Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies

Physica Medica xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Physica Medica journal homepage: http://www.physicamedica.com

Original Paper

Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies Francesca Romana Giglioli a,⇑, Lidia Strigari b, Riccardo Ragona c, Giuseppina R. Borzì d, Elisabetta Cagni e, Claudia Carbonini f, Stefania Clemente g, Rita Consorti h, Randa El Gawhary i, Marco Esposito j, Maria Daniela Falco k, David Fedele l, Christian Fiandra c, Maria Cristina Frassanito m, Valeria Landoni b, Gianfranco Loi n, Elena Lorenzini o, Maria Rosa Malisan p, Carmelo Marino q, Enrico Menghi r, Barbara Nardiello s, Roberta Nigro t, Caterina Oliviero u, Gabriella Pastore v, Mariagrazia Quattrocchi w, Ruggero Ruggieri x, Irene Redaelli y, Giacomo Reggiori z, Serenella Russo j, Elena Villaggi aa, Marta Casati ab, Pietro Mancosu z a

A.O.U. Città della Salute e della Scienza di Torino, Torino, Italy Regina Elena Cancer Center IFO, Roma, Italy c Dep. of Oncology, Radiation Oncology Unit, University of Torino, Italy d REM Radioterapia, Catania, Italy e IRCCS – Arcispedale Santa Maria Nuova, Reggio Emilia, Italy f ASST ‘‘Grande Ospedale Metropolitano Niguarda”, Milano, Italy g Azienda Ospedaliera Universitaria Federico II Napoli, Italy h Fisica Sanitaria ACO San Filippo Neri, Roma, Italy i O. San Pietro Fatebenefratelli, Roma, Italy j Azienda Sanitaria Firenze, Italy k Fondazione Policlinico Tor Vergata, Roma, Italy l Casa di Cura Privata San Rossore, Pisa, Italy m Mater Dei Hospital, C.B.H. Città di Bari Hospital s.p.a., Bari, Italy n A.O.U. Maggiore della Carità di Novara, Italy o U.O. Fisica Sanitaria Azienda USL 1 Massa e Carrara, Italy p Medical Physics Dept, University Hospital of Udine, Italy q Humanitas C.C.O., Catania, Italy r IRCCS – IRST, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori, Meldola, Italy s UPMC San Pietro FBF, Roma, Italy t Ospedale S.Camillo de Lellis, Rieti, Italy u IRCCS CROB Rionero in Vulture (OZ), Italy v Ecomedica Empoli, Italy w S.C. Fisica Sanitaria USL 2 Lucca, Italy x Ospedale Sacro Cuore – don Calabria, Negrar (VR), Italy y A.O. San Gerardo di Monza, Italy z Medical Physics Unit of Radiotherapy Dept. Humanitas Clinical and Research Hospital, Rozzano (MI), Italy aa AUSL Piacenza, Italy ab AOU Careggi, Firenze, Italy b

a r t i c l e

i n f o

Article history: Received 24 December 2015 Received in Revised form 15 March 2016 Accepted 19 March 2016 Available online xxxx

a b s t r a c t Purpose: A large-scale multi-institutional planning comparison on lung cancer SABR is presented with the aim of investigating possible criticism in carrying out retrospective multicentre data analysis from a dosimetric perspective. Methods: Five CT series were sent to the participants. The dose prescription to PTV was 54 Gy in 3 fractions of 18 Gy. The plans were compared in terms of PTV-gEUD2 (generalized Equivalent Uniform Dose equivalent to 2 Gy), mean dose to PTV, Homogeneity Index (PTV-HI), Conformity Index (PTV-CI) and

⇑ Corresponding author at: A.O.U. Città della Salute e della Scienza di Torino, corso Bramante 88/90, Torino, Italy. Tel.: +39 011 6336626; fax: +39 011 6336614. E-mail address: [email protected] (F.R. Giglioli). http://dx.doi.org/10.1016/j.ejmp.2016.03.015 1120-1797/Ó 2016 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.

