a substantial independence of the duration of the overall treatment and an improvement in the treatment effectiveness by increasing the total dose without increasing the dose fraction. Keywords Glioblastoma multiforme · Intrinsic radiosensitivity · Repopulation kinetics · Accelerated proliferation · Hypofractionation

Klinische Strahlenbiologie des Glioblastoms. Schätzung der Tumorkontrollwahrscheinlichkeit von verschiedenen Radiotherapie-Fraktionierungsschemata Zusammenfassung Hintergrund und Zweck.  Schätzung eines strahlenbiologischen Parametersatzes auf der Grundlage klinischer Daten bei Patienten mit einem Glioblastom (GB). Patienten und Methoden.  Eine Reihe klinischer Ergebnisse von Patienten mit GB, die eine Operation sowie eine adjuvante Radiochemotherapie erhielten, wurde analysiert, um strahlenbiologische Parameter für das Tumorkontrollwahrscheinlichkeitsmodell („tumor control probability”, TCP) zu schätzen. Die angewandte analytische/graphische Methode ermöglichte eine Schätzung der intrinsischen Radiosensitivität des Tumors (α), der Regenerationsfähigkeit (β), der Re-Populationsverdopplungszeit (Td) in der ersten Phase und anschließend der Anzahl von clonogenen Zellen (N) sowie des Zeitpunkts einer beschleunigten Proliferation (Tk). Die Ergebnisse

The last step was the estimation of N by using Eq. 2 in which α, α/β, Td, and Tk were fixed, as in the previous estimation, and N was considered a free parameter up to the best fit between the clinical data and the TCP curve was obtained (. Fig. 2a).

wurden verwendet, um eine Hypothese für die Fraktionierung der Strahlendosis zu erheben, welche die lokale Kontrolle erwartungsgemäß signifikant erhöht. Die 95 % Konfidenzintervalle (CI95 %) aller Parameter werden ebenfalls diskutiert. Ergebnisse.  Die angewandte gepoolte Analyse zur Schätzung der Parameter wurde bei insgesamt 559 Patienten durchgeführt, während die Überprüfung der Ergebnisse bei insgesamt 104-Patienten erfolgte. Die besten Schätzungen und CI95% sind α = 0,12 Gy−1 (0,10–0,14), β = 0,015 Gy−2 (0,013–0,020), α/β = 8 Gy (5,0–10,8), Td = 15,4 Tage (13,2– 19,5), N = 1×104 (1,2×103–1×105), und Tk = 37 Tage (29–46). Die erforderliche Dosis für die Repopularisation, die nach 1 Tag (Dprolif) auftritt und nach Tk beginnt, wurde auf 0,30 Gy/ Tag (0,22–0,39) geschätzt.

Clinical outcome data A number of clinical trial outcomes from patients affected by GB and treated with surgery and RT integrated with chemotherapy were analyzed to estimate, and subsequently verify, the consistency of a set of GB radiobiological parameters. The inclusion criteria to select patients for these studies were as follows: patients

Schlussfolgerung.  Die Analyse bestätigt einen hohen Wert für α/β. Darüber hinaus wurden eine hohe intrinsische Radiosensitivität und eine lange Dauer bis zum Beginn der beschleunigten Proliferation sowie eine moderate Repopularisationskinetik festgestellt. Die Ergebnisse zeigen ein hohes Maß an Unabhängigkeit von der Dauer der gesamten Behandlung und eine Verbesserung der Therapiewirksamkeit durch eine Erhöhung der Gesamtdosis, ohne dabei die Dosisfraktion zu erhöhen. Schlüsselwörter Glioblastom · Intrinsische Radiosensitivität · Repopularisationskinetik · Beschleunigte Proliferation · Hypofraktionierung

older than 18 years with newly diagnosed GB, surgery as a first treatment (biopsy, partial or complete excision), RT integrated with chemotherapy (TMZ in most cases, or nitrosoureas, such as BCNU), phase I or II trials with nonstandard dose and fractionation schedules, and availability of data on progression-free survival at 1 year (1-PFS) for each study (individual patient

