Improving interinstitutional and intertechnology consistency of pulmonary SBRT by dose prescription to the mean internal target volume dose

Lung cancer is responsible for the highest number of cancer deaths in males and females worldwide. Surgical resection is standard of care, but growing numbers of patients are medically inoperable due to their age and comorbidities. In patients with untreated early stage non-small cell lung cancer (NSCLC) the median survival is 13 months and the 5‑year cancer-specific survival rate is 16% [1]. In these patients, being inoperable or refusing surgery, the standard of care is stereotactic body radiation therapy (SBRT) [2‐5]. Furthermore, SBRT is increasingly applied for patients with lung metastases in the oligometastatic disease [6‐12].

Despite the fact that the use of SBRT is rapidly increasing, there is high variability in prescribed doses and normalization methods between prospective trials, between institutions and even between practice guidelines, which makes it difficult to compare the truly delivered dose and the treatment outcome between institutions. A recent multicenter planning study from Giglioli et al. [13] showed that the general equivalent uniform planning target volume (PTV) dose varied between 105 and162 Gy if only the dose per fraction was specified without further specification on the dose prescription and normalization method, the dose inhomogeneity and PTV constraints.

Historically an inhomogeneous dose was prescribed to a certain PTV encompassing isodose line, normalization was done on the maximum dose or a representative dose point inside the target volume. This is in agreement with the International Commission on Radiation Units and Measurements (ICRU) reports 50 and 62 [14, 15] which recommend prescription and normalization on a representative point. ICRU 83 [16] for modulated treatment planning recommends dose prescription to the median PTV dose instead of prescribing and reporting the dose to a single point. The new ICRU report 91 [17] recommends for stereotactic treatments to prescribe the dose to the isodose surface that covers an optimal percentage of the PTV. Additionally, it is recommended that the prescription does not only specify the prescribed dose and the normalization method but a comprehensive set of accepted values for target coverage and organ at risk doses. A recent multi-enter planning study from the German Society of Radiation Oncology (DEGRO) working group for Radiosurgery and Stereotactic Radiotherapy [18] showed that interinstitutional variation in the mean PTV dose was reduced by specifying the dose as well as the prescription method; the prescribed dose of 3 × 15 Gy had to cover 95% of the PTV and the allowed D2% was set to 69.2 Gy. However, the variability was still >22%, and there was a large difference between the SBRT techniques. Similarly, all other dosimetric parameters characterizing gross tumor volume (GTV) and PTV dose showed large differences.

Based on the European Society for Radiotherapy and Oncology (ESTRO) and Advisory Committee on Radiation Oncology Practice (ACROP) consensus guidelines for SBRT of peripherally located NSCLC [19], de Jong et al. published recommendations for prescribing and recording taking the ICRU report 91 into account [20]. They showed that even between 8 centers having long-term clinical experience with SBRT significant differences can be seen in the actual planned dose to the PTVs and GTVs.

There is strong retrospective data indicating that this variation in GTV and PTV doses is of clinical relevance. Several studies reported that local tumor control after pulmonary SBRT was significantly associated with the biologically effective dose (BED) at the isocenter and the mean GTV dose [21‐23]. There is consequently a clinical need to better standardize the planning of pulmonary SBRT and to reduce interinstitutional variability. As basis for this study, we postulate a multiparameter dose prescription including dose normalization to the mean ITV dose in combination with specification of more detailed PTV and ITV objectives reduces the interinstitutional variation in ITV and PTV dose.

SBRT is used widely for primary lung tumors such as NSCLC as well as for pulmonary oligometastatic disease [2‐4, 9, 28]. Recommendations for these treatments exist from different organizations [19, 22, 26, 29]. However, even following these, significant differences between studies, institutions and SBRT techniques for doses to target volumes as well as OAR have been published [13, 18, 30, 31].

In the current study, dose prescription to the mean ITV dose combined with additional ITV- and PTV-based planning objectives achieved highly consistent dose distributions within the target volume. Mean and median dose to the PTV varied by less than 3% of the prescribed dose, which is of the order of magnitude as treatment planning for a static phantom [26, 32]. This high consistency was achieved despite the large number of participating institutions (n = 27), the use of heterogeneous planning techniques and planning for all currently available SBRT delivery platforms. We are therefore convinced that the proposed pulmonary SBRT planning and dose prescription methodology is generalizable.

We believe that in particular the use of several DVH-based planning objectives for the ITV and PTV contributed to homogenize dose distribution between centers. Unfortunately, the ICRU Report 91 for stereotactic treatments [17] still recommends only to prescribe to one single DVH point of the PTV and does not give any additional objectives for GTV, CTV or ITV as already discussed in [33]. However, it recommends reporting multiple dose parameters for GTV, CTV, ITV and PTV to make treatment outcome more comparable.