Please cite this article in press as: Giglioli FR et al. Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.03.015

2

Keywords: Stereotactic ablative radiotherapy (SABR) Stereotactic body radiation therapy (SBRT) Lung Multicentric clinical trial Radiobiology Dosimetry

F.R. Giglioli et al. / Physica Medica xxx (2016) xxx–xxx

Gradient Index (PTV-GI). We calculated the maximum dose for each OAR (organ at risk) considered as well as the MLD2 (mean lung dose equivalent to 2 Gy). The data were stratified according to expertise and technology. Results: Twenty-six centers equipped with Linacs, 3DCRT (4% – 1 center), static IMRT (8% – 2 centers), VMAT (76% – 20 centers), CyberKnife (4% – 1 center), and Tomotherapy (8% – 2 centers) collaborated. Significant PTV-gEUD2 differences were observed (range: 105–161 Gy); mean-PTV dose, PTV-HI, PTV-CI, and PTV-GI were, respectively, 56.8 ± 3.4 Gy, 14.2 ± 10.1%, 0.70 ± 0.15, and 4.9 ± 1.9. Significant correlations for PTV-gEUD2 versus PTV-HI, and MLD2 versus PTV-GI, were observed. Conclusions: The differences in terms of PTV-gEUD2 may suggest the inclusion of PTV-gEUD2 calculation for retrospective data inter-comparison. Ó 2016 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.

Introduction Although surgery has generally been considered to be the standard treatment for early stage non-small cell lung cancer (NSCLC), a fair number of patients are unsuitable for the surgical approach [1]. Stereotactic ablative radiotherapy (SABR) is a radiation technique that requires very accurate positioning and advanced image-guidance, allowing the administration of a few, large dose fractions that are able to kill the neoplastic cells through radioablation with very high biologically equivalent doses (BED > 100 Gy) [2]. In particular, the local control findings from early SABR studies on stage I NSCLC were confirmed in multicenter phase II trials with a primary tumor control rate of 97% and a local control rate of 92% at the 3 year follow up [3,4]. In 2012 the Italian Association of Medical Physics (Associazione Italiana di Fisica Medica-AIFM) established a work group with the aim of studying the dosimetric aspects of the SABR technique (AIFM–SABR-WG) as part of a project called ‘‘Dosimetry, physics, and radiobiology of image guided hypo-fractionated ablative radiotherapy”. Treatment outcome can only be interpreted meaningfully with accurate knowledge of the reference dose and dose distribution. However, as demonstrated by many multicenter studies, the prescribed dose seldom corresponds with the planned or delivered dose [5,6]. The aim of this study was to report the results of a large multiinstitution lung cancer SABR plan comparison, and investigate possible criticisms (dose normalization to PTV, technologies required, treatment plan class solutions) concerning the collection of retrospective clinical data and the approaches used for large multicenter studies. Dosimetric data from various treatment planning systems, techniques and planners were analyzed. Materials and methods Study description Computed tomography (CT) data from five patients with stage I NSCLC treated with SABR were selected and anonymized for data sharing. Patients with different tumor locations (close to the chest wall or spinal cord) and dimensions were selected in order to generalize the analysis (see Fig. 1b). A single radiation oncologist determined the target. In particular, the maximum lesion volume was 17 cm3, with mean volume of 13.8 ± 4.8 cm3 (mean ± standard deviation). In all patients, the clinical target volume (CTV) delineation was based on 4D-CT acquisitions. CTV was defined on the maximum inspiration phase (phase 0) and a deformable registration was adopted in order to extend the target to the other phases. Therefore the internal tumor volume (ITV) was defined as the Boolean union of the CTVs. The planning target volume (PTV) was created on the average CT by adding a 3 mm isotropic margin to the ITV bearing in mind the patient set-up. The ipsilateral lung

minus ITV (iLung-ITV), spinal cord, esophagus, heart and ribs were considered as organs at risk (OARs). All structures were intentionally delineated prior to data distribution in order to eliminate inaccuracy and variability of the contouring process. The dose prescription was set at 54 Gy over 3 consecutive days (i.e. 18 Gy per fraction) to PTV. Plans were implemented in order to cover at least 95% of PTV volume with 95% of the prescription dose (i.e. 51.3 Gy). No specific plan normalization procedure and no limitation on maximum dose within the target were requested, bearing in mind the focus of the study: to evaluate the possibility of sharing patients in a retrospective way. Therefore, the planners were asked to adopt the strategies used in their own institutions. For OARs, plans were required to meet the objectives of TG AAPM 101 [7] for relevant organs such as:






iLung-ITV: V20 Gy < 5%; V10 Gy < 10%; V5 Gy < 30%. Spinal cord: maximum dose <18 Gy. Esophagus: maximum dose <25 Gy. Heart: maximum dose <30 Gy. Ribs: maximum dose <37 Gy.