Strahlentherapie und Onkologie X · 2014 

| 3

Original article 0 1.0

1.00 PFS

0 1.00

0.90 0.80

0.90 0.80

0.70

0.70

0.7

0.60

0.60

6 0.6

0.50

0 0.50

0.5

0.40

0 0.40

0.4

0.30 0.20

0.30 0.20

0.3

0.10 0.00 45

EQD2 55

a

65

75

PFS

0.10 0.00 45

PFS

0.9 0.8

0.2 EQD2 55

b

65

75

0.1 0.0 0 50

c

BED 60

70

80

90

100

110

120

Fig. 2 8 Best-fit curve (black) of selected clinical data (white diamonds), plotted along the EQD2. The curve is obtained by fixing the estimated values of α, β, Td, and Tk, and by varying N in Eq. 2. The error bars represent the CI95 % of data obtained by the Greenwood formula (a). Best-fit curve verification. In this graph, the data used to calculate the results (white diamonds) are plotted by using Eq. 2 together with the data used to verify the results (gray diamonds). The EQD2 corrections required to superimpose the clinical data with the best curve were: + 5 % Stupp [28], − 2 % Buckner 1 [5], + 10 % Buckner 2 [5], 0 % Guckenberger [11], + 6 % Terasaki [29], 0 % Morganti [17], − 6 % Balducci [1], − 1 % Massaccesi [15], − 7 % Tsien [30] (b). Best-fit curve and curves corresponding to the extreme clinical data obtained by varying N (gray lines), plotted along the BED. The extreme values of N outline an uncertainty interval smaller than that obtained with the propagation of the uncertainties from the other parameters (dashed lines). The graph also shows the data corresponding to our hypothesized schedule (gray circle) from which a significantly higher increase of disease control is expected (PFS = 0.86) (c). PFS progression-free survival, EQD2 equivalent total dose delivered in fractions of 2 Gy, BED biological effective dose

data, i.e., baseline, treatment, time to failure or censoring, were not available). The exclusion criteria were as follows: studies with heterogeneous histological population (GB plus astrocytoma), palliative approach (e.g., only biopsy or palliative radiotherapy), and unavailable 1-PFS data.

Uncertainties In all the original studies of the survey, the primary end point was 1-PFS, assessed by the Kaplan–Meier method. The PFS 95 % confidence intervals (CI95 %) were obtained by using Greenwood’s formula (with a fixed censoring, because of the absence of individual data on censoring) that puts a standard error on the Kaplan– Meier estimator using the delta method [7]. The CI95 % of α and β were obtained from the CI95 % of PFS in conjunction with the best value of Td (the value where the curves relating to different fractionations intersect). From these uncertainties, the CI95 % of Td was then obtained by propagating the CI95 % of α and β. Subsequently, the uncertainties of Dprolif and Tk were estimated by propagating the CI95 %

4 |  Strahlentherapie und Onkologie X · 2014

of α, β, and Td by using Eq. 6 and Eq. 7, respectively. Finally the uncertainty of N was estimated by propagating the CI95 % of all the estimated parameters and PFS, by using Eq. 2. The uncertainty of N was also graphically compared with the curves corresponding to the extreme clinical data [plotted along the biological effective dose (BED)] obtained by varying N (. Fig. 2c). Moreover, due to the absence of individual patient data, a simulation approach was also used. Data for each study were reconstructed by simulation methods at the patient level, such that the observed PFS rates were matched, and then the simulated data were bootstrapped.

Validation of results The results obtained from the selected clinical data were validated: first, by comparing the placement of the mean normalized slope achieved for PFS = 0.5 (γ50) in the clinical relevant range [8, 19]; second, by repeating the analysis using independent clinical data (validation subset) to test the goodness of the results obtained. The analysis was based on the comparison of γ50 for curves obtained by Eq. 2 from selected clinical data to calcu-

late and to verify the results, respectively (. Fig. 2b). The uncertainties estimated by Greenwood’s formula and propagated for all parameters were compared with those carried out by the simulation of the standard bootstrapping procedure (fixed censoring) using 600 resampled datasets.