The recent study evaluating the difference in dose to GTV and PTV of multiple centers [20] concluded that a multiparametric prescription is needed. The study suggests as minimum requirement a BED10 of 150 Gy as mean ITV dose. However, our study showed that even a higher dose to the ITV is possible.

The single SBRT plan with an inacceptable deviation in CI and 4 plans out of 7 with minor deviations for the CI were observed for one specific planning system. This demonstrates the general problem that volume calculation and also the display of contours and dose may differ significantly between planning systems depending on how calculation voxels are then interpolated. In all cases the planning institution assumed to fulfill the planning objectives when plan evaluation was performed in the respective planning system. This clearly indicates the need that vendors agree on one common way to interpret partial volume effects between voxels and DICOM structures. The other three minor deviations for the conformity index showed no particular pattern and originated from different planning systems. Four out of five minor deviations of the dose to 0.1 ml of the PTV were planned with a 3D static field technique. Using 3D conformal forward planning, it is obviously more difficult to simultaneously control all parameters. Similarly, no pattern was observed for the deviations in ITV and PTV coverage.

All OAR constraints OAR were fulfilled by all institutions and by all SBRT plans for both patients. All institutions followed the ALARA principle and achieved very similar results of OAR sparing, irrespective of the SBRT planning and delivery technique.

The significant difference in the ipsilateral mean lung dose as a function of algorithm is interesting. Nevertheless, they might have to be allocated to the fact that for RRS only MC was used for the dose calculation while for MOD treatments, BT and AAA dominate and for the 3D treatment plans, CC dominates. Even though the deviations we see are small, according to ICRU 91 and other recommendations [17, 26, 29] a type B or MC algorithm, which takes into account the lateral electron scattering in inhomogeneous media, should be used for SBRT, in particular in the lung. In addition, the abovementioned recommendations suggest the use of a calculation grid of 2 mm or smaller, but for 13 plans a calculation grid of 2.5 mm, for 5 plans a calculation grid of 3 mm and for 3 plans a calculation grid of as large as 4 mm was used. In particular the use of a calculation grid larger than 3 mm should be avoided; however, nowadays with sufficient computing power, grid sizes of 2 mm should be feasible in daily routine practice.

Possibly, part of the deviations which were accepted by the planner, could be omitted if regular knowledge-exchange and training was performed on national and international level. This is in line with a survey on the Influence of Institutional Experience and Technological Advances on Outcome of Stereotactic Body Radiation Therapy for Oligometastatic Lung Disease [34] which showed a relation between the local control and the experience of the center, as well as with a recent review on dosimetric multicenter planning comparison studies for SBRT [35] and two other multicentric planning studies for spine SBRT and prostate SBRT [36, 37].

One of the limitations of this study is that only two patients were evaluated which are not representative for all patients. We added to the supplemental material a further study containing 40 patients, where we evaluated the optimal constraints for the planning study in order to have minimal variation in different dosimetric parameters between the different patients. However, these were only planned by one single institution. Therefore, conclusions drawn from this study should be evaluated on more patients in a multicenter setting.

A further limitation is the fact that motion management of different institutions was not evaluated. Nevertheless, a recent detailed 4D dose analysis has indicated negligible difference and variability of GTV mean and near minimum dose between ITV-based and mid-ventilation-based PTV optimization and GTV-based robust optimization provided that normalization is done to the GTV mean dose [38]. Using different motion management strategies might have resulted in smaller doses to the OAR as for this study the target contours were delineated based on an ITV concept. Furthermore, this study relies on the correct dose calculation of the plans, independent from the treatment algorithm used and no dosimetric evaluation of the applicability of the plans was performed. Additionally, the dose to the ITV and not to the GTV was reported in our study, as this would require a full 4D dose calculation taking the range of motion into account; however, it has been shown that there is very close association between mean ITV and GTV dose despite large interpatient variations in GTV volume and motion range [39‐41]. In the cases where no ITV was defined due to a different motion management, prescribing to the mean dose to the GTV might thus be equivalent to prescribing to the mean ITV dose as used in this study while prescribing to the mean PTV dose might not be the optimal strategy.

We evaluated the study according to the recently published Radiotherapy Treatment plannINg study Guidelines (RATING) [42] and achieved a score of 166 out of 186 points (89%), even though the study was conducted before these guidelines had been published.

Analyzing 52 plans from 25 institutions, this planning study demonstrated that dose prescription to the mean internal target volume (ITV) dose in combination with detailed dose–volume histogram (DVH)-based planning objectives for planning target volume (PTV) and ITV achieves highly consistent dose distributions irrespective of the planning institution, and the stereotactic body radiotherapy (SBRT) planning and delivery technologies. We therefore recommend to evaluate the proposed planning approach.