Study design AIFM–SABR-WG proposed its members to collaborate in this project with the aim of including at least one center for each of the usual SABR delivery techniques: 3D static conformal field, IMRT and VMAT by means of Linacs, Tomotherapy or CyberKnife. Treatment planning systems with heterogeneity correction and collapsed-cone convolution models (or superior algorithms) were mandatory for joining the project. The influence of the different calculation modalities, that could affect the dose distributions, were not considered in this study. A less than 3 mm grid size was required for the dose calculation. Each center received:
The five CT series with the DICOM_RT structures.
The relative electron density versus Hounsfield Units calibration curve of the simulation CT.
The constraints and the dose protocol described above. The participants were asked to plan the five cases and send the dose volume DICOM_RT file extracted by the TPS anonymously and to provide a report with relevant data regarding the center (i.e. machine, irradiation technique, TPS, center and planner expertise in SABR). Data analysis The dose volume DICOM_RT files were imported into the VelocityÒ (Varian Medical System) software and superimposed on the patient CTs. Dose Volume Histograms (DVHs) were

Please cite this article in press as: Giglioli FR et al. Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.03.015

F.R. Giglioli et al. / Physica Medica xxx (2016) xxx–xxx

3

Figure 1. (a) PTV, DVHs of a representative patient and (b) axial view of the five planned CT.

calculated and exported after evaluating the consistency of the DVHs calculation. A homemade MatlabÒ program was used for carrying out a dosimetric, equivalent mean lung dose and Equivalent Uniform Dose (EUD) analysis. Regarding the dosimetric parameters, mean dose, D98%, V95% and D2% were considered for PTV. The Homogeneity Index (PTV-HI), Conformity Index (PTV-CI), and Gradient Index (PTV-GI) were also evaluated. PTV-HI was defined as (D2%–D98%)/54 Gy; PTV-CI was defined as V95%PTV/V95%Body; PTV-GI was defined as V50%Body/ V95%PTV. Finally, PTV-gEUD2 was calculated, as equivalent dose parameter, according to the Niemierko formula [8]:

gEUD2 ¼

" N X

#1=a

v i ðEQD2i Þ

a

;

i

In this formula, a is equal to 10 (to be representative of tumors, being the tumor control probability affected by cold spots); vi is the partial volume receiving the dose Di; EQD2 is the equivalent physical dose of 2 Gy per fraction using the linear quadratic (LQ) model:



a þ Di b

EQD2i ¼ Di  

nf

aþ2



;

b

In this formula, nf is the number of fractions (i.e. 3), Di is the dose per fraction (i.e. 18 Gy), a/b = 10 Gy for the PTV [9]. Since this study is a dosimetric comparison, the PTV-gEUD2 could be considered as a metric for describing the differences among the institutions as surrogate of PTV equivalent dose. The OAR parameters were scored for maximum doses and specific values of interest. In particular, the MLD2 was calculated from the differential DVHs assuming a/b = 3 Gy for normal lung tissue. Furthermore, in order to compare the plans from different institutions and normalizations, each plan parameter was evaluated according to a performance index, defined as:

OAR index ¼ ðparameter=PTV-gEUD2 Þmean =ðparameter=PTV-gEUD2 Þi ; where i denotes the ith center. For example the MLD2 performance index is defined as Lung-Index = (MLD2/PTV-gEUD2)mean/(MLD2/ PTV-gEUD2)center. The MLD2 or the other OARs are weighted against the PTV-gEUD2 in order to evaluate the different dose normalization among institutions. This definition establishes the mean value across the centers as a standard; values <1 mean lower performances than the mean indexes. The t-Student (two-side) test and

the analysis of variance (ANOVA) were performed in order to find correlations among parameters; p-values <0.05 were considered significant. The computations were carried out with the STATA statistical package, release 13.0 (STATA Corp., College Station, TX). Results Twenty-six Italian institutions collaborated in this dosimetric intercomparison. The RT technologies adopted in this study were: Linacs (8% static field IMRT, 4% 3DCRT and 76% dynamic arcs-volumetric modulated arc therapy), Tomotherapy (8%) and CyberKnife (4%). The characteristics of the centers are summarized in Table 1. All but five linac-based centers had a MLC width at isocenter 65 mm; at the study date all centers had already treated patients with SABR, even if 15 centers treated less than 100 patients. Figure 1a shows, by way of example, a patient DVHs PTV. Similar results were obtained for all other cases. The mean dosimetric results are reported in Table 2. In detail, regarding PTV, the dose objective (V95% > 95%) was violated four times by three institutions; mean PTV dose averaged over the five patients and the 26 centers was 56.8 ± 3.4 Gy and median of V95% was 99.8% (range 90–100%), PTV D2% and D98% were 60.1 ± 5.8 Gy and 52.4 ± 2.2 Gy. Mean PTV-HI, PTV-CI and PTV-GI values were, respectively, 14.2 ± 10.1%, 0.70 ± 0.15, and 4.9 ± 1.9. In particular, the high relative standard deviations suggest different optimization strategies in target coverage. No correlation was found between PTV-HI and PTV-GI (p = 0.28). The mean PTV-gEUD2 was 133.7 ± 10.3 Gy (range 105–161 Gy). The intra and inter-institutions PTV-gEUD2 distribution variability is shown in Fig. 2. As expected, significant correlation between PTV-gEUD2 and PTV-HI was observed (p < 0.001). A correlation between PTV-gEUD2 and PTV-GI was also observed (p < 0.05): raising the isodose gradient outside the PTV the PTV-gEUD2 increases. Regarding the OARs, dose objectives were respected in almost all cases: all centers were able to fulfill the spinal cord constraint and one center did not respect the ribs constraint for patients n.1 and n.3. The heart and the esophagus were not considered in the dose analysis due to the absence of any criticism regarding the dose values. The mean MLD2 was 11.8 ± 3.5 Gy (range 5.4– 24.7 Gy). Figure 3a shows the Lung-Index for the different centers, and the dotted line corresponds to the MLD2 constraint of 17 Gy [10]. Figure 3b shows the association between the Lung-Index and the technologies used in the study. Furthermore, with the