Results The clinical studies selected to estimate the parameters are presented in the upper section of . Table 1 [5, 11, 28, 29], which includes the study group, study size, fractionation schedule, OTT, and 1-PFS. The pooled analyses summarize the data of 559 patients. The EORTC-NCIC trial was selected because of its therapeutic approach, that is, the current standard of treatment; conversely, the other studies were selected due to the different RT dose scheduling associated with new and old drugs (TMZ and BCNU, respectively). The data used to verify the results (nonstandard RT dose and fractionation schedules) are reported in the lower section of . Table 1 [1, 15, 17, 30]. Overall, the studies selected summarize the data of 104 patients (for Balducci et al., only GB data have been select-

Table 1  Selected clinical data of GB external beam treatment of used to estimate the radioStudy Pts. d (Gy) Data used to estimate the parameters Stupp et al. 2005 287 2.0 Buckner et al. 2006 (arm A) 98 1.8 Buckner et al. 2006 (arm B) 103 1.6 2f/day Guckenberger et al. 2011 45 1.8 2f/day Terasaki et al. 2011 26 3.0 Data used to verify the results Balducci et al. 2010 41 1.8 + 2.4 Morganti et al. 2010 7 2.4 6 2.5 6 2.6 Tsien et al. 2009 38 2.2–2.4 Massaccesi et al. 2012 6 2.4

n

D (Gy)

CT

OTT

1-y PFS

Parameter

30 36 30 30 15

60 64,8 48 54 45

TMZ BCNU BCNU TMZ TMZ

40 48 20 20 15

26.9 % 20.4 % 16.7 % 21.0 % 29.0 %

33 + 4 25 25 25 30 25

69.4 60 62.5 65 66–72 60

TMZ

44

49.0 %

α (Gy-1) β (Gy-2) α/β (Gy) Td (days) Dprolif (Gy) Tk (days) N (clonogens)

TMZ

33

49.1 %

TMZ TMZ

40 33

33.0 % 43.6 %

Pts. number of patients, d dose per fraction, D total radiation dose, CT chemotherapy, OTT overall treatment time, PFS progression-free survival, TMZ temozolomide, BCNU bischloroethylnitrosoure

ed [1] to respect the inclusion and exclusion criteria). Details relating to each subgroup have been reported in the related references. The best estimate and the CI95 % for α, α/β, Td, N, Tk, and Dprolif are shown in . Table 2. . Fig. 1a shows the intersection of the curves that represent different schedules of dose fractionation and .  Fig. 1b shows the corresponding uncertainties. The graph interception shows a high α/β (close to 8 Gy) and a relatively high α value (0.12 Gy−1). The corresponding Td value is 15.4 days. .  Fig. 2a shows the best-fit curve of selected clinical data (upper area of .   Table 1) plotted along the equivalent total dose delivered in fractions of 2 Gy (EQD2). The curve was obtained by fixing the estimated values of α, β, Td, and Tk, and then by varying N in Eq. 2. . Fig. 2b shows the best curve with respect to the selected validation subset of data (lower area of . Table 1). The superimposition of clinical data and best-fit curve was obtained by an acceptable variation in dose (lower than 6 %) along the EQD2 (a dose correction of approximately 10 % was required in only one case). The fitted curve has γ50 = 3.31, which is very close to the mean γ50 of the clinically relevant range (γ50 = 3.20) described in the literature [19]. No noticeable difference between γ50 belonging to the calcu-