Please cite this article in press as: Giglioli FR et al. Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.03.015

4

F.R. Giglioli et al. / Physica Medica xxx (2016) xxx–xxx

Table 1 Characteristic of the centers participating. Machine

TPS

Trilogy True beam Tomo Synergy BM Synergy Clinac Clinac D CyberKnife

3 5 2 5 3 4 3 1

Monaco Raystation Oncentra Eclipse Pinnacle Tomo Tps Iplan Multiplan

4 1 2 11 4 2 1 1

MLC width (mm)

Irradiation technique

Energy

63 4–5 5.5–7 >7

3DCRT Static IMRT VMAT Tomo CyberKnife

6 MV 10 MV

6 15 2 3

Table 2 Dosimetric results. Dosimetric parameter

Values

Mean PTV dose Median of V95% Mean PTV D2% Mean PTV D98% Mean PTV-HI Mean PTV-CI Mean PTV-GI Mean PTV-gEUD2 Mean MLD2 Max dose spine Max dose ribs

56.8 ± 3.4 Gy 99.8% (range 90–100%) 60.1 ± 5.8 Gy 52.4 ± 2.2 Gy 14.2 ± 10.1% 0.70 ± 0.15 4.9 ± 1.9 133.7 Gy (range 105–161 Gy) 11.8 Gy (range 5.4–24.7 Gy) 5.6 ± 3.5 Gy 25.8 ± 8.3 Gy

Figure 2. (a) PTV-gEUD2 versus participant institutions.

same technology, the difference between high experienced centers (>100 patients) and low experienced institution (<100 patients) was evaluated and proved to be significant for 2 out of 9 technologies. The cut-off of 100 patients resulted as the optimal value from a stratification analysis. In these cases a higher level of expertise improves the results. A correlation between Lung-Index and Gradient-Index (p < 0.05) was reported: as expected, due to higher dose conformality to the target, when PTV-GI decreases, i.e. the isodose gradient is higher, the lung sparing performance is better [11]. The Ribs Index analysis is reported in Fig. 4. Inter and intrainstitution variations are shown in Fig. 4a. As before, the dotted line indicates the constraint for ribs (37 Gy as defined in [7]). Figure 4b shows the Ribs-Index for the various technologies: the CyberKnife (technology CYB) showed the best performance for ribs sparing. However, as CyberKnife was utilized in only one institution, no statistical analysis should be performed. Figure 5a reports the distribution of the Spine-Index with the constraint (i.e. 18 Gy) and 5b the correlation with the technologies

1 2 20 2 1

Expertise (number of treated patients) 21 5

<10 11–25 26–50 51–75 76–100 101–150 >150

5 3 4 2 1 2 9

adopted. In this case, no statistical differences were found among them. Finally, the correlation among the TPS model and lung, ribs and spine performance indexes is reported in Fig. 6.

Discussion Modern clinical trials should involve a large amount of patients, which could be difficult for a single institution to recruit. For this reason, a multi-institutional clinical trial may be advantageous as a higher number and more varied range of participants from various geographic locations can be included and it is possible to compare the results achieved by the various centers [12,13]. Therefore before beginning multi-institutional clinical studies, it may be useful to determine and standardize the techniques and policies used by each center which would enable us to compare the data gathered by the various institutions. Peters et al. analyzed the clinical impact of standard fractionated radiotherapy protocol compliance in a large randomized head-and-neck cancer study (853 patients) and demonstrated that it is essential to follow protocol instructions in order to obtain adequate results [14]. In this context, the multicenter inter-comparison of dose distributions for SABR treatments differs to standard fractionation RT for several reasons: firstly because an inhomogeneous dose distribution may be expected within the target, and secondly that innovative accelerators and radiation techniques have been developed such as VMAT and IMRT for Linac technology or other types of new machinery such as Tomotherapy, Vero, and CyberKnife. To our knowledge, this is the first large (i.e. >10 centers) multiinstitutional planning study on lung cancer SABR aimed at investigating possible criticisms in multicentre data comparisons from a dosimetric perspective. We tested the ability to reproduce comparable stereotactic dose distributions with the same clinical intent using the images and contours of five representative patients, with the purpose of evaluating the possibility to collect patients from different institutions retrospectively. In two recent multiinstitutional plan inter-comparison on prostate and liver tumor SABR, significant dosimetric differences between centers, with possible clinical implication, were observed [6,15]. In particular, significant differences were observed on OARs, yet no equivalent dose analysis was carried out. One of the main problems when comparing the SABR plans from different institutions is the varied PTV dose prescription. There are two main approaches regarding dose distribution within the target in SABR: (i) to maintain dose homogeneity within the target, which is usually prescribed at the PTV mean value, and (ii) to prescribe dose at PTV boundaries, without considering the dose heterogeneity inside the target [16]. In conventional radiotherapy, the International Commission on Radiation Units and Measurements (ICRU) Report 62 [17] recommended a uniform dose distribution within the target volume with dose prescribed at a reference point (generally the isocenter). More recently, the ICRU 83 report on IMRT planning [18] suggested using

Please cite this article in press as: Giglioli FR et al. Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.03.015

F.R. Giglioli et al. / Physica Medica xxx (2016) xxx–xxx

5

Figure 3. (a) Lung Index versus participant institutions; (b) Lung Index in function of technology (TRI = Varian Trilogy, TRU1 = Varian True Beam with 2.5 mm leaves, TOM = Tomotherapy, SYNbm = Elekta Synergy Beam Modulator, SYN = Elekta Synergy, CLI = Varian Clinac, BLA = Brainlab, CYB = CyberKnife, TRU2 = Varian True Beam with 5 mm leaves) and the center expertise (Nexp < 100 pts, Yexp P 100 pts).

Figure 4. (a) Ribs Index distribution versus participant institutions; (b) Ribs Index in function of technology (TRI = Varian Trilogy, TRU1 = Varian True Beam with 2.5 mm leaves, TOM = Tomotherapy, SYNbm = Elekta Synergy Beam Modulator, SYN = Elekta Synergy, CLI = Varian Clinac, BLA = Brainlab, CYB = CyberKnife, TRU2 = Varian True Beam with 5 mm leaves).

Figure 5. (a) Spine Index distribution versus participant institutions; (b) Spine Index in function of technology (TRI = Varian Trilogy, TRU1 = Varian True Beam with 2.5 mm leaves, TOM = Tomotherapy, SYNbm = Elekta Synergy Beam Modulator, SYN = Elekta Synergy, CLI = Varian Clinac, BLA = Brainlab, CYB = CyberKnife, TRU2 = Varian True Beam with 5 mm leaves).

Please cite this article in press as: Giglioli FR et al. Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.03.015

6

F.R. Giglioli et al. / Physica Medica xxx (2016) xxx–xxx

Figure 6. Lung Index, Spine Index and Ribs Index versus TPS (ECL = Eclipse, MON = Monaco, PINN = Pinnacle, RAY = Raystation, MAS = Masterplan, iPL = iPlan, TOM = Tomo, CYB = Multiplan CyberKnife).