Table 2  The best estimated parameters

and their uncertainty intervals

biological parameters and to verify the results

lation and verification curves was found (γ50 = 3.31 and γ50 = 3.36, respectively). Therefore, the validation was fully successful. .  Fig.  2c shows the best-fit curve (N = 9.1 · 103) and curves corresponding to the extreme clinical data obtained by varying N (6.0 · 103 to 1.4 · 104) plotted along the BED. The latter outline an uncertainty interval smaller than that obtained with the propagation of the uncertainties from the other parameters (4.0 · 103 to 2.1 · 104). Thus uncertainties propagated on N contain all the experimental points from which the estimates are obtained. The comparison of CI95 % by bootstrap simulation and Greenwood’s formula was made assuming a fixed distribution for censoring (close agreement with the CI95 % of unknown real censoring distribution [9]). As expected, the CI95 % values by Greenwood’s formula were slightly larger than those obtained with bootstrap simulation, and therefore we adopted them with greater caution. In the same figure, the data are shown to correspond to our hypothesis of a schedule where an expected dose of 74.8 Gy in 34 fractions of 2.2 Gy (BED = 92.1 Gy) may result in a significantly higher increase in disease control (PFS = 0.86). . Fig. 3a shows the curve and the CI95 % curves of Tk as a function of dose per fraction d from Eq. 9. . Fig. 3b shows the es-

Best estimate 0.12 0.015 8 15.4 0.30 37 9.1·103

CI95 % 0.10–0.14 0.013–0.020 5.0–10.8 13.2–19.5 0.22–0.39 29–46 4.0·103−2.1·104

timation of the number of clonogens and the relative uncertainties obtained by rearranging Eq. 2 and using all the clinical data [23].

Discussion The current prognosis of GB patients is poor, rarely exceeding a 1-year median survival. However, a small but significant improvement was demonstrated using temozolomide chemotherapy, concurrently and subsequently administered with respect to RT [3, 28]. Nearly all patients eventually die due to local disease progression; however, the improvement of local control is considered to play a fundamental role in this pathology. This statement is also supported by the demonstration that a complete tumor surgical excision (local control) improves the survival of patients compared with biopsy alone [28]. Thus, optimization of the postoperative RT administration might contribute to a reduction in the risk of relapse and to improvements in the patients’ outcome. A radiobiological approach to overcome the poor response of GB to radiation is currently unavailable due to the incomplete understanding of the underlying genetic and biomolecular alterations. Profiling studies of GB based on gene or protein expression have revealed several altered common molecular pathways, resulting in subclassifications of distinct molecular subtypes (classical, mesenchymal, proneural, neural) that are different in terms of their prognosis and response to therapy [31]. This characterization is not currently adopted in common practice. Furthermore, emerging evidence shows the existence of a stem-like cell compartment in GB that demonstrates an increased resisStrahlentherapie und Onkologie X · 2014 

| 5

Original article Tk (days)

1.0E+08

50 0 40 0

1.0E+06

30 0

1.0E+04

20 0

1.0E+02

10 0 0

a

N

1.0E+00 0

1

2

3

4

d (Gy)

b

0

20 0

40

60

80 BED (G (Gy)

Fig. 3 8 Curve of Tk as a function of dose per fraction d from Eq. 9 (a). Number of clonogens from Eq. 2 as function of biological effective dose (BED) corresponding to the best fit for α, β, Td, and Tk (b)

tance to ionizing radiation [4]; therefore, improved mathematical models based on clinical results might be useful. From the results of this study, it is possible to identify some important characteristics of GB when undergoing radiation and chemotherapy. Such characteristics, which can help clinicians to identify an optimal dose and dose fractionation for the prescription, are described as follows: a high fractionation sensitivity α/β (8 Gy) together with a high intrinsic radiosensitivity α (0.12 Gy−1). This corresponds to a low β value (0.015 Gy−2), that represents a high capability of GB cells to repair the radiation damage. Moreover, the fit of clinical data shows a moderate value for the doubling time Td (15.4 days), together with a very long kick-off time for the accelerated repopulation Tk (37 days). This implies a substantial independence of tumor radiation response with respect to the duration of the overall radiation treatment. Consequently, it appears unnecessary to diminish the treatment duration through hypofractionation (doses greater than 2 Gy/fraction) or hyperfractionation (doses less than 2 Gy/fraction) with multiple daily sessions. This statement is confirmed by the results of a recent trial 67 GB patients treated with IMRT (25 Gy in 5 fractions) did not demonstrate an improvement in overall survival or in PFS (1-PFS: 29.42 %), reducing the OTT by an hypofractionated schedule [6]. Moreover, the following consideration should also be taken into account. Generally, a fast repopulation, and thus a strong dependence of the results on the duration of treatment, is explained with a selection of stem cells (more radioresistant to the

6 |  Strahlentherapie und Onkologie X · 2014

radiation because they are normally quiescent) for proliferation during irradiation [21, 25]. The relatively low value we found for N (9.1 · 103) seems to confirm this hypothesis: it is only a small number of the most resistant cells (i.e., stem cells) that determine the clinical outcome [8, 19]. However, given the substantial independence of the results from the OTT, this mechanism appears to be negligible when compared with the mechanism of repair, which should be more pronounced in this cell type. This characteristic can be taken into account in favor of the time required by organs at risk in order to fully repair the radiation damage. Our results demonstrated a strong dependence on total dose; thus, an improvement of these clinical results could be obtained with an increase in the total dose rather than with a reduction of the treatment duration. Based on our estimate of radiobiological parameters, an increase of the total dose up to a BED of approximately 92 Gy (total dose, 74.8 Gy; dose per fraction, 2.2 Gy; 34 fractions) should lead to a TCP greater than 0.85 (. Fig. 2). This result appears to be surprisingly higher than that obtained with standard fractionation ( 30 fractions of 2 Gy (total 60 Gy) with a BED of approximately 74 Gy), which is approximately 0.3. Therefore, such an optimistic prediction by the model requires confirmation. In our opinion, the contributions of several potentially confounding factors that were not necessarily fully considered in our method include: (1) the data from institutions with different patient selection criteria and different modalities of treatment delivery; (2) the possible coex-

istence of different cell types, whose behavior can be explained not by a single set of radiobiological parameters but by the necessary partitioning of the cellular types; (3) the different expression levels of molecular factors among patients, such as MGMT methylation [10, 12]; and (4) other factors, such as hypoxia and reoxygenation, which may have different influences on outcome. The results of the present paper, therefore, deserve further study on larger and more detailed datasets. Nevertheless, our results clearly indicate the possibility for a substantial increase in the effectiveness of radiation treatment by increasing the total dose. In our opinion, such a possibility has not been thoroughly investigated in the past because of the high toxicity observed at doses greater than 60 Gy (30 fractions, BED ~74 Gy, TCP ~0.3), but this was mainly due to the high volumes included in the fields of treatment. Recent studies demonstrated that a high-total-dose reirradiation of reduced volumes is feasible and effective in selected cases [16]. Therefore, high doses (approximately 90 Gy) may improve local control even if survival is unaffected [18]. Moreover, when investigating the effectiveness of a high dose in high-grade gliomas, some RTOG trials (e.g., RTOG 93-05 trial) failed to demonstrate an improved survival and did not definitively assess the role of a stereotactic hypofractionated radiotherapy boost [27]. However, these reports are of limited value for the present study due to some relevant differences (such as a lack of TMZ use, the use of an outdated drug such as BCNU, sample size, and outdated radiotherapy techniques). A more appropriate comparison could be made according to the method described by Panet-Raymond et al. [20], which was also unable to demonstrate any survival advantage over using standard RT-TMZ treatment with a simultaneous integrated IMRT boost. In our opinion, the moderately hypofractionated, high-total-dose treatment schedule described here, which is based on a mathematical model, deserves consideration. Such high tumor doses delivered in an integrated schedule with TMZ might be tested in clinical settings because the surrounding healthy brain would be

spared owing to the accuracy of modern techniques for tumor identification, contouring, and dose delivery to the target volume.

Conclusion In conclusion, the proposed analytical/graphical method described in this manuscript allowed the fit of clinical data, thus providing a self-consistent set of radiobiological parameters for GB. The analysis confirms a high value for α/β with a correspondingly high intrinsic radiosensitivity, which is compatible with a realistic average number of clonogens. Moreover, a long kick-off time for the accelerated repopulation together with a moderate repopulation indicate a substantial independence of the therapeutic results from the duration of the overall radiation treatment. Therefore, these findings demonstrate the possibility to improve the effectiveness of treatment by increasing total dose without the need to greatly increase the dose per fraction. This demonstration, together with modern dose-delivery techniques, may create new conditions for a treatment practice that optimally integrates a high total dose and good tolerance for the organs at risk during GB radiation management.