a dose–volume prescription (i.e. PTV mean dose) without specifically distinguishing between standard fractionation and the SABR approach: this could determine a homogeneous dose distribution also in the case of SABR treatments. In contrast, the best practice guidelines for SABR by the AAPM Task Group 101 allow for dose heterogeneity within the targets [7]. The major advantage of this second approach is the tight target conformity with steep dose fall-off outside the target as shown in many studies [11]. Nevertheless, our data did not support this thesis as nocorrelation between PTV-HI and PTV-GI was obtained (p = 0.28). Conversely, our study showed a positive correlation between PTV-GI and MLD2 sparing (p = 0.01). Bearing in mind the considerations mentioned above, AIFM–SABR-WG recruited planners from institutions irrespective of dose prescription approaches (i.e. homogeneously or heterogeneously) or priorities in plan optimization. As reported by Kavanagh et al. [19], PTV-gEUD2 is a suitable parameter for retrospective or prospective SABR dose description, as it is insensitive to tumor radio-sensitivity and volume. Therefore, PTV-gEUD2 may be useful for comparing dose distributions from numerous institutions, planners and technologies especially if normalization modality is left to the policy center. As expected, our results support that experience does not seem to play a vital role in all technologies and the individual center clinical preferences (homogeneity or heterogeneity prescription) better explain the difference between centers instead than technologies. Bearing in mind the AAPM TG 101 protocol for spinal cord and ribs, the dose requirement value was calculated for each OAR performance-index, and the data concerning MLD2 were published. In this way, it is possible to check if single centers are able to respect the dose constraints. There has been much debate regarding how to keep pneumonia below 20% incidence of Grade 2 complications. In particular, Ricardi et al. [10] retrospectively evaluated various parameters for possible correlations with occurrence of radiation-induced lung injury in patients with primary or secondary lung cancers treated with SABR. The MLD2, as determined by the linear quadratic model, proved to correlate well with the RTOG lung toxicity scores of Grade 2–3 (p = 0.008). MLD2 < 12.1 Gy and 20.1 Gy were observed to avoid RTOG Grade 1, and 2–3 lung toxicities, respectively. Fowler et al. [20] estimated an MLD2 value of 19 Gy for reducing the incidence of pneumonia to <20%. In our series, MLD2 varied from 5.4 Gy to 24.7 Gy. In particular, the limit of 17 Gy, which we selected as safe constraint, was only violated in one case. Lung-Index does not appear to be associated with technologies and is generally insensitive to the expertise of the planners (Fig. 3b); however when there are significant differences

(Tomotherapy and Varian Clinac), the performance of expert planners improves. It is interesting to note the correlation between the Lung-Index and the Gradient-Index: lower Gradient-Index values, corresponding to a steep dose fall-off, affected lung sparing as we had expected. To identify a reference value for PTV-gEUD2 we referred to McCammon et al. [21] that strongly suggested a dose–control relationship within the dose range applied in their study (three-fraction protocol using dose escalation scheme from 10 to 20 Gy/fraction for thoracic and liver tumors) with higher rates of local control achieved with nominal doses of 54–60 Gy, (one and 3-year actuarial local control rates were 100% and 89.9%, respectively), corresponding to a PTV-gEUD > 65.3 Gy and PTVgEUD2 > 152 Gy. In their study, larger tumors were statistically associated with a less efficient local control at the univariate analysis, while a strong trend was reported in the multivariate analysis. The value of 152 Gy appears high if compared to the mean of PTVgEUD2 in our database (133.7 ± 10.3 Gy). To explain this result we should consider that lung tumors were more often kept under control than hepatic tumors and, moreover, the study was carried out with the pencil beam calculation algorithm thus resulting in poor dose estimation as stated by Latifi et al. [22]. For all these reasons, the dose cutoff for PTV-gEUD2 of 152 Gy, appears to be high for small volume lung tumors and this value was only used in the study for comparison purposes. From another point of view, the sterilizing dose could depend on the extension of radio-resistant hypoxic areas, thus higher doses should be used to increase the tumor control [23]. In our investigation the PTV-gEUD2 values ranged widely from 105 to 161 Gy. In particular, the correlation between PTV-gEUD2 and PTV-HI values is of interest: heterogeneity prescription seems to have an effect on the PTV-gEUD2 tumor, as expected, despite a spread on the graph mainly in the central area. Regarding the maximum dose to ribs and spinal cord, we referred to the AAPM TG 101 protocol; it is interesting to note that the constraint for spinal cord (18 Gy) was fulfilled in all cases. Regarding ribs maximum dose <37 Gy, this constraint was only violated in two cases by a single planner. A maximum spinal cord dose lower than 18 Gy in 3 fractions was also proposed by Timmerman et al. [24] and recommended by Milano et al. [25] who reviewed the dose constraints of recent RTOG 0236 and RTOG-0438 studies. No correlation between OARs index and TPS model was found in our data. All but two cases presented low inter-institution variability over the same TPS. Higher variability was only reported in case of Spine Index for Pinnacle and iPlan TPS. In any case, this result is not relevant, being the Spine dose objective achieved by all planners. The individual planner preferences seem to better explain this difference than the TPS model. The linear–quadratic model is widely used for predicting normal tissue toxicity from fractionated radiation (<3–5 Gy/fraction), although its validity is controversial for the higher dose (>7–20 Gy) of SABR. In particular, since this model was obtained from in vitro cell survival assays of cancer cell lines, its predictive capability for the in vivo toxicity of normal tissues may be limited [26] and its findings, in terms of dose constraints at higher doses/ fraction, should be applied with caution. Although the number of patients treated with SABR and the planners’ expertise were not statistically related to the OARIndex distribution, wide spreads in dose distribution to target and OARs were observed. In this regard, our study suggests that detailed planning objectives are required when carrying out multicentre prospective studies, in order to avoid differences in target dose prescription strategies. In this study, the target volumes were intentionally delineated prior to data distribution in order to eliminate inaccuracy and variability of the contouring process (CT scanning, image registration,