Corresponding address P. Pedicini Ph.D. Unit of Nuclear Medicine Department of Radiation and Metabolic Therapies I.R.C.C.S.-Regional-Cancer-Hospital-C.R.O.B 1-Padre-Pio-Street, Rionero-in-Vulture [email protected]

Compliance with ethical guidelines Conflict of interest. P. Pedicini, A. Fiorentino, V. Simeon, P. Tini, C. Chiumento, L.Pirtoli, M. Salvatore, and G. Storto state that there are no conflicts of interest.

References   1. Balducci M, Apicella G, Manfrida S, Mangiola A, Fiorentino A, Azario L, D’Agostino GR, Frascino V, Dinapoli N, Mantini G, Albanese A, de Bonis P, Chiesa S, Valentini V, Anile C, Cellini N (2010) Single-Arm Phase II Study of Conformal Radiation Therapy and Temozolomide plus fractionated Stereotactic Conformal Boost in High grade Gliomas. Strahlen Onkol 186(10):558–564   2. Balducci M, Chiesa S, Diletto B, D’Agostino GR, Mangiola A, Manfrida S, Mantini G, Albanese A, Fiorentino A, Frascino V, De Bari B, Micciche F, De Rose F, Morganti AG, Anile C, Valentini V (2012) Low-dose fractionated radiotherapy and concomitant chemotherapy in glioblastoma multiforme with poor prognosis: a feasibility study. Neuro Oncol 14(1):79–86   3. Balducci M, Fiorentino A, De Bonis P, Chiesa S, Mangiola A, Mattiucci GC, D’Agostino GR, Frascino V, Mantini G, Alitto AR, Colosimo C, Anile C, Valentini V (2013) Concurrent and adjuvant temozolomide-based chemoradiotherapy schedules for glioblastoma. Hypotheses based on two prospective phase II trials. Strahlen Onkol 189(11):926–931   4. Bao S, Wu Q, Mc Lendon RE, Hao Y, Shi Q, Hjelmeland AB, Dewhirst MW, Bigner DD, Rich JN (2006) Glioma stem cells promote radioresistance by preferential activation of the DNA damage response. Nature 444(7120):756–760   5. Buckner JV, Ballman KV, Michalak JC, Burton GV, Cascino TL, Schomberg PJ, Hawkins RB, Scheithauer BW, Sandler HM, Marks RS, O’Fallon JR, North Central Cancer Treatment Group 93–72-52, Southwest Oncology Group 9503 Trials (2006) Phase III Trial of carmustine and cisplatin compared with carmustine alone and standard radiation therapy or accelerated radiation therapy in patients with glioblastoma multiforme: North Central Cancer Treatment Group 93–72-52 and Southwest Oncology Group 9503 Trials. J Clin Oncol 24(24):3871– 3879   6. Ciammella P, Galeandro M, D’Abbiero N, Podgornii A, Pisanello A, Botti A, Cagni E, Iori M, Iotti C (2013) Hypo-fractionated IMRT for patients with newly diagnosed glioblastoma multiforme: a 6 year single institutional experience. Clin Neurol Neurosurg 115(9):1609–1614   7. Collett D (1994) Modelling survival data-in medical research. Chapman and Hall, London, pp 22–26   8. Daşu A, Toma-Daşu I, Fowler JF (2003) Should single distributed parameters be used to explain the steepness of tumour control probability curves? Phys Med Biol 48:387–397   9. Efron B (1981) Censored data and the bootstrap. J Am Stat Ass 76:312–319 10. Fietkau R, Putz F, Lahmer G, Semrau S, Buslei R (2013) Can MGMT promoter methylation status be used as a prognostic and predictive marker for glioblastomamultiforme at the present time?: a word of caution. Strahlenther Onkol 189(12):993– 995 11. Guckenberger M, Mayer M, Buttmann M, Vince GH, Sweeney RA, Flentje M (2011) Prolonged survival when temozolomide is added to accelerated radiotherapy for glioblastoma multiforme. Strahlen Onkol 187(9):548–554 12. Hegi ME, Diserens AC, Gorlia T, Hamou MF, de Tribolet N, Weller M, Kros JM, Hainfellner JA, Mason W, Mariani L, Bromberg JE, Hau P, Mirimanoff RO, Cairncross JG, Janzer RC, Stupp R (2005) MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 352(10):997–1003