Please cite this article in press as: Giglioli FR et al. Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.03.015

F.R. Giglioli et al. / Physica Medica xxx (2016) xxx–xxx

etc.) and allow for comparisons among the DVHs and dosimetric parameters. Furthermore, other factors potentially affecting treatment variability were not considered. As already mentioned for the calculation algorithm, treatment planning systems with heterogeneity correction and collapsed-cone convolution model (or superior algorithm) were mandatory for joining the project, but the influence of the different calculation modalities could affect the dose distributions under specific conditions especially for the lung region [27–29].

[12]

[13]

[14]

Conclusions

[15]

Significant differences were observed for the parameters with possible clinical implications, especially concerning the PTVgEUD2, mainly because the protocol allowed the center to decide how to normalized the dose. This result leads to the consideration that, at present, when sharing patients in a retrospective multicenter clinical trial, the additional PTV-gEUD2 calculation could be required.

[16]

[17]

[18]

[19]

References [1] Cykert S, Dilworth-Anderson P, Monroe MH, Walker P, McGuire FR, CorbieSmith G, et al. Factors associated with decisions to undergo surgery among patients with newly diagnosed early stage lung cancer. J Am Med Assoc 2010;303:2368–76. [2] Onishi H, Araki T, Shirato H, Nagata Y, Hiraoka M, Gomi K, et al. Stereotactic hypofractionated high-dose irradiation for stage I non small cell lung carcinoma: clinical outcomes in 245 subjects in a Japanese multiinstitutional study. Cancer 2004;101:1623–31. [3] Baumann P, Nyman J, Hoyer M, Wennberg B, Gagliardi G, Lax I, et al. Outcome in a prospective phase II trial of medically inoperable stage I non small-cell lung cancer patients treated with stereotactic body radiotherapy. J Clin Oncol 2009;27:3290–6. [4] Timmerman R, Paulus R, Galvin J, Michalski J, Straube W, Bradley J, et al. Stereotactic body radiation therapy for inoperable early stage lung cancer. J Am Med Assoc 2010;303:1070–6. [5] Das IJ, Cheng CW, Chopra KL, Mitra RK, Srivastava SP, Glatstein E. Intensity modulated radiation therapy dose prescription, recording, and delivery: patterns of variability among institutions and treatment planning systems. J Natl Cancer Inst 2008;100(5):300–7. [6] Marino C, Villaggi E, Maggi G, Esposito M, Strigari L, Bonanno E, et al. A Feasibility dosimetric study on prostate cancer: are we ready for a multicenter clinical trial on SBRT? Strahlenther Onkol 2015;191(7):573–81. [7] Benedict S, Yenice K, Followill D, Galvin JM, Hinson W, Kavanagh B, et al. Stereotactic body radiation therapy: the report of AAPM task group 101. Med Phys 2010;37:4078–101. [8] Niemierko A. Reporting and analyzing dose distributions: a concept of equivalent uniform dose. Med Phys 1997;24:103–10. [9] Gay HA, Niemierko A. A free program for calculating Eud-based Ntcp and Tcp in external beam radiotherapy. Phys Med 2007;23:115–257. [10] Ricardi U, Filippi AR, Guarneri A, Giglioli FR, Mantovani C, Fiandra C, et al. Dosimetric predictors of radiation-induced lung injury in stereotactic body radiation therapy. Acta Oncol 2008;27:1–7. [11] Oku Y, Takeda A, Kunieda E, Sudo Y, Oooka Y, Aoki Y, et al. Analysis of suitable prescribed isodose line fitting to planning target volume in stereotactic body

[20] [21]

[22]

[23]

[24]

[25] [26] [27]

[28]

[29]