13. Jones TS, Holland EC (2012) Standard of care therapy for malignant glioma and its effect on tumor and stromal cells. Oncogene 31:1995–2006 14. Lee SW, Fraass BA, Marsh LH, Herbort K, Gebarski SS, Martel MK, Radany EH, Lichter AS, Sandler HM (1999) Patterns of failure following high-dose 3-D conformal radiotherapy for high-grade astrocytomas: a quantitative dosimetric study. Int J Radiat Oncol Biol Phys 43:79–88 15. Massaccesi M, Ferro M, Cilla S, Balducci M, Deodato F, Macchia G, Valentini V, Morganti AG (2013) Accelerated intensity-modulated radiotherapy plus temozolomide in patients with glioblastoma: a phase I dose-escalation study (ISIDE-BT-1). Int J Clin Oncol 18(5):784–791 16. Mizumoto M, Okumura T, Ishikawa E, Yamamoto T, Takano S, Matsumura A, Oshiro Y, Ishikawa H, Sakurai H, Tsuboi K (2013) Reirradiation for recurrent malignant brain tumor with radiotherapy or proton beam therapy. Technical considerations based on experience at a single institution. Strahlenther Onkol 189(8):656–663 17. Morganti AG, Balducci M, Salvati M, Esposito V, Romanelli P, Ferro M, Calista F, Digesù C, Macchia G, Ianiri M, Deodato F, Cilla S, Piermattei A, Valentini V, Cellini N, Cantore GP (2010) A phase I dose-escalation study (ISIDE-BT-1) of accelerated IMRT with temozolomide in patients with glioblastoma. Int J Radiat Oncol Biol Phys 77(1):92–97 18. Nakagawa K, Aoki Y, Fujimaki T, Tago M, Terahara A, Karasawa K, Sakata K, Sasaki Y, Matsutani M, Akanuma A (1998) High-dose conformal radiotherapy influenced the pattern of failure but did not improve survival in glioblastoma multiforme. Int J Radiat Oncol Biol Phys 40(5):1141–1149 19. Okunieff P, Morgan D, Niemierko A et al (1995) Radiation dose-response of human tumor. Int J Radiat Oncol Biol Phys 32:1227–1237 20. Panet-Raymond V, Souhami L, Roberge D, Kavan P, Shakibnia L, Muanza T, Lambert C, Leblanc R, Del Maestro R, Guiot MC, Shenouda G (2009) Accelerated hypofractionated intensity-modulated radiotherapy with concurrent and adjuvant temozolomide for patients with glioblastoma multiforme: a safety and efficacy. Int J Radiat Oncol Biol Phys 73(2):473–478 21. Pedicini P (2013) In regard to Pedicini et al. Int J Radiat Oncol Biol 87(5):858 22. Pedicini P, Nappi A, Strigari L, Jereczek-Fossa BA, Alterio D, Cremonesi M, Botta F, Vischioni B, Caivano R, Fiorentino A, Improta G, Storto G, Benassi M, Orecchia R, Salvatore M (2012) Correlation between EGFR expression and accelerated proliferation during radiotherapy of head and neck squamous cell carcinoma. Radiat Oncol 7:143 23. Pedicini P, Caivano R, Strigari L, Benassi M, Fiorentino A, Fusco V (2013) In regard to Miralbell et al. Re: Dose-fractionation sensitivity of prostate cancer deduced from radiotherapy outcomes of 5969 patients in seven international institutional datasets: alpha/beta = 1.4 (0.9–2.2) Gy. Int J Radiat Oncol Biol Phys 85(1):10–11 24. Pedicini P, Fiorentino A, Improta G, Nappi A, Salvatore M, Storto G (2013) Estimate of the accelerated proliferation by protein tyrosine phosphatase (PTEN) over expression in postoperative radiotherapy of head and neck squamous cell carcinoma. Clin Transl Oncol 15(11):919–924 25. Pedicini P, Strigari L, Benassi M (2013) Estimation of a self-consistent set of radiobiological parameters from hypofractionated versus standard radiation therapy of prostate cancer. Int J Radiat Oncol Biol Phys 85(5):e231–e237

Strahlentherapie und Onkologie X · 2014 

| 7

Original article 26. Roberts SA, Hendry JH (1998) A realistic closedform radiobiological model of clinical tumor-control data incorporating intertumor heterogeneity. Int J Radiat Oncol Biol Phys 41(3):689–699 27. Souhami L, Seiferheld W, Brachman D, Podgorsak EB, Werner-Wasik M, Lustig R, Schultz CJ, Sause W, Okunieff P, Buckner J, Zamorano L, Mehta MP, Curran WJ Jr (2004) Randomized comparison of stereotactic radiosurgery followed by conventional radiotherapy with carmustine to conventional radiotherapy with carmustine for patients with glioblastoma multiforme: report of Radiation Therapy Oncology Group 93-05 protocol. Int J Radiat Oncol Biol Phys 60(3):853–860 28. Stupp R, Mason W, Van der Bent M, Weller M, Fisher B, Taphoorn MJ, Belanger K, Brandes AA, Marosi C, Bogdahn U, Curschmann J, Janzer RC, Ludwin SK, Gorlia T, Allgeier A, Lacombe D, Cairncross JG, Eisenhauer E, Mirimanoff RO, European Organisation for Research and Treatment of Cancer Brain Tumor and Radiotherapy Groups, National Cancer Institute of Canada Clinical Trials Group (2005) Radiotherapy plus concomitant and adjuvant Temozolomide for glioblatoma. N Engl J Med 352(10):987–996 29. Terasaki M, Eto T, Nakashima S, Okada Y, Ogo E, Sugita Y, Tokutomi T, Shigemori M (2011) A pilot study of hypofractionated radiation therapy with temozolomide for adults with glioblastoma multiforme. J Neurooncol 102(2):247–253 30. Tsien C, Moughan J, Michalski JM, Gilbert MR, Purdy J, Simpson J, Kresel JJ, Curran WJ, Diaz A, Mehta MP, Radiation Therapy Oncology Group Trial 9803 (2009) Phase I three-dimensional conformal radiation dose escalation study in newly diagnosed glioblastoma: Radiation Therapy Oncology Group Trial 98-03. Int J Radiat Oncol Biol Phys 73(3):699– 708 31. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, Miller CR, Ding L, Golub T, Mesirov JP, Alexe G, Lawrence M, O’Kelly M, Tamayo P, Weir BA, Gabriel S, Winckler W, Gupta S, Jakkula L, Feiler HS, Hodgson JG, James CD, Sarkaria JN, Brennan C, Kahn A, Spellman PT, Wilson RK, Speed TP, Gray JW, Meyerson M, Getz G, Perou CM, Hayes DN, Cancer Genome Atlas Research Network (2010) An integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR and NF1. Cancer Cell 17(1):98–110 32. Walker MD, Strike TA (1976) An evaluation of methyl-CCNU, BCNU and radiotherapy in the treatment of malignant glioma. Proc Am Assoc Cancer Res 17:63

8 |  Strahlentherapie und Onkologie X · 2014

Clinical radiobiology of glioblastoma multiforme: estimation of tumor control probability from various radiotherapy fractionation schemes.

The aim of this study was to estimate a radiobiological set of parameters from the available clinical data on glioblastoma (GB)...
624KB Sizes 0 Downloads 3 Views