7

radiotherapy using dynamic conformal multiple arc therapy. Pract Radiat Oncol 2011;2:46–53. Matsuo Y, Takayama K, Nagata Y, Takayama K, Sakamoto T, Sakamoto M, et al. Interinstitutional variations in planning for stereotactic body radiation therapy for lung cancer. Int J Radiat Oncol Biol Phys 2007;68:416–25. Nishio T, Kunieda E, Shirato H, Ishikura S, Onishi H, Tateoka K, et al. Dosimetric verification in participating institutions in a stereotactic body radiotherapy trial for stage I non-small cell lung cancer: Japan clinical oncology group trial (JCOG0403). Phys Med Biol 2006;51:5409–17. Peters LJ, O’Sullivan B, Giralt J, Fitzgerald TJ, Trotti A, Bernier J, et al. Critical impact of radiotherapy protocol compliance and quality in the treatment of advanced head and neck cancer: results from TROG 02.02. J Clin Oncol 2010;28 (18):2996–3001. Esposito M, Maggi G, Marino C, Bottalico L, Cagni E, Carbonini C, et al. Multicentre treatment planning inter-comparison in a national context: the liver stereotactic ablative radiotherapy case. Phys Med 2016;32(1):277–83. Widder J, Hollander M, Ubbels JF, Bolt RA, Langendijk JA. Optimizing dose prescription in stereotactic body radiotherapy for lung tumours using Monte Carlo dose calculation. Radiother Oncol 2010;94:42–6. International Commission on Radiation Units and Measurements. ICRU report 62: prescribing, recording, and reporting photon beam therapy (Supplement to ICRU Report 50), 1999. International Commission on Radiation Units and Measurements. ICRU report 83: prescribing, recording, and reporting photon-beam intensity modulated radiation therapy (IMRT). J ICRU 2010;10:1–10. Kavanagh BD, Timmerman RD, Benedict SH, Wu Q, Schefter TE, Stuhr K, et al. How should we describe the radiobiologic effect of extracranial stereotactic radiosurgery: equivalent uniform dose or tumor control probability? Med Phys 2003;30:321–4. Fowler JF, Tomé WA, Fenwick JD, Mehta MP. A challenge to traditional radiation oncology. Int J Radiat Oncol Biol Phys 2004;60:1241–56. McCammon R, Schefter TE, Gaspar LE, Zaemisch R, Gravdahl D, Kavanagh B. Observation of a dose–control relationship for lung and liver tumors after stereotactic body radiation therapy. Int J Radiat Oncol Biol Phys 2009;73:112–8. Latifi K, Oliver J, Baker R, Dilling TJ, Stevens CW, Kim J, et al. Study of 201 non small cell lung cancer patients given stereotactic ablative radiation therapy shows local control dependence on dose calculation algorithm. Int J Radiat Oncol Biol Phys 2014;88:1108–13. Strigari L, Benassi M, Sarnelli A, Polico R, D’Andrea MA. Modified hypoxiabased TCP model to investigate the clinical outcome of stereotactic hypofractionated regimes for early stage non-small-cell lung cancer (NSCLC). Med Phys 2012;39(7):4502–14. Timmerman R, McGarry R, Yiannoutsos C, Papiez L, Tudor K, DeLuca J, et al. Excessive toxicity when treating central tumors in a phase II study of stereotactic body radiation therapy for medically inoperable early-stage lung cancer. J Clin Oncol 2006;24:4833–9. Milano MT, Constine LS, Okunieff P. Normal tissue toxicity after small field hypofractionated stereotactic body radiation. Radiat Oncol 2008;3:36. Glatstein E. Hypofractionation, long-term effects, and the alpha/beta ratio. Int J Radiat Oncol Biol Phys 2008;72:11–2. Aarup LR, Nahum AE, Zacharatou C, Juhler-Nøttrup T, Knöös T, Nyström H, et al. The effect of different lung densities on the accuracy of various radiotherapy dose calculation methods: implications for tumour coverage. Radiother Oncol 2009;91:405–14. Kry SF, Alvarez P, Molineu A, Amador C, Galvin J, Followill DS. Algorithms used in heterogeneous dose calculations show systematic differences as measured with the radiological physics center’s anthropomorphic thorax phantom used for RTOG credentialing. Int J Radiat Oncol Biol Phys 2013;85(1):e95–e100. Sini C, Broggi S, Fiorino C, Cattaneo GM, Calandrino R. Accuracy of dose calculation algorithms for static and rotational IMRT of lung cancer: a phantom study. Phys Med 2015;31(4):382–90.

Please cite this article in press as: Giglioli FR et al. Lung stereotactic ablative body radiotherapy: A large scale multi-institutional planning comparison for interpreting results of multi-institutional studies. Phys. Med. (2016), http://dx.doi.org/10.1016/j.ejmp.2016.03.015